merge: with lataest default branch

.release.cfg

@@ -0,0 +1,16 b''
+[DEFAULT]
+done = false
+
+[task:bump_version]
+done = true
+
+[task:fixes_on_stable]
+done = true
+
+[task:pip2nix_generated]
+done = true
+
+[release]
+state = prepared
+version = 4.27.1
+

configs/logging.ini

@@ -0,0 +1,53 b''
+; #####################
+; LOGGING CONFIGURATION
+; #####################
+; Logging template, used for configure the logging
+; some variables here are replaced by RhodeCode to default values
+
+[loggers]
+keys = root, vcsserver
+
+[handlers]
+keys = console
+
+[formatters]
+keys = generic, json
+
+; #######
+; LOGGERS
+; #######
+[logger_root]
+level = NOTSET
+handlers = console
+
+[logger_vcsserver]
+level = $RC_LOGGING_LEVEL
+handlers =
+qualname = vcsserver
+propagate = 1
+
+; ########
+; HANDLERS
+; ########
+
+[handler_console]
+class = StreamHandler
+args = (sys.stderr, )
+level = $RC_LOGGING_LEVEL
+; To enable JSON formatted logs replace generic with json
+; This allows sending properly formatted logs to grafana loki or elasticsearch
+#formatter = json
+#formatter = generic
+formatter = $RC_LOGGING_FORMATTER
+
+; ##########
+; FORMATTERS
+; ##########
+
+[formatter_generic]
+format = %(asctime)s.%(msecs)03d [%(process)d] %(levelname)-5.5s [%(name)s] %(message)s
+datefmt = %Y-%m-%d %H:%M:%S
+
+[formatter_json]
+format = %(timestamp)s %(levelname)s %(name)s %(message)s %(req_id)s
+class = vcsserver.lib._vendor.jsonlogger.JsonFormatter

pkgs/patches/configparser/pyproject.patch

@@ -0,0 +1,10 b''
+diff -rup configparser-4.0.2-orig/pyproject.toml configparser-4.0.2/pyproject.toml
+--- configparser-4.0.2-orig/pyproject.toml 2021-03-22 21:28:11.000000000 +0100
++++ configparser-4.0.2/pyproject.toml 2021-03-22 21:28:11.000000000 +0100
+@@ -1,5 +1,5 @@
+ [build-system]
+-requires = ["setuptools>=40.7", "wheel", "setuptools_scm>=1.15"]
++requires = ["setuptools<=42.0", "wheel", "setuptools_scm<6.0.0"]
+ build-backend = "setuptools.build_meta"
+
+ [tool.black]

pkgs/patches/importlib_metadata/pyproject.patch

@@ -0,0 +1,7 b''
+diff -rup importlib-metadata-1.6.0-orig/yproject.toml importlib-metadata-1.6.0/pyproject.toml
+--- importlib-metadata-1.6.0-orig/yproject.toml	2021-03-22 22:10:33.000000000 +0100
++++ importlib-metadata-1.6.0/pyproject.toml	2021-03-22 22:11:09.000000000 +0100
+@@ -1,3 +1,3 @@
+ [build-system]
+-requires = ["setuptools>=30.3", "wheel", "setuptools_scm"]
++requires = ["setuptools<42.0", "wheel", "setuptools_scm<6.0.0"]

pkgs/patches/pytest/setuptools.patch

@@ -0,0 +1,12 b''
+diff -rup pytest-4.6.9-orig/setup.py pytest-4.6.9/setup.py
+--- pytest-4.6.9-orig/setup.py	2018-04-10 10:23:04.000000000 +0200
++++  pytest-4.6.9/setup.py	2018-04-10 10:23:34.000000000 +0200
+@@ -24,7 +24,7 @@ def main():
+ def main():
+     setup(
+         use_scm_version={"write_to": "src/_pytest/_version.py"},
+-        setup_requires=["setuptools-scm", "setuptools>=40.0"],
++        setup_requires=["setuptools-scm<6.0.0", "setuptools<=42.0"],
+         package_dir={"": "src"},
+         # fmt: off
+         extras_require={
  No newline at end of file

pkgs/patches/zipp/pyproject.patch

@@ -0,0 +1,10 b''
+diff -rup zip-1.2.0-orig/pyproject.toml zip-1.2.0/pyproject.toml
+--- zip-1.2.0-orig/pyproject.toml	2021-03-23 10:55:37.000000000 +0100
++++ zip-1.2.0/pyproject.toml	2021-03-23 10:56:05.000000000 +0100
+@@ -1,5 +1,5 @@
+ [build-system]
+-requires = ["setuptools>=34.4", "wheel", "setuptools_scm>=1.15"]
++requires = ["setuptools<42.0", "wheel", "setuptools_scm<6.0.0"]
+ build-backend = "setuptools.build_meta"
+
+ [tool.black]

vcsserver/config/settings_maker.py

@@ -0,0 +1,207 b''
+# -*- coding: utf-8 -*-
+
+# Copyright (C) 2010-2020 RhodeCode GmbH
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License, version 3
+# (only), as published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+#
+# This program is dual-licensed. If you wish to learn more about the
+# RhodeCode Enterprise Edition, including its added features, Support services,
+# and proprietary license terms, please see https://rhodecode.com/licenses/
+
+import os
+import textwrap
+import string
+import functools
+import logging
+import tempfile
+import logging.config
+log = logging.getLogger(__name__)
+
+# skip keys, that are set here, so we don't double process those
+set_keys = {
+    '__file__': ''
+}
+
+
+def str2bool(_str):
+    """
+    returns True/False value from given string, it tries to translate the
+    string into boolean
+
+    :param _str: string value to translate into boolean
+    :rtype: boolean
+    :returns: boolean from given string
+    """
+    if _str is None:
+        return False
+    if _str in (True, False):
+        return _str
+    _str = str(_str).strip().lower()
+    return _str in ('t', 'true', 'y', 'yes', 'on', '1')
+
+
+def aslist(obj, sep=None, strip=True):
+    """
+    Returns given string separated by sep as list
+
+    :param obj:
+    :param sep:
+    :param strip:
+    """
+    if isinstance(obj, (basestring,)):
+        if obj in ['', ""]:
+            return []
+
+        lst = obj.split(sep)
+        if strip:
+            lst = [v.strip() for v in lst]
+        return lst
+    elif isinstance(obj, (list, tuple)):
+        return obj
+    elif obj is None:
+        return []
+    else:
+        return [obj]
+
+
+class SettingsMaker(object):
+
+    def __init__(self, app_settings):
+        self.settings = app_settings
+
+    @classmethod
+    def _bool_func(cls, input_val):
+        if isinstance(input_val, unicode):
+            input_val = input_val.encode('utf8')
+        return str2bool(input_val)
+
+    @classmethod
+    def _int_func(cls, input_val):
+        return int(input_val)
+
+    @classmethod
+    def _list_func(cls, input_val, sep=','):
+        return aslist(input_val, sep=sep)
+
+    @classmethod
+    def _string_func(cls, input_val, lower=True):
+        if lower:
+            input_val = input_val.lower()
+        return input_val
+
+    @classmethod
+    def _float_func(cls, input_val):
+        return float(input_val)
+
+    @classmethod
+    def _dir_func(cls, input_val, ensure_dir=False, mode=0o755):
+
+        # ensure we have our dir created
+        if not os.path.isdir(input_val) and ensure_dir:
+            os.makedirs(input_val, mode=mode)
+
+        if not os.path.isdir(input_val):
+            raise Exception('Dir at {} does not exist'.format(input_val))
+        return input_val
+
+    @classmethod
+    def _file_path_func(cls, input_val, ensure_dir=False, mode=0o755):
+        dirname = os.path.dirname(input_val)
+        cls._dir_func(dirname, ensure_dir=ensure_dir)
+        return input_val
+
+    @classmethod
+    def _key_transformator(cls, key):
+        return "{}_{}".format('RC'.upper(), key.upper().replace('.', '_').replace('-', '_'))
+
+    def maybe_env_key(self, key):
+        # now maybe we have this KEY in env, search and use the value with higher priority.
+        transformed_key = self._key_transformator(key)
+        envvar_value = os.environ.get(transformed_key)
+        if envvar_value:
+            log.debug('using `%s` key instead of `%s` key for config', transformed_key, key)
+
+        return envvar_value
+
+    def env_expand(self):
+        replaced = {}
+        for k, v in self.settings.items():
+            if k not in set_keys:
+                envvar_value = self.maybe_env_key(k)
+                if envvar_value:
+                    replaced[k] = envvar_value
+                    set_keys[k] = envvar_value
+
+        # replace ALL keys updated
+        self.settings.update(replaced)
+
+    def enable_logging(self, logging_conf=None, level='INFO', formatter='generic'):
+        """
+        Helper to enable debug on running instance
+        :return:
+        """
+
+        if not str2bool(self.settings.get('logging.autoconfigure')):
+            log.info('logging configuration based on main .ini file')
+            return
+
+        if logging_conf is None:
+            logging_conf = self.settings.get('logging.logging_conf_file') or ''
+
+        if not os.path.isfile(logging_conf):
+            log.error('Unable to setup logging based on %s, '
+                      'file does not exist.... specify path using logging.logging_conf_file= config setting. ', logging_conf)
+            return
+
+        with open(logging_conf, 'rb') as f:
+            ini_template = textwrap.dedent(f.read())
+            ini_template = string.Template(ini_template).safe_substitute(
+                RC_LOGGING_LEVEL=os.environ.get('RC_LOGGING_LEVEL', '') or level,
+                RC_LOGGING_FORMATTER=os.environ.get('RC_LOGGING_FORMATTER', '') or formatter
+            )
+
+        with tempfile.NamedTemporaryFile(prefix='rc_logging_', suffix='.ini', delete=False) as f:
+            log.info('Saved Temporary LOGGING config at %s', f.name)
+            f.write(ini_template)
+
+        logging.config.fileConfig(f.name)
+        os.remove(f.name)
+
+    def make_setting(self, key, default, lower=False, default_when_empty=False, parser=None):
+        input_val = self.settings.get(key, default)
+
+        if default_when_empty and not input_val:
+            # use default value when value is set in the config but it is empty
+            input_val = default
+
+        parser_func = {
+            'bool': self._bool_func,
+            'int': self._int_func,
+            'list': self._list_func,
+            'list:newline': functools.partial(self._list_func, sep='/n'),
+            'list:spacesep': functools.partial(self._list_func, sep=' '),
+            'string': functools.partial(self._string_func, lower=lower),
+            'dir': self._dir_func,
+            'dir:ensured': functools.partial(self._dir_func, ensure_dir=True),
+            'file': self._file_path_func,
+            'file:ensured': functools.partial(self._file_path_func, ensure_dir=True),
+            None: lambda i: i
+        }[parser]
+
+        envvar_value = self.maybe_env_key(key)
+        if envvar_value:
+            input_val = envvar_value
+            set_keys[key] = input_val
+
+        self.settings[key] = parser_func(input_val)
+        return self.settings[key]

vcsserver/lib/_vendor/__init__.py

@@ -0,0 +1,26 b''
+# RhodeCode VCSServer provides access to different vcs backends via network.
+# Copyright (C) 2014-2020 RhodeCode GmbH
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software Foundation,
+# Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
+
+# This package contains non rhodecode licensed packages that are
+# vendored for various reasons
+
+import os
+import sys
+
+vendor_dir = os.path.abspath(os.path.dirname(__file__))
+
+sys.path.append(vendor_dir)

vcsserver/lib/_vendor/jsonlogger/__init__.py

@@ -0,0 +1,243 b''
+'''
+This library is provided to allow standard python logging
+to output log data as JSON formatted strings
+'''
+import logging
+import json
+import re
+from datetime import date, datetime, time, tzinfo, timedelta
+import traceback
+import importlib
+
+from inspect import istraceback
+
+from collections import OrderedDict
+
+
+def _inject_req_id(record, *args, **kwargs):
+    return record
+
+
+ExceptionAwareFormatter = logging.Formatter
+
+
+ZERO = timedelta(0)
+HOUR = timedelta(hours=1)
+
+
+class UTC(tzinfo):
+    """UTC"""
+
+    def utcoffset(self, dt):
+        return ZERO
+
+    def tzname(self, dt):
+        return "UTC"
+
+    def dst(self, dt):
+        return ZERO
+
+utc = UTC()
+
+
+# skip natural LogRecord attributes
+# http://docs.python.org/library/logging.html#logrecord-attributes
+RESERVED_ATTRS = (
+    'args', 'asctime', 'created', 'exc_info', 'exc_text', 'filename',
+    'funcName', 'levelname', 'levelno', 'lineno', 'module',
+    'msecs', 'message', 'msg', 'name', 'pathname', 'process',
+    'processName', 'relativeCreated', 'stack_info', 'thread', 'threadName')
+
+
+def merge_record_extra(record, target, reserved):
+    """
+    Merges extra attributes from LogRecord object into target dictionary
+
+    :param record: logging.LogRecord
+    :param target: dict to update
+    :param reserved: dict or list with reserved keys to skip
+    """
+    for key, value in record.__dict__.items():
+        # this allows to have numeric keys
+        if (key not in reserved
+            and not (hasattr(key, "startswith")
+                     and key.startswith('_'))):
+            target[key] = value
+    return target
+
+
+class JsonEncoder(json.JSONEncoder):
+    """
+    A custom encoder extending the default JSONEncoder
+    """
+
+    def default(self, obj):
+        if isinstance(obj, (date, datetime, time)):
+            return self.format_datetime_obj(obj)
+
+        elif istraceback(obj):
+            return ''.join(traceback.format_tb(obj)).strip()
+
+        elif type(obj) == Exception \
+                or isinstance(obj, Exception) \
+                or type(obj) == type:
+            return str(obj)
+
+        try:
+            return super(JsonEncoder, self).default(obj)
+
+        except TypeError:
+            try:
+                return str(obj)
+
+            except Exception:
+                return None
+
+    def format_datetime_obj(self, obj):
+        return obj.isoformat()
+
+
+class JsonFormatter(ExceptionAwareFormatter):
+    """
+    A custom formatter to format logging records as json strings.
+    Extra values will be formatted as str() if not supported by
+    json default encoder
+    """
+
+    def __init__(self, *args, **kwargs):
+        """
+        :param json_default: a function for encoding non-standard objects
+            as outlined in http://docs.python.org/2/library/json.html
+        :param json_encoder: optional custom encoder
+        :param json_serializer: a :meth:`json.dumps`-compatible callable
+            that will be used to serialize the log record.
+        :param json_indent: an optional :meth:`json.dumps`-compatible numeric value
+            that will be used to customize the indent of the output json.
+        :param prefix: an optional string prefix added at the beginning of
+            the formatted string
+        :param json_indent: indent parameter for json.dumps
+        :param json_ensure_ascii: ensure_ascii parameter for json.dumps
+        :param reserved_attrs: an optional list of fields that will be skipped when
+            outputting json log record. Defaults to all log record attributes:
+            http://docs.python.org/library/logging.html#logrecord-attributes
+        :param timestamp: an optional string/boolean field to add a timestamp when
+            outputting the json log record. If string is passed, timestamp will be added
+            to log record using string as key. If True boolean is passed, timestamp key
+            will be "timestamp". Defaults to False/off.
+        """
+        self.json_default = self._str_to_fn(kwargs.pop("json_default", None))
+        self.json_encoder = self._str_to_fn(kwargs.pop("json_encoder", None))
+        self.json_serializer = self._str_to_fn(kwargs.pop("json_serializer", json.dumps))
+        self.json_indent = kwargs.pop("json_indent", None)
+        self.json_ensure_ascii = kwargs.pop("json_ensure_ascii", True)
+        self.prefix = kwargs.pop("prefix", "")
+        reserved_attrs = kwargs.pop("reserved_attrs", RESERVED_ATTRS)
+        self.reserved_attrs = dict(zip(reserved_attrs, reserved_attrs))
+        self.timestamp = kwargs.pop("timestamp", True)
+
+        # super(JsonFormatter, self).__init__(*args, **kwargs)
+        logging.Formatter.__init__(self, *args, **kwargs)
+        if not self.json_encoder and not self.json_default:
+            self.json_encoder = JsonEncoder
+
+        self._required_fields = self.parse()
+        self._skip_fields = dict(zip(self._required_fields,
+                                     self._required_fields))
+        self._skip_fields.update(self.reserved_attrs)
+
+    def _str_to_fn(self, fn_as_str):
+        """
+        If the argument is not a string, return whatever was passed in.
+        Parses a string such as package.module.function, imports the module
+        and returns the function.
+
+        :param fn_as_str: The string to parse. If not a string, return it.
+        """
+        if not isinstance(fn_as_str, str):
+            return fn_as_str
+
+        path, _, function = fn_as_str.rpartition('.')
+        module = importlib.import_module(path)
+        return getattr(module, function)
+
+    def parse(self):
+        """
+        Parses format string looking for substitutions
+
+        This method is responsible for returning a list of fields (as strings)
+        to include in all log messages.
+        """
+        standard_formatters = re.compile(r'\((.+?)\)', re.IGNORECASE)
+        return standard_formatters.findall(self._fmt)
+
+    def add_fields(self, log_record, record, message_dict):
+        """
+        Override this method to implement custom logic for adding fields.
+        """
+        for field in self._required_fields:
+            log_record[field] = record.__dict__.get(field)
+        log_record.update(message_dict)
+        merge_record_extra(record, log_record, reserved=self._skip_fields)
+
+        if self.timestamp:
+            key = self.timestamp if type(self.timestamp) == str else 'timestamp'
+            log_record[key] = datetime.fromtimestamp(record.created, tz=utc)
+
+    def process_log_record(self, log_record):
+        """
+        Override this method to implement custom logic
+        on the possibly ordered dictionary.
+        """
+        return log_record
+
+    def jsonify_log_record(self, log_record):
+        """Returns a json string of the log record."""
+        return self.json_serializer(log_record,
+                                    default=self.json_default,
+                                    cls=self.json_encoder,
+                                    indent=self.json_indent,
+                                    ensure_ascii=self.json_ensure_ascii)
+
+    def serialize_log_record(self, log_record):
+        """Returns the final representation of the log record."""
+        return "%s%s" % (self.prefix, self.jsonify_log_record(log_record))
+
+    def format(self, record):
+        """Formats a log record and serializes to json"""
+        message_dict = {}
+        # FIXME: logging.LogRecord.msg and logging.LogRecord.message in typeshed
+        #        are always type of str. We shouldn't need to override that.
+        if isinstance(record.msg, dict):
+            message_dict = record.msg
+            record.message = None
+        else:
+            record.message = record.getMessage()
+        # only format time if needed
+        if "asctime" in self._required_fields:
+            record.asctime = self.formatTime(record, self.datefmt)
+
+        # Display formatted exception, but allow overriding it in the
+        # user-supplied dict.
+        if record.exc_info and not message_dict.get('exc_info'):
+            message_dict['exc_info'] = self.formatException(record.exc_info)
+        if not message_dict.get('exc_info') and record.exc_text:
+            message_dict['exc_info'] = record.exc_text
+        # Display formatted record of stack frames
+        # default format is a string returned from :func:`traceback.print_stack`
+        try:
+            if record.stack_info and not message_dict.get('stack_info'):
+                message_dict['stack_info'] = self.formatStack(record.stack_info)
+        except AttributeError:
+            # Python2.7 doesn't have stack_info.
+            pass
+
+        try:
+            log_record = OrderedDict()
+        except NameError:
+            log_record = {}
+
+        _inject_req_id(record, with_prefix=False)
+        self.add_fields(log_record, record, message_dict)
+        log_record = self.process_log_record(log_record)
+
+        return self.serialize_log_record(log_record)

vcsserver/lib/_vendor/redis_lock/__init__.py

@@ -0,0 +1,390 b''
+import sys
+import threading
+import weakref
+from base64 import b64encode
+from logging import getLogger
+from os import urandom
+
+from redis import StrictRedis
+
+__version__ = '3.7.0'
+
+loggers = {
+    k: getLogger("vcsserver." + ".".join((__name__, k)))
+    for k in [
+        "acquire",
+        "refresh.thread.start",
+        "refresh.thread.stop",
+        "refresh.thread.exit",
+        "refresh.start",
+        "refresh.shutdown",
+        "refresh.exit",
+        "release",
+    ]
+}
+
+PY3 = sys.version_info[0] == 3
+
+if PY3:
+    text_type = str
+    binary_type = bytes
+else:
+    text_type = unicode  # noqa
+    binary_type = str
+
+
+# Check if the id match. If not, return an error code.
+UNLOCK_SCRIPT = b"""
+    if redis.call("get", KEYS[1]) ~= ARGV[1] then
+        return 1
+    else
+        redis.call("del", KEYS[2])
+        redis.call("lpush", KEYS[2], 1)
+        redis.call("pexpire", KEYS[2], ARGV[2])
+        redis.call("del", KEYS[1])
+        return 0
+    end
+"""
+
+# Covers both cases when key doesn't exist and doesn't equal to lock's id
+EXTEND_SCRIPT = b"""
+    if redis.call("get", KEYS[1]) ~= ARGV[1] then
+        return 1
+    elseif redis.call("ttl", KEYS[1]) < 0 then
+        return 2
+    else
+        redis.call("expire", KEYS[1], ARGV[2])
+        return 0
+    end
+"""
+
+RESET_SCRIPT = b"""
+    redis.call('del', KEYS[2])
+    redis.call('lpush', KEYS[2], 1)
+    redis.call('pexpire', KEYS[2], ARGV[2])
+    return redis.call('del', KEYS[1])
+"""
+
+RESET_ALL_SCRIPT = b"""
+    local locks = redis.call('keys', 'lock:*')
+    local signal
+    for _, lock in pairs(locks) do
+        signal = 'lock-signal:' .. string.sub(lock, 6)
+        redis.call('del', signal)
+        redis.call('lpush', signal, 1)
+        redis.call('expire', signal, 1)
+        redis.call('del', lock)
+    end
+    return #locks
+"""
+
+
+class AlreadyAcquired(RuntimeError):
+    pass
+
+
+class NotAcquired(RuntimeError):
+    pass
+
+
+class AlreadyStarted(RuntimeError):
+    pass
+
+
+class TimeoutNotUsable(RuntimeError):
+    pass
+
+
+class InvalidTimeout(RuntimeError):
+    pass
+
+
+class TimeoutTooLarge(RuntimeError):
+    pass
+
+
+class NotExpirable(RuntimeError):
+    pass
+
+
+class Lock(object):
+    """
+    A Lock context manager implemented via redis SETNX/BLPOP.
+    """
+    unlock_script = None
+    extend_script = None
+    reset_script = None
+    reset_all_script = None
+
+    def __init__(self, redis_client, name, expire=None, id=None, auto_renewal=False, strict=True, signal_expire=1000):
+        """
+        :param redis_client:
+            An instance of :class:`~StrictRedis`.
+        :param name:
+            The name (redis key) the lock should have.
+        :param expire:
+            The lock expiry time in seconds. If left at the default (None)
+            the lock will not expire.
+        :param id:
+            The ID (redis value) the lock should have. A random value is
+            generated when left at the default.
+
+            Note that if you specify this then the lock is marked as "held". Acquires
+            won't be possible.
+        :param auto_renewal:
+            If set to ``True``, Lock will automatically renew the lock so that it
+            doesn't expire for as long as the lock is held (acquire() called
+            or running in a context manager).
+
+            Implementation note: Renewal will happen using a daemon thread with
+            an interval of ``expire*2/3``. If wishing to use a different renewal
+            time, subclass Lock, call ``super().__init__()`` then set
+            ``self._lock_renewal_interval`` to your desired interval.
+        :param strict:
+            If set ``True`` then the ``redis_client`` needs to be an instance of ``redis.StrictRedis``.
+        :param signal_expire:
+            Advanced option to override signal list expiration in milliseconds. Increase it for very slow clients. Default: ``1000``.
+        """
+        if strict and not isinstance(redis_client, StrictRedis):
+            raise ValueError("redis_client must be instance of StrictRedis. "
+                             "Use strict=False if you know what you're doing.")
+        if auto_renewal and expire is None:
+            raise ValueError("Expire may not be None when auto_renewal is set")
+
+        self._client = redis_client
+
+        if expire:
+            expire = int(expire)
+            if expire < 0:
+                raise ValueError("A negative expire is not acceptable.")
+        else:
+            expire = None
+        self._expire = expire
+
+        self._signal_expire = signal_expire
+        if id is None:
+            self._id = b64encode(urandom(18)).decode('ascii')
+        elif isinstance(id, binary_type):
+            try:
+                self._id = id.decode('ascii')
+            except UnicodeDecodeError:
+                self._id = b64encode(id).decode('ascii')
+        elif isinstance(id, text_type):
+            self._id = id
+        else:
+            raise TypeError("Incorrect type for `id`. Must be bytes/str not %s." % type(id))
+        self._name = 'lock:' + name
+        self._signal = 'lock-signal:' + name
+        self._lock_renewal_interval = (float(expire) * 2 / 3
+                                       if auto_renewal
+                                       else None)
+        self._lock_renewal_thread = None
+
+        self.register_scripts(redis_client)
+
+    @classmethod
+    def register_scripts(cls, redis_client):
+        global reset_all_script
+        if reset_all_script is None:
+            reset_all_script = redis_client.register_script(RESET_ALL_SCRIPT)
+            cls.unlock_script = redis_client.register_script(UNLOCK_SCRIPT)
+            cls.extend_script = redis_client.register_script(EXTEND_SCRIPT)
+            cls.reset_script = redis_client.register_script(RESET_SCRIPT)
+            cls.reset_all_script = redis_client.register_script(RESET_ALL_SCRIPT)
+
+    @property
+    def _held(self):
+        return self.id == self.get_owner_id()
+
+    def reset(self):
+        """
+        Forcibly deletes the lock. Use this with care.
+        """
+        self.reset_script(client=self._client, keys=(self._name, self._signal), args=(self.id, self._signal_expire))
+
+    @property
+    def id(self):
+        return self._id
+
+    def get_owner_id(self):
+        owner_id = self._client.get(self._name)
+        if isinstance(owner_id, binary_type):
+            owner_id = owner_id.decode('ascii', 'replace')
+        return owner_id
+
+    def acquire(self, blocking=True, timeout=None):
+        """
+        :param blocking:
+            Boolean value specifying whether lock should be blocking or not.
+        :param timeout:
+            An integer value specifying the maximum number of seconds to block.
+        """
+        logger = loggers["acquire"]
+
+        logger.debug("Getting blocking: %s acquire on %r ...", blocking, self._name)
+
+        if self._held:
+            owner_id = self.get_owner_id()
+            raise AlreadyAcquired("Already acquired from this Lock instance. Lock id: {}".format(owner_id))
+
+        if not blocking and timeout is not None:
+            raise TimeoutNotUsable("Timeout cannot be used if blocking=False")
+
+        if timeout:
+            timeout = int(timeout)
+            if timeout < 0:
+                raise InvalidTimeout("Timeout (%d) cannot be less than or equal to 0" % timeout)
+
+            if self._expire and not self._lock_renewal_interval and timeout > self._expire:
+                raise TimeoutTooLarge("Timeout (%d) cannot be greater than expire (%d)" % (timeout, self._expire))
+
+        busy = True
+        blpop_timeout = timeout or self._expire or 0
+        timed_out = False
+        while busy:
+            busy = not self._client.set(self._name, self._id, nx=True, ex=self._expire)
+            if busy:
+                if timed_out:
+                    return False
+                elif blocking:
+                    timed_out = not self._client.blpop(self._signal, blpop_timeout) and timeout
+                else:
+                    logger.warning("Failed to get %r.", self._name)
+                    return False
+
+        logger.debug("Got lock for %r.", self._name)
+        if self._lock_renewal_interval is not None:
+            self._start_lock_renewer()
+        return True
+
+    def extend(self, expire=None):
+        """Extends expiration time of the lock.
+
+        :param expire:
+            New expiration time. If ``None`` - `expire` provided during
+            lock initialization will be taken.
+        """
+        if expire:
+            expire = int(expire)
+            if expire < 0:
+                raise ValueError("A negative expire is not acceptable.")
+        elif self._expire is not None:
+            expire = self._expire
+        else:
+            raise TypeError(
+                "To extend a lock 'expire' must be provided as an "
+                "argument to extend() method or at initialization time."
+            )
+
+        error = self.extend_script(client=self._client, keys=(self._name, self._signal), args=(self._id, expire))
+        if error == 1:
+            raise NotAcquired("Lock %s is not acquired or it already expired." % self._name)
+        elif error == 2:
+            raise NotExpirable("Lock %s has no assigned expiration time" % self._name)
+        elif error:
+            raise RuntimeError("Unsupported error code %s from EXTEND script" % error)
+
+    @staticmethod
+    def _lock_renewer(lockref, interval, stop):
+        """
+        Renew the lock key in redis every `interval` seconds for as long
+        as `self._lock_renewal_thread.should_exit` is False.
+        """
+        while not stop.wait(timeout=interval):
+            loggers["refresh.thread.start"].debug("Refreshing lock")
+            lock = lockref()
+            if lock is None:
+                loggers["refresh.thread.stop"].debug(
+                    "The lock no longer exists, stopping lock refreshing"
+                )
+                break
+            lock.extend(expire=lock._expire)
+            del lock
+        loggers["refresh.thread.exit"].debug("Exit requested, stopping lock refreshing")
+
+    def _start_lock_renewer(self):
+        """
+        Starts the lock refresher thread.
+        """
+        if self._lock_renewal_thread is not None:
+            raise AlreadyStarted("Lock refresh thread already started")
+
+        loggers["refresh.start"].debug(
+            "Starting thread to refresh lock every %s seconds",
+            self._lock_renewal_interval
+        )
+        self._lock_renewal_stop = threading.Event()
+        self._lock_renewal_thread = threading.Thread(
+            group=None,
+            target=self._lock_renewer,
+            kwargs={'lockref': weakref.ref(self),
+                    'interval': self._lock_renewal_interval,
+                    'stop': self._lock_renewal_stop}
+        )
+        self._lock_renewal_thread.setDaemon(True)
+        self._lock_renewal_thread.start()
+
+    def _stop_lock_renewer(self):
+        """
+        Stop the lock renewer.
+
+        This signals the renewal thread and waits for its exit.
+        """
+        if self._lock_renewal_thread is None or not self._lock_renewal_thread.is_alive():
+            return
+        loggers["refresh.shutdown"].debug("Signalling the lock refresher to stop")
+        self._lock_renewal_stop.set()
+        self._lock_renewal_thread.join()
+        self._lock_renewal_thread = None
+        loggers["refresh.exit"].debug("Lock refresher has stopped")
+
+    def __enter__(self):
+        acquired = self.acquire(blocking=True)
+        assert acquired, "Lock wasn't acquired, but blocking=True"
+        return self
+
+    def __exit__(self, exc_type=None, exc_value=None, traceback=None):
+        self.release()
+
+    def release(self):
+        """Releases the lock, that was acquired with the same object.
+
+        .. note::
+
+            If you want to release a lock that you acquired in a different place you have two choices:
+
+            * Use ``Lock("name", id=id_from_other_place).release()``
+            * Use ``Lock("name").reset()``
+        """
+        if self._lock_renewal_thread is not None:
+            self._stop_lock_renewer()
+        loggers["release"].debug("Releasing %r.", self._name)
+        error = self.unlock_script(client=self._client, keys=(self._name, self._signal), args=(self._id, self._signal_expire))
+        if error == 1:
+            raise NotAcquired("Lock %s is not acquired or it already expired." % self._name)
+        elif error:
+            raise RuntimeError("Unsupported error code %s from EXTEND script." % error)
+
+    def locked(self):
+        """
+        Return true if the lock is acquired.
+
+        Checks that lock with same name already exists. This method returns true, even if
+        lock have another id.
+        """
+        return self._client.exists(self._name) == 1
+
+
+reset_all_script = None
+
+
+def reset_all(redis_client):
+    """
+    Forcibly deletes all locks if its remains (like a crash reason). Use this with care.
+
+    :param redis_client:
+        An instance of :class:`~StrictRedis`.
+    """
+    Lock.register_scripts(redis_client)
+
+    reset_all_script(client=redis_client)  # noqa

vcsserver/lib/_vendor/statsd/__init__.py

@@ -0,0 +1,52 b''
+from __future__ import absolute_import, division, unicode_literals
+
+import logging
+
+from .stream import TCPStatsClient, UnixSocketStatsClient  # noqa
+from .udp import StatsClient  # noqa
+
+HOST = 'localhost'
+PORT = 8125
+IPV6 = False
+PREFIX = None
+MAXUDPSIZE = 512
+
+log = logging.getLogger('rhodecode.statsd')
+
+
+def statsd_config(config, prefix='statsd.'):
+    _config = {}
+    for key in config.keys():
+        if key.startswith(prefix):
+            _config[key[len(prefix):]] = config[key]
+    return _config
+
+
+def client_from_config(configuration, prefix='statsd.', **kwargs):
+    from pyramid.settings import asbool
+
+    _config = statsd_config(configuration, prefix)
+    statsd_enabled = asbool(_config.pop('enabled', False))
+    if not statsd_enabled:
+        log.debug('statsd client not enabled by statsd.enabled =  flag, skipping...')
+        return
+
+    host = _config.pop('statsd_host', HOST)
+    port = _config.pop('statsd_port', PORT)
+    prefix = _config.pop('statsd_prefix', PREFIX)
+    maxudpsize = _config.pop('statsd_maxudpsize', MAXUDPSIZE)
+    ipv6 = asbool(_config.pop('statsd_ipv6', IPV6))
+    log.debug('configured statsd client %s:%s', host, port)
+
+    try:
+        client = StatsClient(
+            host=host, port=port, prefix=prefix, maxudpsize=maxudpsize, ipv6=ipv6)
+    except Exception:
+        log.exception('StatsD is enabled, but failed to connect to statsd server, fallback: disable statsd')
+        client = None
+
+    return client
+
+
+def get_statsd_client(request):
+    return client_from_config(request.registry.settings)

vcsserver/lib/_vendor/statsd/base.py

@@ -0,0 +1,156 b''
+from __future__ import absolute_import, division, unicode_literals
+
+import re
+import random
+from collections import deque
+from datetime import timedelta
+from repoze.lru import lru_cache
+
+from .timer import Timer
+
+TAG_INVALID_CHARS_RE = re.compile(
+    r"[^\w\d_\-:/\.]",
+    #re.UNICODE
+)
+TAG_INVALID_CHARS_SUBS = "_"
+
+# we save and expose methods called by statsd for discovery
+buckets_dict = {
+
+}
+
+
+@lru_cache(maxsize=500)
+def _normalize_tags_with_cache(tag_list):
+    return [TAG_INVALID_CHARS_RE.sub(TAG_INVALID_CHARS_SUBS, tag) for tag in tag_list]
+
+
+def normalize_tags(tag_list):
+    # We have to turn our input tag list into a non-mutable tuple for it to
+    # be hashable (and thus usable) by the @lru_cache decorator.
+    return _normalize_tags_with_cache(tuple(tag_list))
+
+
+class StatsClientBase(object):
+    """A Base class for various statsd clients."""
+
+    def close(self):
+        """Used to close and clean up any underlying resources."""
+        raise NotImplementedError()
+
+    def _send(self):
+        raise NotImplementedError()
+
+    def pipeline(self):
+        raise NotImplementedError()
+
+    def timer(self, stat, rate=1, tags=None, auto_send=True):
+        """
+        statsd = StatsdClient.statsd
+        with statsd.timer('bucket_name', auto_send=True) as tmr:
+            # This block will be timed.
+            for i in xrange(0, 100000):
+                i ** 2
+        # you can access time here...
+        elapsed_ms = tmr.ms
+        """
+        return Timer(self, stat, rate, tags, auto_send=auto_send)
+
+    def timing(self, stat, delta, rate=1, tags=None, use_decimals=True):
+        """
+        Send new timing information.
+
+        `delta` can be either a number of milliseconds or a timedelta.
+        """
+        if isinstance(delta, timedelta):
+            # Convert timedelta to number of milliseconds.
+            delta = delta.total_seconds() * 1000.
+        if use_decimals:
+            fmt = '%0.6f|ms'
+        else:
+            fmt = '%s|ms'
+        self._send_stat(stat, fmt % delta, rate, tags)
+
+    def incr(self, stat, count=1, rate=1, tags=None):
+        """Increment a stat by `count`."""
+        self._send_stat(stat, '%s|c' % count, rate, tags)
+
+    def decr(self, stat, count=1, rate=1, tags=None):
+        """Decrement a stat by `count`."""
+        self.incr(stat, -count, rate, tags)
+
+    def gauge(self, stat, value, rate=1, delta=False, tags=None):
+        """Set a gauge value."""
+        if value < 0 and not delta:
+            if rate < 1:
+                if random.random() > rate:
+                    return
+            with self.pipeline() as pipe:
+                pipe._send_stat(stat, '0|g', 1)
+                pipe._send_stat(stat, '%s|g' % value, 1)
+        else:
+            prefix = '+' if delta and value >= 0 else ''
+            self._send_stat(stat, '%s%s|g' % (prefix, value), rate, tags)
+
+    def set(self, stat, value, rate=1):
+        """Set a set value."""
+        self._send_stat(stat, '%s|s' % value, rate)
+
+    def histogram(self, stat, value, rate=1, tags=None):
+        """Set a histogram"""
+        self._send_stat(stat, '%s|h' % value, rate, tags)
+
+    def _send_stat(self, stat, value, rate, tags=None):
+        self._after(self._prepare(stat, value, rate, tags))
+
+    def _prepare(self, stat, value, rate, tags=None):
+        global buckets_dict
+        buckets_dict[stat] = 1
+
+        if rate < 1:
+            if random.random() > rate:
+                return
+            value = '%s|@%s' % (value, rate)
+
+        if self._prefix:
+            stat = '%s.%s' % (self._prefix, stat)
+
+        res = '%s:%s%s' % (
+            stat,
+            value,
+            ("|#" + ",".join(normalize_tags(tags))) if tags else "",
+        )
+        return res
+
+    def _after(self, data):
+        if data:
+            self._send(data)
+
+
+class PipelineBase(StatsClientBase):
+
+    def __init__(self, client):
+        self._client = client
+        self._prefix = client._prefix
+        self._stats = deque()
+
+    def _send(self):
+        raise NotImplementedError()
+
+    def _after(self, data):
+        if data is not None:
+            self._stats.append(data)
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, typ, value, tb):
+        self.send()
+
+    def send(self):
+        if not self._stats:
+            return
+        self._send()
+
+    def pipeline(self):
+        return self.__class__(self)

vcsserver/lib/_vendor/statsd/stream.py

@@ -0,0 +1,75 b''
+from __future__ import absolute_import, division, unicode_literals
+
+import socket
+
+from .base import StatsClientBase, PipelineBase
+
+
+class StreamPipeline(PipelineBase):
+    def _send(self):
+        self._client._after('\n'.join(self._stats))
+        self._stats.clear()
+
+
+class StreamClientBase(StatsClientBase):
+    def connect(self):
+        raise NotImplementedError()
+
+    def close(self):
+        if self._sock and hasattr(self._sock, 'close'):
+            self._sock.close()
+        self._sock = None
+
+    def reconnect(self):
+        self.close()
+        self.connect()
+
+    def pipeline(self):
+        return StreamPipeline(self)
+
+    def _send(self, data):
+        """Send data to statsd."""
+        if not self._sock:
+            self.connect()
+        self._do_send(data)
+
+    def _do_send(self, data):
+        self._sock.sendall(data.encode('ascii') + b'\n')
+
+
+class TCPStatsClient(StreamClientBase):
+    """TCP version of StatsClient."""
+
+    def __init__(self, host='localhost', port=8125, prefix=None,
+                 timeout=None, ipv6=False):
+        """Create a new client."""
+        self._host = host
+        self._port = port
+        self._ipv6 = ipv6
+        self._timeout = timeout
+        self._prefix = prefix
+        self._sock = None
+
+    def connect(self):
+        fam = socket.AF_INET6 if self._ipv6 else socket.AF_INET
+        family, _, _, _, addr = socket.getaddrinfo(
+            self._host, self._port, fam, socket.SOCK_STREAM)[0]
+        self._sock = socket.socket(family, socket.SOCK_STREAM)
+        self._sock.settimeout(self._timeout)
+        self._sock.connect(addr)
+
+
+class UnixSocketStatsClient(StreamClientBase):
+    """Unix domain socket version of StatsClient."""
+
+    def __init__(self, socket_path, prefix=None, timeout=None):
+        """Create a new client."""
+        self._socket_path = socket_path
+        self._timeout = timeout
+        self._prefix = prefix
+        self._sock = None
+
+    def connect(self):
+        self._sock = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
+        self._sock.settimeout(self._timeout)
+        self._sock.connect(self._socket_path)

vcsserver/lib/_vendor/statsd/timer.py

@@ -0,0 +1,75 b''
+from __future__ import absolute_import, division, unicode_literals
+
+import functools
+
+# Use timer that's not susceptible to time of day adjustments.
+try:
+    # perf_counter is only present on Py3.3+
+    from time import perf_counter as time_now
+except ImportError:
+    # fall back to using time
+    from time import time as time_now
+
+
+def safe_wraps(wrapper, *args, **kwargs):
+    """Safely wraps partial functions."""
+    while isinstance(wrapper, functools.partial):
+        wrapper = wrapper.func
+    return functools.wraps(wrapper, *args, **kwargs)
+
+
+class Timer(object):
+    """A context manager/decorator for statsd.timing()."""
+
+    def __init__(self, client, stat, rate=1, tags=None, use_decimals=True, auto_send=True):
+        self.client = client
+        self.stat = stat
+        self.rate = rate
+        self.tags = tags
+        self.ms = None
+        self._sent = False
+        self._start_time = None
+        self.use_decimals = use_decimals
+        self.auto_send = auto_send
+
+    def __call__(self, f):
+        """Thread-safe timing function decorator."""
+        @safe_wraps(f)
+        def _wrapped(*args, **kwargs):
+            start_time = time_now()
+            try:
+                return f(*args, **kwargs)
+            finally:
+                elapsed_time_ms = 1000.0 * (time_now() - start_time)
+                self.client.timing(self.stat, elapsed_time_ms, self.rate, self.tags, self.use_decimals)
+                self._sent = True
+        return _wrapped
+
+    def __enter__(self):
+        return self.start()
+
+    def __exit__(self, typ, value, tb):
+        self.stop(send=self.auto_send)
+
+    def start(self):
+        self.ms = None
+        self._sent = False
+        self._start_time = time_now()
+        return self
+
+    def stop(self, send=True):
+        if self._start_time is None:
+            raise RuntimeError('Timer has not started.')
+        dt = time_now() - self._start_time
+        self.ms = 1000.0 * dt  # Convert to milliseconds.
+        if send:
+            self.send()
+        return self
+
+    def send(self):
+        if self.ms is None:
+            raise RuntimeError('No data recorded.')
+        if self._sent:
+            raise RuntimeError('Already sent data.')
+        self._sent = True
+        self.client.timing(self.stat, self.ms, self.rate, self.tags, self.use_decimals)

vcsserver/lib/_vendor/statsd/udp.py

@@ -0,0 +1,55 b''
+from __future__ import absolute_import, division, unicode_literals
+
+import socket
+
+from .base import StatsClientBase, PipelineBase
+
+
+class Pipeline(PipelineBase):
+
+    def __init__(self, client):
+        super(Pipeline, self).__init__(client)
+        self._maxudpsize = client._maxudpsize
+
+    def _send(self):
+        data = self._stats.popleft()
+        while self._stats:
+            # Use popleft to preserve the order of the stats.
+            stat = self._stats.popleft()
+            if len(stat) + len(data) + 1 >= self._maxudpsize:
+                self._client._after(data)
+                data = stat
+            else:
+                data += '\n' + stat
+        self._client._after(data)
+
+
+class StatsClient(StatsClientBase):
+    """A client for statsd."""
+
+    def __init__(self, host='localhost', port=8125, prefix=None,
+                 maxudpsize=512, ipv6=False):
+        """Create a new client."""
+        fam = socket.AF_INET6 if ipv6 else socket.AF_INET
+        family, _, _, _, addr = socket.getaddrinfo(
+            host, port, fam, socket.SOCK_DGRAM)[0]
+        self._addr = addr
+        self._sock = socket.socket(family, socket.SOCK_DGRAM)
+        self._prefix = prefix
+        self._maxudpsize = maxudpsize
+
+    def _send(self, data):
+        """Send data to statsd."""
+        try:
+            self._sock.sendto(data.encode('ascii'), self._addr)
+        except (socket.error, RuntimeError):
+            # No time for love, Dr. Jones!
+            pass
+
+    def close(self):
+        if self._sock and hasattr(self._sock, 'close'):
+            self._sock.close()
+        self._sock = None
+
+    def pipeline(self):
+        return Pipeline(self)

vcsserver/lib/statsd_client.py

@@ -0,0 +1,49 b''
+from vcsserver.lib._vendor.statsd import client_from_config
+
+
+class StatsdClientNotInitialised(Exception):
+    pass
+
+
+class _Singleton(type):
+    """A metaclass that creates a Singleton base class when called."""
+
+    _instances = {}
+
+    def __call__(cls, *args, **kwargs):
+        if cls not in cls._instances:
+            cls._instances[cls] = super(_Singleton, cls).__call__(*args, **kwargs)
+        return cls._instances[cls]
+
+
+class Singleton(_Singleton("SingletonMeta", (object,), {})):
+    pass
+
+
+class StatsdClientClass(Singleton):
+    setup_run = False
+    statsd_client = None
+    statsd = None
+
+    def __getattribute__(self, name):
+
+        if name.startswith("statsd"):
+            if self.setup_run:
+                return super(StatsdClientClass, self).__getattribute__(name)
+            else:
+                return None
+                #raise StatsdClientNotInitialised("requested key was %s" % name)
+
+        return super(StatsdClientClass, self).__getattribute__(name)
+
+    def setup(self, settings):
+        """
+        Initialize the client
+        """
+        statsd = client_from_config(settings)
+        self.statsd = statsd
+        self.statsd_client = statsd
+        self.setup_run = True
+
+
+StatsdClient = StatsdClientClass()

.bumpversion.cfg

@@ -1,6 +1,5 b''
 [bumpversion]
 current_version = 5.0.0
 message = release: Bump version {current_version} to {new_version}
 
 [bumpversion:file:vcsserver/VERSION]
-

.hgtags

@@ -1,63 +1,82 b''
 c6fad7d1e61f22b1f4a4863eff207a04c27e9462 v4.0.0
 77b6e243b4cc5b702c15abd6d737798edbac60dc v4.0.1
 a359c072337fdd8e1e71df72cc520b8a9b042f80 v4.1.0
 49aa7ed030a36b7ceba149a21e587cb5d20b4946 v4.1.1
 f38ed1e1a31dce3c170b4d31585ba43471cf0705 v4.1.2
 21269ba7bafd8f0c77e79dd86a31eb9bce7643d2 v4.2.0
 b53930c918c25b2c8f69ceddc6641e511be27fd3 v4.2.1
 6627ff4119723d8b2b60918e8b1aa49e9f055aab v4.3.0
 d38f2c2b861dde6c4178923f7cf15ea58b85aa92 v4.3.1
 1232313f9e6adac5ce5399c2a891dc1e72b79022 v4.4.0
 cbb9f1d329ae5768379cdec55a62ebdd546c4e27 v4.4.1
 24ffe44a27fcd1c5b6936144e176b9f6dd2f3a17 v4.4.2
 beaeeaa440cd17471110d4621b8816506c0dff4a v4.5.0
 668e5c656f61dd94595611844e1106d1361aa6a7 v4.5.1
 ae0640240cb7a77e6dc8c77e64dd80d79732cb5b v4.5.2
 7af06899f426813583fe60449d7517cc49c15b28 v4.6.0
 8f7f4299bf341b43f94dadafa1ea73d6cea2c9ba v4.6.1
 de00a831a0709ffaac57f948738ea927b97223a9 v4.7.0
 57f527e0646d731768fb5e0fe742b12a35bdc63b v4.7.1
 f9b09787da9845e4a105f4bffdc252099902cefb v4.7.2
 0b7c790b726f08385e6ebdf4f257c905787b9244 v4.8.0
 f4123e725b74d0e82fe89982ab8791a66062e2b3 v4.9.0
 940bac044a0fe1ec839759df81399b50141be720 v4.9.1
 582d9ebbe46bdddac4b26eacae36ee5ecabca267 v4.10.0
 12fbd08d0ab57acce9c0bdccee75633cfa08d7f4 v4.10.1
 78352f95021a9d128f5803fdbca7036daef5dabe v4.10.2
 a47ccfb020cda78c8680e3844aaf0b82b1390f3b v4.10.3
 347ae9ae544bba8deb417995285287a3b6be1611 v4.10.4
 9b257ac49841f850434be0d518baca0827e6c8cc v4.10.5
 e8bf26eea118694edc4ffe50c6c5aa91022bc434 v4.10.6
 71fa9274ba59fb982104f0b9b3d0d024c78675f7 v4.11.0
 92471577ef25636e5babe8001d47fc8e51521522 v4.11.1
 0277edbcda5a8d075e1e41a95bcee6dcf21f3f77 v4.11.2
 6c5ecbf0778ef870e5b23d9fad5340135b563356 v4.11.3
 be788a89a939ebd63606220064bd624fa9d5c9c9 v4.11.4
 15c90a04098a373ac761fab07695fd80dde3bcdb v4.11.5
 77aff155b3251cc00394a49f5e8f2c99e33149a7 v4.11.6
 c218a1ce5d370c2e671d42a91684b3fc2c91b81d v4.12.0
 80085fb846cc948195a5c76b579ca34cbc49b59b v4.12.1
 346f04fc8a18df3235defbe6e71bd552c0d46481 v4.12.2
 764fdd752322f3e0c13ea00957f2d548bf4363a7 v4.12.3
 b58038974a5cecbb9c100d32ad2e4c68582f1a78 v4.12.4
 e1d42d92a0fec0c80b56c82f37bc7b5472613706 v4.13.0
 c3ded3ff17e9bb2a47002a808984a7a946f58a1c v4.13.1
 7ff81aa47b1b40cdef9dd5bcdd439f59c269db3d v4.13.2
 628a08e6aaeff2c3f9e0e268e854f870e6778e53 v4.13.3
 941d675f10cfa7d774815bfacfb37085751b7a0d v4.14.0
 75e11d32c0be0a457198f07888e7ef650cfa6888 v4.14.1
 6c6f49fda0191c4641dcd43aa0d4376b8b728d40 v4.15.0
 184dea5e01c36e6474c83d3bb34719cdfec22b0d v4.15.1
 a4dc3669345553582296b2ce1485229a6c6f0522 v4.15.2
 d2a4a1a66f204668841da1cdccfa29083e1ef7a3 v4.16.0
 744cf8f2c8f23051978fc293404bf475cc5a31f6 v4.16.1
 e68aff93ce4ad11fea13420e914f7dfb05c39566 v4.16.2
 647aeff9752dc1aa00796fa280d0d2ce2f511bc9 v4.17.0
 5e0c2990e095bba1dc903cf0e6ef6ac035e0ccf9 v4.17.1
 8a824544d95037d76d99b104b5d2363858101d53 v4.17.2
 ccd806a2d9482f61bd7e8956a02a28eb24a1d46a v4.17.3
 e533ca02ccc205189b7bad9f227a312212772022 v4.17.4
 ba6a6dc9ecd7fd8b1dcd6eb0c4ee0210e897c426 v4.18.0
 17bc818b41bcf6883b9ff0da31f01d8c2a5d0781 v4.18.1
 1e9f12aa01f82c335abc9017efe94ce1c30b52ba v4.18.2
 f4cc6b3c5680bdf4541d7d442fbb7086640fb547 v4.18.3
+5dc0277e4f77bd4cc3042d99625bb5d3ba480c8c v4.19.0
+3a815eeb1b1efa340dda9b81a8da3cf24a7d605b v4.19.1
+8841da3680fba841e5a54ebccd8ca56c078f7553 v4.19.2
+4b0dec7fd80b1ca38e5073e5e562a5a450f73669 v4.19.3
+1485aa75ffe1b1ec48352dce7b7492d92f85e95f v4.20.0
+5b740274011766ef2f73803cc196d081e1e7f1d4 v4.20.1
+5a7835234e2c45e8fb8184c60f548a64b5842af8 v4.21.0
+26af88343015f8b89d5a66f92bc7547c51fcf0df v4.22.0
+cf54e5f700fe5dc50af1a1bdf5197c18cf52105f v4.23.0
+179d989bcfe02c6227f9f6aa9236cbbe1c14c400 v4.23.1
+383aee8b1652affaa26aefe336a89ee366b2b26d v4.23.2
+bc1a8141cc51fc23c455ebc50c6609c810b46f8d v4.24.0
+530a1c03caabc806ea1ef34605f8f67f18c70e55 v4.24.1
+5908ae65cee1043982e1b26d7b618af5fcfebbb3 v4.25.0
+cce8bcdf75090d5943a1e9706fe5212d7b5d1fa1 v4.25.1
+8610c4bf846c63bbc95d3ddfb53fadaaa9c7aa42 v4.25.2
+d46b7d1be72c76c9f9aaeab6a342951d54459f49 v4.26.0
+6fba0daab1e20a9e18fb70fa59bd21753e0a5b90 v4.27.0
+6195da4fc454087173918ae59cae946289458676 v4.27.1

configs/development.ini

@@ -1,237 +1,275 b''
 ## -*- coding: utf-8 -*-
 
 ; #################################
 ; RHODECODE VCSSERVER CONFIGURATION
 ; #################################
 
 [server:main]
 ; COMMON HOST/IP CONFIG
 host = 0.0.0.0
 port = 9900
 
 ; ##################################################
 ; WAITRESS WSGI SERVER - Recommended for Development
 ; ##################################################
 
 ; use server type
 use = egg:waitress#main
 
 ; number of worker threads
 threads = 5
 
 ; MAX BODY SIZE 100GB
 max_request_body_size = 107374182400
 
 ; Use poll instead of select, fixes file descriptors limits problems.
 ; May not work on old windows systems.
 asyncore_use_poll = true
 
 
 ; ###########################
 ; GUNICORN APPLICATION SERVER
 ; ###########################
 
-; run with gunicorn --log-config rhodecode.ini --paste rhodecode.ini
+; run with gunicorn --paste rhodecode.ini
 
 ; Module to use, this setting shouldn't be changed
 #use = egg:gunicorn#main
 
 ; Sets the number of process workers. More workers means more concurrent connections
 ; RhodeCode can handle at the same time. Each additional worker also it increases
 ; memory usage as each has it's own set of caches.
 ; Recommended value is (2 * NUMBER_OF_CPUS + 1), eg 2CPU = 5 workers, but no more
 ; than 8-10 unless for really big deployments .e.g 700-1000 users.
 ; `instance_id = *` must be set in the [app:main] section below (which is the default)
 ; when using more than 1 worker.
 #workers = 2
 
 ; Gunicorn access log level
 #loglevel = info
 
 ; Process name visible in process list
 #proc_name = rhodecode_vcsserver
 
 ; Type of worker class, one of `sync`, `gevent`
 ; currently `sync` is the only option allowed.
 #worker_class = sync
 
 ; The maximum number of simultaneous clients. Valid only for gevent
 #worker_connections = 10
 
 ; Max number of requests that worker will handle before being gracefully restarted.
 ; Prevents memory leaks, jitter adds variability so not all workers are restarted at once.
 #max_requests = 1000
 #max_requests_jitter = 30
 
 ; Amount of time a worker can spend with handling a request before it
 ; gets killed and restarted. By default set to 21600 (6hrs)
 ; Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h)
 #timeout = 21600
 
 ; The maximum size of HTTP request line in bytes.
 ; 0 for unlimited
 #limit_request_line = 0
 
 ; Limit the number of HTTP headers fields in a request.
 ; By default this value is 100 and can't be larger than 32768.
 #limit_request_fields = 32768
 
 ; Limit the allowed size of an HTTP request header field.
 ; Value is a positive number or 0.
 ; Setting it to 0 will allow unlimited header field sizes.
 #limit_request_field_size = 0
 
 ; Timeout for graceful workers restart.
 ; After receiving a restart signal, workers have this much time to finish
 ; serving requests. Workers still alive after the timeout (starting from the
 ; receipt of the restart signal) are force killed.
 ; Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h)
-#graceful_timeout = 3600
+#graceful_timeout = 21600
 
 # The number of seconds to wait for requests on a Keep-Alive connection.
 # Generally set in the 1-5 seconds range.
 #keepalive = 2
 
 ; Maximum memory usage that each worker can use before it will receive a
 ; graceful restart signal 0 = memory monitoring is disabled
 ; Examples: 268435456 (256MB), 536870912 (512MB)
 ; 1073741824 (1GB), 2147483648 (2GB), 4294967296 (4GB)
 #memory_max_usage = 0
 
 ; How often in seconds to check for memory usage for each gunicorn worker
 #memory_usage_check_interval = 60
 
 ; Threshold value for which we don't recycle worker if GarbageCollection
 ; frees up enough resources. Before each restart we try to run GC on worker
 ; in case we get enough free memory after that, restart will not happen.
 #memory_usage_recovery_threshold = 0.8
 
 
 [app:main]
 ; The %(here)s variable will be replaced with the absolute path of parent directory
 ; of this file
+; Each option in the app:main can be override by an environmental variable
+;
+;To override an option:
+;
+;RC_<KeyName>
+;Everything should be uppercase, . and - should be replaced by _.
+;For example, if you have these configuration settings:
+;rc_cache.repo_object.backend = foo
+;can be overridden by
+;export RC_CACHE_REPO_OBJECT_BACKEND=foo
+
 use = egg:rhodecode-vcsserver
 
 
 ; #############
 ; DEBUG OPTIONS
 ; #############
 
 # During development the we want to have the debug toolbar enabled
 pyramid.includes =
     pyramid_debugtoolbar
 
 debugtoolbar.hosts = 0.0.0.0/0
 debugtoolbar.exclude_prefixes =
     /css
     /fonts
     /images
     /js
 
 ; #################
 ; END DEBUG OPTIONS
 ; #################
 
 ; Pyramid default locales, we need this to be set
-pyramid.default_locale_name = en
+#pyramid.default_locale_name = en
 
 ; default locale used by VCS systems
-locale = en_US.UTF-8
+#locale = en_US.UTF-8
 
 ; path to binaries for vcsserver, it should be set by the installer
-; at installation time, e.g /home/user/vcsserver-1/profile/bin
+; at installation time, e.g /home/user/.rccontrol/vcsserver-1/profile/bin
 ; it can also be a path to nix-build output in case of development
 core.binary_dir = ""
 
 ; Custom exception store path, defaults to TMPDIR
 ; This is used to store exception from RhodeCode in shared directory
 #exception_tracker.store_path =
 
 ; #############
 ; DOGPILE CACHE
 ; #############
 
 ; Default cache dir for caches. Putting this into a ramdisk can boost performance.
 ; eg. /tmpfs/data_ramdisk, however this directory might require large amount of space
-cache_dir = %(here)s/data
+#cache_dir = %(here)s/data
 
 ; ***************************************
 ; `repo_object` cache, default file based
 ; ***************************************
 
 ; `repo_object` cache settings for vcs methods for repositories
-rc_cache.repo_object.backend = dogpile.cache.rc.file_namespace
+#rc_cache.repo_object.backend = dogpile.cache.rc.file_namespace
 
 ; cache auto-expires after N seconds
 ; Examples: 86400 (1Day), 604800 (7Days), 1209600 (14Days), 2592000 (30days), 7776000 (90Days)
-rc_cache.repo_object.expiration_time = 2592000
+#rc_cache.repo_object.expiration_time = 2592000
 
 ; file cache store path. Defaults to `cache_dir =` value or tempdir if both values are not set
-#rc_cache.repo_object.arguments.filename = /tmp/vcsserver_cache.db
+#rc_cache.repo_object.arguments.filename = /tmp/vcsserver_cache_repo_object.db
 
 ; ***********************************************************
 ; `repo_object` cache with redis backend
 ; recommended for larger instance, and for better performance
 ; ***********************************************************
 
 ; `repo_object` cache settings for vcs methods for repositories
 #rc_cache.repo_object.backend = dogpile.cache.rc.redis_msgpack
 
 ; cache auto-expires after N seconds
 ; Examples: 86400 (1Day), 604800 (7Days), 1209600 (14Days), 2592000 (30days), 7776000 (90Days)
 #rc_cache.repo_object.expiration_time = 2592000
 
 ; redis_expiration_time needs to be greater then expiration_time
 #rc_cache.repo_object.arguments.redis_expiration_time = 3592000
 
 #rc_cache.repo_object.arguments.host = localhost
 #rc_cache.repo_object.arguments.port = 6379
 #rc_cache.repo_object.arguments.db = 5
 #rc_cache.repo_object.arguments.socket_timeout = 30
 ; more Redis options: https://dogpilecache.sqlalchemy.org/en/latest/api.html#redis-backends
 #rc_cache.repo_object.arguments.distributed_lock = true
 
+; auto-renew lock to prevent stale locks, slower but safer. Use only if problems happen
+#rc_cache.repo_object.arguments.lock_auto_renewal = true
+
+; Statsd client config, this is used to send metrics to statsd
+; We recommend setting statsd_exported and scrape them using Promethues
+#statsd.enabled = false
+#statsd.statsd_host = 0.0.0.0
+#statsd.statsd_port = 8125
+#statsd.statsd_prefix =
+#statsd.statsd_ipv6 = false
+
+; configure logging automatically at server startup set to false
+; to use the below custom logging config.
+; RC_LOGGING_FORMATTER
+; RC_LOGGING_LEVEL
+; env variables can control the settings for logging in case of autoconfigure
+
+#logging.autoconfigure = true
+
+; specify your own custom logging config file to configure logging
+#logging.logging_conf_file = /path/to/custom_logging.ini
 
 ; #####################
 ; LOGGING CONFIGURATION
 ; #####################
+
 [loggers]
 keys = root, vcsserver
 
 [handlers]
 keys = console
 
 [formatters]
-keys = generic
+keys = generic, json
 
 ; #######
 ; LOGGERS
 ; #######
 [logger_root]
 level = NOTSET
 handlers = console
 
 [logger_vcsserver]
 level = DEBUG
 handlers =
 qualname = vcsserver
 propagate = 1
 
-
 ; ########
 ; HANDLERS
 ; ########
 
 [handler_console]
 class = StreamHandler
 args = (sys.stderr, )
 level = DEBUG
+; To enable JSON formatted logs replace 'generic' with 'json'
+; This allows sending properly formatted logs to grafana loki or elasticsearch
 formatter = generic
 
 ; ##########
 ; FORMATTERS
 ; ##########
 
 [formatter_generic]
 format = %(asctime)s.%(msecs)03d [%(process)d] %(levelname)-5.5s [%(name)s] %(message)s
 datefmt = %Y-%m-%d %H:%M:%S
+
+[formatter_json]
+format = %(timestamp)s %(levelname)s %(name)s %(message)s %(req_id)s
+class = vcsserver.lib._vendor.jsonlogger.JsonFormatter

configs/gunicorn_config.py

@@ -1,262 +1,393 b''
 """
 Gunicorn config extension and hooks. This config file adds some extra settings and memory management.
 Gunicorn configuration should be managed by .ini files entries of RhodeCode or VCSServer
 """
 
 import gc
 import os
 import sys
 import math
 import time
 import threading
 import traceback
 import random
+import socket
 from gunicorn.glogging import Logger
 
 
 def get_workers():
     import multiprocessing
     return multiprocessing.cpu_count() * 2 + 1
 
 # GLOBAL
 errorlog = '-'
 accesslog = '-'
 
 
 # SERVER MECHANICS
 # None == system temp dir
 # worker_tmp_dir is recommended to be set to some tmpfs
 worker_tmp_dir = None
 tmp_upload_dir = None
 
+#reuse_port = True
+
 # Custom log format
+#access_log_format = (
+#    '%(t)s %(p)s INFO  [GNCRN] %(h)-15s rqt:%(L)s %(s)s %(b)-6s "%(m)s:%(U)s %(q)s" usr:%(u)s "%(f)s" "%(a)s"')
+
+# loki format for easier parsing in grafana
 access_log_format = (
-    '%(t)s %(p)s INFO  [GNCRN] %(h)-15s rqt:%(L)s %(s)s %(b)-6s "%(m)s:%(U)s %(q)s" usr:%(u)s "%(f)s" "%(a)s"')
+    'time="%(t)s" pid=%(p)s level="INFO" type="[GNCRN]" ip="%(h)-15s" rqt="%(L)s" response_code="%(s)s" response_bytes="%(b)-6s" uri="%(m)s:%(U)s %(q)s" user=":%(u)s" user_agent="%(a)s"')
 
 # self adjust workers based on CPU count
 # workers = get_workers()
 
 
 def _get_process_rss(pid=None):
     try:
         import psutil
         if pid:
             proc = psutil.Process(pid)
         else:
             proc = psutil.Process()
         return proc.memory_info().rss
     except Exception:
         return None
 
 
 def _get_config(ini_path):
-    import configparser
+        import configparser
 
     try:
         config = configparser.RawConfigParser()
         config.read(ini_path)
         return config
     except Exception:
         return None
 
 
 def _time_with_offset(memory_usage_check_interval):
     return time.time() - random.randint(0, memory_usage_check_interval/2.0)
 
 
 def pre_fork(server, worker):
     pass
 
 
 def post_fork(server, worker):
 
     # memory spec defaults
     _memory_max_usage = 0
     _memory_usage_check_interval = 60
     _memory_usage_recovery_threshold = 0.8
 
     ini_path = os.path.abspath(server.cfg.paste)
     conf = _get_config(ini_path)
 
     section = 'server:main'
     if conf and conf.has_section(section):
 
         if conf.has_option(section, 'memory_max_usage'):
             _memory_max_usage = conf.getint(section, 'memory_max_usage')
 
         if conf.has_option(section, 'memory_usage_check_interval'):
             _memory_usage_check_interval = conf.getint(section, 'memory_usage_check_interval')
 
         if conf.has_option(section, 'memory_usage_recovery_threshold'):
             _memory_usage_recovery_threshold = conf.getfloat(section, 'memory_usage_recovery_threshold')
 
-    worker._memory_max_usage = _memory_max_usage
-    worker._memory_usage_check_interval = _memory_usage_check_interval
-    worker._memory_usage_recovery_threshold = _memory_usage_recovery_threshold
+    worker._memory_max_usage = int(os.environ.get('RC_GUNICORN_MEMORY_MAX_USAGE', '')
+                                   or _memory_max_usage)
+    worker._memory_usage_check_interval = int(os.environ.get('RC_GUNICORN_MEMORY_USAGE_CHECK_INTERVAL', '')
+                                              or _memory_usage_check_interval)
+    worker._memory_usage_recovery_threshold = float(os.environ.get('RC_GUNICORN_MEMORY_USAGE_RECOVERY_THRESHOLD', '')
+                                                    or _memory_usage_recovery_threshold)
 
     # register memory last check time, with some random offset so we don't recycle all
     # at once
     worker._last_memory_check_time = _time_with_offset(_memory_usage_check_interval)
 
     if _memory_max_usage:
         server.log.info("[%-10s] WORKER spawned with max memory set at %s", worker.pid,
                         _format_data_size(_memory_max_usage))
     else:
         server.log.info("[%-10s] WORKER spawned", worker.pid)
 
 
 def pre_exec(server):
     server.log.info("Forked child, re-executing.")
 
 
 def on_starting(server):
     server_lbl = '{} {}'.format(server.proc_name, server.address)
     server.log.info("Server %s is starting.", server_lbl)
 
 
 def when_ready(server):
     server.log.info("Server %s is ready. Spawning workers", server)
 
 
 def on_reload(server):
     pass
 
 
 def _format_data_size(size, unit="B", precision=1, binary=True):
     """Format a number using SI units (kilo, mega, etc.).
 
     ``size``: The number as a float or int.
 
     ``unit``: The unit name in plural form. Examples: "bytes", "B".
 
     ``precision``: How many digits to the right of the decimal point. Default
     is 1.  0 suppresses the decimal point.
 
     ``binary``: If false, use base-10 decimal prefixes (kilo = K = 1000).
     If true, use base-2 binary prefixes (kibi = Ki = 1024).
 
     ``full_name``: If false (default), use the prefix abbreviation ("k" or
     "Ki").  If true, use the full prefix ("kilo" or "kibi"). If false,
     use abbreviation ("k" or "Ki").
 
     """
 
     if not binary:
         base = 1000
         multiples = ('', 'k', 'M', 'G', 'T', 'P', 'E', 'Z', 'Y')
     else:
         base = 1024
         multiples = ('', 'Ki', 'Mi', 'Gi', 'Ti', 'Pi', 'Ei', 'Zi', 'Yi')
 
     sign = ""
     if size > 0:
         m = int(math.log(size, base))
     elif size < 0:
         sign = "-"
         size = -size
         m = int(math.log(size, base))
     else:
         m = 0
     if m > 8:
         m = 8
 
     if m == 0:
         precision = '%.0f'
     else:
         precision = '%%.%df' % precision
 
     size = precision % (size / math.pow(base, m))
 
     return '%s%s %s%s' % (sign, size.strip(), multiples[m], unit)
 
 
 def _check_memory_usage(worker):
     memory_max_usage = worker._memory_max_usage
     if not memory_max_usage:
         return
 
     memory_usage_check_interval = worker._memory_usage_check_interval
     memory_usage_recovery_threshold = memory_max_usage * worker._memory_usage_recovery_threshold
 
     elapsed = time.time() - worker._last_memory_check_time
     if elapsed > memory_usage_check_interval:
         mem_usage = _get_process_rss()
         if mem_usage and mem_usage > memory_max_usage:
             worker.log.info(
                 "memory usage %s > %s, forcing gc",
                 _format_data_size(mem_usage), _format_data_size(memory_max_usage))
             # Try to clean it up by forcing a full collection.
             gc.collect()
             mem_usage = _get_process_rss()
             if mem_usage > memory_usage_recovery_threshold:
                 # Didn't clean up enough, we'll have to terminate.
                 worker.log.warning(
                     "memory usage %s > %s after gc, quitting",
                     _format_data_size(mem_usage), _format_data_size(memory_max_usage))
                 # This will cause worker to auto-restart itself
                 worker.alive = False
         worker._last_memory_check_time = time.time()
 
 
 def worker_int(worker):
     worker.log.info("[%-10s] worker received INT or QUIT signal", worker.pid)
 
     # get traceback info, on worker crash
     id2name = dict([(th.ident, th.name) for th in threading.enumerate()])
     code = []
     for thread_id, stack in sys._current_frames().items():
         code.append(
             "\n# Thread: %s(%d)" % (id2name.get(thread_id, ""), thread_id))
         for fname, lineno, name, line in traceback.extract_stack(stack):
             code.append('File: "%s", line %d, in %s' % (fname, lineno, name))
             if line:
                 code.append("  %s" % (line.strip()))
     worker.log.debug("\n".join(code))
 
 
 def worker_abort(worker):
     worker.log.info("[%-10s] worker received SIGABRT signal", worker.pid)
 
 
 def worker_exit(server, worker):
     worker.log.info("[%-10s] worker exit", worker.pid)
 
 
 def child_exit(server, worker):
     worker.log.info("[%-10s] worker child exit", worker.pid)
 
 
 def pre_request(worker, req):
     worker.start_time = time.time()
     worker.log.debug(
         "GNCRN PRE  WORKER [cnt:%s]: %s %s", worker.nr, req.method, req.path)
 
 
 def post_request(worker, req, environ, resp):
     total_time = time.time() - worker.start_time
     # Gunicorn sometimes has problems with reading the status_code
     status_code = getattr(resp, 'status_code', '')
     worker.log.debug(
         "GNCRN POST WORKER [cnt:%s]: %s %s resp: %s, Load Time: %.4fs",
         worker.nr, req.method, req.path, status_code, total_time)
     _check_memory_usage(worker)
 
 
+def _filter_proxy(ip):
+    """
+    Passed in IP addresses in HEADERS can be in a special format of multiple
+    ips. Those comma separated IPs are passed from various proxies in the
+    chain of request processing. The left-most being the original client.
+    We only care about the first IP which came from the org. client.
+
+    :param ip: ip string from headers
+    """
+    if ',' in ip:
+        _ips = ip.split(',')
+        _first_ip = _ips[0].strip()
+        return _first_ip
+    return ip
+
+
+def _filter_port(ip):
+    """
+    Removes a port from ip, there are 4 main cases to handle here.
+    - ipv4 eg. 127.0.0.1
+    - ipv6 eg. ::1
+    - ipv4+port eg. 127.0.0.1:8080
+    - ipv6+port eg. [::1]:8080
+
+    :param ip:
+    """
+    def is_ipv6(ip_addr):
+        if hasattr(socket, 'inet_pton'):
+            try:
+                socket.inet_pton(socket.AF_INET6, ip_addr)
+            except socket.error:
+                return False
+        else:
+            return False
+        return True
+
+    if ':' not in ip:  # must be ipv4 pure ip
+        return ip
+
+    if '[' in ip and ']' in ip:  # ipv6 with port
+        return ip.split(']')[0][1:].lower()
+
+    # must be ipv6 or ipv4 with port
+    if is_ipv6(ip):
+        return ip
+    else:
+        ip, _port = ip.split(':')[:2]  # means ipv4+port
+        return ip
+
+
+def get_ip_addr(environ):
+    proxy_key = 'HTTP_X_REAL_IP'
+    proxy_key2 = 'HTTP_X_FORWARDED_FOR'
+    def_key = 'REMOTE_ADDR'
+    _filters = lambda x: _filter_port(_filter_proxy(x))
+
+    ip = environ.get(proxy_key)
+    if ip:
+        return _filters(ip)
+
+    ip = environ.get(proxy_key2)
+    if ip:
+        return _filters(ip)
+
+    ip = environ.get(def_key, '0.0.0.0')
+    return _filters(ip)
+
+
 class RhodeCodeLogger(Logger):
     """
     Custom Logger that allows some customization that gunicorn doesn't allow
     """
 
     datefmt = r"%Y-%m-%d %H:%M:%S"
 
     def __init__(self, cfg):
         Logger.__init__(self, cfg)
 
     def now(self):
         """ return date in RhodeCode Log format """
         now = time.time()
         msecs = int((now - long(now)) * 1000)
         return time.strftime(self.datefmt, time.localtime(now)) + '.{0:03d}'.format(msecs)
 
+    def atoms(self, resp, req, environ, request_time):
+        """ Gets atoms for log formatting.
+        """
+        status = resp.status
+        if isinstance(status, str):
+            status = status.split(None, 1)[0]
+        atoms = {
+            'h': get_ip_addr(environ),
+            'l': '-',
+            'u': self._get_user(environ) or '-',
+            't': self.now(),
+            'r': "%s %s %s" % (environ['REQUEST_METHOD'],
+                               environ['RAW_URI'],
+                               environ["SERVER_PROTOCOL"]),
+            's': status,
+            'm': environ.get('REQUEST_METHOD'),
+            'U': environ.get('PATH_INFO'),
+            'q': environ.get('QUERY_STRING'),
+            'H': environ.get('SERVER_PROTOCOL'),
+            'b': getattr(resp, 'sent', None) is not None and str(resp.sent) or '-',
+            'B': getattr(resp, 'sent', None),
+            'f': environ.get('HTTP_REFERER', '-'),
+            'a': environ.get('HTTP_USER_AGENT', '-'),
+            'T': request_time.seconds,
+            'D': (request_time.seconds * 1000000) + request_time.microseconds,
+            'M': (request_time.seconds * 1000) + int(request_time.microseconds/1000),
+            'L': "%d.%06d" % (request_time.seconds, request_time.microseconds),
+            'p': "<%s>" % os.getpid()
+        }
+
+        # add request headers
+        if hasattr(req, 'headers'):
+            req_headers = req.headers
+        else:
+            req_headers = req
+
+        if hasattr(req_headers, "items"):
+            req_headers = req_headers.items()
+
+        atoms.update({"{%s}i" % k.lower(): v for k, v in req_headers})
+
+        resp_headers = resp.headers
+        if hasattr(resp_headers, "items"):
+            resp_headers = resp_headers.items()
+
+        # add response headers
+        atoms.update({"{%s}o" % k.lower(): v for k, v in resp_headers})
+
+        # add environ variables
+        environ_variables = environ.items()
+        atoms.update({"{%s}e" % k.lower(): v for k, v in environ_variables})
+
+        return atoms
 
 logger_class = RhodeCodeLogger

configs/production.ini

@@ -1,200 +1,238 b''
 ## -*- coding: utf-8 -*-
 
 ; #################################
 ; RHODECODE VCSSERVER CONFIGURATION
 ; #################################
 
 [server:main]
 ; COMMON HOST/IP CONFIG
 host = 127.0.0.1
 port = 9900
 
 
 ; ###########################
 ; GUNICORN APPLICATION SERVER
 ; ###########################
 
-; run with gunicorn --log-config rhodecode.ini --paste rhodecode.ini
+; run with gunicorn --paste rhodecode.ini
 
 ; Module to use, this setting shouldn't be changed
 use = egg:gunicorn#main
 
 ; Sets the number of process workers. More workers means more concurrent connections
 ; RhodeCode can handle at the same time. Each additional worker also it increases
 ; memory usage as each has it's own set of caches.
 ; Recommended value is (2 * NUMBER_OF_CPUS + 1), eg 2CPU = 5 workers, but no more
 ; than 8-10 unless for really big deployments .e.g 700-1000 users.
 ; `instance_id = *` must be set in the [app:main] section below (which is the default)
 ; when using more than 1 worker.
 workers = 2
 
 ; Gunicorn access log level
 loglevel = info
 
 ; Process name visible in process list
 proc_name = rhodecode_vcsserver
 
 ; Type of worker class, one of `sync`, `gevent`
 ; currently `sync` is the only option allowed.
 worker_class = sync
 
 ; The maximum number of simultaneous clients. Valid only for gevent
 worker_connections = 10
 
 ; Max number of requests that worker will handle before being gracefully restarted.
 ; Prevents memory leaks, jitter adds variability so not all workers are restarted at once.
 max_requests = 1000
 max_requests_jitter = 30
 
 ; Amount of time a worker can spend with handling a request before it
 ; gets killed and restarted. By default set to 21600 (6hrs)
 ; Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h)
 timeout = 21600
 
 ; The maximum size of HTTP request line in bytes.
 ; 0 for unlimited
 limit_request_line = 0
 
 ; Limit the number of HTTP headers fields in a request.
 ; By default this value is 100 and can't be larger than 32768.
 limit_request_fields = 32768
 
 ; Limit the allowed size of an HTTP request header field.
 ; Value is a positive number or 0.
 ; Setting it to 0 will allow unlimited header field sizes.
 limit_request_field_size = 0
 
 ; Timeout for graceful workers restart.
 ; After receiving a restart signal, workers have this much time to finish
 ; serving requests. Workers still alive after the timeout (starting from the
 ; receipt of the restart signal) are force killed.
 ; Examples: 1800 (30min), 3600 (1hr), 7200 (2hr), 43200 (12h)
-graceful_timeout = 3600
+graceful_timeout = 21600
 
 # The number of seconds to wait for requests on a Keep-Alive connection.
 # Generally set in the 1-5 seconds range.
 keepalive = 2
 
 ; Maximum memory usage that each worker can use before it will receive a
 ; graceful restart signal 0 = memory monitoring is disabled
 ; Examples: 268435456 (256MB), 536870912 (512MB)
 ; 1073741824 (1GB), 2147483648 (2GB), 4294967296 (4GB)
 memory_max_usage = 0
 
 ; How often in seconds to check for memory usage for each gunicorn worker
 memory_usage_check_interval = 60
 
 ; Threshold value for which we don't recycle worker if GarbageCollection
 ; frees up enough resources. Before each restart we try to run GC on worker
 ; in case we get enough free memory after that, restart will not happen.
 memory_usage_recovery_threshold = 0.8
 
 
 [app:main]
 ; The %(here)s variable will be replaced with the absolute path of parent directory
 ; of this file
+; Each option in the app:main can be override by an environmental variable
+;
+;To override an option:
+;
+;RC_<KeyName>
+;Everything should be uppercase, . and - should be replaced by _.
+;For example, if you have these configuration settings:
+;rc_cache.repo_object.backend = foo
+;can be overridden by
+;export RC_CACHE_REPO_OBJECT_BACKEND=foo
+
 use = egg:rhodecode-vcsserver
 
 ; Pyramid default locales, we need this to be set
-pyramid.default_locale_name = en
+#pyramid.default_locale_name = en
 
 ; default locale used by VCS systems
-locale = en_US.UTF-8
+#locale = en_US.UTF-8
 
 ; path to binaries for vcsserver, it should be set by the installer
-; at installation time, e.g /home/user/vcsserver-1/profile/bin
+; at installation time, e.g /home/user/.rccontrol/vcsserver-1/profile/bin
 ; it can also be a path to nix-build output in case of development
 core.binary_dir = ""
 
 ; Custom exception store path, defaults to TMPDIR
 ; This is used to store exception from RhodeCode in shared directory
 #exception_tracker.store_path =
 
 ; #############
 ; DOGPILE CACHE
 ; #############
 
 ; Default cache dir for caches. Putting this into a ramdisk can boost performance.
 ; eg. /tmpfs/data_ramdisk, however this directory might require large amount of space
-cache_dir = %(here)s/data
+#cache_dir = %(here)s/data
 
 ; ***************************************
 ; `repo_object` cache, default file based
 ; ***************************************
 
 ; `repo_object` cache settings for vcs methods for repositories
-rc_cache.repo_object.backend = dogpile.cache.rc.file_namespace
+#rc_cache.repo_object.backend = dogpile.cache.rc.file_namespace
 
 ; cache auto-expires after N seconds
 ; Examples: 86400 (1Day), 604800 (7Days), 1209600 (14Days), 2592000 (30days), 7776000 (90Days)
-rc_cache.repo_object.expiration_time = 2592000
+#rc_cache.repo_object.expiration_time = 2592000
 
 ; file cache store path. Defaults to `cache_dir =` value or tempdir if both values are not set
-#rc_cache.repo_object.arguments.filename = /tmp/vcsserver_cache.db
+#rc_cache.repo_object.arguments.filename = /tmp/vcsserver_cache_repo_object.db
 
 ; ***********************************************************
 ; `repo_object` cache with redis backend
 ; recommended for larger instance, and for better performance
 ; ***********************************************************
 
 ; `repo_object` cache settings for vcs methods for repositories
 #rc_cache.repo_object.backend = dogpile.cache.rc.redis_msgpack
 
 ; cache auto-expires after N seconds
 ; Examples: 86400 (1Day), 604800 (7Days), 1209600 (14Days), 2592000 (30days), 7776000 (90Days)
 #rc_cache.repo_object.expiration_time = 2592000
 
 ; redis_expiration_time needs to be greater then expiration_time
 #rc_cache.repo_object.arguments.redis_expiration_time = 3592000
 
 #rc_cache.repo_object.arguments.host = localhost
 #rc_cache.repo_object.arguments.port = 6379
 #rc_cache.repo_object.arguments.db = 5
 #rc_cache.repo_object.arguments.socket_timeout = 30
 ; more Redis options: https://dogpilecache.sqlalchemy.org/en/latest/api.html#redis-backends
 #rc_cache.repo_object.arguments.distributed_lock = true
 
+; auto-renew lock to prevent stale locks, slower but safer. Use only if problems happen
+#rc_cache.repo_object.arguments.lock_auto_renewal = true
+
+; Statsd client config, this is used to send metrics to statsd
+; We recommend setting statsd_exported and scrape them using Promethues
+#statsd.enabled = false
+#statsd.statsd_host = 0.0.0.0
+#statsd.statsd_port = 8125
+#statsd.statsd_prefix =
+#statsd.statsd_ipv6 = false
+
+; configure logging automatically at server startup set to false
+; to use the below custom logging config.
+; RC_LOGGING_FORMATTER
+; RC_LOGGING_LEVEL
+; env variables can control the settings for logging in case of autoconfigure
+
+#logging.autoconfigure = true
+
+; specify your own custom logging config file to configure logging
+#logging.logging_conf_file = /path/to/custom_logging.ini
 
 ; #####################
 ; LOGGING CONFIGURATION
 ; #####################
+
 [loggers]
 keys = root, vcsserver
 
 [handlers]
 keys = console
 
 [formatters]
-keys = generic
+keys = generic, json
 
 ; #######
 ; LOGGERS
 ; #######
 [logger_root]
 level = NOTSET
 handlers = console
 
 [logger_vcsserver]
-level = DEBUG
+level = INFO
 handlers =
 qualname = vcsserver
 propagate = 1
 
-
 ; ########
 ; HANDLERS
 ; ########
 
 [handler_console]
 class = StreamHandler
 args = (sys.stderr, )
 level = INFO
+; To enable JSON formatted logs replace 'generic' with 'json'
+; This allows sending properly formatted logs to grafana loki or elasticsearch
 formatter = generic
 
 ; ##########
 ; FORMATTERS
 ; ##########
 
 [formatter_generic]
 format = %(asctime)s.%(msecs)03d [%(process)d] %(levelname)-5.5s [%(name)s] %(message)s
 datefmt = %Y-%m-%d %H:%M:%S
+
+[formatter_json]
+format = %(timestamp)s %(levelname)s %(name)s %(message)s %(req_id)s
+class = vcsserver.lib._vendor.jsonlogger.JsonFormatter

setup.py

@@ -1,134 +1,163 b''
 # -*- coding: utf-8 -*-
+
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2019 RodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 # Import early to make sure things are patched up properly
 from setuptools import setup, find_packages
 
 import os
+import re
 import sys
 import pkgutil
 import platform
 import codecs
 
-try:  # for pip >= 10
+import pip
+
+pip_major_version = int(pip.__version__.split(".")[0])
+if pip_major_version >= 20:
     from pip._internal.req import parse_requirements
-except ImportError:  # for pip <= 9.0.3
+    from pip._internal.network.session import PipSession
+elif pip_major_version >= 10:
+    from pip._internal.req import parse_requirements
+    from pip._internal.download import PipSession
+else:
     from pip.req import parse_requirements
+    from pip.download import PipSession
+
 
-try:  # for pip >= 10
-    from pip._internal.download import PipSession
-except ImportError:  # for pip <= 9.0.3
-    from pip.download import PipSession
+def get_package_name(req_object):
+    package_name = None
+    try:
+        from pip._internal.req.constructors import install_req_from_parsed_requirement
+    except ImportError:
+        install_req_from_parsed_requirement = None
+
+    # In 20.1 of pip, the requirements object changed
+    if hasattr(req_object, 'req'):
+        package_name = req_object.req.name
+
+    if package_name is None:
+        if install_req_from_parsed_requirement:
+            package = install_req_from_parsed_requirement(req_object)
+            package_name = package.req.name
+
+    if package_name is None:
+        # fallback for older pip
+        package_name = re.split('===|<=|!=|==|>=|~=|<|>', req_object.requirement)[0]
+
+    return package_name
 
 
 if sys.version_info < (2, 7):
     raise Exception('VCSServer requires Python 2.7 or later')
 
 here = os.path.abspath(os.path.dirname(__file__))
 
 # defines current platform
 __platform__ = platform.system()
 __license__ = 'GPL V3'
 __author__ = 'RhodeCode GmbH'
 __url__ = 'https://code.rhodecode.com'
 is_windows = __platform__ in ('Windows',)
 
 
 def _get_requirements(req_filename, exclude=None, extras=None):
     extras = extras or []
     exclude = exclude or []
 
     try:
         parsed = parse_requirements(
             os.path.join(here, req_filename), session=PipSession())
     except TypeError:
         # try pip < 6.0.0, that doesn't support session
         parsed = parse_requirements(os.path.join(here, req_filename))
 
     requirements = []
-    for ir in parsed:
-        if ir.req and ir.name not in exclude:
-            requirements.append(str(ir.req))
+    for int_req in parsed:
+        req_name = get_package_name(int_req)
+        if req_name not in exclude:
+            requirements.append(req_name)
     return requirements + extras
 
 
 # requirements extract
-setup_requirements = ['pytest-runner']
+setup_requirements = []
 install_requirements = _get_requirements(
     'requirements.txt', exclude=['setuptools'])
 test_requirements = _get_requirements(
     'requirements_test.txt', extras=['configobj'])
 
 
 def get_version():
     version = pkgutil.get_data('vcsserver', 'VERSION')
     return version.decode().strip()
 
 
 # additional files that goes into package itself
 package_data = {
     '': ['*.txt', '*.rst'],
     'configs': ['*.ini'],
     'vcsserver': ['VERSION'],
 }
 
 description = 'Version Control System Server'
 keywords = ' '.join(['Version Control System'])
 
 # README/DESCRIPTION generation
 readme_file = 'README.rst'
 changelog_file = 'CHANGES.rst'
 try:
     long_description = codecs.open(readme_file).read() + '\n\n' + \
                        codecs.open(changelog_file).read()
 except IOError as err:
     sys.stderr.write(
         "[WARNING] Cannot find file specified as long_description (%s)\n "
         "or changelog (%s) skipping that file" % (readme_file, changelog_file))
     long_description = description
 
 
 setup(
     name='rhodecode-vcsserver',
     version=get_version(),
     description=description,
     long_description=long_description,
     keywords=keywords,
     license=__license__,
     author=__author__,
     author_email='[e-mail]',
     url=__url__,
     setup_requires=setup_requirements,
     install_requires=install_requirements,
     tests_require=test_requirements,
     zip_safe=False,
     packages=find_packages(exclude=["docs", "tests*"]),
     package_data=package_data,
     include_package_data=True,
     classifiers=[
         'Development Status :: 6 - Mature',
         'Intended Audience :: Developers',
         'Operating System :: OS Independent',
         'Topic :: Software Development :: Version Control',
         'License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)',
         'Programming Language :: Python :: 2.7',
     ],
     entry_points={
         'paste.app_factory': ['main=vcsserver.http_main:main']
     },
 )

vcsserver/base.py

@@ -1,76 +1,132 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
-
+import os
 import sys
 import traceback
 import logging
 import urllib.parse
 
 from vcsserver.lib.rc_cache import region_meta
+
+from vcsserver import exceptions
+from vcsserver.exceptions import NoContentException
+from vcsserver.hgcompat import (archival)
+
 log = logging.getLogger(__name__)
 
 
 class RepoFactory(object):
     """
     Utility to create instances of repository
 
     It provides internal caching of the `repo` object based on
     the :term:`call context`.
     """
     repo_type = None
 
     def __init__(self):
         self._cache_region = region_meta.dogpile_cache_regions['repo_object']
 
     def _create_config(self, path, config):
         config = {}
         return config
 
     def _create_repo(self, wire, create):
         raise NotImplementedError()
 
     def repo(self, wire, create=False):
         raise NotImplementedError()
 
 
 def obfuscate_qs(query_string):
     if query_string is None:
         return None
 
     parsed = []
     for k, v in urllib.parse.parse_qsl(query_string, keep_blank_values=True):
         if k in ['auth_token', 'api_key']:
             v = "*****"
         parsed.append((k, v))
 
     return '&'.join('{}{}'.format(
         k, '={}'.format(v) if v else '') for k, v in parsed)
 
 
 def raise_from_original(new_type):
     """
     Raise a new exception type with original args and traceback.
     """
     exc_type, exc_value, exc_traceback = sys.exc_info()
     new_exc = new_type(*exc_value.args)
     # store the original traceback into the new exc
     new_exc._org_exc_tb = traceback.format_exc(exc_traceback)
 
     try:
         raise new_exc.with_traceback(exc_traceback)
     finally:
         del exc_traceback
+
+
+class ArchiveNode(object):
+    def __init__(self, path, mode, is_link, raw_bytes):
+        self.path = path
+        self.mode = mode
+        self.is_link = is_link
+        self.raw_bytes = raw_bytes
+
+
+def archive_repo(walker, archive_dest_path, kind, mtime, archive_at_path,
+                 archive_dir_name, commit_id, write_metadata=True, extra_metadata=None):
+    """
+    walker should be a file walker, for example:
+        def walker():
+            for file_info in files:
+                yield ArchiveNode(fn, mode, is_link, ctx[fn].data)
+    """
+    extra_metadata = extra_metadata or {}
+
+    if kind == "tgz":
+        archiver = archival.tarit(archive_dest_path, mtime, "gz")
+    elif kind == "tbz2":
+        archiver = archival.tarit(archive_dest_path, mtime, "bz2")
+    elif kind == 'zip':
+        archiver = archival.zipit(archive_dest_path, mtime)
+    else:
+        raise exceptions.ArchiveException()(
+            'Remote does not support: "%s" archive type.' % kind)
+
+    for f in walker(commit_id, archive_at_path):
+        f_path = os.path.join(archive_dir_name, f.path.lstrip('/'))
+        try:
+            archiver.addfile(f_path, f.mode, f.is_link, f.raw_bytes())
+        except NoContentException:
+            # NOTE(marcink): this is a special case for SVN so we can create "empty"
+            # directories which arent supported by archiver
+            archiver.addfile(os.path.join(f_path, '.dir'), f.mode, f.is_link, '')
+
+    if write_metadata:
+        metadata = dict([
+            ('commit_id', commit_id),
+            ('mtime', mtime),
+        ])
+        metadata.update(extra_metadata)
+
+        meta = ["%s:%s" % (f_name, value) for f_name, value in metadata.items()]
+        f_path = os.path.join(archive_dir_name, '.archival.txt')
+        archiver.addfile(f_path, 0o644, False, '\n'.join(meta))
+
+    return archiver.done()

vcsserver/exceptions.py

@@ -1,121 +1,125 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 """
 Special exception handling over the wire.
 
 Since we cannot assume that our client is able to import our exception classes,
 this module provides a "wrapping" mechanism to raise plain exceptions
 which contain an extra attribute `_vcs_kind` to allow a client to distinguish
 different error conditions.
 """
 
 from pyramid.httpexceptions import HTTPLocked, HTTPForbidden
 
 
 def _make_exception(kind, org_exc, *args):
     """
     Prepares a base `Exception` instance to be sent over the wire.
 
     To give our caller a hint what this is about, it will attach an attribute
     `_vcs_kind` to the exception.
     """
     exc = Exception(*args)
     exc._vcs_kind = kind
     exc._org_exc = org_exc
     exc._org_exc_tb = getattr(org_exc, '_org_exc_tb', '')
     return exc
 
 
 def AbortException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('abort', org_exc, *args)
     return _make_exception_wrapper
 
 
 def ArchiveException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('archive', org_exc, *args)
     return _make_exception_wrapper
 
 
 def LookupException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('lookup', org_exc, *args)
     return _make_exception_wrapper
 
 
 def VcsException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('error', org_exc, *args)
     return _make_exception_wrapper
 
 
 def RepositoryLockedException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('repo_locked', org_exc, *args)
     return _make_exception_wrapper
 
 
 def RepositoryBranchProtectedException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('repo_branch_protected', org_exc, *args)
     return _make_exception_wrapper
 
 
 def RequirementException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('requirement', org_exc, *args)
     return _make_exception_wrapper
 
 
 def UnhandledException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('unhandled', org_exc, *args)
     return _make_exception_wrapper
 
 
 def URLError(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('url_error', org_exc, *args)
     return _make_exception_wrapper
 
 
 def SubrepoMergeException(org_exc=None):
     def _make_exception_wrapper(*args):
         return _make_exception('subrepo_merge_error', org_exc, *args)
     return _make_exception_wrapper
 
 
 class HTTPRepoLocked(HTTPLocked):
     """
     Subclass of HTTPLocked response that allows to set the title and status
     code via constructor arguments.
     """
     def __init__(self, title, status_code=None, **kwargs):
         self.code = status_code or HTTPLocked.code
         self.title = title
         super(HTTPRepoLocked, self).__init__(**kwargs)
 
 
 class HTTPRepoBranchProtected(HTTPForbidden):
     def __init__(self, *args, **kwargs):
         super(HTTPForbidden, self).__init__(*args, **kwargs)
 
 
 class RefNotFoundException(KeyError):
     pass
+
+
+class NoContentException(ValueError):
+    pass

vcsserver/git.py

@@ -1,1192 +1,1281 b''
-# RhodeCode VCSServer provides access to different vcs backends via network.
+ RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import collections
 import logging
 import os
 import posixpath as vcspath
 import re
 import stat
 import traceback
 import urllib.request, urllib.parse, urllib.error
 import urllib.request, urllib.error, urllib.parse
 from functools import wraps
 
 import more_itertools
 import pygit2
 from pygit2 import Repository as LibGit2Repo
+from pygit2 import index as LibGit2Index
 from dulwich import index, objects
 from dulwich.client import HttpGitClient, LocalGitClient
 from dulwich.errors import (
     NotGitRepository, ChecksumMismatch, WrongObjectException,
     MissingCommitError, ObjectMissing, HangupException,
     UnexpectedCommandError)
 from dulwich.repo import Repo as DulwichRepo
 from dulwich.server import update_server_info
 
 from vcsserver import exceptions, settings, subprocessio
 from vcsserver.utils import safe_str, safe_int, safe_unicode
-from vcsserver.base import RepoFactory, obfuscate_qs
+from vcsserver.base import RepoFactory, obfuscate_qs, ArchiveNode, archive_repo
 from vcsserver.hgcompat import (
     hg_url as url_parser, httpbasicauthhandler, httpdigestauthhandler)
 from vcsserver.git_lfs.lib import LFSOidStore
 from vcsserver.vcs_base import RemoteBase
 
 DIR_STAT = stat.S_IFDIR
 FILE_MODE = stat.S_IFMT
 GIT_LINK = objects.S_IFGITLINK
 PEELED_REF_MARKER = '^{}'
 
 
 log = logging.getLogger(__name__)
 
 
 def str_to_dulwich(value):
     """
     Dulwich 0.10.1a requires `unicode` objects to be passed in.
     """
     return value.decode(settings.WIRE_ENCODING)
 
 
 def reraise_safe_exceptions(func):
     """Converts Dulwich exceptions to something neutral."""
 
     @wraps(func)
     def wrapper(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except (ChecksumMismatch, WrongObjectException, MissingCommitError, ObjectMissing,) as e:
             exc = exceptions.LookupException(org_exc=e)
             raise exc(safe_str(e))
         except (HangupException, UnexpectedCommandError) as e:
             exc = exceptions.VcsException(org_exc=e)
             raise exc(safe_str(e))
         except Exception as e:
             # NOTE(marcink): becuase of how dulwich handles some exceptions
             # (KeyError on empty repos), we cannot track this and catch all
             # exceptions, it's an exceptions from other handlers
             #if not hasattr(e, '_vcs_kind'):
                 #log.exception("Unhandled exception in git remote call")
                 #raise_from_original(exceptions.UnhandledException)
             raise
     return wrapper
 
 
 class Repo(DulwichRepo):
     """
     A wrapper for dulwich Repo class.
 
     Since dulwich is sometimes keeping .idx file descriptors open, it leads to
     "Too many open files" error. We need to close all opened file descriptors
     once the repo object is destroyed.
     """
     def __del__(self):
         if hasattr(self, 'object_store'):
             self.close()
 
 
 class Repository(LibGit2Repo):
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         self.free()
 
 
 class GitFactory(RepoFactory):
     repo_type = 'git'
 
     def _create_repo(self, wire, create, use_libgit2=False):
         if use_libgit2:
             return Repository(wire['path'])
         else:
             repo_path = str_to_dulwich(wire['path'])
             return Repo(repo_path)
 
     def repo(self, wire, create=False, use_libgit2=False):
         """
         Get a repository instance for the given path.
         """
         return self._create_repo(wire, create, use_libgit2)
 
     def repo_libgit2(self, wire):
         return self.repo(wire, use_libgit2=True)
 
 
 class GitRemote(RemoteBase):
 
     def __init__(self, factory):
         self._factory = factory
         self._bulk_methods = {
             "date": self.date,
             "author": self.author,
             "branch": self.branch,
             "message": self.message,
             "parents": self.parents,
             "_commit": self.revision,
         }
 
     def _wire_to_config(self, wire):
         if 'config' in wire:
             return dict([(x[0] + '_' + x[1], x[2]) for x in wire['config']])
         return {}
 
     def _remote_conf(self, config):
         params = [
             '-c', 'core.askpass=""',
         ]
         ssl_cert_dir = config.get('vcs_ssl_dir')
         if ssl_cert_dir:
             params.extend(['-c', 'http.sslCAinfo={}'.format(ssl_cert_dir)])
         return params
 
     @reraise_safe_exceptions
     def discover_git_version(self):
         stdout, _ = self.run_git_command(
             {}, ['--version'], _bare=True, _safe=True)
         prefix = 'git version'
         if stdout.startswith(prefix):
             stdout = stdout[len(prefix):]
         return stdout.strip()
 
     @reraise_safe_exceptions
     def is_empty(self, wire):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
 
             try:
                 has_head = repo.head.name
                 if has_head:
                     return False
 
                 # NOTE(marcink): check again using more expensive method
                 return repo.is_empty
             except Exception:
                 pass
 
             return True
 
     @reraise_safe_exceptions
     def assert_correct_path(self, wire):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _assert_correct_path(_context_uid, _repo_id):
             try:
                 repo_init = self._factory.repo_libgit2(wire)
                 with repo_init as repo:
                     pass
             except pygit2.GitError:
                 path = wire.get('path')
                 tb = traceback.format_exc()
                 log.debug("Invalid Git path `%s`, tb: %s", path, tb)
                 return False
 
             return True
         return _assert_correct_path(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def bare(self, wire):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             return repo.is_bare
 
     @reraise_safe_exceptions
     def blob_as_pretty_string(self, wire, sha):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             blob_obj = repo[sha]
             blob = blob_obj.data
             return blob
 
     @reraise_safe_exceptions
     def blob_raw_length(self, wire, sha):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _blob_raw_length(_repo_id, _sha):
 
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 blob = repo[sha]
                 return blob.size
 
         return _blob_raw_length(repo_id, sha)
 
     def _parse_lfs_pointer(self, raw_content):
 
         spec_string = 'version https://git-lfs.github.com/spec'
         if raw_content and raw_content.startswith(spec_string):
             pattern = re.compile(r"""
             (?:\n)?
             ^version[ ]https://git-lfs\.github\.com/spec/(?P<spec_ver>v\d+)\n
             ^oid[ ] sha256:(?P<oid_hash>[0-9a-f]{64})\n
             ^size[ ](?P<oid_size>[0-9]+)\n
             (?:\n)?
             """, re.VERBOSE | re.MULTILINE)
             match = pattern.match(raw_content)
             if match:
                 return match.groupdict()
 
         return {}
 
     @reraise_safe_exceptions
     def is_large_file(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
 
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _is_large_file(_repo_id, _sha):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 blob = repo[commit_id]
                 if blob.is_binary:
                     return {}
 
                 return self._parse_lfs_pointer(blob.data)
 
         return _is_large_file(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def is_binary(self, wire, tree_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
 
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _is_binary(_repo_id, _tree_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 blob_obj = repo[tree_id]
                 return blob_obj.is_binary
 
         return _is_binary(repo_id, tree_id)
 
     @reraise_safe_exceptions
     def in_largefiles_store(self, wire, oid):
         conf = self._wire_to_config(wire)
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             repo_name = repo.path
 
         store_location = conf.get('vcs_git_lfs_store_location')
         if store_location:
 
             store = LFSOidStore(
                 oid=oid, repo=repo_name, store_location=store_location)
             return store.has_oid()
 
         return False
 
     @reraise_safe_exceptions
     def store_path(self, wire, oid):
         conf = self._wire_to_config(wire)
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             repo_name = repo.path
 
         store_location = conf.get('vcs_git_lfs_store_location')
         if store_location:
             store = LFSOidStore(
                 oid=oid, repo=repo_name, store_location=store_location)
             return store.oid_path
         raise ValueError('Unable to fetch oid with path {}'.format(oid))
 
     @reraise_safe_exceptions
     def bulk_request(self, wire, rev, pre_load):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _bulk_request(_repo_id, _rev, _pre_load):
             result = {}
             for attr in pre_load:
                 try:
                     method = self._bulk_methods[attr]
                     args = [wire, rev]
                     result[attr] = method(*args)
                 except KeyError as e:
                     raise exceptions.VcsException(e)(
                         "Unknown bulk attribute: %s" % attr)
             return result
 
         return _bulk_request(repo_id, rev, sorted(pre_load))
 
     def _build_opener(self, url):
         handlers = []
         url_obj = url_parser(url)
         _, authinfo = url_obj.authinfo()
 
         if authinfo:
             # create a password manager
             passmgr = urllib.request.HTTPPasswordMgrWithDefaultRealm()
             passmgr.add_password(*authinfo)
 
             handlers.extend((httpbasicauthhandler(passmgr),
                              httpdigestauthhandler(passmgr)))
 
         return urllib.request.build_opener(*handlers)
 
     def _type_id_to_name(self, type_id):
         return {
             1: b'commit',
             2: b'tree',
             3: b'blob',
             4: b'tag'
         }[type_id]
 
     @reraise_safe_exceptions
     def check_url(self, url, config):
         url_obj = url_parser(url)
         test_uri, _ = url_obj.authinfo()
         url_obj.passwd = '*****' if url_obj.passwd else url_obj.passwd
         url_obj.query = obfuscate_qs(url_obj.query)
         cleaned_uri = str(url_obj)
         log.info("Checking URL for remote cloning/import: %s", cleaned_uri)
 
         if not test_uri.endswith('info/refs'):
             test_uri = test_uri.rstrip('/') + '/info/refs'
 
         o = self._build_opener(url)
         o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git
 
         q = {"service": 'git-upload-pack'}
         qs = '?%s' % urllib.parse.urlencode(q)
         cu = "%s%s" % (test_uri, qs)
         req = urllib.request.Request(cu, None, {})
 
         try:
             log.debug("Trying to open URL %s", cleaned_uri)
             resp = o.open(req)
             if resp.code != 200:
                 raise exceptions.URLError()('Return Code is not 200')
         except Exception as e:
             log.warning("URL cannot be opened: %s", cleaned_uri, exc_info=True)
             # means it cannot be cloned
             raise exceptions.URLError(e)("[%s] org_exc: %s" % (cleaned_uri, e))
 
         # now detect if it's proper git repo
         gitdata = resp.read()
         if 'service=git-upload-pack' in gitdata:
             pass
         elif re.findall(r'[0-9a-fA-F]{40}\s+refs', gitdata):
             # old style git can return some other format !
             pass
         else:
             raise exceptions.URLError()(
                 "url [%s] does not look like an git" % (cleaned_uri,))
 
         return True
 
     @reraise_safe_exceptions
     def clone(self, wire, url, deferred, valid_refs, update_after_clone):
         # TODO(marcink): deprecate this method. Last i checked we don't use it anymore
         remote_refs = self.pull(wire, url, apply_refs=False)
         repo = self._factory.repo(wire)
         if isinstance(valid_refs, list):
             valid_refs = tuple(valid_refs)
 
         for k in remote_refs:
             # only parse heads/tags and skip so called deferred tags
             if k.startswith(valid_refs) and not k.endswith(deferred):
                 repo[k] = remote_refs[k]
 
         if update_after_clone:
             # we want to checkout HEAD
             repo["HEAD"] = remote_refs["HEAD"]
             index.build_index_from_tree(repo.path, repo.index_path(),
                                         repo.object_store, repo["HEAD"].tree)
 
     @reraise_safe_exceptions
     def branch(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _branch(_context_uid, _repo_id, _commit_id):
             regex = re.compile('^refs/heads')
 
             def filter_with(ref):
                 return regex.match(ref[0]) and ref[1] == _commit_id
 
             branches = list(filter(filter_with, list(self.get_refs(wire).items())))
             return [x[0].split('refs/heads/')[-1] for x in branches]
 
         return _branch(context_uid, repo_id, commit_id)
 
     @reraise_safe_exceptions
     def commit_branches(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _commit_branches(_context_uid, _repo_id, _commit_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 branches = [x for x in repo.branches.with_commit(_commit_id)]
                 return branches
 
         return _commit_branches(context_uid, repo_id, commit_id)
 
     @reraise_safe_exceptions
     def add_object(self, wire, content):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             blob = objects.Blob()
             blob.set_raw_string(content)
             repo.object_store.add_object(blob)
             return blob.id
 
     # TODO: this is quite complex, check if that can be simplified
     @reraise_safe_exceptions
     def commit(self, wire, commit_data, branch, commit_tree, updated, removed):
+        # Defines the root tree
+        class _Root(object):
+            def __repr__(self):
+                return 'ROOT TREE'
+        ROOT = _Root()
+
         repo = self._factory.repo(wire)
         object_store = repo.object_store
 
         # Create tree and populates it with blobs
-        commit_tree = commit_tree and repo[commit_tree] or objects.Tree()
+
+        if commit_tree and repo[commit_tree]:
+            git_commit = repo[commit_data['parents'][0]]
+            commit_tree = repo[git_commit.tree]  # root tree
+        else:
+            commit_tree = objects.Tree()
 
         for node in updated:
             # Compute subdirs if needed
             dirpath, nodename = vcspath.split(node['path'])
             dirnames = list(map(safe_str, dirpath and dirpath.split('/') or []))
             parent = commit_tree
             ancestors = [('', parent)]
 
             # Tries to dig for the deepest existing tree
             while dirnames:
                 curdir = dirnames.pop(0)
                 try:
                     dir_id = parent[curdir][1]
                 except KeyError:
                     # put curdir back into dirnames and stops
                     dirnames.insert(0, curdir)
                     break
                 else:
                     # If found, updates parent
                     parent = repo[dir_id]
                     ancestors.append((curdir, parent))
             # Now parent is deepest existing tree and we need to create
             # subtrees for dirnames (in reverse order)
             # [this only applies for nodes from added]
             new_trees = []
 
             blob = objects.Blob.from_string(node['content'])
 
             if dirnames:
                 # If there are trees which should be created we need to build
                 # them now (in reverse order)
                 reversed_dirnames = list(reversed(dirnames))
                 curtree = objects.Tree()
                 curtree[node['node_path']] = node['mode'], blob.id
                 new_trees.append(curtree)
                 for dirname in reversed_dirnames[:-1]:
                     newtree = objects.Tree()
                     newtree[dirname] = (DIR_STAT, curtree.id)
                     new_trees.append(newtree)
                     curtree = newtree
                 parent[reversed_dirnames[-1]] = (DIR_STAT, curtree.id)
             else:
                 parent.add(name=node['node_path'], mode=node['mode'], hexsha=blob.id)
 
             new_trees.append(parent)
             # Update ancestors
             reversed_ancestors = reversed(
                 [(a[1], b[1], b[0]) for a, b in zip(ancestors, ancestors[1:])])
             for parent, tree, path in reversed_ancestors:
                 parent[path] = (DIR_STAT, tree.id)
                 object_store.add_object(tree)
 
             object_store.add_object(blob)
             for tree in new_trees:
                 object_store.add_object(tree)
 
         for node_path in removed:
             paths = node_path.split('/')
-            tree = commit_tree
-            trees = [tree]
+            tree = commit_tree  # start with top-level
+            trees = [{'tree': tree, 'path': ROOT}]
             # Traverse deep into the forest...
+            # resolve final tree by iterating the path.
+            # e.g a/b/c.txt will get
+            # - root as tree then
+            # - 'a' as tree,
+            # - 'b' as tree,
+            # - stop at c as blob.
             for path in paths:
                 try:
                     obj = repo[tree[path][1]]
                     if isinstance(obj, objects.Tree):
-                        trees.append(obj)
+                        trees.append({'tree': obj, 'path': path})
                         tree = obj
                 except KeyError:
                     break
+            #PROBLEM:
+            """
+            We're not editing same reference tree object
+            """
             # Cut down the blob and all rotten trees on the way back...
-            for path, tree in reversed(list(zip(paths, trees))):
-                del tree[path]
-                if tree:
+            for path, tree_data in reversed(list(zip(paths, trees))):
+                tree = tree_data['tree']
+                tree.__delitem__(path)
+                # This operation edits the tree, we need to mark new commit back
+
+                if len(tree) > 0:
                     # This tree still has elements - don't remove it or any
                     # of it's parents
                     break
 
         object_store.add_object(commit_tree)
 
         # Create commit
         commit = objects.Commit()
         commit.tree = commit_tree.id
         for k, v in commit_data.items():
             setattr(commit, k, v)
         object_store.add_object(commit)
 
         self.create_branch(wire, branch, commit.id)
 
         # dulwich set-ref
         ref = 'refs/heads/%s' % branch
         repo.refs[ref] = commit.id
 
         return commit.id
 
     @reraise_safe_exceptions
     def pull(self, wire, url, apply_refs=True, refs=None, update_after=False):
         if url != 'default' and '://' not in url:
             client = LocalGitClient(url)
         else:
             url_obj = url_parser(url)
             o = self._build_opener(url)
             url, _ = url_obj.authinfo()
             client = HttpGitClient(base_url=url, opener=o)
         repo = self._factory.repo(wire)
 
         determine_wants = repo.object_store.determine_wants_all
         if refs:
             def determine_wants_requested(references):
                 return [references[r] for r in references if r in refs]
             determine_wants = determine_wants_requested
 
         try:
             remote_refs = client.fetch(
                 path=url, target=repo, determine_wants=determine_wants)
         except NotGitRepository as e:
             log.warning(
                 'Trying to fetch from "%s" failed, not a Git repository.', url)
             # Exception can contain unicode which we convert
             raise exceptions.AbortException(e)(repr(e))
 
         # mikhail: client.fetch() returns all the remote refs, but fetches only
         # refs filtered by `determine_wants` function. We need to filter result
         # as well
         if refs:
             remote_refs = {k: remote_refs[k] for k in remote_refs if k in refs}
 
         if apply_refs:
             # TODO: johbo: Needs proper test coverage with a git repository
             # that contains a tag object, so that we would end up with
             # a peeled ref at this point.
             for k in remote_refs:
                 if k.endswith(PEELED_REF_MARKER):
                     log.debug("Skipping peeled reference %s", k)
                     continue
                 repo[k] = remote_refs[k]
 
             if refs and not update_after:
                 # mikhail: explicitly set the head to the last ref.
-                repo['HEAD'] = remote_refs[refs[-1]]
+                repo["HEAD"] = remote_refs[refs[-1]]
 
         if update_after:
             # we want to checkout HEAD
             repo["HEAD"] = remote_refs["HEAD"]
             index.build_index_from_tree(repo.path, repo.index_path(),
                                         repo.object_store, repo["HEAD"].tree)
         return remote_refs
 
     @reraise_safe_exceptions
     def sync_fetch(self, wire, url, refs=None, all_refs=False):
         repo = self._factory.repo(wire)
         if refs and not isinstance(refs, (list, tuple)):
             refs = [refs]
 
         config = self._wire_to_config(wire)
         # get all remote refs we'll use to fetch later
         cmd = ['ls-remote']
         if not all_refs:
             cmd += ['--heads', '--tags']
         cmd += [url]
         output, __ = self.run_git_command(
             wire, cmd, fail_on_stderr=False,
             _copts=self._remote_conf(config),
             extra_env={'GIT_TERMINAL_PROMPT': '0'})
 
         remote_refs = collections.OrderedDict()
         fetch_refs = []
 
         for ref_line in output.splitlines():
             sha, ref = ref_line.split('\t')
             sha = sha.strip()
             if ref in remote_refs:
                 # duplicate, skip
                 continue
             if ref.endswith(PEELED_REF_MARKER):
                 log.debug("Skipping peeled reference %s", ref)
                 continue
             # don't sync HEAD
             if ref in ['HEAD']:
                 continue
 
             remote_refs[ref] = sha
 
             if refs and sha in refs:
                 # we filter fetch using our specified refs
                 fetch_refs.append('{}:{}'.format(ref, ref))
             elif not refs:
                 fetch_refs.append('{}:{}'.format(ref, ref))
         log.debug('Finished obtaining fetch refs, total: %s', len(fetch_refs))
 
         if fetch_refs:
             for chunk in more_itertools.chunked(fetch_refs, 1024 * 4):
                 fetch_refs_chunks = list(chunk)
                 log.debug('Fetching %s refs from import url', len(fetch_refs_chunks))
                 _out, _err = self.run_git_command(
                     wire, ['fetch', url, '--force', '--prune', '--'] + fetch_refs_chunks,
                     fail_on_stderr=False,
                     _copts=self._remote_conf(config),
                     extra_env={'GIT_TERMINAL_PROMPT': '0'})
 
         return remote_refs
 
     @reraise_safe_exceptions
     def sync_push(self, wire, url, refs=None):
         if not self.check_url(url, wire):
             return
         config = self._wire_to_config(wire)
         self._factory.repo(wire)
         self.run_git_command(
             wire, ['push', url, '--mirror'], fail_on_stderr=False,
             _copts=self._remote_conf(config),
             extra_env={'GIT_TERMINAL_PROMPT': '0'})
 
     @reraise_safe_exceptions
     def get_remote_refs(self, wire, url):
         repo = Repo(url)
         return repo.get_refs()
 
     @reraise_safe_exceptions
     def get_description(self, wire):
         repo = self._factory.repo(wire)
         return repo.get_description()
 
     @reraise_safe_exceptions
     def get_missing_revs(self, wire, rev1, rev2, path2):
         repo = self._factory.repo(wire)
         LocalGitClient(thin_packs=False).fetch(path2, repo)
 
         wire_remote = wire.copy()
         wire_remote['path'] = path2
         repo_remote = self._factory.repo(wire_remote)
         LocalGitClient(thin_packs=False).fetch(wire["path"], repo_remote)
 
         revs = [
             x.commit.id
             for x in repo_remote.get_walker(include=[rev2], exclude=[rev1])]
         return revs
 
     @reraise_safe_exceptions
     def get_object(self, wire, sha, maybe_unreachable=False):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_object(_context_uid, _repo_id, _sha):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
 
                 missing_commit_err = 'Commit {} does not exist for `{}`'.format(sha, wire['path'])
                 try:
                     commit = repo.revparse_single(sha)
                 except KeyError:
                     # NOTE(marcink): KeyError doesn't give us any meaningful information
                     # here, we instead give something more explicit
                     e = exceptions.RefNotFoundException('SHA: %s not found', sha)
                     raise exceptions.LookupException(e)(missing_commit_err)
                 except ValueError as e:
                     raise exceptions.LookupException(e)(missing_commit_err)
 
                 is_tag = False
                 if isinstance(commit, pygit2.Tag):
                     commit = repo.get(commit.target)
                     is_tag = True
 
                 check_dangling = True
                 if is_tag:
                     check_dangling = False
 
                 if check_dangling and maybe_unreachable:
                     check_dangling = False
 
                 # we used a reference and it parsed means we're not having a dangling commit
                 if sha != commit.hex:
                     check_dangling = False
 
                 if check_dangling:
                     # check for dangling commit
                     for branch in repo.branches.with_commit(commit.hex):
                         if branch:
                             break
                     else:
                         # NOTE(marcink): Empty error doesn't give us any meaningful information
                         # here, we instead give something more explicit
                         e = exceptions.RefNotFoundException('SHA: %s not found in branches', sha)
                         raise exceptions.LookupException(e)(missing_commit_err)
 
                 commit_id = commit.hex
                 type_id = commit.type_str
 
                 return {
                     'id': commit_id,
                     'type': self._type_id_to_name(type_id),
                     'commit_id': commit_id,
                     'idx': 0
                 }
 
         return _get_object(context_uid, repo_id, sha)
 
     @reraise_safe_exceptions
     def get_refs(self, wire):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_refs(_context_uid, _repo_id):
 
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 regex = re.compile('^refs/(heads|tags)/')
                 return {x.name: x.target.hex for x in
                         [ref for ref in repo.listall_reference_objects() if regex.match(ref.name)]}
 
         return _get_refs(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def get_branch_pointers(self, wire):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_branch_pointers(_context_uid, _repo_id):
 
             repo_init = self._factory.repo_libgit2(wire)
             regex = re.compile('^refs/heads')
             with repo_init as repo:
                 branches = [ref for ref in repo.listall_reference_objects() if regex.match(ref.name)]
                 return {x.target.hex: x.shorthand for x in branches}
 
         return _get_branch_pointers(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def head(self, wire, show_exc=True):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _head(_context_uid, _repo_id, _show_exc):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 try:
                     return repo.head.peel().hex
                 except Exception:
                     if show_exc:
                         raise
         return _head(context_uid, repo_id, show_exc)
 
     @reraise_safe_exceptions
     def init(self, wire):
         repo_path = str_to_dulwich(wire['path'])
         self.repo = Repo.init(repo_path)
 
     @reraise_safe_exceptions
     def init_bare(self, wire):
         repo_path = str_to_dulwich(wire['path'])
         self.repo = Repo.init_bare(repo_path)
 
     @reraise_safe_exceptions
     def revision(self, wire, rev):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _revision(_context_uid, _repo_id, _rev):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 commit = repo[rev]
                 obj_data = {
                     'id': commit.id.hex,
                 }
                 # tree objects itself don't have tree_id attribute
                 if hasattr(commit, 'tree_id'):
                     obj_data['tree'] = commit.tree_id.hex
 
                 return obj_data
         return _revision(context_uid, repo_id, rev)
 
     @reraise_safe_exceptions
     def date(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _date(_repo_id, _commit_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 commit = repo[commit_id]
 
                 if hasattr(commit, 'commit_time'):
                     commit_time, commit_time_offset = commit.commit_time, commit.commit_time_offset
                 else:
                     commit = commit.get_object()
                     commit_time, commit_time_offset = commit.commit_time, commit.commit_time_offset
 
                 # TODO(marcink): check dulwich difference of offset vs timezone
                 return [commit_time, commit_time_offset]
         return _date(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def author(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _author(_repo_id, _commit_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 commit = repo[commit_id]
 
                 if hasattr(commit, 'author'):
                     author = commit.author
                 else:
                     author = commit.get_object().author
 
                 if author.email:
                     return "{} <{}>".format(author.name, author.email)
 
                 try:
                     return "{}".format(author.name)
                 except Exception:
                     return "{}".format(safe_unicode(author.raw_name))
 
         return _author(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def message(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _message(_repo_id, _commit_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 commit = repo[commit_id]
                 return commit.message
         return _message(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def parents(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _parents(_repo_id, _commit_id):
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 commit = repo[commit_id]
                 if hasattr(commit, 'parent_ids'):
                     parent_ids = commit.parent_ids
                 else:
                     parent_ids = commit.get_object().parent_ids
 
                 return [x.hex for x in parent_ids]
         return _parents(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def children(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _children(_repo_id, _commit_id):
             output, __ = self.run_git_command(
                 wire, ['rev-list', '--all', '--children'])
 
             child_ids = []
             pat = re.compile(r'^%s' % commit_id)
             for l in output.splitlines():
                 if pat.match(l):
                     found_ids = l.split(' ')[1:]
                     child_ids.extend(found_ids)
 
             return child_ids
         return _children(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def set_refs(self, wire, key, value):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             repo.references.create(key, value, force=True)
 
     @reraise_safe_exceptions
     def create_branch(self, wire, branch_name, commit_id, force=False):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             commit = repo[commit_id]
 
             if force:
                 repo.branches.local.create(branch_name, commit, force=force)
             elif not repo.branches.get(branch_name):
                 # create only if that branch isn't existing
                 repo.branches.local.create(branch_name, commit, force=force)
 
     @reraise_safe_exceptions
     def remove_ref(self, wire, key):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             repo.references.delete(key)
 
     @reraise_safe_exceptions
     def tag_remove(self, wire, tag_name):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             key = 'refs/tags/{}'.format(tag_name)
             repo.references.delete(key)
 
     @reraise_safe_exceptions
     def tree_changes(self, wire, source_id, target_id):
         # TODO(marcink): remove this seems it's only used by tests
         repo = self._factory.repo(wire)
         source = repo[source_id].tree if source_id else None
         target = repo[target_id].tree
         result = repo.object_store.tree_changes(source, target)
         return list(result)
 
     @reraise_safe_exceptions
     def tree_and_type_for_path(self, wire, commit_id, path):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _tree_and_type_for_path(_context_uid, _repo_id, _commit_id, _path):
             repo_init = self._factory.repo_libgit2(wire)
 
             with repo_init as repo:
                 commit = repo[commit_id]
                 try:
                     tree = commit.tree[path]
                 except KeyError:
                     return None, None, None
 
                 return tree.id.hex, tree.type_str, tree.filemode
         return _tree_and_type_for_path(context_uid, repo_id, commit_id, path)
 
     @reraise_safe_exceptions
     def tree_items(self, wire, tree_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _tree_items(_repo_id, _tree_id):
 
             repo_init = self._factory.repo_libgit2(wire)
             with repo_init as repo:
                 try:
                     tree = repo[tree_id]
                 except KeyError:
                     raise ObjectMissing('No tree with id: {}'.format(tree_id))
 
                 result = []
                 for item in tree:
                     item_sha = item.hex
                     item_mode = item.filemode
                     item_type = item.type_str
 
                     if item_type == 'commit':
                         # NOTE(marcink): submodules we translate to 'link' for backward compat
                         item_type = 'link'
 
                     result.append((item.name, item_mode, item_sha, item_type))
                 return result
         return _tree_items(repo_id, tree_id)
 
     @reraise_safe_exceptions
     def diff_2(self, wire, commit_id_1, commit_id_2, file_filter, opt_ignorews, context):
         """
         Old version that uses subprocess to call diff
         """
 
         flags = [
             '-U%s' % context, '--patch',
             '--binary',
             '--find-renames',
             '--no-indent-heuristic',
             # '--indent-heuristic',
             #'--full-index',
             #'--abbrev=40'
         ]
 
         if opt_ignorews:
             flags.append('--ignore-all-space')
 
         if commit_id_1 == self.EMPTY_COMMIT:
             cmd = ['show'] + flags + [commit_id_2]
         else:
             cmd = ['diff'] + flags + [commit_id_1, commit_id_2]
 
         if file_filter:
             cmd.extend(['--', file_filter])
 
         diff, __ = self.run_git_command(wire, cmd)
         # If we used 'show' command, strip first few lines (until actual diff
         # starts)
         if commit_id_1 == self.EMPTY_COMMIT:
             lines = diff.splitlines()
             x = 0
             for line in lines:
                 if line.startswith('diff'):
                     break
                 x += 1
             # Append new line just like 'diff' command do
             diff = '\n'.join(lines[x:]) + '\n'
         return diff
 
     @reraise_safe_exceptions
     def diff(self, wire, commit_id_1, commit_id_2, file_filter, opt_ignorews, context):
         repo_init = self._factory.repo_libgit2(wire)
         with repo_init as repo:
             swap = True
             flags = 0
             flags |= pygit2.GIT_DIFF_SHOW_BINARY
 
             if opt_ignorews:
                 flags |= pygit2.GIT_DIFF_IGNORE_WHITESPACE
 
             if commit_id_1 == self.EMPTY_COMMIT:
                 comm1 = repo[commit_id_2]
                 diff_obj = comm1.tree.diff_to_tree(
                     flags=flags, context_lines=context, swap=swap)
 
             else:
                 comm1 = repo[commit_id_2]
                 comm2 = repo[commit_id_1]
                 diff_obj = comm1.tree.diff_to_tree(
                     comm2.tree, flags=flags, context_lines=context, swap=swap)
             similar_flags = 0
             similar_flags |= pygit2.GIT_DIFF_FIND_RENAMES
             diff_obj.find_similar(flags=similar_flags)
 
             if file_filter:
                 for p in diff_obj:
                     if p.delta.old_file.path == file_filter:
                         return p.patch or ''
                 # fo matching path == no diff
                 return ''
             return diff_obj.patch or ''
 
     @reraise_safe_exceptions
     def node_history(self, wire, commit_id, path, limit):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _node_history(_context_uid, _repo_id, _commit_id, _path, _limit):
             # optimize for n==1, rev-list is much faster for that use-case
             if limit == 1:
                 cmd = ['rev-list', '-1', commit_id, '--', path]
             else:
                 cmd = ['log']
                 if limit:
                     cmd.extend(['-n', str(safe_int(limit, 0))])
                 cmd.extend(['--pretty=format: %H', '-s', commit_id, '--', path])
 
             output, __ = self.run_git_command(wire, cmd)
             commit_ids = re.findall(r'[0-9a-fA-F]{40}', output)
 
             return [x for x in commit_ids]
         return _node_history(context_uid, repo_id, commit_id, path, limit)
 
     @reraise_safe_exceptions
     def node_annotate(self, wire, commit_id, path):
 
         cmd = ['blame', '-l', '--root', '-r', commit_id, '--', path]
         # -l     ==> outputs long shas (and we need all 40 characters)
         # --root ==> doesn't put '^' character for boundaries
         # -r commit_id ==> blames for the given commit
         output, __ = self.run_git_command(wire, cmd)
 
         result = []
         for i, blame_line in enumerate(output.split('\n')[:-1]):
             line_no = i + 1
             commit_id, line = re.split(r' ', blame_line, 1)
             result.append((line_no, commit_id, line))
         return result
 
     @reraise_safe_exceptions
     def update_server_info(self, wire):
         repo = self._factory.repo(wire)
         update_server_info(repo)
 
     @reraise_safe_exceptions
     def get_all_commit_ids(self, wire):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_all_commit_ids(_context_uid, _repo_id):
 
             cmd = ['rev-list', '--reverse', '--date-order', '--branches', '--tags']
             try:
                 output, __ = self.run_git_command(wire, cmd)
                 return output.splitlines()
             except Exception:
                 # Can be raised for empty repositories
                 return []
         return _get_all_commit_ids(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def run_git_command(self, wire, cmd, **opts):
         path = wire.get('path', None)
 
         if path and os.path.isdir(path):
             opts['cwd'] = path
 
         if '_bare' in opts:
             _copts = []
             del opts['_bare']
         else:
             _copts = ['-c', 'core.quotepath=false', ]
         safe_call = False
         if '_safe' in opts:
             # no exc on failure
             del opts['_safe']
             safe_call = True
 
         if '_copts' in opts:
             _copts.extend(opts['_copts'] or [])
             del opts['_copts']
 
         gitenv = os.environ.copy()
         gitenv.update(opts.pop('extra_env', {}))
         # need to clean fix GIT_DIR !
         if 'GIT_DIR' in gitenv:
             del gitenv['GIT_DIR']
         gitenv['GIT_CONFIG_NOGLOBAL'] = '1'
         gitenv['GIT_DISCOVERY_ACROSS_FILESYSTEM'] = '1'
 
         cmd = [settings.GIT_EXECUTABLE] + _copts + cmd
         _opts = {'env': gitenv, 'shell': False}
 
         proc = None
         try:
             _opts.update(opts)
             proc = subprocessio.SubprocessIOChunker(cmd, **_opts)
 
             return ''.join(proc), ''.join(proc.error)
         except (EnvironmentError, OSError) as err:
             cmd = ' '.join(cmd)  # human friendly CMD
             tb_err = ("Couldn't run git command (%s).\n"
                       "Original error was:%s\n"
                       "Call options:%s\n"
                       % (cmd, err, _opts))
             log.exception(tb_err)
             if safe_call:
                 return '', err
             else:
                 raise exceptions.VcsException()(tb_err)
         finally:
             if proc:
                 proc.close()
 
     @reraise_safe_exceptions
     def install_hooks(self, wire, force=False):
         from vcsserver.hook_utils import install_git_hooks
         bare = self.bare(wire)
         path = wire['path']
         return install_git_hooks(path, bare, force_create=force)
 
     @reraise_safe_exceptions
     def get_hooks_info(self, wire):
         from vcsserver.hook_utils import (
             get_git_pre_hook_version, get_git_post_hook_version)
         bare = self.bare(wire)
         path = wire['path']
         return {
             'pre_version': get_git_pre_hook_version(path, bare),
             'post_version': get_git_post_hook_version(path, bare),
         }
+
+    @reraise_safe_exceptions
+    def set_head_ref(self, wire, head_name):
+        log.debug('Setting refs/head to `%s`', head_name)
+        cmd = ['symbolic-ref', '"HEAD"', '"refs/heads/%s"' % head_name]
+        output, __ = self.run_git_command(wire, cmd)
+        return [head_name] + output.splitlines()
+
+    @reraise_safe_exceptions
+    def archive_repo(self, wire, archive_dest_path, kind, mtime, archive_at_path,
+                     archive_dir_name, commit_id):
+
+        def file_walker(_commit_id, path):
+            repo_init = self._factory.repo_libgit2(wire)
+
+            with repo_init as repo:
+                commit = repo[commit_id]
+
+                if path in ['', '/']:
+                    tree = commit.tree
+                else:
+                    tree = commit.tree[path.rstrip('/')]
+                    tree_id = tree.id.hex
+                    try:
+                        tree = repo[tree_id]
+                    except KeyError:
+                        raise ObjectMissing('No tree with id: {}'.format(tree_id))
+
+                index = LibGit2Index.Index()
+                index.read_tree(tree)
+                file_iter = index
+
+                for fn in file_iter:
+                    file_path = fn.path
+                    mode = fn.mode
+                    is_link = stat.S_ISLNK(mode)
+                    if mode == pygit2.GIT_FILEMODE_COMMIT:
+                        log.debug('Skipping path %s as a commit node', file_path)
+                        continue
+                    yield ArchiveNode(file_path, mode, is_link, repo[fn.hex].read_raw)
+
+        return archive_repo(file_walker, archive_dest_path, kind, mtime, archive_at_path,
+                            archive_dir_name, commit_id)

vcsserver/hg.py

@@ -1,1009 +1,1047 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
-
+import functools
 import io
 import logging
+import os
 import stat
 import urllib.request, urllib.parse, urllib.error
 import urllib.request, urllib.error, urllib.parse
 import traceback
 
 from hgext import largefiles, rebase, purge
 from hgext.strip import strip as hgext_strip
 from mercurial import commands
 from mercurial import unionrepo
 from mercurial import verify
 from mercurial import repair
 
 import vcsserver
 from vcsserver import exceptions
-from vcsserver.base import RepoFactory, obfuscate_qs, raise_from_original
+from vcsserver.base import RepoFactory, obfuscate_qs, raise_from_original, archive_repo, ArchiveNode
 from vcsserver.hgcompat import (
     archival, bin, clone, config as hgconfig, diffopts, hex, get_ctx,
     hg_url as url_parser, httpbasicauthhandler, httpdigestauthhandler,
     makepeer, instance, match, memctx, exchange, memfilectx, nullrev, hg_merge,
     patch, peer, revrange, ui, hg_tag, Abort, LookupError, RepoError,
-    RepoLookupError, InterventionRequired, RequirementError)
+    RepoLookupError, InterventionRequired, RequirementError,
+    alwaysmatcher, patternmatcher, hgutil)
 from vcsserver.vcs_base import RemoteBase
 
 log = logging.getLogger(__name__)
 
 
 def make_ui_from_config(repo_config):
 
     class LoggingUI(ui.ui):
         def status(self, *msg, **opts):
             log.info(' '.join(msg).rstrip('\n'))
             super(LoggingUI, self).status(*msg, **opts)
 
         def warn(self, *msg, **opts):
             log.warn(' '.join(msg).rstrip('\n'))
             super(LoggingUI, self).warn(*msg, **opts)
 
         def error(self, *msg, **opts):
             log.error(' '.join(msg).rstrip('\n'))
             super(LoggingUI, self).error(*msg, **opts)
 
         def note(self, *msg, **opts):
             log.info(' '.join(msg).rstrip('\n'))
             super(LoggingUI, self).note(*msg, **opts)
 
         def debug(self, *msg, **opts):
             log.debug(' '.join(msg).rstrip('\n'))
             super(LoggingUI, self).debug(*msg, **opts)
 
     baseui = LoggingUI()
 
     # clean the baseui object
     baseui._ocfg = hgconfig.config()
     baseui._ucfg = hgconfig.config()
     baseui._tcfg = hgconfig.config()
 
     for section, option, value in repo_config:
         baseui.setconfig(section, option, value)
 
     # make our hgweb quiet so it doesn't print output
     baseui.setconfig('ui', 'quiet', 'true')
 
     baseui.setconfig('ui', 'paginate', 'never')
     # for better Error reporting of Mercurial
     baseui.setconfig('ui', 'message-output', 'stderr')
 
     # force mercurial to only use 1 thread, otherwise it may try to set a
     # signal in a non-main thread, thus generating a ValueError.
     baseui.setconfig('worker', 'numcpus', 1)
 
     # If there is no config for the largefiles extension, we explicitly disable
     # it here. This overrides settings from repositories hgrc file. Recent
     # mercurial versions enable largefiles in hgrc on clone from largefile
     # repo.
     if not baseui.hasconfig('extensions', 'largefiles'):
         log.debug('Explicitly disable largefiles extension for repo.')
         baseui.setconfig('extensions', 'largefiles', '!')
 
     return baseui
 
 
 def reraise_safe_exceptions(func):
     """Decorator for converting mercurial exceptions to something neutral."""
 
     def wrapper(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except (Abort, InterventionRequired) as e:
             raise_from_original(exceptions.AbortException(e))
         except RepoLookupError as e:
             raise_from_original(exceptions.LookupException(e))
         except RequirementError as e:
             raise_from_original(exceptions.RequirementException(e))
         except RepoError as e:
             raise_from_original(exceptions.VcsException(e))
         except LookupError as e:
             raise_from_original(exceptions.LookupException(e))
         except Exception as e:
             if not hasattr(e, '_vcs_kind'):
                 log.exception("Unhandled exception in hg remote call")
                 raise_from_original(exceptions.UnhandledException(e))
 
             raise
     return wrapper
 
 
 class MercurialFactory(RepoFactory):
     repo_type = 'hg'
 
     def _create_config(self, config, hooks=True):
         if not hooks:
             hooks_to_clean = frozenset((
                 'changegroup.repo_size', 'preoutgoing.pre_pull',
                 'outgoing.pull_logger', 'prechangegroup.pre_push'))
             new_config = []
             for section, option, value in config:
                 if section == 'hooks' and option in hooks_to_clean:
                     continue
                 new_config.append((section, option, value))
             config = new_config
 
         baseui = make_ui_from_config(config)
         return baseui
 
     def _create_repo(self, wire, create):
         baseui = self._create_config(wire["config"])
         return instance(baseui, wire["path"], create)
 
     def repo(self, wire, create=False):
         """
         Get a repository instance for the given path.
         """
         return self._create_repo(wire, create)
 
 
 def patch_ui_message_output(baseui):
     baseui.setconfig('ui', 'quiet', 'false')
     output = io.BytesIO()
 
     def write(data, **unused_kwargs):
         output.write(data)
 
     baseui.status = write
     baseui.write = write
     baseui.warn = write
     baseui.debug = write
 
     return baseui, output
 
 
 class HgRemote(RemoteBase):
 
     def __init__(self, factory):
         self._factory = factory
         self._bulk_methods = {
             "affected_files": self.ctx_files,
             "author": self.ctx_user,
             "branch": self.ctx_branch,
             "children": self.ctx_children,
             "date": self.ctx_date,
             "message": self.ctx_description,
             "parents": self.ctx_parents,
             "status": self.ctx_status,
             "obsolete": self.ctx_obsolete,
             "phase": self.ctx_phase,
             "hidden": self.ctx_hidden,
             "_file_paths": self.ctx_list,
         }
 
     def _get_ctx(self, repo, ref):
         return get_ctx(repo, ref)
 
     @reraise_safe_exceptions
     def discover_hg_version(self):
         from mercurial import util
         return util.version()
 
     @reraise_safe_exceptions
     def is_empty(self, wire):
         repo = self._factory.repo(wire)
 
         try:
             return len(repo) == 0
         except Exception:
             log.exception("failed to read object_store")
             return False
 
     @reraise_safe_exceptions
-    def archive_repo(self, archive_path, mtime, file_info, kind):
-        if kind == "tgz":
-            archiver = archival.tarit(archive_path, mtime, "gz")
-        elif kind == "tbz2":
-            archiver = archival.tarit(archive_path, mtime, "bz2")
-        elif kind == 'zip':
-            archiver = archival.zipit(archive_path, mtime)
-        else:
-            raise exceptions.ArchiveException()(
-                'Remote does not support: "%s".' % kind)
-
-        for f_path, f_mode, f_is_link, f_content in file_info:
-            archiver.addfile(f_path, f_mode, f_is_link, f_content)
-        archiver.done()
-
-    @reraise_safe_exceptions
     def bookmarks(self, wire):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _bookmarks(_context_uid, _repo_id):
             repo = self._factory.repo(wire)
             return dict(repo._bookmarks)
 
         return _bookmarks(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def branches(self, wire, normal, closed):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _branches(_context_uid, _repo_id, _normal, _closed):
             repo = self._factory.repo(wire)
             iter_branches = repo.branchmap().iterbranches()
             bt = {}
             for branch_name, _heads, tip, is_closed in iter_branches:
                 if normal and not is_closed:
                     bt[branch_name] = tip
                 if closed and is_closed:
                     bt[branch_name] = tip
 
             return bt
 
         return _branches(context_uid, repo_id, normal, closed)
 
     @reraise_safe_exceptions
     def bulk_request(self, wire, commit_id, pre_load):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _bulk_request(_repo_id, _commit_id, _pre_load):
             result = {}
             for attr in pre_load:
                 try:
                     method = self._bulk_methods[attr]
                     result[attr] = method(wire, commit_id)
                 except KeyError as e:
                     raise exceptions.VcsException(e)(
                         'Unknown bulk attribute: "%s"' % attr)
             return result
 
         return _bulk_request(repo_id, commit_id, sorted(pre_load))
 
     @reraise_safe_exceptions
     def ctx_branch(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_branch(_repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return ctx.branch()
         return _ctx_branch(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_date(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_date(_repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return ctx.date()
         return _ctx_date(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_description(self, wire, revision):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         return ctx.description()
 
     @reraise_safe_exceptions
     def ctx_files(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_files(_repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return ctx.files()
 
         return _ctx_files(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_list(self, path, revision):
         repo = self._factory.repo(path)
         ctx = self._get_ctx(repo, revision)
         return list(ctx)
 
     @reraise_safe_exceptions
     def ctx_parents(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_parents(_repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return [parent.hex() for parent in ctx.parents()
                     if not (parent.hidden() or parent.obsolete())]
 
         return _ctx_parents(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_children(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_children(_repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return [child.hex() for child in ctx.children()
                     if not (child.hidden() or child.obsolete())]
 
         return _ctx_children(repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_phase(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_phase(_context_uid, _repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             # public=0, draft=1, secret=3
             return ctx.phase()
         return _ctx_phase(context_uid, repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_obsolete(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_obsolete(_context_uid, _repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return ctx.obsolete()
         return _ctx_obsolete(context_uid, repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_hidden(self, wire, commit_id):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _ctx_hidden(_context_uid, _repo_id, _commit_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             return ctx.hidden()
         return _ctx_hidden(context_uid, repo_id, commit_id)
 
     @reraise_safe_exceptions
     def ctx_substate(self, wire, revision):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         return ctx.substate
 
     @reraise_safe_exceptions
     def ctx_status(self, wire, revision):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         status = repo[ctx.p1().node()].status(other=ctx.node())
         # object of status (odd, custom named tuple in mercurial) is not
         # correctly serializable, we make it a list, as the underling
         # API expects this to be a list
         return list(status)
 
     @reraise_safe_exceptions
     def ctx_user(self, wire, revision):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         return ctx.user()
 
     @reraise_safe_exceptions
     def check_url(self, url, config):
         _proto = None
         if '+' in url[:url.find('://')]:
             _proto = url[0:url.find('+')]
             url = url[url.find('+') + 1:]
         handlers = []
         url_obj = url_parser(url)
         test_uri, authinfo = url_obj.authinfo()
         url_obj.passwd = '*****' if url_obj.passwd else url_obj.passwd
         url_obj.query = obfuscate_qs(url_obj.query)
 
         cleaned_uri = str(url_obj)
         log.info("Checking URL for remote cloning/import: %s", cleaned_uri)
 
         if authinfo:
             # create a password manager
             passmgr = urllib.request.HTTPPasswordMgrWithDefaultRealm()
             passmgr.add_password(*authinfo)
 
             handlers.extend((httpbasicauthhandler(passmgr),
                              httpdigestauthhandler(passmgr)))
 
         o = urllib.request.build_opener(*handlers)
         o.addheaders = [('Content-Type', 'application/mercurial-0.1'),
                         ('Accept', 'application/mercurial-0.1')]
 
         q = {"cmd": 'between'}
         q.update({'pairs': "%s-%s" % ('0' * 40, '0' * 40)})
         qs = '?%s' % urllib.parse.urlencode(q)
         cu = "%s%s" % (test_uri, qs)
         req = urllib.request.Request(cu, None, {})
 
         try:
             log.debug("Trying to open URL %s", cleaned_uri)
             resp = o.open(req)
             if resp.code != 200:
                 raise exceptions.URLError()('Return Code is not 200')
         except Exception as e:
             log.warning("URL cannot be opened: %s", cleaned_uri, exc_info=True)
             # means it cannot be cloned
             raise exceptions.URLError(e)("[%s] org_exc: %s" % (cleaned_uri, e))
 
         # now check if it's a proper hg repo, but don't do it for svn
         try:
             if _proto == 'svn':
                 pass
             else:
                 # check for pure hg repos
                 log.debug(
                     "Verifying if URL is a Mercurial repository: %s",
                     cleaned_uri)
                 ui = make_ui_from_config(config)
                 peer_checker = makepeer(ui, url)
                 peer_checker.lookup('tip')
         except Exception as e:
             log.warning("URL is not a valid Mercurial repository: %s",
                         cleaned_uri)
             raise exceptions.URLError(e)(
                 "url [%s] does not look like an hg repo org_exc: %s"
                 % (cleaned_uri, e))
 
         log.info("URL is a valid Mercurial repository: %s", cleaned_uri)
         return True
 
     @reraise_safe_exceptions
     def diff(self, wire, commit_id_1, commit_id_2, file_filter, opt_git, opt_ignorews, context):
         repo = self._factory.repo(wire)
 
         if file_filter:
             match_filter = match(file_filter[0], '', [file_filter[1]])
         else:
             match_filter = file_filter
         opts = diffopts(git=opt_git, ignorews=opt_ignorews, context=context, showfunc=1)
 
         try:
             return "".join(patch.diff(
                 repo, node1=commit_id_1, node2=commit_id_2, match=match_filter, opts=opts))
         except RepoLookupError as e:
             raise exceptions.LookupException(e)()
 
     @reraise_safe_exceptions
     def node_history(self, wire, revision, path, limit):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _node_history(_context_uid, _repo_id, _revision, _path, _limit):
             repo = self._factory.repo(wire)
 
             ctx = self._get_ctx(repo, revision)
             fctx = ctx.filectx(path)
 
             def history_iter():
                 limit_rev = fctx.rev()
                 for obj in reversed(list(fctx.filelog())):
                     obj = fctx.filectx(obj)
                     ctx = obj.changectx()
                     if ctx.hidden() or ctx.obsolete():
                         continue
 
                     if limit_rev >= obj.rev():
                         yield obj
 
             history = []
             for cnt, obj in enumerate(history_iter()):
                 if limit and cnt >= limit:
                     break
                 history.append(hex(obj.node()))
 
             return [x for x in history]
         return _node_history(context_uid, repo_id, revision, path, limit)
 
     @reraise_safe_exceptions
     def node_history_untill(self, wire, revision, path, limit):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _node_history_until(_context_uid, _repo_id):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, revision)
             fctx = ctx.filectx(path)
 
             file_log = list(fctx.filelog())
             if limit:
                 # Limit to the last n items
                 file_log = file_log[-limit:]
 
             return [hex(fctx.filectx(cs).node()) for cs in reversed(file_log)]
         return _node_history_until(context_uid, repo_id, revision, path, limit)
 
     @reraise_safe_exceptions
     def fctx_annotate(self, wire, revision, path):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         fctx = ctx.filectx(path)
 
         result = []
         for i, annotate_obj in enumerate(fctx.annotate(), 1):
             ln_no = i
             sha = hex(annotate_obj.fctx.node())
             content = annotate_obj.text
             result.append((ln_no, sha, content))
         return result
 
     @reraise_safe_exceptions
     def fctx_node_data(self, wire, revision, path):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         fctx = ctx.filectx(path)
         return fctx.data()
 
     @reraise_safe_exceptions
     def fctx_flags(self, wire, commit_id, path):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _fctx_flags(_repo_id, _commit_id, _path):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             fctx = ctx.filectx(path)
             return fctx.flags()
 
         return _fctx_flags(repo_id, commit_id, path)
 
     @reraise_safe_exceptions
     def fctx_size(self, wire, commit_id, path):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _fctx_size(_repo_id, _revision, _path):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, commit_id)
             fctx = ctx.filectx(path)
             return fctx.size()
         return _fctx_size(repo_id, commit_id, path)
 
     @reraise_safe_exceptions
     def get_all_commit_ids(self, wire, name):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_all_commit_ids(_context_uid, _repo_id, _name):
             repo = self._factory.repo(wire)
             repo = repo.filtered(name)
             revs = [hex(x[7]) for x in repo.changelog.index]
             return revs
         return _get_all_commit_ids(context_uid, repo_id, name)
 
     @reraise_safe_exceptions
     def get_config_value(self, wire, section, name, untrusted=False):
         repo = self._factory.repo(wire)
         return repo.ui.config(section, name, untrusted=untrusted)
 
     @reraise_safe_exceptions
     def is_large_file(self, wire, commit_id, path):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _is_large_file(_context_uid, _repo_id, _commit_id, _path):
             return largefiles.lfutil.isstandin(path)
 
         return _is_large_file(context_uid, repo_id, commit_id, path)
 
     @reraise_safe_exceptions
     def is_binary(self, wire, revision, path):
         cache_on, context_uid, repo_id = self._cache_on(wire)
 
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _is_binary(_repo_id, _sha, _path):
             repo = self._factory.repo(wire)
             ctx = self._get_ctx(repo, revision)
             fctx = ctx.filectx(path)
             return fctx.isbinary()
 
         return _is_binary(repo_id, revision, path)
 
     @reraise_safe_exceptions
     def in_largefiles_store(self, wire, sha):
         repo = self._factory.repo(wire)
         return largefiles.lfutil.instore(repo, sha)
 
     @reraise_safe_exceptions
     def in_user_cache(self, wire, sha):
         repo = self._factory.repo(wire)
         return largefiles.lfutil.inusercache(repo.ui, sha)
 
     @reraise_safe_exceptions
     def store_path(self, wire, sha):
         repo = self._factory.repo(wire)
         return largefiles.lfutil.storepath(repo, sha)
 
     @reraise_safe_exceptions
     def link(self, wire, sha, path):
         repo = self._factory.repo(wire)
         largefiles.lfutil.link(
             largefiles.lfutil.usercachepath(repo.ui, sha), path)
 
     @reraise_safe_exceptions
     def localrepository(self, wire, create=False):
         self._factory.repo(wire, create=create)
 
     @reraise_safe_exceptions
     def lookup(self, wire, revision, both):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _lookup(_context_uid, _repo_id, _revision, _both):
 
             repo = self._factory.repo(wire)
             rev = _revision
             if isinstance(rev, int):
                 # NOTE(marcink):
                 # since Mercurial doesn't support negative indexes properly
                 # we need to shift accordingly by one to get proper index, e.g
                 # repo[-1] => repo[-2]
                 # repo[0]  => repo[-1]
                 if rev <= 0:
                     rev = rev + -1
             try:
                 ctx = self._get_ctx(repo, rev)
             except (TypeError, RepoLookupError) as e:
                 e._org_exc_tb = traceback.format_exc()
                 raise exceptions.LookupException(e)(rev)
             except LookupError as e:
                 e._org_exc_tb = traceback.format_exc()
                 raise exceptions.LookupException(e)(e.name)
 
             if not both:
                 return ctx.hex()
 
             ctx = repo[ctx.hex()]
             return ctx.hex(), ctx.rev()
 
         return _lookup(context_uid, repo_id, revision, both)
 
     @reraise_safe_exceptions
     def sync_push(self, wire, url):
         if not self.check_url(url, wire['config']):
             return
 
         repo = self._factory.repo(wire)
 
         # Disable any prompts for this repo
         repo.ui.setconfig('ui', 'interactive', 'off', '-y')
 
         bookmarks = list(dict(repo._bookmarks).keys())
         remote = peer(repo, {}, url)
         # Disable any prompts for this remote
         remote.ui.setconfig('ui', 'interactive', 'off', '-y')
 
         return exchange.push(
             repo, remote, newbranch=True, bookmarks=bookmarks).cgresult
 
     @reraise_safe_exceptions
     def revision(self, wire, rev):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, rev)
         return ctx.rev()
 
     @reraise_safe_exceptions
     def rev_range(self, wire, commit_filter):
         cache_on, context_uid, repo_id = self._cache_on(wire)
 
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _rev_range(_context_uid, _repo_id, _filter):
             repo = self._factory.repo(wire)
             revisions = [rev for rev in revrange(repo, commit_filter)]
             return revisions
 
         return _rev_range(context_uid, repo_id, sorted(commit_filter))
 
     @reraise_safe_exceptions
     def rev_range_hash(self, wire, node):
         repo = self._factory.repo(wire)
 
         def get_revs(repo, rev_opt):
             if rev_opt:
                 revs = revrange(repo, rev_opt)
                 if len(revs) == 0:
                     return (nullrev, nullrev)
                 return max(revs), min(revs)
             else:
                 return len(repo) - 1, 0
 
         stop, start = get_revs(repo, [node + ':'])
         revs = [hex(repo[r].node()) for r in range(start, stop + 1)]
         return revs
 
     @reraise_safe_exceptions
     def revs_from_revspec(self, wire, rev_spec, *args, **kwargs):
         other_path = kwargs.pop('other_path', None)
 
         # case when we want to compare two independent repositories
         if other_path and other_path != wire["path"]:
             baseui = self._factory._create_config(wire["config"])
             repo = unionrepo.makeunionrepository(baseui, other_path, wire["path"])
         else:
             repo = self._factory.repo(wire)
         return list(repo.revs(rev_spec, *args))
 
     @reraise_safe_exceptions
     def verify(self, wire,):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
 
         baseui, output = patch_ui_message_output(baseui)
 
         repo.ui = baseui
         verify.verify(repo)
         return output.getvalue()
 
     @reraise_safe_exceptions
     def hg_update_cache(self, wire,):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         baseui, output = patch_ui_message_output(baseui)
 
         repo.ui = baseui
         with repo.wlock(), repo.lock():
             repo.updatecaches(full=True)
 
         return output.getvalue()
 
     @reraise_safe_exceptions
     def hg_rebuild_fn_cache(self, wire,):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         baseui, output = patch_ui_message_output(baseui)
 
         repo.ui = baseui
 
         repair.rebuildfncache(baseui, repo)
 
         return output.getvalue()
 
     @reraise_safe_exceptions
     def tags(self, wire):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _tags(_context_uid, _repo_id):
             repo = self._factory.repo(wire)
             return repo.tags()
 
         return _tags(context_uid, repo_id)
 
     @reraise_safe_exceptions
     def update(self, wire, node=None, clean=False):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         commands.update(baseui, repo, node=node, clean=clean)
 
     @reraise_safe_exceptions
     def identify(self, wire):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         output = io.BytesIO()
         baseui.write = output.write
         # This is required to get a full node id
         baseui.debugflag = True
         commands.identify(baseui, repo, id=True)
 
         return output.getvalue()
 
     @reraise_safe_exceptions
     def heads(self, wire, branch=None):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         output = io.BytesIO()
 
         def write(data, **unused_kwargs):
             output.write(data)
 
         baseui.write = write
         if branch:
             args = [branch]
         else:
             args = []
         commands.heads(baseui, repo, template='{node} ', *args)
 
         return output.getvalue()
 
     @reraise_safe_exceptions
     def ancestor(self, wire, revision1, revision2):
         repo = self._factory.repo(wire)
         changelog = repo.changelog
         lookup = repo.lookup
         a = changelog.ancestor(lookup(revision1), lookup(revision2))
         return hex(a)
 
     @reraise_safe_exceptions
     def clone(self, wire, source, dest, update_after_clone=False, hooks=True):
         baseui = self._factory._create_config(wire["config"], hooks=hooks)
         clone(baseui, source, dest, noupdate=not update_after_clone)
 
     @reraise_safe_exceptions
     def commitctx(self, wire, message, parents, commit_time, commit_timezone, user, files, extra, removed, updated):
 
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         publishing = baseui.configbool('phases', 'publish')
         if publishing:
             new_commit = 'public'
         else:
             new_commit = 'draft'
 
         def _filectxfn(_repo, ctx, path):
             """
             Marks given path as added/changed/removed in a given _repo. This is
             for internal mercurial commit function.
             """
 
             # check if this path is removed
             if path in removed:
                 # returning None is a way to mark node for removal
                 return None
 
             # check if this path is added
             for node in updated:
                 if node['path'] == path:
                     return memfilectx(
                         _repo,
                         changectx=ctx,
                         path=node['path'],
                         data=node['content'],
                         islink=False,
                         isexec=bool(node['mode'] & stat.S_IXUSR),
                         copysource=False)
 
             raise exceptions.AbortException()(
                 "Given path haven't been marked as added, "
                 "changed or removed (%s)" % path)
 
         with repo.ui.configoverride({('phases', 'new-commit'): new_commit}):
 
             commit_ctx = memctx(
                 repo=repo,
                 parents=parents,
                 text=message,
                 files=files,
                 filectxfn=_filectxfn,
                 user=user,
                 date=(commit_time, commit_timezone),
                 extra=extra)
 
             n = repo.commitctx(commit_ctx)
             new_id = hex(n)
 
             return new_id
 
     @reraise_safe_exceptions
     def pull(self, wire, url, commit_ids=None):
         repo = self._factory.repo(wire)
         # Disable any prompts for this repo
         repo.ui.setconfig('ui', 'interactive', 'off', '-y')
 
         remote = peer(repo, {}, url)
         # Disable any prompts for this remote
         remote.ui.setconfig('ui', 'interactive', 'off', '-y')
 
         if commit_ids:
             commit_ids = [bin(commit_id) for commit_id in commit_ids]
 
         return exchange.pull(
             repo, remote, heads=commit_ids, force=None).cgresult
 
     @reraise_safe_exceptions
     def pull_cmd(self, wire, source, bookmark=None, branch=None, revision=None, hooks=True):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'], hooks=hooks)
 
         # Mercurial internally has a lot of logic that checks ONLY if
         # option is defined, we just pass those if they are defined then
         opts = {}
         if bookmark:
             opts['bookmark'] = bookmark
         if branch:
             opts['branch'] = branch
         if revision:
             opts['rev'] = revision
 
         commands.pull(baseui, repo, source, **opts)
 
     @reraise_safe_exceptions
     def push(self, wire, revisions, dest_path, hooks=True, push_branches=False):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'], hooks=hooks)
         commands.push(baseui, repo, dest=dest_path, rev=revisions,
                       new_branch=push_branches)
 
     @reraise_safe_exceptions
     def strip(self, wire, revision, update, backup):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         hgext_strip(
             repo.baseui, repo, ctx.node(), update=update, backup=backup)
 
     @reraise_safe_exceptions
     def get_unresolved_files(self, wire):
         repo = self._factory.repo(wire)
 
         log.debug('Calculating unresolved files for repo: %s', repo)
         output = io.BytesIO()
 
         def write(data, **unused_kwargs):
             output.write(data)
 
         baseui = self._factory._create_config(wire['config'])
         baseui.write = write
 
         commands.resolve(baseui, repo, list=True)
         unresolved = output.getvalue().splitlines(0)
         return unresolved
 
     @reraise_safe_exceptions
     def merge(self, wire, revision):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         repo.ui.setconfig('ui', 'merge', 'internal:dump')
 
         # In case of sub repositories are used mercurial prompts the user in
         # case of merge conflicts or different sub repository sources. By
         # setting the interactive flag to `False` mercurial doesn't prompt the
         # used but instead uses a default value.
         repo.ui.setconfig('ui', 'interactive', False)
         commands.merge(baseui, repo, rev=revision)
 
     @reraise_safe_exceptions
     def merge_state(self, wire):
         repo = self._factory.repo(wire)
         repo.ui.setconfig('ui', 'merge', 'internal:dump')
 
         # In case of sub repositories are used mercurial prompts the user in
         # case of merge conflicts or different sub repository sources. By
         # setting the interactive flag to `False` mercurial doesn't prompt the
         # used but instead uses a default value.
         repo.ui.setconfig('ui', 'interactive', False)
         ms = hg_merge.mergestate(repo)
         return [x for x in ms.unresolved()]
 
     @reraise_safe_exceptions
     def commit(self, wire, message, username, close_branch=False):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         repo.ui.setconfig('ui', 'username', username)
         commands.commit(baseui, repo, message=message, close_branch=close_branch)
 
     @reraise_safe_exceptions
     def rebase(self, wire, source=None, dest=None, abort=False):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         repo.ui.setconfig('ui', 'merge', 'internal:dump')
         # In case of sub repositories are used mercurial prompts the user in
         # case of merge conflicts or different sub repository sources. By
         # setting the interactive flag to `False` mercurial doesn't prompt the
         # used but instead uses a default value.
         repo.ui.setconfig('ui', 'interactive', False)
         rebase.rebase(baseui, repo, base=source, dest=dest, abort=abort, keep=not abort)
 
     @reraise_safe_exceptions
     def tag(self, wire, name, revision, message, local, user, tag_time, tag_timezone):
         repo = self._factory.repo(wire)
         ctx = self._get_ctx(repo, revision)
         node = ctx.node()
 
         date = (tag_time, tag_timezone)
         try:
             hg_tag.tag(repo, name, node, message, local, user, date)
         except Abort as e:
             log.exception("Tag operation aborted")
             # Exception can contain unicode which we convert
             raise exceptions.AbortException(e)(repr(e))
 
     @reraise_safe_exceptions
     def bookmark(self, wire, bookmark, revision=None):
         repo = self._factory.repo(wire)
         baseui = self._factory._create_config(wire['config'])
         commands.bookmark(baseui, repo, bookmark, rev=revision, force=True)
 
     @reraise_safe_exceptions
     def install_hooks(self, wire, force=False):
         # we don't need any special hooks for Mercurial
         pass
 
     @reraise_safe_exceptions
     def get_hooks_info(self, wire):
         return {
             'pre_version': vcsserver.__version__,
             'post_version': vcsserver.__version__,
         }
+
+    @reraise_safe_exceptions
+    def set_head_ref(self, wire, head_name):
+        pass
+
+    @reraise_safe_exceptions
+    def archive_repo(self, wire, archive_dest_path, kind, mtime, archive_at_path,
+                     archive_dir_name, commit_id):
+
+        def file_walker(_commit_id, path):
+            repo = self._factory.repo(wire)
+            ctx = repo[_commit_id]
+            is_root = path in ['', '/']
+            if is_root:
+                matcher = alwaysmatcher(badfn=None)
+            else:
+                matcher = patternmatcher('', [(b'glob', path+'/**', b'')], badfn=None)
+            file_iter = ctx.manifest().walk(matcher)
+
+            for fn in file_iter:
+                file_path = fn
+                flags = ctx.flags(fn)
+                mode = b'x' in flags and 0o755 or 0o644
+                is_link = b'l' in flags
+
+                yield ArchiveNode(file_path, mode, is_link, ctx[fn].data)
+
+        return archive_repo(file_walker, archive_dest_path, kind, mtime, archive_at_path,
+                            archive_dir_name, commit_id)
+

vcsserver/hgcompat.py

@@ -1,79 +1,79 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 """
 Mercurial libs compatibility
 """
 
 import mercurial
 from mercurial import demandimport
 # patch demandimport, due to bug in mercurial when it always triggers
 # demandimport.enable()
 demandimport.enable = lambda *args, **kwargs: 1
 
 from mercurial import ui
 from mercurial import patch
 from mercurial import config
 from mercurial import extensions
 from mercurial import scmutil
 from mercurial import archival
 from mercurial import discovery
 from mercurial import unionrepo
 from mercurial import localrepo
 from mercurial import merge as hg_merge
 from mercurial import subrepo
 from mercurial import subrepoutil
 from mercurial import tags as hg_tag
-
+from mercurial import util as hgutil
 from mercurial.commands import clone, nullid, pull
 from mercurial.context import memctx, memfilectx
 from mercurial.error import (
     LookupError, RepoError, RepoLookupError, Abort, InterventionRequired,
     RequirementError, ProgrammingError)
 from mercurial.hgweb import hgweb_mod
 from mercurial.localrepo import instance
-from mercurial.match import match
+from mercurial.match import match, alwaysmatcher, patternmatcher
 from mercurial.mdiff import diffopts
 from mercurial.node import bin, hex
 from mercurial.encoding import tolocal
 from mercurial.discovery import findcommonoutgoing
 from mercurial.hg import peer
 from mercurial.httppeer import makepeer
 from mercurial.util import url as hg_url
 from mercurial.scmutil import revrange, revsymbol
 from mercurial.node import nullrev
 from mercurial import exchange
 from hgext import largefiles
 
 # those authnadlers are patched for python 2.6.5 bug an
 # infinit looping when given invalid resources
 from mercurial.url import httpbasicauthhandler, httpdigestauthhandler
 
 
 def get_ctx(repo, ref):
     try:
         ctx = repo[ref]
-    except ProgrammingError:
+    except (ProgrammingError, TypeError):
         # we're unable to find the rev using a regular lookup, we fallback
         # to slower, but backward compat revsymbol usage
         ctx = revsymbol(repo, ref)
     except (LookupError, RepoLookupError):
         # Similar case as above but only for refs that are not numeric
         if isinstance(ref, (int, long)):
             raise
         ctx = revsymbol(repo, ref)
     return ctx

vcsserver/hooks.py

@@ -1,729 +1,729 b''
 # -*- coding: utf-8 -*-
 
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import io
 import os
 import sys
 import logging
 import collections
 import importlib
 import base64
 
 from http.client import HTTPConnection
 
 
 import mercurial.scmutil
 import mercurial.node
 import simplejson as json
 
 from vcsserver import exceptions, subprocessio, settings
 
 log = logging.getLogger(__name__)
 
 
 class HooksHttpClient(object):
     connection = None
 
     def __init__(self, hooks_uri):
         self.hooks_uri = hooks_uri
 
     def __call__(self, method, extras):
         connection = HTTPConnection(self.hooks_uri)
         body = self._serialize(method, extras)
         try:
             connection.request('POST', '/', body)
         except Exception:
-            log.error('Connection failed on %s', connection)
+            log.error('Hooks calling Connection failed on %s', connection.__dict__)
             raise
         response = connection.getresponse()
 
         response_data = response.read()
 
         try:
             return json.loads(response_data)
         except Exception:
             log.exception('Failed to decode hook response json data. '
                           'response_code:%s, raw_data:%s',
                           response.status, response_data)
             raise
 
     def _serialize(self, hook_name, extras):
         data = {
             'method': hook_name,
             'extras': extras
         }
         return json.dumps(data)
 
 
 class HooksDummyClient(object):
     def __init__(self, hooks_module):
         self._hooks_module = importlib.import_module(hooks_module)
 
     def __call__(self, hook_name, extras):
         with self._hooks_module.Hooks() as hooks:
             return getattr(hooks, hook_name)(extras)
 
 
 class HooksShadowRepoClient(object):
 
     def __call__(self, hook_name, extras):
         return {'output': '', 'status': 0}
 
 
 class RemoteMessageWriter(object):
     """Writer base class."""
     def write(self, message):
         raise NotImplementedError()
 
 
 class HgMessageWriter(RemoteMessageWriter):
     """Writer that knows how to send messages to mercurial clients."""
 
     def __init__(self, ui):
         self.ui = ui
 
     def write(self, message):
         # TODO: Check why the quiet flag is set by default.
         old = self.ui.quiet
         self.ui.quiet = False
         self.ui.status(message.encode('utf-8'))
         self.ui.quiet = old
 
 
 class GitMessageWriter(RemoteMessageWriter):
     """Writer that knows how to send messages to git clients."""
 
     def __init__(self, stdout=None):
         self.stdout = stdout or sys.stdout
 
     def write(self, message):
         self.stdout.write(message.encode('utf-8'))
 
 
 class SvnMessageWriter(RemoteMessageWriter):
     """Writer that knows how to send messages to svn clients."""
 
     def __init__(self, stderr=None):
         # SVN needs data sent to stderr for back-to-client messaging
         self.stderr = stderr or sys.stderr
 
     def write(self, message):
         self.stderr.write(message.encode('utf-8'))
 
 
 def _handle_exception(result):
     exception_class = result.get('exception')
     exception_traceback = result.get('exception_traceback')
 
     if exception_traceback:
         log.error('Got traceback from remote call:%s', exception_traceback)
 
     if exception_class == 'HTTPLockedRC':
         raise exceptions.RepositoryLockedException()(*result['exception_args'])
     elif exception_class == 'HTTPBranchProtected':
         raise exceptions.RepositoryBranchProtectedException()(*result['exception_args'])
     elif exception_class == 'RepositoryError':
         raise exceptions.VcsException()(*result['exception_args'])
     elif exception_class:
         raise Exception('Got remote exception "%s" with args "%s"' %
                         (exception_class, result['exception_args']))
 
 
 def _get_hooks_client(extras):
     hooks_uri = extras.get('hooks_uri')
     is_shadow_repo = extras.get('is_shadow_repo')
     if hooks_uri:
         return HooksHttpClient(extras['hooks_uri'])
     elif is_shadow_repo:
         return HooksShadowRepoClient()
     else:
         return HooksDummyClient(extras['hooks_module'])
 
 
 def _call_hook(hook_name, extras, writer):
     hooks_client = _get_hooks_client(extras)
     log.debug('Hooks, using client:%s', hooks_client)
     result = hooks_client(hook_name, extras)
     log.debug('Hooks got result: %s', result)
 
     _handle_exception(result)
     writer.write(result['output'])
 
     return result['status']
 
 
 def _extras_from_ui(ui):
     hook_data = ui.config('rhodecode', 'RC_SCM_DATA')
     if not hook_data:
         # maybe it's inside environ ?
         env_hook_data = os.environ.get('RC_SCM_DATA')
         if env_hook_data:
             hook_data = env_hook_data
 
     extras = {}
     if hook_data:
         extras = json.loads(hook_data)
     return extras
 
 
 def _rev_range_hash(repo, node, check_heads=False):
     from vcsserver.hgcompat import get_ctx
 
     commits = []
     revs = []
     start = get_ctx(repo, node).rev()
     end = len(repo)
     for rev in range(start, end):
         revs.append(rev)
         ctx = get_ctx(repo, rev)
         commit_id = mercurial.node.hex(ctx.node())
         branch = ctx.branch()
         commits.append((commit_id, branch))
 
     parent_heads = []
     if check_heads:
         parent_heads = _check_heads(repo, start, end, revs)
     return commits, parent_heads
 
 
 def _check_heads(repo, start, end, commits):
     from vcsserver.hgcompat import get_ctx
     changelog = repo.changelog
     parents = set()
 
     for new_rev in commits:
         for p in changelog.parentrevs(new_rev):
             if p == mercurial.node.nullrev:
                 continue
             if p < start:
                 parents.add(p)
 
     for p in parents:
         branch = get_ctx(repo, p).branch()
         # The heads descending from that parent, on the same branch
         parent_heads = set([p])
         reachable = set([p])
         for x in range(p + 1, end):
             if get_ctx(repo, x).branch() != branch:
                 continue
             for pp in changelog.parentrevs(x):
                 if pp in reachable:
                     reachable.add(x)
                     parent_heads.discard(pp)
                     parent_heads.add(x)
         # More than one head? Suggest merging
         if len(parent_heads) > 1:
             return list(parent_heads)
 
     return []
 
 
 def _get_git_env():
     env = {}
     for k, v in os.environ.items():
         if k.startswith('GIT'):
             env[k] = v
 
     # serialized version
     return [(k, v) for k, v in env.items()]
 
 
 def _get_hg_env(old_rev, new_rev, txnid, repo_path):
     env = {}
     for k, v in os.environ.items():
         if k.startswith('HG'):
             env[k] = v
 
     env['HG_NODE'] = old_rev
     env['HG_NODE_LAST'] = new_rev
     env['HG_TXNID'] = txnid
     env['HG_PENDING'] = repo_path
 
     return [(k, v) for k, v in env.items()]
 
 
 def repo_size(ui, repo, **kwargs):
     extras = _extras_from_ui(ui)
     return _call_hook('repo_size', extras, HgMessageWriter(ui))
 
 
 def pre_pull(ui, repo, **kwargs):
     extras = _extras_from_ui(ui)
     return _call_hook('pre_pull', extras, HgMessageWriter(ui))
 
 
 def pre_pull_ssh(ui, repo, **kwargs):
     extras = _extras_from_ui(ui)
     if extras and extras.get('SSH'):
         return pre_pull(ui, repo, **kwargs)
     return 0
 
 
 def post_pull(ui, repo, **kwargs):
     extras = _extras_from_ui(ui)
     return _call_hook('post_pull', extras, HgMessageWriter(ui))
 
 
 def post_pull_ssh(ui, repo, **kwargs):
     extras = _extras_from_ui(ui)
     if extras and extras.get('SSH'):
         return post_pull(ui, repo, **kwargs)
     return 0
 
 
 def pre_push(ui, repo, node=None, **kwargs):
     """
     Mercurial pre_push hook
     """
     extras = _extras_from_ui(ui)
     detect_force_push = extras.get('detect_force_push')
 
     rev_data = []
     if node and kwargs.get('hooktype') == 'pretxnchangegroup':
         branches = collections.defaultdict(list)
         commits, _heads = _rev_range_hash(repo, node, check_heads=detect_force_push)
         for commit_id, branch in commits:
             branches[branch].append(commit_id)
 
         for branch, commits in branches.items():
             old_rev = kwargs.get('node_last') or commits[0]
             rev_data.append({
                 'total_commits': len(commits),
                 'old_rev': old_rev,
                 'new_rev': commits[-1],
                 'ref': '',
                 'type': 'branch',
                 'name': branch,
             })
 
         for push_ref in rev_data:
             push_ref['multiple_heads'] = _heads
 
             repo_path = os.path.join(
                 extras.get('repo_store', ''), extras.get('repository', ''))
             push_ref['hg_env'] = _get_hg_env(
                 old_rev=push_ref['old_rev'],
                 new_rev=push_ref['new_rev'], txnid=kwargs.get('txnid'),
                 repo_path=repo_path)
 
     extras['hook_type'] = kwargs.get('hooktype', 'pre_push')
     extras['commit_ids'] = rev_data
 
     return _call_hook('pre_push', extras, HgMessageWriter(ui))
 
 
 def pre_push_ssh(ui, repo, node=None, **kwargs):
     extras = _extras_from_ui(ui)
     if extras.get('SSH'):
         return pre_push(ui, repo, node, **kwargs)
 
     return 0
 
 
 def pre_push_ssh_auth(ui, repo, node=None, **kwargs):
     """
     Mercurial pre_push hook for SSH
     """
     extras = _extras_from_ui(ui)
     if extras.get('SSH'):
         permission = extras['SSH_PERMISSIONS']
 
         if 'repository.write' == permission or 'repository.admin' == permission:
             return 0
 
         # non-zero ret code
         return 1
 
     return 0
 
 
 def post_push(ui, repo, node, **kwargs):
     """
     Mercurial post_push hook
     """
     extras = _extras_from_ui(ui)
 
     commit_ids = []
     branches = []
     bookmarks = []
     tags = []
 
     commits, _heads = _rev_range_hash(repo, node)
     for commit_id, branch in commits:
         commit_ids.append(commit_id)
         if branch not in branches:
             branches.append(branch)
 
     if hasattr(ui, '_rc_pushkey_branches'):
         bookmarks = ui._rc_pushkey_branches
 
     extras['hook_type'] = kwargs.get('hooktype', 'post_push')
     extras['commit_ids'] = commit_ids
     extras['new_refs'] = {
         'branches': branches,
         'bookmarks': bookmarks,
         'tags': tags
     }
 
     return _call_hook('post_push', extras, HgMessageWriter(ui))
 
 
 def post_push_ssh(ui, repo, node, **kwargs):
     """
     Mercurial post_push hook for SSH
     """
     if _extras_from_ui(ui).get('SSH'):
         return post_push(ui, repo, node, **kwargs)
     return 0
 
 
 def key_push(ui, repo, **kwargs):
     from vcsserver.hgcompat import get_ctx
     if kwargs['new'] != '0' and kwargs['namespace'] == 'bookmarks':
         # store new bookmarks in our UI object propagated later to post_push
         ui._rc_pushkey_branches = get_ctx(repo, kwargs['key']).bookmarks()
     return
 
 
 # backward compat
 log_pull_action = post_pull
 
 # backward compat
 log_push_action = post_push
 
 
 def handle_git_pre_receive(unused_repo_path, unused_revs, unused_env):
     """
     Old hook name: keep here for backward compatibility.
 
     This is only required when the installed git hooks are not upgraded.
     """
     pass
 
 
 def handle_git_post_receive(unused_repo_path, unused_revs, unused_env):
     """
     Old hook name: keep here for backward compatibility.
 
     This is only required when the installed git hooks are not upgraded.
     """
     pass
 
 
 HookResponse = collections.namedtuple('HookResponse', ('status', 'output'))
 
 
 def git_pre_pull(extras):
     """
     Pre pull hook.
 
     :param extras: dictionary containing the keys defined in simplevcs
     :type extras: dict
 
     :return: status code of the hook. 0 for success.
     :rtype: int
     """
     if 'pull' not in extras['hooks']:
         return HookResponse(0, '')
 
     stdout = io.BytesIO()
     try:
         status = _call_hook('pre_pull', extras, GitMessageWriter(stdout))
     except Exception as error:
         status = 128
         stdout.write('ERROR: %s\n' % str(error))
 
     return HookResponse(status, stdout.getvalue())
 
 
 def git_post_pull(extras):
     """
     Post pull hook.
 
     :param extras: dictionary containing the keys defined in simplevcs
     :type extras: dict
 
     :return: status code of the hook. 0 for success.
     :rtype: int
     """
     if 'pull' not in extras['hooks']:
         return HookResponse(0, '')
 
     stdout = io.BytesIO()
     try:
         status = _call_hook('post_pull', extras, GitMessageWriter(stdout))
     except Exception as error:
         status = 128
         stdout.write('ERROR: %s\n' % error)
 
     return HookResponse(status, stdout.getvalue())
 
 
 def _parse_git_ref_lines(revision_lines):
     rev_data = []
     for revision_line in revision_lines or []:
         old_rev, new_rev, ref = revision_line.strip().split(' ')
         ref_data = ref.split('/', 2)
         if ref_data[1] in ('tags', 'heads'):
             rev_data.append({
                 # NOTE(marcink):
                 # we're unable to tell total_commits for git at this point
                 # but we set the variable for consistency with GIT
                 'total_commits': -1,
                 'old_rev': old_rev,
                 'new_rev': new_rev,
                 'ref': ref,
                 'type': ref_data[1],
                 'name': ref_data[2],
             })
     return rev_data
 
 
 def git_pre_receive(unused_repo_path, revision_lines, env):
     """
     Pre push hook.
 
     :param extras: dictionary containing the keys defined in simplevcs
     :type extras: dict
 
     :return: status code of the hook. 0 for success.
     :rtype: int
     """
     extras = json.loads(env['RC_SCM_DATA'])
     rev_data = _parse_git_ref_lines(revision_lines)
     if 'push' not in extras['hooks']:
         return 0
     empty_commit_id = '0' * 40
 
     detect_force_push = extras.get('detect_force_push')
 
     for push_ref in rev_data:
         # store our git-env which holds the temp store
         push_ref['git_env'] = _get_git_env()
         push_ref['pruned_sha'] = ''
         if not detect_force_push:
             # don't check for forced-push when we don't need to
             continue
 
         type_ = push_ref['type']
         new_branch = push_ref['old_rev'] == empty_commit_id
         delete_branch = push_ref['new_rev'] == empty_commit_id
         if type_ == 'heads' and not (new_branch or delete_branch):
             old_rev = push_ref['old_rev']
             new_rev = push_ref['new_rev']
             cmd = [settings.GIT_EXECUTABLE, 'rev-list', old_rev, '^{}'.format(new_rev)]
             stdout, stderr = subprocessio.run_command(
                 cmd, env=os.environ.copy())
             # means we're having some non-reachable objects, this forced push was used
             if stdout:
                 push_ref['pruned_sha'] = stdout.splitlines()
 
     extras['hook_type'] = 'pre_receive'
     extras['commit_ids'] = rev_data
     return _call_hook('pre_push', extras, GitMessageWriter())
 
 
 def git_post_receive(unused_repo_path, revision_lines, env):
     """
     Post push hook.
 
     :param extras: dictionary containing the keys defined in simplevcs
     :type extras: dict
 
     :return: status code of the hook. 0 for success.
     :rtype: int
     """
     extras = json.loads(env['RC_SCM_DATA'])
     if 'push' not in extras['hooks']:
         return 0
 
     rev_data = _parse_git_ref_lines(revision_lines)
 
     git_revs = []
 
     # N.B.(skreft): it is ok to just call git, as git before calling a
     # subcommand sets the PATH environment variable so that it point to the
     # correct version of the git executable.
     empty_commit_id = '0' * 40
     branches = []
     tags = []
     for push_ref in rev_data:
         type_ = push_ref['type']
 
         if type_ == 'heads':
             if push_ref['old_rev'] == empty_commit_id:
                 # starting new branch case
                 if push_ref['name'] not in branches:
                     branches.append(push_ref['name'])
 
                 # Fix up head revision if needed
                 cmd = [settings.GIT_EXECUTABLE, 'show', 'HEAD']
                 try:
                     subprocessio.run_command(cmd, env=os.environ.copy())
                 except Exception:
-                    cmd = [settings.GIT_EXECUTABLE, 'symbolic-ref', 'HEAD',
-                           'refs/heads/%s' % push_ref['name']]
+                    cmd = [settings.GIT_EXECUTABLE, 'symbolic-ref', '"HEAD"',
+                           '"refs/heads/%s"' % push_ref['name']]
                     print("Setting default branch to %s" % push_ref['name'])
                     subprocessio.run_command(cmd, env=os.environ.copy())
 
                 cmd = [settings.GIT_EXECUTABLE, 'for-each-ref',
                        '--format=%(refname)', 'refs/heads/*']
                 stdout, stderr = subprocessio.run_command(
                     cmd, env=os.environ.copy())
                 heads = stdout
                 heads = heads.replace(push_ref['ref'], '')
                 heads = ' '.join(head for head
                                  in heads.splitlines() if head) or '.'
                 cmd = [settings.GIT_EXECUTABLE, 'log', '--reverse',
                        '--pretty=format:%H', '--', push_ref['new_rev'],
                        '--not', heads]
                 stdout, stderr = subprocessio.run_command(
                     cmd, env=os.environ.copy())
                 git_revs.extend(stdout.splitlines())
             elif push_ref['new_rev'] == empty_commit_id:
                 # delete branch case
                 git_revs.append('delete_branch=>%s' % push_ref['name'])
             else:
                 if push_ref['name'] not in branches:
                     branches.append(push_ref['name'])
 
                 cmd = [settings.GIT_EXECUTABLE, 'log',
                        '{old_rev}..{new_rev}'.format(**push_ref),
                        '--reverse', '--pretty=format:%H']
                 stdout, stderr = subprocessio.run_command(
                     cmd, env=os.environ.copy())
                 git_revs.extend(stdout.splitlines())
         elif type_ == 'tags':
             if push_ref['name'] not in tags:
                 tags.append(push_ref['name'])
             git_revs.append('tag=>%s' % push_ref['name'])
 
     extras['hook_type'] = 'post_receive'
     extras['commit_ids'] = git_revs
     extras['new_refs'] = {
         'branches': branches,
         'bookmarks': [],
         'tags': tags,
     }
 
     if 'repo_size' in extras['hooks']:
         try:
             _call_hook('repo_size', extras, GitMessageWriter())
         except:
             pass
 
     return _call_hook('post_push', extras, GitMessageWriter())
 
 
 def _get_extras_from_txn_id(path, txn_id):
     extras = {}
     try:
         cmd = [settings.SVNLOOK_EXECUTABLE, 'pget',
                '-t', txn_id,
                '--revprop', path, 'rc-scm-extras']
         stdout, stderr = subprocessio.run_command(
             cmd, env=os.environ.copy())
         extras = json.loads(base64.urlsafe_b64decode(stdout))
     except Exception:
         log.exception('Failed to extract extras info from txn_id')
 
     return extras
 
 
 def _get_extras_from_commit_id(commit_id, path):
     extras = {}
     try:
         cmd = [settings.SVNLOOK_EXECUTABLE, 'pget',
                '-r', commit_id,
                '--revprop', path, 'rc-scm-extras']
         stdout, stderr = subprocessio.run_command(
             cmd, env=os.environ.copy())
         extras = json.loads(base64.urlsafe_b64decode(stdout))
     except Exception:
         log.exception('Failed to extract extras info from commit_id')
 
     return extras
 
 
 def svn_pre_commit(repo_path, commit_data, env):
     path, txn_id = commit_data
     branches = []
     tags = []
 
     if env.get('RC_SCM_DATA'):
         extras = json.loads(env['RC_SCM_DATA'])
     else:
         # fallback method to read from TXN-ID stored data
         extras = _get_extras_from_txn_id(path, txn_id)
         if not extras:
             return 0
 
     extras['hook_type'] = 'pre_commit'
     extras['commit_ids'] = [txn_id]
     extras['txn_id'] = txn_id
     extras['new_refs'] = {
         'total_commits': 1,
         'branches': branches,
         'bookmarks': [],
         'tags': tags,
     }
 
     return _call_hook('pre_push', extras, SvnMessageWriter())
 
 
 def svn_post_commit(repo_path, commit_data, env):
     """
     commit_data is path, rev, txn_id
     """
     if len(commit_data) == 3:
         path, commit_id, txn_id = commit_data
     elif len(commit_data) == 2:
         log.error('Failed to extract txn_id from commit_data using legacy method. '
                   'Some functionality might be limited')
         path, commit_id = commit_data
         txn_id = None
 
     branches = []
     tags = []
 
     if env.get('RC_SCM_DATA'):
         extras = json.loads(env['RC_SCM_DATA'])
     else:
         # fallback method to read from TXN-ID stored data
         extras = _get_extras_from_commit_id(commit_id, path)
         if not extras:
             return 0
 
     extras['hook_type'] = 'post_commit'
     extras['commit_ids'] = [commit_id]
     extras['txn_id'] = txn_id
     extras['new_refs'] = {
         'branches': branches,
         'bookmarks': [],
         'tags': tags,
         'total_commits': 1,
     }
 
     if 'repo_size' in extras['hooks']:
         try:
             _call_hook('repo_size', extras, SvnMessageWriter())
         except Exception:
             pass
 
     return _call_hook('post_push', extras, SvnMessageWriter())

vcsserver/http_main.py

@@ -1,702 +1,740 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import os
 import sys
 import base64
 import locale
 import logging
 import uuid
+import time
 import wsgiref.util
 import traceback
 import tempfile
+import psutil
+
 from itertools import chain
 from io import StringIO
 
 import simplejson as json
 import msgpack
 import configparser
+
 from pyramid.config import Configurator
 from pyramid.settings import asbool, aslist
 from pyramid.wsgi import wsgiapp
 from pyramid.response import Response
 
+from vcsserver.config.settings_maker import SettingsMaker
 from vcsserver.utils import safe_int
+from vcsserver.lib.statsd_client import StatsdClient
 
 log = logging.getLogger(__name__)
 
 # due to Mercurial/glibc2.27 problems we need to detect if locale settings are
 # causing problems and "fix" it in case they do and fallback to LC_ALL = C
 
 try:
     locale.setlocale(locale.LC_ALL, '')
 except locale.Error as e:
     log.error(
         'LOCALE ERROR: failed to set LC_ALL, fallback to LC_ALL=C, org error: %s', e)
     os.environ['LC_ALL'] = 'C'
 
+
 import vcsserver
 from vcsserver import remote_wsgi, scm_app, settings, hgpatches
 from vcsserver.git_lfs.app import GIT_LFS_CONTENT_TYPE, GIT_LFS_PROTO_PAT
 from vcsserver.echo_stub import remote_wsgi as remote_wsgi_stub
 from vcsserver.echo_stub.echo_app import EchoApp
 from vcsserver.exceptions import HTTPRepoLocked, HTTPRepoBranchProtected
 from vcsserver.lib.exc_tracking import store_exception
 from vcsserver.server import VcsServer
 
 strict_vcs = True
 
 git_import_err = None
 try:
     from vcsserver.git import GitFactory, GitRemote
 except ImportError as e:
     GitFactory = None
     GitRemote = None
     git_import_err = e
     if strict_vcs:
         raise
 
 
 hg_import_err = None
 try:
     from vcsserver.hg import MercurialFactory, HgRemote
 except ImportError as e:
     MercurialFactory = None
     HgRemote = None
     hg_import_err = e
     if strict_vcs:
         raise
 
 
 svn_import_err = None
 try:
     from vcsserver.svn import SubversionFactory, SvnRemote
 except ImportError as e:
     SubversionFactory = None
     SvnRemote = None
     svn_import_err = e
     if strict_vcs:
         raise
 
 
 def _is_request_chunked(environ):
     stream = environ.get('HTTP_TRANSFER_ENCODING', '') == 'chunked'
     return stream
 
 
-def _int_setting(settings, name, default):
-    settings[name] = int(settings.get(name, default))
-    return settings[name]
-
-
-def _bool_setting(settings, name, default):
-    input_val = settings.get(name, default)
-    if isinstance(input_val, str):
-        input_val = input_val.encode('utf8')
-    settings[name] = asbool(input_val)
-    return settings[name]
-
-
-def _list_setting(settings, name, default):
-    raw_value = settings.get(name, default)
-
-    # Otherwise we assume it uses pyramids space/newline separation.
-    settings[name] = aslist(raw_value)
-    return settings[name]
-
-
-def _string_setting(settings, name, default, lower=True, default_when_empty=False):
-    value = settings.get(name, default)
-
-    if default_when_empty and not value:
-        # use default value when value is empty
-        value = default
-
-    if lower:
-        value = value.lower()
-    settings[name] = value
-    return settings[name]
+def log_max_fd():
+    try:
+        maxfd = psutil.Process().rlimit(psutil.RLIMIT_NOFILE)[1]
+        log.info('Max file descriptors value: %s', maxfd)
+    except Exception:
+        pass
 
 
 class VCS(object):
     def __init__(self, locale_conf=None, cache_config=None):
         self.locale = locale_conf
         self.cache_config = cache_config
         self._configure_locale()
 
+        log_max_fd()
+
         if GitFactory and GitRemote:
             git_factory = GitFactory()
             self._git_remote = GitRemote(git_factory)
         else:
             log.error("Git client import failed: %s", git_import_err)
 
         if MercurialFactory and HgRemote:
             hg_factory = MercurialFactory()
             self._hg_remote = HgRemote(hg_factory)
         else:
             log.error("Mercurial client import failed: %s", hg_import_err)
 
         if SubversionFactory and SvnRemote:
             svn_factory = SubversionFactory()
 
             # hg factory is used for svn url validation
             hg_factory = MercurialFactory()
             self._svn_remote = SvnRemote(svn_factory, hg_factory=hg_factory)
         else:
             log.error("Subversion client import failed: %s", svn_import_err)
 
         self._vcsserver = VcsServer()
 
     def _configure_locale(self):
         if self.locale:
             log.info('Settings locale: `LC_ALL` to %s', self.locale)
         else:
             log.info('Configuring locale subsystem based on environment variables')
         try:
             # If self.locale is the empty string, then the locale
             # module will use the environment variables. See the
             # documentation of the package `locale`.
             locale.setlocale(locale.LC_ALL, self.locale)
 
             language_code, encoding = locale.getlocale()
             log.info(
                 'Locale set to language code "%s" with encoding "%s".',
                 language_code, encoding)
         except locale.Error:
             log.exception('Cannot set locale, not configuring the locale system')
 
 
 class WsgiProxy(object):
     def __init__(self, wsgi):
         self.wsgi = wsgi
 
     def __call__(self, environ, start_response):
         input_data = environ['wsgi.input'].read()
         input_data = msgpack.unpackb(input_data)
 
         error = None
         try:
             data, status, headers = self.wsgi.handle(
                 input_data['environment'], input_data['input_data'],
                 *input_data['args'], **input_data['kwargs'])
         except Exception as e:
             data, status, headers = [], None, None
             error = {
                 'message': str(e),
                 '_vcs_kind': getattr(e, '_vcs_kind', None)
             }
 
         start_response(200, {})
         return self._iterator(error, status, headers, data)
 
     def _iterator(self, error, status, headers, data):
         initial_data = [
             error,
             status,
             headers,
         ]
 
         for d in chain(initial_data, data):
             yield msgpack.packb(d)
 
 
 def not_found(request):
     return {'status': '404 NOT FOUND'}
 
 
 class VCSViewPredicate(object):
     def __init__(self, val, config):
         self.remotes = val
 
     def text(self):
         return 'vcs view method = %s' % (list(self.remotes.keys()),)
 
     phash = text
 
     def __call__(self, context, request):
         """
         View predicate that returns true if given backend is supported by
         defined remotes.
         """
         backend = request.matchdict.get('backend')
         return backend in self.remotes
 
 
 class HTTPApplication(object):
     ALLOWED_EXCEPTIONS = ('KeyError', 'URLError')
 
     remote_wsgi = remote_wsgi
     _use_echo_app = False
 
     def __init__(self, settings=None, global_config=None):
-        self._sanitize_settings_and_apply_defaults(settings)
 
         self.config = Configurator(settings=settings)
+        # Init our statsd at very start
+        self.config.registry.statsd = StatsdClient.statsd
+
         self.global_config = global_config
         self.config.include('vcsserver.lib.rc_cache')
 
         settings_locale = settings.get('locale', '') or 'en_US.UTF-8'
         vcs = VCS(locale_conf=settings_locale, cache_config=settings)
         self._remotes = {
             'hg': vcs._hg_remote,
             'git': vcs._git_remote,
             'svn': vcs._svn_remote,
             'server': vcs._vcsserver,
         }
         if settings.get('dev.use_echo_app', 'false').lower() == 'true':
             self._use_echo_app = True
             log.warning("Using EchoApp for VCS operations.")
             self.remote_wsgi = remote_wsgi_stub
 
         self._configure_settings(global_config, settings)
+
         self._configure()
 
     def _configure_settings(self, global_config, app_settings):
         """
         Configure the settings module.
         """
         settings_merged = global_config.copy()
         settings_merged.update(app_settings)
 
         git_path = app_settings.get('git_path', None)
         if git_path:
             settings.GIT_EXECUTABLE = git_path
         binary_dir = app_settings.get('core.binary_dir', None)
         if binary_dir:
             settings.BINARY_DIR = binary_dir
 
         # Store the settings to make them available to other modules.
         vcsserver.PYRAMID_SETTINGS = settings_merged
         vcsserver.CONFIG = settings_merged
 
-    def _sanitize_settings_and_apply_defaults(self, settings):
-        temp_store = tempfile.gettempdir()
-        default_cache_dir = os.path.join(temp_store, 'rc_cache')
-
-        # save default, cache dir, and use it for all backends later.
-        default_cache_dir = _string_setting(
-            settings,
-            'cache_dir',
-            default_cache_dir, lower=False, default_when_empty=True)
-
-        # ensure we have our dir created
-        if not os.path.isdir(default_cache_dir):
-            os.makedirs(default_cache_dir, mode=0o755)
-
-        # exception store cache
-        _string_setting(
-            settings,
-            'exception_tracker.store_path',
-            temp_store, lower=False, default_when_empty=True)
-
-        # repo_object cache
-        _string_setting(
-            settings,
-            'rc_cache.repo_object.backend',
-            'dogpile.cache.rc.file_namespace', lower=False)
-        _int_setting(
-            settings,
-            'rc_cache.repo_object.expiration_time',
-            30 * 24 * 60 * 60)
-        _string_setting(
-            settings,
-            'rc_cache.repo_object.arguments.filename',
-            os.path.join(default_cache_dir, 'vcsserver_cache_1'), lower=False)
-
     def _configure(self):
         self.config.add_renderer(name='msgpack', factory=self._msgpack_renderer_factory)
 
         self.config.add_route('service', '/_service')
         self.config.add_route('status', '/status')
         self.config.add_route('hg_proxy', '/proxy/hg')
         self.config.add_route('git_proxy', '/proxy/git')
 
         # rpc methods
         self.config.add_route('vcs', '/{backend}')
 
         # streaming rpc remote methods
         self.config.add_route('vcs_stream', '/{backend}/stream')
 
         # vcs operations clone/push as streaming
         self.config.add_route('stream_git', '/stream/git/*repo_name')
         self.config.add_route('stream_hg', '/stream/hg/*repo_name')
 
         self.config.add_view(self.status_view, route_name='status', renderer='json')
         self.config.add_view(self.service_view, route_name='service', renderer='msgpack')
 
         self.config.add_view(self.hg_proxy(), route_name='hg_proxy')
         self.config.add_view(self.git_proxy(), route_name='git_proxy')
         self.config.add_view(self.vcs_view, route_name='vcs', renderer='msgpack',
                              vcs_view=self._remotes)
         self.config.add_view(self.vcs_stream_view, route_name='vcs_stream',
                              vcs_view=self._remotes)
 
         self.config.add_view(self.hg_stream(), route_name='stream_hg')
         self.config.add_view(self.git_stream(), route_name='stream_git')
 
         self.config.add_view_predicate('vcs_view', VCSViewPredicate)
 
         self.config.add_notfound_view(not_found, renderer='json')
 
         self.config.add_view(self.handle_vcs_exception, context=Exception)
 
         self.config.add_tween(
             'vcsserver.tweens.request_wrapper.RequestWrapperTween',
         )
         self.config.add_request_method(
             'vcsserver.lib.request_counter.get_request_counter',
             'request_count')
 
     def wsgi_app(self):
         return self.config.make_wsgi_app()
 
     def _vcs_view_params(self, request):
         remote = self._remotes[request.matchdict['backend']]
         payload = msgpack.unpackb(request.body, use_list=True)
         method = payload.get('method')
         params = payload['params']
         wire = params.get('wire')
         args = params.get('args')
         kwargs = params.get('kwargs')
         context_uid = None
 
         if wire:
             try:
                 wire['context'] = context_uid = uuid.UUID(wire['context'])
             except KeyError:
                 pass
             args.insert(0, wire)
         repo_state_uid = wire.get('repo_state_uid') if wire else None
 
         # NOTE(marcink): trading complexity for slight performance
         if log.isEnabledFor(logging.DEBUG):
             no_args_methods = [
-                'archive_repo'
+
             ]
             if method in no_args_methods:
                 call_args = ''
             else:
                 call_args = args[1:]
 
-            log.debug('method requested:%s with args:%s kwargs:%s context_uid: %s, repo_state_uid:%s',
+            log.debug('Method requested:`%s` with args:%s kwargs:%s context_uid: %s, repo_state_uid:%s',
                       method, call_args, kwargs, context_uid, repo_state_uid)
 
+        statsd = request.registry.statsd
+        if statsd:
+            statsd.incr(
+                'vcsserver_method_total', tags=[
+                    "method:{}".format(method),
+                ])
         return payload, remote, method, args, kwargs
 
     def vcs_view(self, request):
 
         payload, remote, method, args, kwargs = self._vcs_view_params(request)
         payload_id = payload.get('id')
 
         try:
             resp = getattr(remote, method)(*args, **kwargs)
         except Exception as e:
             exc_info = list(sys.exc_info())
             exc_type, exc_value, exc_traceback = exc_info
 
             org_exc = getattr(e, '_org_exc', None)
             org_exc_name = None
             org_exc_tb = ''
             if org_exc:
                 org_exc_name = org_exc.__class__.__name__
                 org_exc_tb = getattr(e, '_org_exc_tb', '')
                 # replace our "faked" exception with our org
                 exc_info[0] = org_exc.__class__
                 exc_info[1] = org_exc
 
             should_store_exc = True
             if org_exc:
                 def get_exc_fqn(_exc_obj):
                     module_name = getattr(org_exc.__class__, '__module__', 'UNKNOWN')
                     return module_name + '.' + org_exc_name
 
                 exc_fqn = get_exc_fqn(org_exc)
 
                 if exc_fqn in ['mercurial.error.RepoLookupError',
                                'vcsserver.exceptions.RefNotFoundException']:
                     should_store_exc = False
 
             if should_store_exc:
-                store_exception(id(exc_info), exc_info)
+                store_exception(id(exc_info), exc_info, request_path=request.path)
 
             tb_info = ''.join(
                 traceback.format_exception(exc_type, exc_value, exc_traceback))
 
             type_ = e.__class__.__name__
             if type_ not in self.ALLOWED_EXCEPTIONS:
                 type_ = None
 
             resp = {
                 'id': payload_id,
                 'error': {
                     'message': e.message,
                     'traceback': tb_info,
                     'org_exc': org_exc_name,
                     'org_exc_tb': org_exc_tb,
                     'type': type_
                 }
             }
+
             try:
                 resp['error']['_vcs_kind'] = getattr(e, '_vcs_kind', None)
             except AttributeError:
                 pass
         else:
             resp = {
                 'id': payload_id,
                 'result': resp
             }
 
         return resp
 
     def vcs_stream_view(self, request):
         payload, remote, method, args, kwargs = self._vcs_view_params(request)
         # this method has a stream: marker we remove it here
         method = method.split('stream:')[-1]
         chunk_size = safe_int(payload.get('chunk_size')) or 4096
 
         try:
             resp = getattr(remote, method)(*args, **kwargs)
         except Exception as e:
             raise
 
         def get_chunked_data(method_resp):
             stream = StringIO(method_resp)
             while 1:
                 chunk = stream.read(chunk_size)
                 if not chunk:
                     break
                 yield chunk
 
         response = Response(app_iter=get_chunked_data(resp))
         response.content_type = 'application/octet-stream'
 
         return response
 
     def status_view(self, request):
         import vcsserver
         return {'status': 'OK', 'vcsserver_version': vcsserver.__version__,
                 'pid': os.getpid()}
 
     def service_view(self, request):
         import vcsserver
 
         payload = msgpack.unpackb(request.body, use_list=True)
         server_config, app_config = {}, {}
 
         try:
             path = self.global_config['__file__']
             config = configparser.RawConfigParser()
 
             config.read(path)
 
             if config.has_section('server:main'):
                 server_config = dict(config.items('server:main'))
             if config.has_section('app:main'):
                 app_config = dict(config.items('app:main'))
 
         except Exception:
             log.exception('Failed to read .ini file for display')
 
         environ = list(os.environ.items())
 
         resp = {
             'id': payload.get('id'),
             'result': dict(
                 version=vcsserver.__version__,
                 config=server_config,
                 app_config=app_config,
                 environ=environ,
                 payload=payload,
             )
         }
         return resp
 
     def _msgpack_renderer_factory(self, info):
         def _render(value, system):
             request = system.get('request')
             if request is not None:
                 response = request.response
                 ct = response.content_type
                 if ct == response.default_content_type:
                     response.content_type = 'application/x-msgpack'
             return msgpack.packb(value)
         return _render
 
     def set_env_from_config(self, environ, config):
         dict_conf = {}
         try:
             for elem in config:
                 if elem[0] == 'rhodecode':
                     dict_conf = json.loads(elem[2])
                     break
         except Exception:
             log.exception('Failed to fetch SCM CONFIG')
             return
 
         username = dict_conf.get('username')
         if username:
             environ['REMOTE_USER'] = username
             # mercurial specific, some extension api rely on this
             environ['HGUSER'] = username
 
         ip = dict_conf.get('ip')
         if ip:
             environ['REMOTE_HOST'] = ip
 
         if _is_request_chunked(environ):
             # set the compatibility flag for webob
             environ['wsgi.input_terminated'] = True
 
     def hg_proxy(self):
         @wsgiapp
         def _hg_proxy(environ, start_response):
             app = WsgiProxy(self.remote_wsgi.HgRemoteWsgi())
             return app(environ, start_response)
         return _hg_proxy
 
     def git_proxy(self):
         @wsgiapp
         def _git_proxy(environ, start_response):
             app = WsgiProxy(self.remote_wsgi.GitRemoteWsgi())
             return app(environ, start_response)
         return _git_proxy
 
     def hg_stream(self):
         if self._use_echo_app:
             @wsgiapp
             def _hg_stream(environ, start_response):
                 app = EchoApp('fake_path', 'fake_name', None)
                 return app(environ, start_response)
             return _hg_stream
         else:
             @wsgiapp
             def _hg_stream(environ, start_response):
                 log.debug('http-app: handling hg stream')
                 repo_path = environ['HTTP_X_RC_REPO_PATH']
                 repo_name = environ['HTTP_X_RC_REPO_NAME']
                 packed_config = base64.b64decode(
                     environ['HTTP_X_RC_REPO_CONFIG'])
                 config = msgpack.unpackb(packed_config)
                 app = scm_app.create_hg_wsgi_app(
                     repo_path, repo_name, config)
 
                 # Consistent path information for hgweb
                 environ['PATH_INFO'] = environ['HTTP_X_RC_PATH_INFO']
                 environ['REPO_NAME'] = repo_name
                 self.set_env_from_config(environ, config)
 
                 log.debug('http-app: starting app handler '
                           'with %s and process request', app)
                 return app(environ, ResponseFilter(start_response))
             return _hg_stream
 
     def git_stream(self):
         if self._use_echo_app:
             @wsgiapp
             def _git_stream(environ, start_response):
                 app = EchoApp('fake_path', 'fake_name', None)
                 return app(environ, start_response)
             return _git_stream
         else:
             @wsgiapp
             def _git_stream(environ, start_response):
                 log.debug('http-app: handling git stream')
                 repo_path = environ['HTTP_X_RC_REPO_PATH']
                 repo_name = environ['HTTP_X_RC_REPO_NAME']
                 packed_config = base64.b64decode(
                     environ['HTTP_X_RC_REPO_CONFIG'])
                 config = msgpack.unpackb(packed_config)
 
                 environ['PATH_INFO'] = environ['HTTP_X_RC_PATH_INFO']
                 self.set_env_from_config(environ, config)
 
                 content_type = environ.get('CONTENT_TYPE', '')
 
                 path = environ['PATH_INFO']
                 is_lfs_request = GIT_LFS_CONTENT_TYPE in content_type
                 log.debug(
                     'LFS: Detecting if request `%s` is LFS server path based '
                     'on content type:`%s`, is_lfs:%s',
                     path, content_type, is_lfs_request)
 
                 if not is_lfs_request:
                     # fallback detection by path
                     if GIT_LFS_PROTO_PAT.match(path):
                         is_lfs_request = True
                     log.debug(
                         'LFS: fallback detection by path of: `%s`, is_lfs:%s',
                         path, is_lfs_request)
 
                 if is_lfs_request:
                     app = scm_app.create_git_lfs_wsgi_app(
                         repo_path, repo_name, config)
                 else:
                     app = scm_app.create_git_wsgi_app(
                         repo_path, repo_name, config)
 
                 log.debug('http-app: starting app handler '
                           'with %s and process request', app)
 
                 return app(environ, start_response)
 
             return _git_stream
 
     def handle_vcs_exception(self, exception, request):
         _vcs_kind = getattr(exception, '_vcs_kind', '')
         if _vcs_kind == 'repo_locked':
             # Get custom repo-locked status code if present.
             status_code = request.headers.get('X-RC-Locked-Status-Code')
             return HTTPRepoLocked(
                 title=exception.message, status_code=status_code)
 
         elif _vcs_kind == 'repo_branch_protected':
             # Get custom repo-branch-protected status code if present.
             return HTTPRepoBranchProtected(title=exception.message)
 
         exc_info = request.exc_info
         store_exception(id(exc_info), exc_info)
 
         traceback_info = 'unavailable'
         if request.exc_info:
             exc_type, exc_value, exc_tb = request.exc_info
             traceback_info = ''.join(traceback.format_exception(exc_type, exc_value, exc_tb))
 
         log.error(
             'error occurred handling this request for path: %s, \n tb: %s',
             request.path, traceback_info)
+
+        statsd = request.registry.statsd
+        if statsd:
+            exc_type = "{}.{}".format(exception.__class__.__module__, exception.__class__.__name__)
+            statsd.incr('vcsserver_exception_total',
+                        tags=["type:{}".format(exc_type)])
         raise exception
 
 
 class ResponseFilter(object):
 
     def __init__(self, start_response):
         self._start_response = start_response
 
     def __call__(self, status, response_headers, exc_info=None):
         headers = tuple(
             (h, v) for h, v in response_headers
             if not wsgiref.util.is_hop_by_hop(h))
         return self._start_response(status, headers, exc_info)
 
 
+def sanitize_settings_and_apply_defaults(global_config, settings):
+    global_settings_maker = SettingsMaker(global_config)
+    settings_maker = SettingsMaker(settings)
+
+    settings_maker.make_setting('logging.autoconfigure', False, parser='bool')
+
+    logging_conf = os.path.join(os.path.dirname(global_config.get('__file__')), 'logging.ini')
+    settings_maker.enable_logging(logging_conf)
+
+    # Default includes, possible to change as a user
+    pyramid_includes = settings_maker.make_setting('pyramid.includes', [], parser='list:newline')
+    log.debug("Using the following pyramid.includes: %s", pyramid_includes)
+
+    settings_maker.make_setting('__file__', global_config.get('__file__'))
+
+    settings_maker.make_setting('pyramid.default_locale_name', 'en')
+    settings_maker.make_setting('locale', 'en_US.UTF-8')
+
+    settings_maker.make_setting('core.binary_dir', '')
+
+    temp_store = tempfile.gettempdir()
+    default_cache_dir = os.path.join(temp_store, 'rc_cache')
+    # save default, cache dir, and use it for all backends later.
+    default_cache_dir = settings_maker.make_setting(
+        'cache_dir',
+        default=default_cache_dir, default_when_empty=True,
+        parser='dir:ensured')
+
+    # exception store cache
+    settings_maker.make_setting(
+        'exception_tracker.store_path',
+        default=os.path.join(default_cache_dir, 'exc_store'), default_when_empty=True,
+        parser='dir:ensured'
+    )
+
+    # repo_object cache defaults
+    settings_maker.make_setting(
+        'rc_cache.repo_object.backend',
+        default='dogpile.cache.rc.file_namespace',
+        parser='string')
+    settings_maker.make_setting(
+        'rc_cache.repo_object.expiration_time',
+        default=30 * 24 * 60 * 60,  # 30days
+        parser='int')
+    settings_maker.make_setting(
+        'rc_cache.repo_object.arguments.filename',
+        default=os.path.join(default_cache_dir, 'vcsserver_cache_repo_object.db'),
+        parser='string')
+
+    # statsd
+    settings_maker.make_setting('statsd.enabled', False, parser='bool')
+    settings_maker.make_setting('statsd.statsd_host', 'statsd-exporter', parser='string')
+    settings_maker.make_setting('statsd.statsd_port', 9125, parser='int')
+    settings_maker.make_setting('statsd.statsd_prefix', '')
+    settings_maker.make_setting('statsd.statsd_ipv6', False, parser='bool')
+
+    settings_maker.env_expand()
+
+
 def main(global_config, **settings):
+    start_time = time.time()
+    log.info('Pyramid app config starting')
+
     if MercurialFactory:
         hgpatches.patch_largefiles_capabilities()
         hgpatches.patch_subrepo_type_mapping()
 
-    app = HTTPApplication(settings=settings, global_config=global_config)
-    return app.wsgi_app()
+    # Fill in and sanitize the defaults & do ENV expansion
+    sanitize_settings_and_apply_defaults(global_config, settings)
+
+    # init and bootstrap StatsdClient
+    StatsdClient.setup(settings)
+
+    pyramid_app = HTTPApplication(settings=settings, global_config=global_config).wsgi_app()
+    total_time = time.time() - start_time
+    log.info('Pyramid app `%s` created and configured in %.2fs',
+             getattr(pyramid_app, 'func_name', 'pyramid_app'), total_time)
+    return pyramid_app
+
+

vcsserver/lib/exc_tracking.py

@@ -1,167 +1,173 b''
 # -*- coding: utf-8 -*-
 
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 
 import os
 import time
 import datetime
 import msgpack
 import logging
 import traceback
 import tempfile
 
 log = logging.getLogger(__name__)
 
 # NOTE: Any changes should be synced with exc_tracking at rhodecode.lib.exc_tracking
 global_prefix = 'vcsserver'
 exc_store_dir_name = 'rc_exception_store_v1'
 
 
 def exc_serialize(exc_id, tb, exc_type):
 
     data = {
         'version': 'v1',
         'exc_id': exc_id,
         'exc_utc_date': datetime.datetime.utcnow().isoformat(),
         'exc_timestamp': repr(time.time()),
         'exc_message': tb,
         'exc_type': exc_type,
     }
     return msgpack.packb(data), data
 
 
 def exc_unserialize(tb):
     return msgpack.unpackb(tb)
 
 
 def get_exc_store():
     """
     Get and create exception store if it's not existing
     """
     import vcsserver as app
 
     exc_store_dir = app.CONFIG.get('exception_tracker.store_path', '') or tempfile.gettempdir()
     _exc_store_path = os.path.join(exc_store_dir, exc_store_dir_name)
 
     _exc_store_path = os.path.abspath(_exc_store_path)
     if not os.path.isdir(_exc_store_path):
         os.makedirs(_exc_store_path)
         log.debug('Initializing exceptions store at %s', _exc_store_path)
     return _exc_store_path
 
 
-def _store_exception(exc_id, exc_info, prefix):
+def _store_exception(exc_id, exc_info, prefix, request_path=''):
     exc_type, exc_value, exc_traceback = exc_info
 
     tb = ''.join(traceback.format_exception(
         exc_type, exc_value, exc_traceback, None))
 
     detailed_tb = getattr(exc_value, '_org_exc_tb', None)
 
     if detailed_tb:
         if isinstance(detailed_tb, str):
             remote_tb = [detailed_tb]
 
         tb += (
             '\n+++ BEG SOURCE EXCEPTION +++\n\n'
             '{}\n'
             '+++ END SOURCE EXCEPTION +++\n'
             ''.format('\n'.join(remote_tb))
         )
 
         # Avoid that remote_tb also appears in the frame
         del remote_tb
 
     exc_type_name = exc_type.__name__
     exc_store_path = get_exc_store()
     exc_data, org_data = exc_serialize(exc_id, tb, exc_type_name)
     exc_pref_id = '{}_{}_{}'.format(exc_id, prefix, org_data['exc_timestamp'])
     if not os.path.isdir(exc_store_path):
         os.makedirs(exc_store_path)
     stored_exc_path = os.path.join(exc_store_path, exc_pref_id)
     with open(stored_exc_path, 'wb') as f:
         f.write(exc_data)
     log.debug('Stored generated exception %s as: %s', exc_id, stored_exc_path)
 
+    log.error(
+        'error occurred handling this request.\n'
+        'Path: `%s`, tb: %s',
+        request_path, tb)
 
-def store_exception(exc_id, exc_info, prefix=global_prefix):
+
+def store_exception(exc_id, exc_info, prefix=global_prefix, request_path=''):
     """
     Example usage::
 
         exc_info = sys.exc_info()
         store_exception(id(exc_info), exc_info)
     """
 
     try:
-        _store_exception(exc_id=exc_id, exc_info=exc_info, prefix=prefix)
+        _store_exception(exc_id=exc_id, exc_info=exc_info, prefix=prefix,
+                         request_path=request_path)
     except Exception:
         log.exception('Failed to store exception `%s` information', exc_id)
         # there's no way this can fail, it will crash server badly if it does.
         pass
 
 
 def _find_exc_file(exc_id, prefix=global_prefix):
     exc_store_path = get_exc_store()
     if prefix:
         exc_id = '{}_{}'.format(exc_id, prefix)
     else:
         # search without a prefix
         exc_id = '{}'.format(exc_id)
 
     # we need to search the store for such start pattern as above
     for fname in os.listdir(exc_store_path):
         if fname.startswith(exc_id):
             exc_id = os.path.join(exc_store_path, fname)
             break
         continue
     else:
         exc_id = None
 
     return exc_id
 
 
 def _read_exception(exc_id, prefix):
     exc_id_file_path = _find_exc_file(exc_id=exc_id, prefix=prefix)
     if exc_id_file_path:
         with open(exc_id_file_path, 'rb') as f:
             return exc_unserialize(f.read())
     else:
         log.debug('Exception File `%s` not found', exc_id_file_path)
     return None
 
 
 def read_exception(exc_id, prefix=global_prefix):
     try:
         return _read_exception(exc_id=exc_id, prefix=prefix)
     except Exception:
         log.exception('Failed to read exception `%s` information', exc_id)
         # there's no way this can fail, it will crash server badly if it does.
     return None
 
 
 def delete_exception(exc_id, prefix=global_prefix):
     try:
         exc_id_file_path = _find_exc_file(exc_id, prefix=prefix)
         if exc_id_file_path:
             os.remove(exc_id_file_path)
 
     except Exception:
         log.exception('Failed to remove exception `%s` information', exc_id)
         # there's no way this can fail, it will crash server badly if it does.
         pass

vcsserver/lib/rc_cache/__init__.py

@@ -1,72 +1,79 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import logging
 from dogpile.cache import register_backend
 
 register_backend(
     "dogpile.cache.rc.memory_lru", "vcsserver.lib.rc_cache.backends",
     "LRUMemoryBackend")
 
 register_backend(
     "dogpile.cache.rc.file_namespace", "vcsserver.lib.rc_cache.backends",
     "FileNamespaceBackend")
 
 register_backend(
     "dogpile.cache.rc.redis", "vcsserver.lib.rc_cache.backends",
     "RedisPickleBackend")
 
 register_backend(
     "dogpile.cache.rc.redis_msgpack", "vcsserver.lib.rc_cache.backends",
     "RedisMsgPackBackend")
 
 
 log = logging.getLogger(__name__)
 
 from . import region_meta
-from .utils import (get_default_cache_settings, backend_key_generator, make_region)
+from .utils import (
+    get_default_cache_settings, backend_key_generator, get_or_create_region,
+    clear_cache_namespace, make_region)
 
 
 def configure_dogpile_cache(settings):
     cache_dir = settings.get('cache_dir')
     if cache_dir:
         region_meta.dogpile_config_defaults['cache_dir'] = cache_dir
 
     rc_cache_data = get_default_cache_settings(settings, prefixes=['rc_cache.'])
 
     # inspect available namespaces
     avail_regions = set()
     for key in rc_cache_data.keys():
         namespace_name = key.split('.', 1)[0]
-        avail_regions.add(namespace_name)
-    log.debug('dogpile: found following cache regions: %s', avail_regions)
+        if namespace_name in avail_regions:
+            continue
 
-    # register them into namespace
-    for region_name in avail_regions:
+        avail_regions.add(namespace_name)
+        log.debug('dogpile: found following cache regions: %s', namespace_name)
+
         new_region = make_region(
-            name=region_name,
+            name=namespace_name,
             function_key_generator=None
         )
 
-        new_region.configure_from_config(settings, 'rc_cache.{}.'.format(region_name))
+        new_region.configure_from_config(settings, 'rc_cache.{}.'.format(namespace_name))
         new_region.function_key_generator = backend_key_generator(new_region.actual_backend)
-        log.debug('dogpile: registering a new region %s[%s]', region_name, new_region.__dict__)
-        region_meta.dogpile_cache_regions[region_name] = new_region
+        if log.isEnabledFor(logging.DEBUG):
+            region_args = dict(backend=new_region.actual_backend.__class__,
+                               region_invalidator=new_region.region_invalidator.__class__)
+            log.debug('dogpile: registering a new region `%s` %s', namespace_name, region_args)
+
+        region_meta.dogpile_cache_regions[namespace_name] = new_region
 
 
 def includeme(config):
     configure_dogpile_cache(config.registry.settings)

vcsserver/lib/rc_cache/backends.py

@@ -1,253 +1,329 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import time
 import errno
 import logging
 
 import msgpack
 import redis
 
 from dogpile.cache.api import CachedValue
 from dogpile.cache.backends import memory as memory_backend
 from dogpile.cache.backends import file as file_backend
 from dogpile.cache.backends import redis as redis_backend
 from dogpile.cache.backends.file import NO_VALUE, FileLock
 from dogpile.cache.util import memoized_property
 
+from pyramid.settings import asbool
+
 from vcsserver.lib.memory_lru_dict import LRUDict, LRUDictDebug
+from vcsserver.utils import safe_str, safe_unicode
 
 
 _default_max_size = 1024
 
 log = logging.getLogger(__name__)
 
 
 class LRUMemoryBackend(memory_backend.MemoryBackend):
     key_prefix = 'lru_mem_backend'
     pickle_values = False
 
     def __init__(self, arguments):
         max_size = arguments.pop('max_size', _default_max_size)
 
         LRUDictClass = LRUDict
         if arguments.pop('log_key_count', None):
             LRUDictClass = LRUDictDebug
 
         arguments['cache_dict'] = LRUDictClass(max_size)
         super(LRUMemoryBackend, self).__init__(arguments)
 
     def delete(self, key):
         try:
             del self._cache[key]
         except KeyError:
             # we don't care if key isn't there at deletion
             pass
 
     def delete_multi(self, keys):
         for key in keys:
             self.delete(key)
 
 
 class PickleSerializer(object):
 
     def _dumps(self, value, safe=False):
         try:
             return pickle.dumps(value)
         except Exception:
             if safe:
                 return NO_VALUE
             else:
                 raise
 
     def _loads(self, value, safe=True):
         try:
             return pickle.loads(value)
         except Exception:
             if safe:
                 return NO_VALUE
             else:
                 raise
 
 
 class MsgPackSerializer(object):
 
     def _dumps(self, value, safe=False):
         try:
             return msgpack.packb(value)
         except Exception:
             if safe:
                 return NO_VALUE
             else:
                 raise
 
     def _loads(self, value, safe=True):
         """
         pickle maintained the `CachedValue` wrapper of the tuple
         msgpack does not, so it must be added back in.
        """
         try:
             value = msgpack.unpackb(value, use_list=False)
             return CachedValue(*value)
         except Exception:
             if safe:
                 return NO_VALUE
             else:
                 raise
 
 
 import fcntl
 flock_org = fcntl.flock
 
 
 class CustomLockFactory(FileLock):
 
     pass
 
 
 class FileNamespaceBackend(PickleSerializer, file_backend.DBMBackend):
     key_prefix = 'file_backend'
 
     def __init__(self, arguments):
         arguments['lock_factory'] = CustomLockFactory
-        super(FileNamespaceBackend, self).__init__(arguments)
+        db_file = arguments.get('filename')
+
+        log.debug('initialing %s DB in %s', self.__class__.__name__, db_file)
+        try:
+            super(FileNamespaceBackend, self).__init__(arguments)
+        except Exception:
+            log.exception('Failed to initialize db at: %s', db_file)
+            raise
 
     def __repr__(self):
         return '{} `{}`'.format(self.__class__, self.filename)
 
     def list_keys(self, prefix=''):
         prefix = '{}:{}'.format(self.key_prefix, prefix)
 
         def cond(v):
             if not prefix:
                 return True
 
             if v.startswith(prefix):
                 return True
             return False
 
         with self._dbm_file(True) as dbm:
-
-            return filter(cond, dbm.keys())
+            try:
+                return filter(cond, dbm.keys())
+            except Exception:
+                log.error('Failed to fetch DBM keys from DB: %s', self.get_store())
+                raise
 
     def get_store(self):
         return self.filename
 
-    def get(self, key):
+    def _dbm_get(self, key):
         with self._dbm_file(False) as dbm:
             if hasattr(dbm, 'get'):
                 value = dbm.get(key, NO_VALUE)
             else:
                 # gdbm objects lack a .get method
                 try:
                     value = dbm[key]
                 except KeyError:
                     value = NO_VALUE
             if value is not NO_VALUE:
                 value = self._loads(value)
             return value
 
+    def get(self, key):
+        try:
+            return self._dbm_get(key)
+        except Exception:
+            log.error('Failed to fetch DBM key %s from DB: %s', key, self.get_store())
+            raise
+
     def set(self, key, value):
         with self._dbm_file(True) as dbm:
             dbm[key] = self._dumps(value)
 
     def set_multi(self, mapping):
         with self._dbm_file(True) as dbm:
             for key, value in mapping.items():
                 dbm[key] = self._dumps(value)
 
 
 class BaseRedisBackend(redis_backend.RedisBackend):
+    key_prefix = ''
+
+    def __init__(self, arguments):
+        super(BaseRedisBackend, self).__init__(arguments)
+        self._lock_timeout = self.lock_timeout
+        self._lock_auto_renewal = asbool(arguments.pop("lock_auto_renewal", True))
+
+        if self._lock_auto_renewal and not self._lock_timeout:
+            # set default timeout for auto_renewal
+            self._lock_timeout = 30
 
     def _create_client(self):
         args = {}
 
         if self.url is not None:
             args.update(url=self.url)
 
         else:
             args.update(
                 host=self.host, password=self.password,
                 port=self.port, db=self.db
             )
 
         connection_pool = redis.ConnectionPool(**args)
 
         return redis.StrictRedis(connection_pool=connection_pool)
 
     def list_keys(self, prefix=''):
         prefix = '{}:{}*'.format(self.key_prefix, prefix)
         return self.client.keys(prefix)
 
     def get_store(self):
         return self.client.connection_pool
 
     def get(self, key):
         value = self.client.get(key)
         if value is None:
             return NO_VALUE
         return self._loads(value)
 
     def get_multi(self, keys):
         if not keys:
             return []
         values = self.client.mget(keys)
         loads = self._loads
         return [
             loads(v) if v is not None else NO_VALUE
             for v in values]
 
     def set(self, key, value):
         if self.redis_expiration_time:
             self.client.setex(key, self.redis_expiration_time,
                               self._dumps(value))
         else:
             self.client.set(key, self._dumps(value))
 
     def set_multi(self, mapping):
         dumps = self._dumps
         mapping = dict(
             (k, dumps(v))
             for k, v in mapping.items()
         )
 
         if not self.redis_expiration_time:
             self.client.mset(mapping)
         else:
             pipe = self.client.pipeline()
             for key, value in mapping.items():
                 pipe.setex(key, self.redis_expiration_time, value)
             pipe.execute()
 
     def get_mutex(self, key):
-        u = redis_backend.u
         if self.distributed_lock:
-            lock_key = u('_lock_{0}').format(key)
-            log.debug('Trying to acquire Redis lock for key %s', lock_key)
-            return self.client.lock(lock_key, self.lock_timeout, self.lock_sleep)
+            lock_key = u'_lock_{0}'.format(safe_unicode(key))
+            return get_mutex_lock(self.client, lock_key, self._lock_timeout,
+                                  auto_renewal=self._lock_auto_renewal)
         else:
             return None
 
 
 class RedisPickleBackend(PickleSerializer, BaseRedisBackend):
     key_prefix = 'redis_pickle_backend'
     pass
 
 
 class RedisMsgPackBackend(MsgPackSerializer, BaseRedisBackend):
     key_prefix = 'redis_msgpack_backend'
     pass
+
+
+def get_mutex_lock(client, lock_key, lock_timeout, auto_renewal=False):
+    import redis_lock
+
+    class _RedisLockWrapper(object):
+        """LockWrapper for redis_lock"""
+
+        @classmethod
+        def get_lock(cls):
+            return redis_lock.Lock(
+                redis_client=client,
+                name=lock_key,
+                expire=lock_timeout,
+                auto_renewal=auto_renewal,
+                strict=True,
+            )
+
+        def __repr__(self):
+            return "{}:{}".format(self.__class__.__name__, lock_key)
+
+        def __str__(self):
+            return "{}:{}".format(self.__class__.__name__, lock_key)
+
+        def __init__(self):
+            self.lock = self.get_lock()
+            self.lock_key = lock_key
+
+        def acquire(self, wait=True):
+            log.debug('Trying to acquire Redis lock for key %s', self.lock_key)
+            try:
+                acquired = self.lock.acquire(wait)
+                log.debug('Got lock for key %s, %s', self.lock_key, acquired)
+                return acquired
+            except redis_lock.AlreadyAcquired:
+                return False
+            except redis_lock.AlreadyStarted:
+                # refresh thread exists, but it also means we acquired the lock
+                return True
+
+        def release(self):
+            try:
+                self.lock.release()
+            except redis_lock.NotAcquired:
+                pass
+
+    return _RedisLockWrapper()

vcsserver/lib/rc_cache/utils.py

@@ -1,153 +1,263 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import os
+import time
 import logging
 import functools
-from decorator import decorate
 
 from dogpile.cache import CacheRegion
 from dogpile.cache.util import compat
 
 from vcsserver.utils import safe_str, sha1
 
+from vcsserver.lib.rc_cache import region_meta
 
 log = logging.getLogger(__name__)
 
 
 class RhodeCodeCacheRegion(CacheRegion):
 
     def conditional_cache_on_arguments(
             self, namespace=None,
             expiration_time=None,
             should_cache_fn=None,
             to_str=compat.string_type,
             function_key_generator=None,
             condition=True):
         """
         Custom conditional decorator, that will not touch any dogpile internals if
         condition isn't meet. This works a bit different than should_cache_fn
         And it's faster in cases we don't ever want to compute cached values
         """
         expiration_time_is_callable = compat.callable(expiration_time)
 
         if function_key_generator is None:
             function_key_generator = self.function_key_generator
 
+        # workaround for py2 and cython problems, this block should be removed
+        # once we've migrated to py3
+        if 'cython' == 'cython':
+            def decorator(fn):
+                if to_str is compat.string_type:
+                    # backwards compatible
+                    key_generator = function_key_generator(namespace, fn)
+                else:
+                    key_generator = function_key_generator(namespace, fn, to_str=to_str)
+
+                @functools.wraps(fn)
+                def decorate(*arg, **kw):
+                    key = key_generator(*arg, **kw)
+
+                    @functools.wraps(fn)
+                    def creator():
+                        return fn(*arg, **kw)
+
+                    if not condition:
+                        return creator()
+
+                    timeout = expiration_time() if expiration_time_is_callable \
+                        else expiration_time
+
+                    return self.get_or_create(key, creator, timeout, should_cache_fn)
+
+                def invalidate(*arg, **kw):
+                    key = key_generator(*arg, **kw)
+                    self.delete(key)
+
+                def set_(value, *arg, **kw):
+                    key = key_generator(*arg, **kw)
+                    self.set(key, value)
+
+                def get(*arg, **kw):
+                    key = key_generator(*arg, **kw)
+                    return self.get(key)
+
+                def refresh(*arg, **kw):
+                    key = key_generator(*arg, **kw)
+                    value = fn(*arg, **kw)
+                    self.set(key, value)
+                    return value
+
+                decorate.set = set_
+                decorate.invalidate = invalidate
+                decorate.refresh = refresh
+                decorate.get = get
+                decorate.original = fn
+                decorate.key_generator = key_generator
+                decorate.__wrapped__ = fn
+
+                return decorate
+            return decorator
+
         def get_or_create_for_user_func(key_generator, user_func, *arg, **kw):
 
             if not condition:
-                log.debug('Calling un-cached func:%s', user_func.func_name)
-                return user_func(*arg, **kw)
+                log.debug('Calling un-cached method:%s', user_func.func_name)
+                start = time.time()
+                result = user_func(*arg, **kw)
+                total = time.time() - start
+                log.debug('un-cached method:%s took %.4fs', user_func.func_name, total)
+                return result
 
             key = key_generator(*arg, **kw)
 
             timeout = expiration_time() if expiration_time_is_callable \
                 else expiration_time
 
-            log.debug('Calling cached fn:%s', user_func.func_name)
+            log.debug('Calling cached method:`%s`', user_func.func_name)
             return self.get_or_create(key, user_func, timeout, should_cache_fn, (arg, kw))
 
         def cache_decorator(user_func):
             if to_str is compat.string_type:
                 # backwards compatible
                 key_generator = function_key_generator(namespace, user_func)
             else:
                 key_generator = function_key_generator(namespace, user_func, to_str=to_str)
 
             def refresh(*arg, **kw):
                 """
                 Like invalidate, but regenerates the value instead
                 """
                 key = key_generator(*arg, **kw)
                 value = user_func(*arg, **kw)
                 self.set(key, value)
                 return value
 
             def invalidate(*arg, **kw):
                 key = key_generator(*arg, **kw)
                 self.delete(key)
 
             def set_(value, *arg, **kw):
                 key = key_generator(*arg, **kw)
                 self.set(key, value)
 
             def get(*arg, **kw):
                 key = key_generator(*arg, **kw)
                 return self.get(key)
 
             user_func.set = set_
             user_func.invalidate = invalidate
             user_func.get = get
             user_func.refresh = refresh
             user_func.key_generator = key_generator
             user_func.original = user_func
 
             # Use `decorate` to preserve the signature of :param:`user_func`.
-
-            return decorate(user_func, functools.partial(
+            return decorator.decorate(user_func, functools.partial(
                 get_or_create_for_user_func, key_generator))
 
         return cache_decorator
 
 
 def make_region(*arg, **kw):
     return RhodeCodeCacheRegion(*arg, **kw)
 
 
 def get_default_cache_settings(settings, prefixes=None):
     prefixes = prefixes or []
     cache_settings = {}
     for key in settings.keys():
         for prefix in prefixes:
             if key.startswith(prefix):
                 name = key.split(prefix)[1].strip()
                 val = settings[key]
                 if isinstance(val, compat.string_types):
                     val = val.strip()
                 cache_settings[name] = val
     return cache_settings
 
 
 def compute_key_from_params(*args):
     """
     Helper to compute key from given params to be used in cache manager
     """
     return sha1("_".join(map(safe_str, args)))
 
 
 def backend_key_generator(backend):
     """
     Special wrapper that also sends over the backend to the key generator
     """
     def wrapper(namespace, fn):
         return key_generator(backend, namespace, fn)
     return wrapper
 
 
 def key_generator(backend, namespace, fn):
     fname = fn.__name__
 
     def generate_key(*args):
         backend_prefix = getattr(backend, 'key_prefix', None) or 'backend_prefix'
         namespace_pref = namespace or 'default_namespace'
         arg_key = compute_key_from_params(*args)
         final_key = "{}:{}:{}_{}".format(backend_prefix, namespace_pref, fname, arg_key)
 
         return final_key
 
     return generate_key
+
+
+def get_or_create_region(region_name, region_namespace=None):
+    from vcsserver.lib.rc_cache.backends import FileNamespaceBackend
+    region_obj = region_meta.dogpile_cache_regions.get(region_name)
+    if not region_obj:
+        raise EnvironmentError(
+            'Region `{}` not in configured: {}.'.format(
+                region_name, region_meta.dogpile_cache_regions.keys()))
+
+    region_uid_name = '{}:{}'.format(region_name, region_namespace)
+    if isinstance(region_obj.actual_backend, FileNamespaceBackend):
+        region_exist = region_meta.dogpile_cache_regions.get(region_namespace)
+        if region_exist:
+            log.debug('Using already configured region: %s', region_namespace)
+            return region_exist
+        cache_dir = region_meta.dogpile_config_defaults['cache_dir']
+        expiration_time = region_obj.expiration_time
+
+        if not os.path.isdir(cache_dir):
+            os.makedirs(cache_dir)
+        new_region = make_region(
+            name=region_uid_name,
+            function_key_generator=backend_key_generator(region_obj.actual_backend)
+        )
+        namespace_filename = os.path.join(
+            cache_dir, "{}.cache.dbm".format(region_namespace))
+        # special type that allows 1db per namespace
+        new_region.configure(
+            backend='dogpile.cache.rc.file_namespace',
+            expiration_time=expiration_time,
+            arguments={"filename": namespace_filename}
+        )
+
+        # create and save in region caches
+        log.debug('configuring new region: %s', region_uid_name)
+        region_obj = region_meta.dogpile_cache_regions[region_namespace] = new_region
+
+    return region_obj
+
+
+def clear_cache_namespace(cache_region, cache_namespace_uid, invalidate=False):
+    region = get_or_create_region(cache_region, cache_namespace_uid)
+    cache_keys = region.backend.list_keys(prefix=cache_namespace_uid)
+    num_delete_keys = len(cache_keys)
+    if invalidate:
+        region.invalidate(hard=False)
+    else:
+        if num_delete_keys:
+            region.delete_multi(cache_keys)
+    return num_delete_keys

vcsserver/svn.py

@@ -1,791 +1,867 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 
 
 import os
 import subprocess
+import time
 from urllib.error import URLError
 import urllib.parse
 import logging
 import posixpath as vcspath
 import io
 import urllib.request, urllib.parse, urllib.error
 import traceback
 
 import svn.client
 import svn.core
 import svn.delta
 import svn.diff
 import svn.fs
 import svn.repos
 
 from vcsserver import svn_diff, exceptions, subprocessio, settings
-from vcsserver.base import RepoFactory, raise_from_original
+from vcsserver.base import RepoFactory, raise_from_original, ArchiveNode, archive_repo
+from vcsserver.exceptions import NoContentException
+from vcsserver.utils import safe_str
 from vcsserver.vcs_base import RemoteBase
 
 log = logging.getLogger(__name__)
 
 
 svn_compatible_versions_map = {
     'pre-1.4-compatible': '1.3',
     'pre-1.5-compatible': '1.4',
     'pre-1.6-compatible': '1.5',
     'pre-1.8-compatible': '1.7',
     'pre-1.9-compatible': '1.8',
 }
 
 current_compatible_version = '1.12'
 
 
 def reraise_safe_exceptions(func):
     """Decorator for converting svn exceptions to something neutral."""
     def wrapper(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except Exception as e:
             if not hasattr(e, '_vcs_kind'):
                 log.exception("Unhandled exception in svn remote call")
                 raise_from_original(exceptions.UnhandledException(e))
             raise
     return wrapper
 
 
 class SubversionFactory(RepoFactory):
     repo_type = 'svn'
 
     def _create_repo(self, wire, create, compatible_version):
         path = svn.core.svn_path_canonicalize(wire['path'])
         if create:
             fs_config = {'compatible-version': current_compatible_version}
             if compatible_version:
 
                 compatible_version_string = \
                     svn_compatible_versions_map.get(compatible_version) \
                     or compatible_version
                 fs_config['compatible-version'] = compatible_version_string
 
             log.debug('Create SVN repo with config "%s"', fs_config)
             repo = svn.repos.create(path, "", "", None, fs_config)
         else:
             repo = svn.repos.open(path)
 
         log.debug('Got SVN object: %s', repo)
         return repo
 
     def repo(self, wire, create=False, compatible_version=None):
         """
         Get a repository instance for the given path.
         """
         return self._create_repo(wire, create, compatible_version)
 
 
 NODE_TYPE_MAPPING = {
     svn.core.svn_node_file: 'file',
     svn.core.svn_node_dir: 'dir',
 }
 
 
 class SvnRemote(RemoteBase):
 
     def __init__(self, factory, hg_factory=None):
         self._factory = factory
         # TODO: Remove once we do not use internal Mercurial objects anymore
         # for subversion
         self._hg_factory = hg_factory
 
     @reraise_safe_exceptions
     def discover_svn_version(self):
         try:
             import svn.core
             svn_ver = svn.core.SVN_VERSION
         except ImportError:
             svn_ver = None
         return svn_ver
 
     @reraise_safe_exceptions
     def is_empty(self, wire):
 
         try:
             return self.lookup(wire, -1) == 0
         except Exception:
             log.exception("failed to read object_store")
             return False
 
     def check_url(self, url, config_items):
         # this can throw exception if not installed, but we detect this
         from hgsubversion import svnrepo
 
         baseui = self._hg_factory._create_config(config_items)
         # uuid function get's only valid UUID from proper repo, else
         # throws exception
         try:
             svnrepo.svnremoterepo(baseui, url).svn.uuid
         except Exception:
             tb = traceback.format_exc()
             log.debug("Invalid Subversion url: `%s`, tb: %s", url, tb)
             raise URLError(
                 '"%s" is not a valid Subversion source url.' % (url, ))
         return True
 
     def is_path_valid_repository(self, wire, path):
 
         # NOTE(marcink):  short circuit the check for SVN repo
         # the repos.open might be expensive to check, but we have one cheap
         # pre condition that we can use, to check for 'format' file
 
         if not os.path.isfile(os.path.join(path, 'format')):
             return False
 
         try:
             svn.repos.open(path)
         except svn.core.SubversionException:
             tb = traceback.format_exc()
             log.debug("Invalid Subversion path `%s`, tb: %s", path, tb)
             return False
         return True
 
     @reraise_safe_exceptions
     def verify(self, wire,):
         repo_path = wire['path']
         if not self.is_path_valid_repository(wire, repo_path):
             raise Exception(
                 "Path %s is not a valid Subversion repository." % repo_path)
 
         cmd = ['svnadmin', 'info', repo_path]
         stdout, stderr = subprocessio.run_command(cmd)
         return stdout
 
     def lookup(self, wire, revision):
         if revision not in [-1, None, 'HEAD']:
             raise NotImplementedError
         repo = self._factory.repo(wire)
         fs_ptr = svn.repos.fs(repo)
         head = svn.fs.youngest_rev(fs_ptr)
         return head
 
     def lookup_interval(self, wire, start_ts, end_ts):
         repo = self._factory.repo(wire)
         fsobj = svn.repos.fs(repo)
         start_rev = None
         end_rev = None
         if start_ts:
             start_ts_svn = apr_time_t(start_ts)
             start_rev = svn.repos.dated_revision(repo, start_ts_svn) + 1
         else:
             start_rev = 1
         if end_ts:
             end_ts_svn = apr_time_t(end_ts)
             end_rev = svn.repos.dated_revision(repo, end_ts_svn)
         else:
             end_rev = svn.fs.youngest_rev(fsobj)
         return start_rev, end_rev
 
     def revision_properties(self, wire, revision):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _revision_properties(_repo_id, _revision):
             repo = self._factory.repo(wire)
             fs_ptr = svn.repos.fs(repo)
             return svn.fs.revision_proplist(fs_ptr, revision)
         return _revision_properties(repo_id, revision)
 
     def revision_changes(self, wire, revision):
 
         repo = self._factory.repo(wire)
         fsobj = svn.repos.fs(repo)
         rev_root = svn.fs.revision_root(fsobj, revision)
 
         editor = svn.repos.ChangeCollector(fsobj, rev_root)
         editor_ptr, editor_baton = svn.delta.make_editor(editor)
         base_dir = ""
         send_deltas = False
         svn.repos.replay2(
             rev_root, base_dir, svn.core.SVN_INVALID_REVNUM, send_deltas,
             editor_ptr, editor_baton, None)
 
         added = []
         changed = []
         removed = []
 
         # TODO: CHANGE_ACTION_REPLACE: Figure out where it belongs
         for path, change in editor.changes.items():
             # TODO: Decide what to do with directory nodes. Subversion can add
             # empty directories.
 
             if change.item_kind == svn.core.svn_node_dir:
                 continue
             if change.action in [svn.repos.CHANGE_ACTION_ADD]:
                 added.append(path)
             elif change.action in [svn.repos.CHANGE_ACTION_MODIFY,
                                    svn.repos.CHANGE_ACTION_REPLACE]:
                 changed.append(path)
             elif change.action in [svn.repos.CHANGE_ACTION_DELETE]:
                 removed.append(path)
             else:
                 raise NotImplementedError(
                     "Action %s not supported on path %s" % (
                         change.action, path))
 
         changes = {
             'added': added,
             'changed': changed,
             'removed': removed,
         }
         return changes
 
     @reraise_safe_exceptions
     def node_history(self, wire, path, revision, limit):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _assert_correct_path(_context_uid, _repo_id, _path, _revision, _limit):
             cross_copies = False
             repo = self._factory.repo(wire)
             fsobj = svn.repos.fs(repo)
             rev_root = svn.fs.revision_root(fsobj, revision)
 
             history_revisions = []
             history = svn.fs.node_history(rev_root, path)
             history = svn.fs.history_prev(history, cross_copies)
             while history:
                 __, node_revision = svn.fs.history_location(history)
                 history_revisions.append(node_revision)
                 if limit and len(history_revisions) >= limit:
                     break
                 history = svn.fs.history_prev(history, cross_copies)
             return history_revisions
         return _assert_correct_path(context_uid, repo_id, path, revision, limit)
 
     def node_properties(self, wire, path, revision):
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _node_properties(_repo_id, _path, _revision):
             repo = self._factory.repo(wire)
             fsobj = svn.repos.fs(repo)
             rev_root = svn.fs.revision_root(fsobj, revision)
             return svn.fs.node_proplist(rev_root, path)
         return _node_properties(repo_id, path, revision)
 
     def file_annotate(self, wire, path, revision):
-        abs_path = 'file://' + urllib.request.pathname2url(
+        abs_path = 'file://' + urllib.pathname2url(
             vcspath.join(wire['path'], path))
         file_uri = svn.core.svn_path_canonicalize(abs_path)
 
         start_rev = svn_opt_revision_value_t(0)
         peg_rev = svn_opt_revision_value_t(revision)
         end_rev = peg_rev
 
         annotations = []
 
         def receiver(line_no, revision, author, date, line, pool):
             annotations.append((line_no, revision, line))
 
         # TODO: Cannot use blame5, missing typemap function in the swig code
         try:
             svn.client.blame2(
                 file_uri, peg_rev, start_rev, end_rev,
                 receiver, svn.client.create_context())
         except svn.core.SubversionException as exc:
             log.exception("Error during blame operation.")
             raise Exception(
                 "Blame not supported or file does not exist at path %s. "
                 "Error %s." % (path, exc))
 
         return annotations
 
     def get_node_type(self, wire, path, revision=None):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_node_type(_repo_id, _path, _revision):
             repo = self._factory.repo(wire)
             fs_ptr = svn.repos.fs(repo)
             if _revision is None:
                 _revision = svn.fs.youngest_rev(fs_ptr)
             root = svn.fs.revision_root(fs_ptr, _revision)
             node = svn.fs.check_path(root, path)
             return NODE_TYPE_MAPPING.get(node, None)
         return _get_node_type(repo_id, path, revision)
 
     def get_nodes(self, wire, path, revision=None):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_nodes(_repo_id, _path, _revision):
             repo = self._factory.repo(wire)
             fsobj = svn.repos.fs(repo)
             if _revision is None:
                 _revision = svn.fs.youngest_rev(fsobj)
             root = svn.fs.revision_root(fsobj, _revision)
             entries = svn.fs.dir_entries(root, path)
             result = []
             for entry_path, entry_info in entries.items():
                 result.append(
                     (entry_path, NODE_TYPE_MAPPING.get(entry_info.kind, None)))
             return result
         return _get_nodes(repo_id, path, revision)
 
     def get_file_content(self, wire, path, rev=None):
         repo = self._factory.repo(wire)
         fsobj = svn.repos.fs(repo)
         if rev is None:
             rev = svn.fs.youngest_revision(fsobj)
         root = svn.fs.revision_root(fsobj, rev)
         content = svn.core.Stream(svn.fs.file_contents(root, path))
         return content.read()
 
     def get_file_size(self, wire, path, revision=None):
 
         cache_on, context_uid, repo_id = self._cache_on(wire)
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _get_file_size(_repo_id, _path, _revision):
             repo = self._factory.repo(wire)
             fsobj = svn.repos.fs(repo)
             if _revision is None:
                 _revision = svn.fs.youngest_revision(fsobj)
             root = svn.fs.revision_root(fsobj, _revision)
             size = svn.fs.file_length(root, path)
             return size
         return _get_file_size(repo_id, path, revision)
 
     def create_repository(self, wire, compatible_version=None):
         log.info('Creating Subversion repository in path "%s"', wire['path'])
         self._factory.repo(wire, create=True,
                            compatible_version=compatible_version)
 
     def get_url_and_credentials(self, src_url):
         obj = urllib.parse.urlparse(src_url)
         username = obj.username or None
         password = obj.password or None
         return username, password, src_url
 
     def import_remote_repository(self, wire, src_url):
         repo_path = wire['path']
         if not self.is_path_valid_repository(wire, repo_path):
             raise Exception(
                 "Path %s is not a valid Subversion repository." % repo_path)
 
         username, password, src_url = self.get_url_and_credentials(src_url)
         rdump_cmd = ['svnrdump', 'dump', '--non-interactive',
                      '--trust-server-cert-failures=unknown-ca']
         if username and password:
             rdump_cmd += ['--username', username, '--password', password]
         rdump_cmd += [src_url]
 
         rdump = subprocess.Popen(
             rdump_cmd,
             stdout=subprocess.PIPE, stderr=subprocess.PIPE)
         load = subprocess.Popen(
             ['svnadmin', 'load', repo_path], stdin=rdump.stdout)
 
         # TODO: johbo: This can be a very long operation, might be better
         # to track some kind of status and provide an api to check if the
         # import is done.
         rdump.wait()
         load.wait()
 
         log.debug('Return process ended with code: %s', rdump.returncode)
         if rdump.returncode != 0:
             errors = rdump.stderr.read()
             log.error('svnrdump dump failed: statuscode %s: message: %s',
                       rdump.returncode, errors)
             reason = 'UNKNOWN'
             if 'svnrdump: E230001:' in errors:
                 reason = 'INVALID_CERTIFICATE'
 
             if reason == 'UNKNOWN':
                 reason = 'UNKNOWN:{}'.format(errors)
             raise Exception(
                 'Failed to dump the remote repository from %s. Reason:%s' % (
                     src_url, reason))
         if load.returncode != 0:
             raise Exception(
                 'Failed to load the dump of remote repository from %s.' %
                 (src_url, ))
 
     def commit(self, wire, message, author, timestamp, updated, removed):
         assert isinstance(message, str)
         assert isinstance(author, str)
 
         repo = self._factory.repo(wire)
         fsobj = svn.repos.fs(repo)
 
         rev = svn.fs.youngest_rev(fsobj)
         txn = svn.repos.fs_begin_txn_for_commit(repo, rev, author, message)
         txn_root = svn.fs.txn_root(txn)
 
         for node in updated:
             TxnNodeProcessor(node, txn_root).update()
         for node in removed:
             TxnNodeProcessor(node, txn_root).remove()
 
         commit_id = svn.repos.fs_commit_txn(repo, txn)
 
         if timestamp:
             apr_time = apr_time_t(timestamp)
             ts_formatted = svn.core.svn_time_to_cstring(apr_time)
             svn.fs.change_rev_prop(fsobj, commit_id, 'svn:date', ts_formatted)
 
         log.debug('Committed revision "%s" to "%s".', commit_id, wire['path'])
         return commit_id
 
     def diff(self, wire, rev1, rev2, path1=None, path2=None,
              ignore_whitespace=False, context=3):
 
         wire.update(cache=False)
         repo = self._factory.repo(wire)
         diff_creator = SvnDiffer(
             repo, rev1, path1, rev2, path2, ignore_whitespace, context)
         try:
             return diff_creator.generate_diff()
         except svn.core.SubversionException as e:
             log.exception(
                 "Error during diff operation operation. "
                 "Path might not exist %s, %s" % (path1, path2))
             return ""
 
     @reraise_safe_exceptions
     def is_large_file(self, wire, path):
         return False
 
     @reraise_safe_exceptions
     def is_binary(self, wire, rev, path):
         cache_on, context_uid, repo_id = self._cache_on(wire)
 
-        @self.region.conditional_cache_on_arguments(condition=cache_on)
+        region = self._region(wire)
+        @region.conditional_cache_on_arguments(condition=cache_on)
         def _is_binary(_repo_id, _rev, _path):
             raw_bytes = self.get_file_content(wire, path, rev)
             return raw_bytes and '\0' in raw_bytes
 
         return _is_binary(repo_id, rev, path)
 
     @reraise_safe_exceptions
     def run_svn_command(self, wire, cmd, **opts):
         path = wire.get('path', None)
 
         if path and os.path.isdir(path):
             opts['cwd'] = path
 
-        safe_call = False
-        if '_safe' in opts:
-            safe_call = True
+        safe_call = opts.pop('_safe', False)
 
         svnenv = os.environ.copy()
         svnenv.update(opts.pop('extra_env', {}))
 
         _opts = {'env': svnenv, 'shell': False}
 
         try:
             _opts.update(opts)
             p = subprocessio.SubprocessIOChunker(cmd, **_opts)
 
             return ''.join(p), ''.join(p.error)
         except (EnvironmentError, OSError) as err:
+            if safe_call:
+                return '', safe_str(err).strip()
+            else:
             cmd = ' '.join(cmd)  # human friendly CMD
             tb_err = ("Couldn't run svn command (%s).\n"
                       "Original error was:%s\n"
                       "Call options:%s\n"
                       % (cmd, err, _opts))
             log.exception(tb_err)
-            if safe_call:
-                return '', err
-            else:
                 raise exceptions.VcsException()(tb_err)
 
     @reraise_safe_exceptions
     def install_hooks(self, wire, force=False):
         from vcsserver.hook_utils import install_svn_hooks
         repo_path = wire['path']
         binary_dir = settings.BINARY_DIR
         executable = None
         if binary_dir:
             executable = os.path.join(binary_dir, 'python')
         return install_svn_hooks(
             repo_path, executable=executable, force_create=force)
 
     @reraise_safe_exceptions
     def get_hooks_info(self, wire):
         from vcsserver.hook_utils import (
             get_svn_pre_hook_version, get_svn_post_hook_version)
         repo_path = wire['path']
         return {
             'pre_version': get_svn_pre_hook_version(repo_path),
             'post_version': get_svn_post_hook_version(repo_path),
         }
 
+    @reraise_safe_exceptions
+    def set_head_ref(self, wire, head_name):
+        pass
+
+    @reraise_safe_exceptions
+    def archive_repo(self, wire, archive_dest_path, kind, mtime, archive_at_path,
+                     archive_dir_name, commit_id):
+
+        def walk_tree(root, root_dir, _commit_id):
+            """
+            Special recursive svn repo walker
+            """
+
+            filemode_default = 0o100644
+            filemode_executable = 0o100755
+
+            file_iter = svn.fs.dir_entries(root, root_dir)
+            for f_name in file_iter:
+                f_type = NODE_TYPE_MAPPING.get(file_iter[f_name].kind, None)
+
+                if f_type == 'dir':
+                    # return only DIR, and then all entries in that dir
+                    yield os.path.join(root_dir, f_name), {'mode': filemode_default}, f_type
+                    new_root = os.path.join(root_dir, f_name)
+                    for _f_name, _f_data, _f_type in walk_tree(root, new_root, _commit_id):
+                        yield _f_name, _f_data, _f_type
+                else:
+                    f_path = os.path.join(root_dir, f_name).rstrip('/')
+                    prop_list = svn.fs.node_proplist(root, f_path)
+
+                    f_mode = filemode_default
+                    if prop_list.get('svn:executable'):
+                        f_mode = filemode_executable
+
+                    f_is_link = False
+                    if prop_list.get('svn:special'):
+                        f_is_link = True
+
+                    data = {
+                        'is_link': f_is_link,
+                        'mode': f_mode,
+                        'content_stream': svn.core.Stream(svn.fs.file_contents(root, f_path)).read
+                    }
+
+                    yield f_path, data, f_type
+
+        def file_walker(_commit_id, path):
+            repo = self._factory.repo(wire)
+            root = svn.fs.revision_root(svn.repos.fs(repo), int(commit_id))
+
+            def no_content():
+                raise NoContentException()
+
+            for f_name, f_data, f_type in walk_tree(root, path, _commit_id):
+                file_path = f_name
+
+                if f_type == 'dir':
+                    mode = f_data['mode']
+                    yield ArchiveNode(file_path, mode, False, no_content)
+                else:
+                    mode = f_data['mode']
+                    is_link = f_data['is_link']
+                    data_stream = f_data['content_stream']
+                    yield ArchiveNode(file_path, mode, is_link, data_stream)
+
+        return archive_repo(file_walker, archive_dest_path, kind, mtime, archive_at_path,
+                            archive_dir_name, commit_id)
+
 
 class SvnDiffer(object):
     """
     Utility to create diffs based on difflib and the Subversion api
     """
 
     binary_content = False
 
     def __init__(
             self, repo, src_rev, src_path, tgt_rev, tgt_path,
             ignore_whitespace, context):
         self.repo = repo
         self.ignore_whitespace = ignore_whitespace
         self.context = context
 
         fsobj = svn.repos.fs(repo)
 
         self.tgt_rev = tgt_rev
         self.tgt_path = tgt_path or ''
         self.tgt_root = svn.fs.revision_root(fsobj, tgt_rev)
         self.tgt_kind = svn.fs.check_path(self.tgt_root, self.tgt_path)
 
         self.src_rev = src_rev
         self.src_path = src_path or self.tgt_path
         self.src_root = svn.fs.revision_root(fsobj, src_rev)
         self.src_kind = svn.fs.check_path(self.src_root, self.src_path)
 
         self._validate()
 
     def _validate(self):
         if (self.tgt_kind != svn.core.svn_node_none and
                 self.src_kind != svn.core.svn_node_none and
                 self.src_kind != self.tgt_kind):
             # TODO: johbo: proper error handling
             raise Exception(
                 "Source and target are not compatible for diff generation. "
                 "Source type: %s, target type: %s" %
                 (self.src_kind, self.tgt_kind))
 
     def generate_diff(self):
         buf = io.StringIO()
         if self.tgt_kind == svn.core.svn_node_dir:
             self._generate_dir_diff(buf)
         else:
             self._generate_file_diff(buf)
         return buf.getvalue()
 
     def _generate_dir_diff(self, buf):
         editor = DiffChangeEditor()
         editor_ptr, editor_baton = svn.delta.make_editor(editor)
         svn.repos.dir_delta2(
             self.src_root,
             self.src_path,
             '',  # src_entry
             self.tgt_root,
             self.tgt_path,
             editor_ptr, editor_baton,
             authorization_callback_allow_all,
             False,  # text_deltas
             svn.core.svn_depth_infinity,  # depth
             False,  # entry_props
             False,  # ignore_ancestry
         )
 
         for path, __, change in sorted(editor.changes):
             self._generate_node_diff(
                 buf, change, path, self.tgt_path, path, self.src_path)
 
     def _generate_file_diff(self, buf):
         change = None
         if self.src_kind == svn.core.svn_node_none:
             change = "add"
         elif self.tgt_kind == svn.core.svn_node_none:
             change = "delete"
         tgt_base, tgt_path = vcspath.split(self.tgt_path)
         src_base, src_path = vcspath.split(self.src_path)
         self._generate_node_diff(
             buf, change, tgt_path, tgt_base, src_path, src_base)
 
     def _generate_node_diff(
             self, buf, change, tgt_path, tgt_base, src_path, src_base):
 
         if self.src_rev == self.tgt_rev and tgt_base == src_base:
             # makes consistent behaviour with git/hg to return empty diff if
             # we compare same revisions
             return
 
         tgt_full_path = vcspath.join(tgt_base, tgt_path)
         src_full_path = vcspath.join(src_base, src_path)
 
         self.binary_content = False
         mime_type = self._get_mime_type(tgt_full_path)
 
         if mime_type and not mime_type.startswith('text'):
             self.binary_content = True
             buf.write("=" * 67 + '\n')
             buf.write("Cannot display: file marked as a binary type.\n")
             buf.write("svn:mime-type = %s\n" % mime_type)
         buf.write("Index: %s\n" % (tgt_path, ))
         buf.write("=" * 67 + '\n')
         buf.write("diff --git a/%(tgt_path)s b/%(tgt_path)s\n" % {
             'tgt_path': tgt_path})
 
         if change == 'add':
             # TODO: johbo: SVN is missing a zero here compared to git
             buf.write("new file mode 10644\n")
 
             #TODO(marcink): intro to binary detection of svn patches
             # if self.binary_content:
             #     buf.write('GIT binary patch\n')
 
             buf.write("--- /dev/null\t(revision 0)\n")
             src_lines = []
         else:
             if change == 'delete':
                 buf.write("deleted file mode 10644\n")
 
             #TODO(marcink): intro to binary detection of svn patches
             # if self.binary_content:
             #     buf.write('GIT binary patch\n')
 
             buf.write("--- a/%s\t(revision %s)\n" % (
                 src_path, self.src_rev))
             src_lines = self._svn_readlines(self.src_root, src_full_path)
 
         if change == 'delete':
             buf.write("+++ /dev/null\t(revision %s)\n" % (self.tgt_rev, ))
             tgt_lines = []
         else:
             buf.write("+++ b/%s\t(revision %s)\n" % (
                 tgt_path, self.tgt_rev))
             tgt_lines = self._svn_readlines(self.tgt_root, tgt_full_path)
 
         if not self.binary_content:
             udiff = svn_diff.unified_diff(
                 src_lines, tgt_lines, context=self.context,
                 ignore_blank_lines=self.ignore_whitespace,
                 ignore_case=False,
                 ignore_space_changes=self.ignore_whitespace)
             buf.writelines(udiff)
 
     def _get_mime_type(self, path):
         try:
             mime_type = svn.fs.node_prop(
                 self.tgt_root, path, svn.core.SVN_PROP_MIME_TYPE)
         except svn.core.SubversionException:
             mime_type = svn.fs.node_prop(
                 self.src_root, path, svn.core.SVN_PROP_MIME_TYPE)
         return mime_type
 
     def _svn_readlines(self, fs_root, node_path):
         if self.binary_content:
             return []
         node_kind = svn.fs.check_path(fs_root, node_path)
         if node_kind not in (
                 svn.core.svn_node_file, svn.core.svn_node_symlink):
             return []
         content = svn.core.Stream(
             svn.fs.file_contents(fs_root, node_path)).read()
         return content.splitlines(True)
 
 
 class DiffChangeEditor(svn.delta.Editor):
     """
     Records changes between two given revisions
     """
 
     def __init__(self):
         self.changes = []
 
     def delete_entry(self, path, revision, parent_baton, pool=None):
         self.changes.append((path, None, 'delete'))
 
     def add_file(
             self, path, parent_baton, copyfrom_path, copyfrom_revision,
             file_pool=None):
         self.changes.append((path, 'file', 'add'))
 
     def open_file(self, path, parent_baton, base_revision, file_pool=None):
         self.changes.append((path, 'file', 'change'))
 
 
 def authorization_callback_allow_all(root, path, pool):
     return True
 
 
 class TxnNodeProcessor(object):
     """
     Utility to process the change of one node within a transaction root.
 
     It encapsulates the knowledge of how to add, update or remove
     a node for a given transaction root. The purpose is to support the method
     `SvnRemote.commit`.
     """
 
     def __init__(self, node, txn_root):
         assert isinstance(node['path'], str)
 
         self.node = node
         self.txn_root = txn_root
 
     def update(self):
         self._ensure_parent_dirs()
         self._add_file_if_node_does_not_exist()
         self._update_file_content()
         self._update_file_properties()
 
     def remove(self):
         svn.fs.delete(self.txn_root, self.node['path'])
         # TODO: Clean up directory if empty
 
     def _ensure_parent_dirs(self):
         curdir = vcspath.dirname(self.node['path'])
         dirs_to_create = []
         while not self._svn_path_exists(curdir):
             dirs_to_create.append(curdir)
             curdir = vcspath.dirname(curdir)
 
         for curdir in reversed(dirs_to_create):
             log.debug('Creating missing directory "%s"', curdir)
             svn.fs.make_dir(self.txn_root, curdir)
 
     def _svn_path_exists(self, path):
         path_status = svn.fs.check_path(self.txn_root, path)
         return path_status != svn.core.svn_node_none
 
     def _add_file_if_node_does_not_exist(self):
         kind = svn.fs.check_path(self.txn_root, self.node['path'])
         if kind == svn.core.svn_node_none:
             svn.fs.make_file(self.txn_root, self.node['path'])
 
     def _update_file_content(self):
         assert isinstance(self.node['content'], str)
         handler, baton = svn.fs.apply_textdelta(
             self.txn_root, self.node['path'], None, None)
         svn.delta.svn_txdelta_send_string(self.node['content'], handler, baton)
 
     def _update_file_properties(self):
         properties = self.node.get('properties', {})
         for key, value in properties.items():
             svn.fs.change_node_prop(
                 self.txn_root, self.node['path'], key, value)
 
 
 def apr_time_t(timestamp):
     """
     Convert a Python timestamp into APR timestamp type apr_time_t
     """
     return timestamp * 1E6
 
 
 def svn_opt_revision_value_t(num):
     """
     Put `num` into a `svn_opt_revision_value_t` structure.
     """
     value = svn.core.svn_opt_revision_value_t()
     value.number = num
     revision = svn.core.svn_opt_revision_t()
     revision.kind = svn.core.svn_opt_revision_number
     revision.value = value
     return revision

vcsserver/tests/conftest.py

@@ -1,57 +1,56 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import socket
-
 import pytest
 
 
 def pytest_addoption(parser):
     parser.addoption(
-        '--repeat', type=int, default=100,
+        '--perf-repeat-vcs', type=int, default=100,
         help="Number of repetitions in performance tests.")
 
 
 @pytest.fixture(scope='session')
 def repeat(request):
     """
     The number of repetitions is based on this fixture.
 
     Slower calls may divide it by 10 or 100. It is chosen in a way so that the
     tests are not too slow in our default test suite.
     """
-    return request.config.getoption('--repeat')
+    return request.config.getoption('--perf-repeat-vcs')
 
 
 @pytest.fixture(scope='session')
 def vcsserver_port(request):
     port = get_available_port()
     print('Using vcsserver port %s' % (port, ))
     return port
 
 
 def get_available_port():
     family = socket.AF_INET
     socktype = socket.SOCK_STREAM
     host = '127.0.0.1'
 
     mysocket = socket.socket(family, socktype)
     mysocket.bind((host, 0))
     port = mysocket.getsockname()[1]
     mysocket.close()
     del mysocket
     return port

vcsserver/tests/test_git.py

@@ -1,160 +1,160 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import inspect
 
 import pytest
 import dulwich.errors
 from mock import Mock, patch
 
 from vcsserver import git
 
 
 SAMPLE_REFS = {
     'HEAD': 'fd627b9e0dd80b47be81af07c4a98518244ed2f7',
     'refs/tags/v0.1.9': '341d28f0eec5ddf0b6b77871e13c2bbd6bec685c',
     'refs/tags/v0.1.8': '74ebce002c088b8a5ecf40073db09375515ecd68',
     'refs/tags/v0.1.1': 'e6ea6d16e2f26250124a1f4b4fe37a912f9d86a0',
     'refs/tags/v0.1.3': '5a3a8fb005554692b16e21dee62bf02667d8dc3e',
 }
 
 
 @pytest.fixture
 def git_remote():
     """
     A GitRemote instance with a mock factory.
     """
     factory = Mock()
     remote = git.GitRemote(factory)
     return remote
 
 
 def test_discover_git_version(git_remote):
     version = git_remote.discover_git_version()
     assert version
 
 
 class TestGitFetch(object):
     def setup(self):
         self.mock_repo = Mock()
         factory = Mock()
         factory.repo = Mock(return_value=self.mock_repo)
         self.remote_git = git.GitRemote(factory)
 
     def test_fetches_all_when_no_commit_ids_specified(self):
         def side_effect(determine_wants, *args, **kwargs):
             determine_wants(SAMPLE_REFS)
 
         with patch('dulwich.client.LocalGitClient.fetch') as mock_fetch:
             mock_fetch.side_effect = side_effect
             self.remote_git.pull(wire={}, url='/tmp/', apply_refs=False)
             determine_wants = self.mock_repo.object_store.determine_wants_all
             determine_wants.assert_called_once_with(SAMPLE_REFS)
 
     def test_fetches_specified_commits(self):
         selected_refs = {
             'refs/tags/v0.1.8': '74ebce002c088b8a5ecf40073db09375515ecd68',
             'refs/tags/v0.1.3': '5a3a8fb005554692b16e21dee62bf02667d8dc3e',
         }
 
         def side_effect(determine_wants, *args, **kwargs):
             result = determine_wants(SAMPLE_REFS)
             assert sorted(result) == sorted(selected_refs.values())
             return result
 
         with patch('dulwich.client.LocalGitClient.fetch') as mock_fetch:
             mock_fetch.side_effect = side_effect
             self.remote_git.pull(
                 wire={}, url='/tmp/', apply_refs=False,
                 refs=selected_refs.keys())
             determine_wants = self.mock_repo.object_store.determine_wants_all
             assert determine_wants.call_count == 0
 
     def test_get_remote_refs(self):
         factory = Mock()
         remote_git = git.GitRemote(factory)
         url = 'http://example.com/test/test.git'
         sample_refs = {
             'refs/tags/v0.1.8': '74ebce002c088b8a5ecf40073db09375515ecd68',
             'refs/tags/v0.1.3': '5a3a8fb005554692b16e21dee62bf02667d8dc3e',
         }
 
         with patch('vcsserver.git.Repo', create=False) as mock_repo:
             mock_repo().get_refs.return_value = sample_refs
             remote_refs = remote_git.get_remote_refs(wire={}, url=url)
             mock_repo().get_refs.assert_called_once_with()
             assert remote_refs == sample_refs
 
 
 class TestReraiseSafeExceptions(object):
 
     def test_method_decorated_with_reraise_safe_exceptions(self):
         factory = Mock()
         git_remote = git.GitRemote(factory)
 
         def fake_function():
             return None
 
         decorator = git.reraise_safe_exceptions(fake_function)
 
         methods = inspect.getmembers(git_remote, predicate=inspect.ismethod)
         for method_name, method in methods:
-            if not method_name.startswith('_'):
+            if not method_name.startswith('_') and method_name not in ['vcsserver_invalidate_cache']:
                 assert method.im_func.__code__ == decorator.__code__
 
     @pytest.mark.parametrize('side_effect, expected_type', [
         (dulwich.errors.ChecksumMismatch('0000000', 'deadbeef'), 'lookup'),
         (dulwich.errors.NotCommitError('deadbeef'), 'lookup'),
         (dulwich.errors.MissingCommitError('deadbeef'), 'lookup'),
         (dulwich.errors.ObjectMissing('deadbeef'), 'lookup'),
         (dulwich.errors.HangupException(), 'error'),
         (dulwich.errors.UnexpectedCommandError('test-cmd'), 'error'),
     ])
     def test_safe_exceptions_reraised(self, side_effect, expected_type):
         @git.reraise_safe_exceptions
         def fake_method():
             raise side_effect
 
         with pytest.raises(Exception) as exc_info:
             fake_method()
         assert type(exc_info.value) == Exception
         assert exc_info.value._vcs_kind == expected_type
 
 
 class TestDulwichRepoWrapper(object):
     def test_calls_close_on_delete(self):
         isdir_patcher = patch('dulwich.repo.os.path.isdir', return_value=True)
         with isdir_patcher:
             repo = git.Repo('/tmp/abcde')
         with patch.object(git.DulwichRepo, 'close') as close_mock:
             del repo
         close_mock.assert_called_once_with()
 
 
 class TestGitFactory(object):
     def test_create_repo_returns_dulwich_wrapper(self):
 
         with patch('vcsserver.lib.rc_cache.region_meta.dogpile_cache_regions') as mock:
             mock.side_effect = {'repo_objects': ''}
             factory = git.GitFactory()
             wire = {
                 'path': '/tmp/abcde'
             }
             isdir_patcher = patch('dulwich.repo.os.path.isdir', return_value=True)
             with isdir_patcher:
                 result = factory._create_repo(wire, True)
             assert isinstance(result, git.Repo)

vcsserver/tests/test_hg.py

@@ -1,108 +1,108 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import inspect
 import sys
 import traceback
 
 import pytest
 from mercurial.error import LookupError
 from mock import Mock, MagicMock, patch
 
 from vcsserver import exceptions, hg, hgcompat
 
 
 class TestDiff(object):
     def test_raising_safe_exception_when_lookup_failed(self):
 
         factory = Mock()
         hg_remote = hg.HgRemote(factory)
         with patch('mercurial.patch.diff') as diff_mock:
             diff_mock.side_effect = LookupError(
                 'deadbeef', 'index', 'message')
             with pytest.raises(Exception) as exc_info:
                 hg_remote.diff(
                     wire={}, commit_id_1='deadbeef', commit_id_2='deadbee1',
                     file_filter=None, opt_git=True, opt_ignorews=True,
                     context=3)
             assert type(exc_info.value) == Exception
             assert exc_info.value._vcs_kind == 'lookup'
 
 
 class TestReraiseSafeExceptions(object):
     def test_method_decorated_with_reraise_safe_exceptions(self):
         factory = Mock()
         hg_remote = hg.HgRemote(factory)
         methods = inspect.getmembers(hg_remote, predicate=inspect.ismethod)
         decorator = hg.reraise_safe_exceptions(None)
         for method_name, method in methods:
-            if not method_name.startswith('_'):
+            if not method_name.startswith('_') and method_name not in ['vcsserver_invalidate_cache']:
                 assert method.im_func.__code__ == decorator.__code__
 
     @pytest.mark.parametrize('side_effect, expected_type', [
         (hgcompat.Abort(), 'abort'),
         (hgcompat.InterventionRequired(), 'abort'),
         (hgcompat.RepoLookupError(), 'lookup'),
         (hgcompat.LookupError('deadbeef', 'index', 'message'), 'lookup'),
         (hgcompat.RepoError(), 'error'),
         (hgcompat.RequirementError(), 'requirement'),
     ])
     def test_safe_exceptions_reraised(self, side_effect, expected_type):
         @hg.reraise_safe_exceptions
         def fake_method():
             raise side_effect
 
         with pytest.raises(Exception) as exc_info:
             fake_method()
         assert type(exc_info.value) == Exception
         assert exc_info.value._vcs_kind == expected_type
 
     def test_keeps_original_traceback(self):
         @hg.reraise_safe_exceptions
         def fake_method():
             try:
                 raise hgcompat.Abort()
             except:
                 self.original_traceback = traceback.format_tb(
                     sys.exc_info()[2])
                 raise
 
         try:
             fake_method()
         except Exception:
             new_traceback = traceback.format_tb(sys.exc_info()[2])
 
         new_traceback_tail = new_traceback[-len(self.original_traceback):]
         assert new_traceback_tail == self.original_traceback
 
     def test_maps_unknow_exceptions_to_unhandled(self):
         @hg.reraise_safe_exceptions
         def stub_method():
             raise ValueError('stub')
 
         with pytest.raises(Exception) as exc_info:
             stub_method()
         assert exc_info.value._vcs_kind == 'unhandled'
 
     def test_does_not_map_known_exceptions(self):
         @hg.reraise_safe_exceptions
         def stub_method():
             raise exceptions.LookupException()('stub')
 
         with pytest.raises(Exception) as exc_info:
             stub_method()
         assert exc_info.value._vcs_kind == 'lookup'

vcsserver/tests/test_http_performance.py

@@ -1,39 +1,42 b''
 """
 Tests used to profile the HTTP based implementation.
 """
 
 import pytest
 import webtest
 
 from vcsserver.http_main import main
 
 
 @pytest.fixture
 def vcs_app():
     stub_settings = {
         'dev.use_echo_app': 'true',
         'locale': 'en_US.UTF-8',
     }
-    vcs_app = main({}, **stub_settings)
+    stub_global_conf = {
+        '__file__': ''
+    }
+    vcs_app = main(stub_global_conf, **stub_settings)
     app = webtest.TestApp(vcs_app)
     return app
 
 
 @pytest.fixture(scope='module')
 def data():
     one_kb = 'x' * 1024
     return one_kb * 1024 * 10
 
 
 def test_http_app_streaming_with_data(data, repeat, vcs_app):
     app = vcs_app
     for x in range(repeat / 10):
         response = app.post('/stream/git/', params=data)
         assert response.status_code == 200
 
 
 def test_http_app_streaming_no_data(repeat, vcs_app):
     app = vcs_app
     for x in range(repeat / 10):
         response = app.post('/stream/git/')
         assert response.status_code == 200

vcsserver/tests/test_main_http.py

@@ -1,57 +1,57 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import mock
 import pytest
 
 from vcsserver import http_main
 from vcsserver.base import obfuscate_qs
 
 
 @mock.patch('vcsserver.http_main.VCS', mock.Mock())
 @mock.patch('vcsserver.hgpatches.patch_largefiles_capabilities')
 def test_applies_largefiles_patch(patch_largefiles_capabilities):
-    http_main.main({})
+    http_main.main({'__file__': ''})
     patch_largefiles_capabilities.assert_called_once_with()
 
 
 @mock.patch('vcsserver.http_main.VCS', mock.Mock())
 @mock.patch('vcsserver.http_main.MercurialFactory', None)
 @mock.patch(
     'vcsserver.hgpatches.patch_largefiles_capabilities',
     mock.Mock(side_effect=Exception("Must not be called")))
 def test_applies_largefiles_patch_only_if_mercurial_is_available():
-    http_main.main({})
+    http_main.main({'__file__': ''})
 
 
 @pytest.mark.parametrize('given, expected', [
     ('bad', 'bad'),
     ('query&foo=bar', 'query&foo=bar'),
     ('equery&auth_token=bar', 'equery&auth_token=*****'),
     ('a;b;c;query&foo=bar&auth_token=secret',
      'a&b&c&query&foo=bar&auth_token=*****'),
     ('', ''),
     (None, None),
     ('foo=bar', 'foo=bar'),
     ('auth_token=secret', 'auth_token=*****'),
     ('auth_token=secret&api_key=secret2',
      'auth_token=*****&api_key=*****'),
     ('auth_token=secret&api_key=secret2&param=value',
      'auth_token=*****&api_key=*****&param=value'),
 ])
 def test_obfuscate_qs(given, expected):
     assert expected == obfuscate_qs(given)

vcsserver/tweens/request_wrapper.py

@@ -1,64 +1,107 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
 import time
 import logging
 
 import vcsserver
 from vcsserver.utils import safe_str
 
 
 log = logging.getLogger(__name__)
 
 
-def get_access_path(request):
-    environ = request.environ
-    return environ.get('PATH_INFO')
+def get_access_path(environ):
+    path = environ.get('PATH_INFO')
+    return path
 
 
 def get_user_agent(environ):
     return environ.get('HTTP_USER_AGENT')
 
 
+def get_vcs_method(environ):
+    return environ.get('HTTP_X_RC_METHOD')
+
+
+def get_vcs_repo(environ):
+    return environ.get('HTTP_X_RC_REPO_NAME')
+
+
 class RequestWrapperTween(object):
     def __init__(self, handler, registry):
         self.handler = handler
         self.registry = registry
 
         # one-time configuration code goes here
 
     def __call__(self, request):
         start = time.time()
+        log.debug('Starting request time measurement')
+        response = None
+
+        ua = get_user_agent(request.environ)
+        vcs_method = get_vcs_method(request.environ)
+        repo_name = get_vcs_repo(request.environ)
+
         try:
             response = self.handler(request)
         finally:
-            end = time.time()
-            total = end - start
             count = request.request_count()
             _ver_ = vcsserver.__version__
+            _path = safe_str(get_access_path(request.environ))
+            ip = '127.0.0.1'
+            match_route = request.matched_route.name if request.matched_route else "NOT_FOUND"
+            resp_code = getattr(response, 'status_code', 'UNDEFINED')
+
+            total = time.time() - start
+
+            _view_path = "{}/{}@{}".format(_path, vcs_method, repo_name)
             log.info(
                 'Req[%4s] IP: %s %s Request to %s time: %.4fs [%s], VCSServer %s',
-                count, '127.0.0.1', request.environ.get('REQUEST_METHOD'),
-                safe_str(get_access_path(request)), total, get_user_agent(request.environ), _ver_)
+                count, ip, request.environ.get('REQUEST_METHOD'),
+                _view_path, total, ua, _ver_,
+                extra={"time": total, "ver": _ver_, "code": resp_code,
+                       "path": _path, "view_name": match_route, "user_agent": ua,
+                       "vcs_method": vcs_method, "repo_name": repo_name}
+            )
+
+            statsd = request.registry.statsd
+            if statsd:
+                match_route = request.matched_route.name if request.matched_route else _path
+                elapsed_time_ms = round(1000.0 * total)  # use ms only
+                statsd.timing(
+                    "vcsserver_req_timing.histogram", elapsed_time_ms,
+                    tags=[
+                        "view_name:{}".format(match_route),
+                        "code:{}".format(resp_code)
+                    ],
+                    use_decimals=False
+                )
+                statsd.incr(
+                    "vcsserver_req_total", tags=[
+                        "view_name:{}".format(match_route),
+                        "code:{}".format(resp_code)
+                    ])
 
         return response
 
 
 def includeme(config):
     config.add_tween(
         'vcsserver.tweens.request_wrapper.RequestWrapperTween',
     )

vcsserver/vcs_base.py

@@ -1,32 +1,46 b''
 # RhodeCode VCSServer provides access to different vcs backends via network.
 # Copyright (C) 2014-2020 RhodeCode GmbH
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 3 of the License, or
 # (at your option) any later version.
 #
 # This program is distributed in the hope that it will be useful,
 # but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 # GNU General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
 # along with this program; if not, write to the Free Software Foundation,
 # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301  USA
 
+from vcsserver.lib import rc_cache
 
 class RemoteBase(object):
     EMPTY_COMMIT = '0' * 40
 
-    @property
-    def region(self):
-        return self._factory._cache_region
+    def _region(self, wire):
+        cache_repo_id = wire.get('cache_repo_id', '')
+        cache_namespace_uid = 'cache_repo.{}'.format(cache_repo_id)
+        return rc_cache.get_or_create_region('repo_object', cache_namespace_uid)
 
     def _cache_on(self, wire):
         context = wire.get('context', '')
         context_uid = '{}'.format(context)
         repo_id = wire.get('repo_id', '')
         cache = wire.get('cache', True)
         cache_on = context and cache
         return cache_on, context_uid, repo_id
+
+    def vcsserver_invalidate_cache(self, wire, delete):
+        from vcsserver.lib import rc_cache
+        repo_id = wire.get('repo_id', '')
+        cache_repo_id = wire.get('cache_repo_id', '')
+        cache_namespace_uid = 'cache_repo.{}'.format(cache_repo_id)
+
+        if delete:
+            rc_cache.clear_cache_namespace(
+                'repo_object', cache_namespace_uid, invalidate=True)
+
+        return {'invalidated': {'repo_id': repo_id, 'delete': delete}}