##// END OF EJS Templates
fix(jupyter): added handling of v3 format issues. Fixes RCCE-30
fix(jupyter): added handling of v3 format issues. Fixes RCCE-30

File last commit:

r5092:d0d88608 default
r5265:3705b0c6 default
Show More
nodes.py
960 lines | 29.9 KiB | text/x-python | PythonLexer
# Copyright (C) 2014-2023 RhodeCode GmbH
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License, version 3
# (only), as published by the Free Software Foundation.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
# This program is dual-licensed. If you wish to learn more about the
# RhodeCode Enterprise Edition, including its added features, Support services,
# and proprietary license terms, please see https://rhodecode.com/licenses/
"""
Module holding everything related to vcs nodes, with vcs2 architecture.
"""
import functools
import os
import stat
from zope.cachedescriptors.property import Lazy as LazyProperty
from rhodecode.config.conf import LANGUAGES_EXTENSIONS_MAP
from rhodecode.lib.str_utils import safe_str, safe_bytes
from rhodecode.lib.hash_utils import md5
from rhodecode.lib.vcs import path as vcspath
from rhodecode.lib.vcs.backends.base import EmptyCommit, FILEMODE_DEFAULT
from rhodecode.lib.vcs.conf.mtypes import get_mimetypes_db
from rhodecode.lib.vcs.exceptions import NodeError, RemovedFileNodeError
LARGEFILE_PREFIX = '.hglf'
class NodeKind:
SUBMODULE = -1
DIR = 1
FILE = 2
LARGEFILE = 3
class NodeState:
ADDED = 'added'
CHANGED = 'changed'
NOT_CHANGED = 'not changed'
REMOVED = 'removed'
#TODO: not sure if that should be bytes or str ?
# most probably bytes because content should be bytes and we check it
BIN_BYTE_MARKER = b'\0'
class NodeGeneratorBase(object):
"""
Base class for removed added and changed filenodes, it's a lazy generator
class that will create filenodes only on iteration or call
The len method doesn't need to create filenodes at all
"""
def __init__(self, current_paths, cs):
self.cs = cs
self.current_paths = current_paths
def __call__(self):
return [n for n in self]
def __getitem__(self, key):
if isinstance(key, slice):
for p in self.current_paths[key.start:key.stop]:
yield self.cs.get_node(p)
def __len__(self):
return len(self.current_paths)
def __iter__(self):
for p in self.current_paths:
yield self.cs.get_node(p)
class AddedFileNodesGenerator(NodeGeneratorBase):
"""
Class holding added files for current commit
"""
class ChangedFileNodesGenerator(NodeGeneratorBase):
"""
Class holding changed files for current commit
"""
class RemovedFileNodesGenerator(NodeGeneratorBase):
"""
Class holding removed files for current commit
"""
def __iter__(self):
for p in self.current_paths:
yield RemovedFileNode(path=safe_bytes(p))
def __getitem__(self, key):
if isinstance(key, slice):
for p in self.current_paths[key.start:key.stop]:
yield RemovedFileNode(path=safe_bytes(p))
@functools.total_ordering
class Node(object):
"""
Simplest class representing file or directory on repository. SCM backends
should use ``FileNode`` and ``DirNode`` subclasses rather than ``Node``
directly.
Node's ``path`` cannot start with slash as we operate on *relative* paths
only. Moreover, every single node is identified by the ``path`` attribute,
so it cannot end with slash, too. Otherwise, path could lead to mistakes.
"""
# RTLO marker allows swapping text, and certain
# security attacks could be used with this
RTLO_MARKER = "\u202E"
commit = None
def __init__(self, path: bytes, kind):
self._validate_path(path) # can throw exception if path is invalid
self.bytes_path = path.rstrip(b'/') # store for __repr__
self.path = safe_str(self.bytes_path) # we store paths as str
if self.bytes_path == b'' and kind != NodeKind.DIR:
raise NodeError("Only DirNode and its subclasses may be "
"initialized with empty path")
self.kind = kind
if self.is_root() and not self.is_dir():
raise NodeError("Root node cannot be FILE kind")
def __eq__(self, other):
if type(self) is not type(other):
return False
for attr in ['name', 'path', 'kind']:
if getattr(self, attr) != getattr(other, attr):
return False
if self.is_file():
# FileNode compare, we need to fallback to content compare
return None
else:
# For DirNode's check without entering each dir
self_nodes_paths = list(sorted(n.path for n in self.nodes))
other_nodes_paths = list(sorted(n.path for n in self.nodes))
if self_nodes_paths != other_nodes_paths:
return False
return True
def __lt__(self, other):
if self.kind < other.kind:
return True
if self.kind > other.kind:
return False
if self.path < other.path:
return True
if self.path > other.path:
return False
# def __cmp__(self, other):
# """
# Comparator using name of the node, needed for quick list sorting.
# """
#
# kind_cmp = cmp(self.kind, other.kind)
# if kind_cmp:
# if isinstance(self, SubModuleNode):
# # we make submodules equal to dirnode for "sorting" purposes
# return NodeKind.DIR
# return kind_cmp
# return cmp(self.name, other.name)
def __repr__(self):
maybe_path = getattr(self, 'path', 'UNKNOWN_PATH')
return f'<{self.__class__.__name__} {maybe_path!r}>'
def __str__(self):
return self.name
def _validate_path(self, path: bytes):
self._assert_bytes(path)
if path.startswith(b'/'):
raise NodeError(
f"Cannot initialize Node objects with slash at "
f"the beginning as only relative paths are supported. "
f"Got {path}")
def _assert_bytes(self, value):
if not isinstance(value, bytes):
raise TypeError(f"Bytes required as input, got {type(value)} of {value}.")
@LazyProperty
def parent(self):
parent_path = self.get_parent_path()
if parent_path:
if self.commit:
return self.commit.get_node(parent_path)
return DirNode(parent_path)
return None
@LazyProperty
def str_path(self) -> str:
return safe_str(self.path)
@LazyProperty
def has_rtlo(self):
"""Detects if a path has right-to-left-override marker"""
return self.RTLO_MARKER in self.str_path
@LazyProperty
def dir_path(self):
"""
Returns name of the directory from full path of this vcs node. Empty
string is returned if there's no directory in the path
"""
_parts = self.path.rstrip('/').rsplit('/', 1)
if len(_parts) == 2:
return _parts[0]
return ''
@LazyProperty
def name(self):
"""
Returns name of the node so if its path
then only last part is returned.
"""
return self.path.rstrip('/').split('/')[-1]
@property
def kind(self):
return self._kind
@kind.setter
def kind(self, kind):
if hasattr(self, '_kind'):
raise NodeError("Cannot change node's kind")
else:
self._kind = kind
# Post setter check (path's trailing slash)
if self.path.endswith('/'):
raise NodeError("Node's path cannot end with slash")
def get_parent_path(self) -> bytes:
"""
Returns node's parent path or empty string if node is root.
"""
if self.is_root():
return b''
str_path = vcspath.dirname(self.path.rstrip('/')) + '/'
return safe_bytes(str_path)
def is_file(self):
"""
Returns ``True`` if node's kind is ``NodeKind.FILE``, ``False``
otherwise.
"""
return self.kind == NodeKind.FILE
def is_dir(self):
"""
Returns ``True`` if node's kind is ``NodeKind.DIR``, ``False``
otherwise.
"""
return self.kind == NodeKind.DIR
def is_root(self):
"""
Returns ``True`` if node is a root node and ``False`` otherwise.
"""
return self.kind == NodeKind.DIR and self.path == ''
def is_submodule(self):
"""
Returns ``True`` if node's kind is ``NodeKind.SUBMODULE``, ``False``
otherwise.
"""
return self.kind == NodeKind.SUBMODULE
def is_largefile(self):
"""
Returns ``True`` if node's kind is ``NodeKind.LARGEFILE``, ``False``
otherwise
"""
return self.kind == NodeKind.LARGEFILE
def is_link(self):
if self.commit:
return self.commit.is_link(self.path)
return False
@LazyProperty
def added(self):
return self.state is NodeState.ADDED
@LazyProperty
def changed(self):
return self.state is NodeState.CHANGED
@LazyProperty
def not_changed(self):
return self.state is NodeState.NOT_CHANGED
@LazyProperty
def removed(self):
return self.state is NodeState.REMOVED
class FileNode(Node):
"""
Class representing file nodes.
:attribute: path: path to the node, relative to repository's root
:attribute: content: if given arbitrary sets content of the file
:attribute: commit: if given, first time content is accessed, callback
:attribute: mode: stat mode for a node. Default is `FILEMODE_DEFAULT`.
"""
_filter_pre_load = []
def __init__(self, path: bytes, content: bytes | None = None, commit=None, mode=None, pre_load=None):
"""
Only one of ``content`` and ``commit`` may be given. Passing both
would raise ``NodeError`` exception.
:param path: relative path to the node
:param content: content may be passed to constructor
:param commit: if given, will use it to lazily fetch content
:param mode: ST_MODE (i.e. 0100644)
"""
if content and commit:
raise NodeError("Cannot use both content and commit")
super().__init__(path, kind=NodeKind.FILE)
self.commit = commit
if content and not isinstance(content, bytes):
# File content is one thing that inherently must be bytes
# we support passing str too, and convert the content
content = safe_bytes(content)
self._content = content
self._mode = mode or FILEMODE_DEFAULT
self._set_bulk_properties(pre_load)
def __eq__(self, other):
eq = super().__eq__(other)
if eq is not None:
return eq
return self.content == other.content
def __hash__(self):
raw_id = getattr(self.commit, 'raw_id', '')
return hash((self.path, raw_id))
def __lt__(self, other):
lt = super().__lt__(other)
if lt is not None:
return lt
return self.content < other.content
def __repr__(self):
short_id = getattr(self.commit, 'short_id', '')
return f'<{self.__class__.__name__} path={self.path!r}, short_id={short_id}>'
def _set_bulk_properties(self, pre_load):
if not pre_load:
return
pre_load = [entry for entry in pre_load
if entry not in self._filter_pre_load]
if not pre_load:
return
remote = self.commit.get_remote()
result = remote.bulk_file_request(self.commit.raw_id, self.path, pre_load)
for attr, value in result.items():
if attr == "flags":
self.__dict__['mode'] = safe_str(value)
elif attr == "size":
self.__dict__['size'] = value
elif attr == "data":
self.__dict__['_content'] = value
elif attr == "is_binary":
self.__dict__['is_binary'] = value
elif attr == "md5":
self.__dict__['md5'] = value
else:
raise ValueError(f'Unsupported attr in bulk_property: {attr}')
@LazyProperty
def mode(self):
"""
Returns lazily mode of the FileNode. If `commit` is not set, would
use value given at initialization or `FILEMODE_DEFAULT` (default).
"""
if self.commit:
mode = self.commit.get_file_mode(self.path)
else:
mode = self._mode
return mode
@LazyProperty
def raw_bytes(self) -> bytes:
"""
Returns lazily the raw bytes of the FileNode.
"""
if self.commit:
if self._content is None:
self._content = self.commit.get_file_content(self.path)
content = self._content
else:
content = self._content
return content
def content_uncached(self):
"""
Returns lazily content of the FileNode.
"""
if self.commit:
content = self.commit.get_file_content(self.path)
else:
content = self._content
return content
def stream_bytes(self):
"""
Returns an iterator that will stream the content of the file directly from
vcsserver without loading it to memory.
"""
if self.commit:
return self.commit.get_file_content_streamed(self.path)
raise NodeError("Cannot retrieve stream_bytes without related commit attribute")
def metadata_uncached(self):
"""
Returns md5, binary flag of the file node, without any cache usage.
"""
content = self.content_uncached()
is_binary = bool(content and BIN_BYTE_MARKER in content)
size = 0
if content:
size = len(content)
return is_binary, md5(content), size, content
@LazyProperty
def content(self) -> bytes:
"""
Returns lazily content of the FileNode.
"""
content = self.raw_bytes
if content and not isinstance(content, bytes):
raise ValueError(f'Content is of type {type(content)} instead of bytes')
return content
@LazyProperty
def str_content(self) -> str:
return safe_str(self.raw_bytes)
@LazyProperty
def size(self):
if self.commit:
return self.commit.get_file_size(self.path)
raise NodeError(
"Cannot retrieve size of the file without related "
"commit attribute")
@LazyProperty
def message(self):
if self.commit:
return self.last_commit.message
raise NodeError(
"Cannot retrieve message of the file without related "
"commit attribute")
@LazyProperty
def last_commit(self):
if self.commit:
pre_load = ["author", "date", "message", "parents"]
return self.commit.get_path_commit(self.path, pre_load=pre_load)
raise NodeError(
"Cannot retrieve last commit of the file without "
"related commit attribute")
def get_mimetype(self):
"""
Mimetype is calculated based on the file's content. If ``_mimetype``
attribute is available, it will be returned (backends which store
mimetypes or can easily recognize them, should set this private
attribute to indicate that type should *NOT* be calculated).
"""
if hasattr(self, '_mimetype'):
if (isinstance(self._mimetype, (tuple, list)) and
len(self._mimetype) == 2):
return self._mimetype
else:
raise NodeError('given _mimetype attribute must be an 2 '
'element list or tuple')
db = get_mimetypes_db()
mtype, encoding = db.guess_type(self.name)
if mtype is None:
if not self.is_largefile() and self.is_binary:
mtype = 'application/octet-stream'
encoding = None
else:
mtype = 'text/plain'
encoding = None
# try with pygments
try:
from pygments.lexers import get_lexer_for_filename
mt = get_lexer_for_filename(self.name).mimetypes
except Exception:
mt = None
if mt:
mtype = mt[0]
return mtype, encoding
@LazyProperty
def mimetype(self):
"""
Wrapper around full mimetype info. It returns only type of fetched
mimetype without the encoding part. use get_mimetype function to fetch
full set of (type,encoding)
"""
return self.get_mimetype()[0]
@LazyProperty
def mimetype_main(self):
return self.mimetype.split('/')[0]
@classmethod
def get_lexer(cls, filename, content=None):
from pygments import lexers
extension = filename.split('.')[-1]
lexer = None
try:
lexer = lexers.guess_lexer_for_filename(
filename, content, stripnl=False)
except lexers.ClassNotFound:
pass
# try our EXTENSION_MAP
if not lexer:
try:
lexer_class = LANGUAGES_EXTENSIONS_MAP.get(extension)
if lexer_class:
lexer = lexers.get_lexer_by_name(lexer_class[0])
except lexers.ClassNotFound:
pass
if not lexer:
lexer = lexers.TextLexer(stripnl=False)
return lexer
@LazyProperty
def lexer(self):
"""
Returns pygment's lexer class. Would try to guess lexer taking file's
content, name and mimetype.
"""
# TODO: this is more proper, but super heavy on investigating the type based on the content
#self.get_lexer(self.name, self.content)
return self.get_lexer(self.name)
@LazyProperty
def lexer_alias(self):
"""
Returns first alias of the lexer guessed for this file.
"""
return self.lexer.aliases[0]
@LazyProperty
def history(self):
"""
Returns a list of commit for this file in which the file was changed
"""
if self.commit is None:
raise NodeError('Unable to get commit for this FileNode')
return self.commit.get_path_history(self.path)
@LazyProperty
def annotate(self):
"""
Returns a list of three element tuples with lineno, commit and line
"""
if self.commit is None:
raise NodeError('Unable to get commit for this FileNode')
pre_load = ["author", "date", "message", "parents"]
return self.commit.get_file_annotate(self.path, pre_load=pre_load)
@LazyProperty
def state(self):
if not self.commit:
raise NodeError(
"Cannot check state of the node if it's not "
"linked with commit")
elif self.path in (node.path for node in self.commit.added):
return NodeState.ADDED
elif self.path in (node.path for node in self.commit.changed):
return NodeState.CHANGED
else:
return NodeState.NOT_CHANGED
@LazyProperty
def is_binary(self):
"""
Returns True if file has binary content.
"""
if self.commit:
return self.commit.is_node_binary(self.path)
else:
raw_bytes = self._content
return bool(raw_bytes and BIN_BYTE_MARKER in raw_bytes)
@LazyProperty
def md5(self):
"""
Returns md5 of the file node.
"""
if self.commit:
return self.commit.node_md5_hash(self.path)
else:
raw_bytes = self._content
# TODO: this sucks, we're computing md5 on potentially super big stream data...
return md5(raw_bytes)
@LazyProperty
def extension(self):
"""Returns filenode extension"""
return self.name.split('.')[-1]
@property
def is_executable(self):
"""
Returns ``True`` if file has executable flag turned on.
"""
return bool(self.mode & stat.S_IXUSR)
def get_largefile_node(self):
"""
Try to return a Mercurial FileNode from this node. It does internal
checks inside largefile store, if that file exist there it will
create special instance of LargeFileNode which can get content from
LF store.
"""
if self.commit:
return self.commit.get_largefile_node(self.path)
def count_lines(self, content: str | bytes, count_empty=False):
if isinstance(content, str):
newline_marker = '\n'
elif isinstance(content, bytes):
newline_marker = b'\n'
else:
raise ValueError('content must be bytes or str got {type(content)} instead')
if count_empty:
all_lines = 0
empty_lines = 0
for line in content.splitlines(True):
if line == newline_marker:
empty_lines += 1
all_lines += 1
return all_lines, all_lines - empty_lines
else:
# fast method
empty_lines = all_lines = content.count(newline_marker)
if all_lines == 0 and content:
# one-line without a newline
empty_lines = all_lines = 1
return all_lines, empty_lines
def lines(self, count_empty=False):
all_lines, empty_lines = 0, 0
if not self.is_binary:
content = self.content
all_lines, empty_lines = self.count_lines(content, count_empty=count_empty)
return all_lines, empty_lines
class RemovedFileNode(FileNode):
"""
Dummy FileNode class - trying to access any public attribute except path,
name, kind or state (or methods/attributes checking those two) would raise
RemovedFileNodeError.
"""
ALLOWED_ATTRIBUTES = [
'name', 'path', 'state', 'is_root', 'is_file', 'is_dir', 'kind',
'added', 'changed', 'not_changed', 'removed', 'bytes_path'
]
def __init__(self, path):
"""
:param path: relative path to the node
"""
super().__init__(path=path)
def __getattribute__(self, attr):
if attr.startswith('_') or attr in RemovedFileNode.ALLOWED_ATTRIBUTES:
return super().__getattribute__(attr)
raise RemovedFileNodeError(f"Cannot access attribute {attr} on RemovedFileNode. Not in allowed attributes")
@LazyProperty
def state(self):
return NodeState.REMOVED
class DirNode(Node):
"""
DirNode stores list of files and directories within this node.
Nodes may be used standalone but within repository context they
lazily fetch data within same repository's commit.
"""
def __init__(self, path, nodes=(), commit=None, default_pre_load=None):
"""
Only one of ``nodes`` and ``commit`` may be given. Passing both
would raise ``NodeError`` exception.
:param path: relative path to the node
:param nodes: content may be passed to constructor
:param commit: if given, will use it to lazily fetch content
"""
if nodes and commit:
raise NodeError("Cannot use both nodes and commit")
super().__init__(path, NodeKind.DIR)
self.commit = commit
self._nodes = nodes
self.default_pre_load = default_pre_load or ['is_binary', 'size']
def __iter__(self):
yield from self.nodes
def __eq__(self, other):
eq = super().__eq__(other)
if eq is not None:
return eq
# check without entering each dir
self_nodes_paths = list(sorted(n.path for n in self.nodes))
other_nodes_paths = list(sorted(n.path for n in self.nodes))
return self_nodes_paths == other_nodes_paths
def __lt__(self, other):
lt = super().__lt__(other)
if lt is not None:
return lt
# check without entering each dir
self_nodes_paths = list(sorted(n.path for n in self.nodes))
other_nodes_paths = list(sorted(n.path for n in self.nodes))
return self_nodes_paths < other_nodes_paths
@LazyProperty
def content(self):
raise NodeError(f"{self} represents a dir and has no `content` attribute")
@LazyProperty
def nodes(self):
if self.commit:
nodes = self.commit.get_nodes(self.path, pre_load=self.default_pre_load)
else:
nodes = self._nodes
self._nodes_dict = {node.path: node for node in nodes}
return sorted(nodes)
@LazyProperty
def files(self):
return sorted(node for node in self.nodes if node.is_file())
@LazyProperty
def dirs(self):
return sorted(node for node in self.nodes if node.is_dir())
def get_node(self, path):
"""
Returns node from within this particular ``DirNode``, so it is now
allowed to fetch, i.e. node located at 'docs/api/index.rst' from node
'docs'. In order to access deeper nodes one must fetch nodes between
them first - this would work::
docs = root.get_node('docs')
docs.get_node('api').get_node('index.rst')
:param: path - relative to the current node
.. note::
To access lazily (as in example above) node have to be initialized
with related commit object - without it node is out of
context and may know nothing about anything else than nearest
(located at same level) nodes.
"""
try:
path = path.rstrip('/')
if path == '':
raise NodeError("Cannot retrieve node without path")
self.nodes # access nodes first in order to set _nodes_dict
paths = path.split('/')
if len(paths) == 1:
if not self.is_root():
path = '/'.join((self.path, paths[0]))
else:
path = paths[0]
return self._nodes_dict[path]
elif len(paths) > 1:
if self.commit is None:
raise NodeError("Cannot access deeper nodes without commit")
else:
path1, path2 = paths[0], '/'.join(paths[1:])
return self.get_node(path1).get_node(path2)
else:
raise KeyError
except KeyError:
raise NodeError(f"Node does not exist at {path}")
@LazyProperty
def state(self):
raise NodeError("Cannot access state of DirNode")
@LazyProperty
def size(self):
size = 0
for root, dirs, files in self.commit.walk(self.path):
for f in files:
size += f.size
return size
@LazyProperty
def last_commit(self):
if self.commit:
pre_load = ["author", "date", "message", "parents"]
return self.commit.get_path_commit(self.path, pre_load=pre_load)
raise NodeError(
"Cannot retrieve last commit of the file without "
"related commit attribute")
def __repr__(self):
short_id = getattr(self.commit, 'short_id', '')
return f'<{self.__class__.__name__} {self.path!r} @ {short_id}>'
class RootNode(DirNode):
"""
DirNode being the root node of the repository.
"""
def __init__(self, nodes=(), commit=None):
super().__init__(path=b'', nodes=nodes, commit=commit)
def __repr__(self):
return f'<{self.__class__.__name__}>'
class SubModuleNode(Node):
"""
represents a SubModule of Git or SubRepo of Mercurial
"""
is_binary = False
size = 0
def __init__(self, name, url=None, commit=None, alias=None):
self.path = name
self.kind = NodeKind.SUBMODULE
self.alias = alias
# we have to use EmptyCommit here since this can point to svn/git/hg
# submodules we cannot get from repository
self.commit = EmptyCommit(str(commit), alias=alias)
self.url = url or self._extract_submodule_url()
def __repr__(self):
short_id = getattr(self.commit, 'short_id', '')
return f'<{self.__class__.__name__} {self.path!r} @ {short_id}>'
def _extract_submodule_url(self):
# TODO: find a way to parse gits submodule file and extract the
# linking URL
return self.path
@LazyProperty
def name(self):
"""
Returns name of the node so if its path
then only last part is returned.
"""
org = safe_str(self.path.rstrip('/').split('/')[-1])
return f'{org} @ {self.commit.short_id}'
class LargeFileNode(FileNode):
def __init__(self, path, url=None, commit=None, alias=None, org_path=None):
self._validate_path(path) # can throw exception if path is invalid
self.org_path = org_path # as stored in VCS as LF pointer
self.bytes_path = path.rstrip(b'/') # store for __repr__
self.path = safe_str(self.bytes_path) # we store paths as str
self.kind = NodeKind.LARGEFILE
self.alias = alias
self._content = b''
def _validate_path(self, path: bytes):
"""
we override check since the LargeFileNode path is system absolute, but we check for bytes only
"""
self._assert_bytes(path)
def __repr__(self):
return f'<{self.__class__.__name__} {self.org_path} -> {self.path!r}>'
@LazyProperty
def size(self):
return os.stat(self.path).st_size
@LazyProperty
def raw_bytes(self):
with open(self.path, 'rb') as f:
content = f.read()
return content
@LazyProperty
def name(self):
"""
Overwrites name to be the org lf path
"""
return self.org_path
def stream_bytes(self):
with open(self.path, 'rb') as stream:
while True:
data = stream.read(16 * 1024)
if not data:
break
yield data