diff --git a/.travis.yml b/.travis.yml index da03b82b..8e7713ef 100644 --- a/.travis.yml +++ b/.travis.yml @@ -1,67 +1,68 @@ language: python sudo: false cache: pip python: - 3.5 - 3.6 - 3.7 - 3.8 - pypy3.5 env: - PYTHONHASHSEED=random TEST_REQUIRE="gevent greenlet geventhttpclient fastimport" PURE=false matrix: include: - python: pypy env: TEST_REQUIRE=fastimport - python: 3.7 env: TEST_REQUIRE=fastimport dist: xenial sudo: true - python: 3.8 env: TEST_REQUIRE=fastimport dist: xenial sudo: true - python: 3.8 env: TEST_REQUIRE=fastimport dist: xenial sudo: true - python: 3.6 env: PURE=true install: - travis_retry pip install -U pip coverage codecov flake8 $TEST_REQUIRE script: - if [ $PURE = false ]; then python setup.py build_ext -i; fi - python -m coverage run -p -m unittest dulwich.tests.test_suite # Style - make style + - make typing - if [ $PURE = true ]; then SETUP_ARGS=--pure; fi - python setup.py $SETUP_ARGS bdist_wheel after_success: - python -m coverage combine - codecov deploy: provider: pypi user: dulwich-bot password: secure: Q8DDDojBugQWzXvmmEQiU90UkVPk+OYoFZwv1H9LYpQ4u5CfwQNWpf8qXYhlGMdr/gzWaSWsqLvgWLpzfkvqS4Vyk2bO9mr+dSskfD8uwc82LiiL9CNd/NY03CjH9RaFgVMD/+exMjY/yCtlyH1jL4kjgOyNnC+x4B37CliZHcE= skip_cleanup: true skip_existing: true file_glob: true file: - dist/dulwich*.whl - dist/dulwich*.tar.gz on: tags: true repo: dulwich/dulwich diff --git a/Makefile b/Makefile index c4efce29..c0c7047d 100644 --- a/Makefile +++ b/Makefile @@ -1,66 +1,69 @@ PYTHON = python3 PYFLAKES = pyflakes PEP8 = pep8 FLAKE8 ?= flake8 SETUP = $(PYTHON) setup.py TESTRUNNER ?= unittest RUNTEST = PYTHONHASHSEED=random PYTHONPATH=$(shell pwd)$(if $(PYTHONPATH),:$(PYTHONPATH),) $(PYTHON) -m $(TESTRUNNER) $(TEST_OPTIONS) COVERAGE = python3-coverage DESTDIR=/ all: build doc:: sphinx sphinx:: $(MAKE) -C docs html build:: $(SETUP) build $(SETUP) build_ext -i install:: $(SETUP) install --root="$(DESTDIR)" check:: build $(RUNTEST) dulwich.tests.test_suite check-tutorial:: build $(RUNTEST) dulwich.tests.tutorial_test_suite check-nocompat:: build $(RUNTEST) dulwich.tests.nocompat_test_suite check-compat:: build $(RUNTEST) dulwich.tests.compat_test_suite check-pypy:: clean $(MAKE) check-noextensions PYTHON=pypy check-noextensions:: clean $(RUNTEST) dulwich.tests.test_suite check-all: check check-pypy check-noextensions +typing: + mypy dulwich + clean:: $(SETUP) clean --all rm -f dulwich/*.so flakes: $(PYFLAKES) dulwich pep8: $(PEP8) dulwich style: $(FLAKE8) before-push: check git diff origin/master | $(PEP8) --diff coverage: $(COVERAGE) run -m unittest dulwich.tests.test_suite dulwich.contrib.test_suite coverage-html: coverage $(COVERAGE) html diff --git a/dulwich/diff_tree.py b/dulwich/diff_tree.py index d81fa19e..f4cc38ac 100644 --- a/dulwich/diff_tree.py +++ b/dulwich/diff_tree.py @@ -1,623 +1,627 @@ # diff_tree.py -- Utilities for diffing files and trees. # Copyright (C) 2010 Google, Inc. # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Utilities for diffing files and trees.""" from collections import ( defaultdict, namedtuple, ) from io import BytesIO from itertools import chain import stat from dulwich.objects import ( S_ISGITLINK, TreeEntry, ) # TreeChange type constants. CHANGE_ADD = 'add' CHANGE_MODIFY = 'modify' CHANGE_DELETE = 'delete' CHANGE_RENAME = 'rename' CHANGE_COPY = 'copy' CHANGE_UNCHANGED = 'unchanged' RENAME_CHANGE_TYPES = (CHANGE_RENAME, CHANGE_COPY) _NULL_ENTRY = TreeEntry(None, None, None) _MAX_SCORE = 100 RENAME_THRESHOLD = 60 MAX_FILES = 200 REWRITE_THRESHOLD = None class TreeChange(namedtuple('TreeChange', ['type', 'old', 'new'])): """Named tuple a single change between two trees.""" @classmethod def add(cls, new): return cls(CHANGE_ADD, _NULL_ENTRY, new) @classmethod def delete(cls, old): return cls(CHANGE_DELETE, old, _NULL_ENTRY) def _tree_entries(path, tree): result = [] if not tree: return result for entry in tree.iteritems(name_order=True): result.append(entry.in_path(path)) return result def _merge_entries(path, tree1, tree2): """Merge the entries of two trees. Args: path: A path to prepend to all tree entry names. tree1: The first Tree object to iterate, or None. tree2: The second Tree object to iterate, or None. Returns: A list of pairs of TreeEntry objects for each pair of entries in the trees. If an entry exists in one tree but not the other, the other entry will have all attributes set to None. If neither entry's path is None, they are guaranteed to match. """ entries1 = _tree_entries(path, tree1) entries2 = _tree_entries(path, tree2) i1 = i2 = 0 len1 = len(entries1) len2 = len(entries2) result = [] while i1 < len1 and i2 < len2: entry1 = entries1[i1] entry2 = entries2[i2] if entry1.path < entry2.path: result.append((entry1, _NULL_ENTRY)) i1 += 1 elif entry1.path > entry2.path: result.append((_NULL_ENTRY, entry2)) i2 += 1 else: result.append((entry1, entry2)) i1 += 1 i2 += 1 for i in range(i1, len1): result.append((entries1[i], _NULL_ENTRY)) for i in range(i2, len2): result.append((_NULL_ENTRY, entries2[i])) return result def _is_tree(entry): mode = entry.mode if mode is None: return False return stat.S_ISDIR(mode) def walk_trees(store, tree1_id, tree2_id, prune_identical=False): """Recursively walk all the entries of two trees. Iteration is depth-first pre-order, as in e.g. os.walk. Args: store: An ObjectStore for looking up objects. tree1_id: The SHA of the first Tree object to iterate, or None. tree2_id: The SHA of the second Tree object to iterate, or None. param prune_identical: If True, identical subtrees will not be walked. Returns: Iterator over Pairs of TreeEntry objects for each pair of entries in the trees and their subtrees recursively. If an entry exists in one tree but not the other, the other entry will have all attributes set to None. If neither entry's path is None, they are guaranteed to match. """ # This could be fairly easily generalized to >2 trees if we find a use # case. mode1 = tree1_id and stat.S_IFDIR or None mode2 = tree2_id and stat.S_IFDIR or None todo = [(TreeEntry(b'', mode1, tree1_id), TreeEntry(b'', mode2, tree2_id))] while todo: entry1, entry2 = todo.pop() is_tree1 = _is_tree(entry1) is_tree2 = _is_tree(entry2) if prune_identical and is_tree1 and is_tree2 and entry1 == entry2: continue tree1 = is_tree1 and store[entry1.sha] or None tree2 = is_tree2 and store[entry2.sha] or None path = entry1.path or entry2.path todo.extend(reversed(_merge_entries(path, tree1, tree2))) yield entry1, entry2 def _skip_tree(entry, include_trees): if entry.mode is None or (not include_trees and stat.S_ISDIR(entry.mode)): return _NULL_ENTRY return entry def tree_changes(store, tree1_id, tree2_id, want_unchanged=False, rename_detector=None, include_trees=False, change_type_same=False): """Find the differences between the contents of two trees. Args: store: An ObjectStore for looking up objects. tree1_id: The SHA of the source tree. tree2_id: The SHA of the target tree. want_unchanged: If True, include TreeChanges for unmodified entries as well. include_trees: Whether to include trees rename_detector: RenameDetector object for detecting renames. change_type_same: Whether to report change types in the same entry or as delete+add. Returns: Iterator over TreeChange instances for each change between the source and target tree. """ if (rename_detector is not None and tree1_id is not None and tree2_id is not None): for change in rename_detector.changes_with_renames( tree1_id, tree2_id, want_unchanged=want_unchanged, include_trees=include_trees): yield change return entries = walk_trees(store, tree1_id, tree2_id, prune_identical=(not want_unchanged)) for entry1, entry2 in entries: if entry1 == entry2 and not want_unchanged: continue # Treat entries for trees as missing. entry1 = _skip_tree(entry1, include_trees) entry2 = _skip_tree(entry2, include_trees) if entry1 != _NULL_ENTRY and entry2 != _NULL_ENTRY: if (stat.S_IFMT(entry1.mode) != stat.S_IFMT(entry2.mode) and not change_type_same): # File type changed: report as delete/add. yield TreeChange.delete(entry1) entry1 = _NULL_ENTRY change_type = CHANGE_ADD elif entry1 == entry2: change_type = CHANGE_UNCHANGED else: change_type = CHANGE_MODIFY elif entry1 != _NULL_ENTRY: change_type = CHANGE_DELETE elif entry2 != _NULL_ENTRY: change_type = CHANGE_ADD else: # Both were None because at least one was a tree. continue yield TreeChange(change_type, entry1, entry2) def _all_eq(seq, key, value): for e in seq: if key(e) != value: return False return True def _all_same(seq, key): return _all_eq(seq[1:], key, key(seq[0])) def tree_changes_for_merge(store, parent_tree_ids, tree_id, rename_detector=None): """Get the tree changes for a merge tree relative to all its parents. Args: store: An ObjectStore for looking up objects. parent_tree_ids: An iterable of the SHAs of the parent trees. tree_id: The SHA of the merge tree. rename_detector: RenameDetector object for detecting renames. Returns: Iterator over lists of TreeChange objects, one per conflicted path in the merge. Each list contains one element per parent, with the TreeChange for that path relative to that parent. An element may be None if it never existed in one parent and was deleted in two others. A path is only included in the output if it is a conflict, i.e. its SHA in the merge tree is not found in any of the parents, or in the case of deletes, if not all of the old SHAs match. """ all_parent_changes = [tree_changes(store, t, tree_id, rename_detector=rename_detector) for t in parent_tree_ids] num_parents = len(parent_tree_ids) changes_by_path = defaultdict(lambda: [None] * num_parents) # Organize by path. for i, parent_changes in enumerate(all_parent_changes): for change in parent_changes: if change.type == CHANGE_DELETE: path = change.old.path else: path = change.new.path changes_by_path[path][i] = change def old_sha(c): return c.old.sha def change_type(c): return c.type # Yield only conflicting changes. for _, changes in sorted(changes_by_path.items()): assert len(changes) == num_parents have = [c for c in changes if c is not None] if _all_eq(have, change_type, CHANGE_DELETE): if not _all_same(have, old_sha): yield changes elif not _all_same(have, change_type): yield changes elif None not in changes: # If no change was found relative to one parent, that means the SHA # must have matched the SHA in that parent, so it is not a # conflict. yield changes _BLOCK_SIZE = 64 def _count_blocks(obj): """Count the blocks in an object. Splits the data into blocks either on lines or <=64-byte chunks of lines. Args: obj: The object to count blocks for. Returns: A dict of block hashcode -> total bytes occurring. """ block_counts = defaultdict(int) block = BytesIO() n = 0 # Cache attrs as locals to avoid expensive lookups in the inner loop. block_write = block.write block_seek = block.seek block_truncate = block.truncate block_getvalue = block.getvalue for c in chain(*obj.as_raw_chunks()): c = c.to_bytes(1, 'big') block_write(c) n += 1 if c == b'\n' or n == _BLOCK_SIZE: value = block_getvalue() block_counts[hash(value)] += len(value) block_seek(0) block_truncate() n = 0 if n > 0: last_block = block_getvalue() block_counts[hash(last_block)] += len(last_block) return block_counts def _common_bytes(blocks1, blocks2): """Count the number of common bytes in two block count dicts. Args: block1: The first dict of block hashcode -> total bytes. block2: The second dict of block hashcode -> total bytes. Returns: The number of bytes in common between blocks1 and blocks2. This is only approximate due to possible hash collisions. """ # Iterate over the smaller of the two dicts, since this is symmetrical. if len(blocks1) > len(blocks2): blocks1, blocks2 = blocks2, blocks1 score = 0 for block, count1 in blocks1.items(): count2 = blocks2.get(block) if count2: score += min(count1, count2) return score def _similarity_score(obj1, obj2, block_cache=None): """Compute a similarity score for two objects. Args: obj1: The first object to score. obj2: The second object to score. block_cache: An optional dict of SHA to block counts to cache results between calls. Returns: The similarity score between the two objects, defined as the number of bytes in common between the two objects divided by the maximum size, scaled to the range 0-100. """ if block_cache is None: block_cache = {} if obj1.id not in block_cache: block_cache[obj1.id] = _count_blocks(obj1) if obj2.id not in block_cache: block_cache[obj2.id] = _count_blocks(obj2) common_bytes = _common_bytes(block_cache[obj1.id], block_cache[obj2.id]) max_size = max(obj1.raw_length(), obj2.raw_length()) if not max_size: return _MAX_SCORE return int(float(common_bytes) * _MAX_SCORE / max_size) def _tree_change_key(entry): # Sort by old path then new path. If only one exists, use it for both keys. path1 = entry.old.path path2 = entry.new.path if path1 is None: path1 = path2 if path2 is None: path2 = path1 return (path1, path2) class RenameDetector(object): """Object for handling rename detection between two trees.""" def __init__(self, store, rename_threshold=RENAME_THRESHOLD, max_files=MAX_FILES, rewrite_threshold=REWRITE_THRESHOLD, find_copies_harder=False): """Initialize the rename detector. Args: store: An ObjectStore for looking up objects. rename_threshold: The threshold similarity score for considering an add/delete pair to be a rename/copy; see _similarity_score. max_files: The maximum number of adds and deletes to consider, or None for no limit. The detector is guaranteed to compare no more than max_files ** 2 add/delete pairs. This limit is provided because rename detection can be quadratic in the project size. If the limit is exceeded, no content rename detection is attempted. rewrite_threshold: The threshold similarity score below which a modify should be considered a delete/add, or None to not break modifies; see _similarity_score. find_copies_harder: If True, consider unmodified files when detecting copies. """ self._store = store self._rename_threshold = rename_threshold self._rewrite_threshold = rewrite_threshold self._max_files = max_files self._find_copies_harder = find_copies_harder self._want_unchanged = False def _reset(self): self._adds = [] self._deletes = [] self._changes = [] def _should_split(self, change): if (self._rewrite_threshold is None or change.type != CHANGE_MODIFY or change.old.sha == change.new.sha): return False old_obj = self._store[change.old.sha] new_obj = self._store[change.new.sha] return _similarity_score(old_obj, new_obj) < self._rewrite_threshold def _add_change(self, change): if change.type == CHANGE_ADD: self._adds.append(change) elif change.type == CHANGE_DELETE: self._deletes.append(change) elif self._should_split(change): self._deletes.append(TreeChange.delete(change.old)) self._adds.append(TreeChange.add(change.new)) elif ((self._find_copies_harder and change.type == CHANGE_UNCHANGED) or change.type == CHANGE_MODIFY): # Treat all modifies as potential deletes for rename detection, # but don't split them (to avoid spurious renames). Setting # find_copies_harder means we treat unchanged the same as # modified. self._deletes.append(change) else: self._changes.append(change) def _collect_changes(self, tree1_id, tree2_id): want_unchanged = self._find_copies_harder or self._want_unchanged for change in tree_changes(self._store, tree1_id, tree2_id, want_unchanged=want_unchanged, include_trees=self._include_trees): self._add_change(change) def _prune(self, add_paths, delete_paths): self._adds = [a for a in self._adds if a.new.path not in add_paths] self._deletes = [d for d in self._deletes if d.old.path not in delete_paths] def _find_exact_renames(self): add_map = defaultdict(list) for add in self._adds: add_map[add.new.sha].append(add.new) delete_map = defaultdict(list) for delete in self._deletes: # Keep track of whether the delete was actually marked as a delete. # If not, it needs to be marked as a copy. is_delete = delete.type == CHANGE_DELETE delete_map[delete.old.sha].append((delete.old, is_delete)) add_paths = set() delete_paths = set() for sha, sha_deletes in delete_map.items(): sha_adds = add_map[sha] for (old, is_delete), new in zip(sha_deletes, sha_adds): if stat.S_IFMT(old.mode) != stat.S_IFMT(new.mode): continue if is_delete: delete_paths.add(old.path) add_paths.add(new.path) new_type = is_delete and CHANGE_RENAME or CHANGE_COPY self._changes.append(TreeChange(new_type, old, new)) num_extra_adds = len(sha_adds) - len(sha_deletes) # TODO(dborowitz): Less arbitrary way of dealing with extra copies. old = sha_deletes[0][0] if num_extra_adds > 0: for new in sha_adds[-num_extra_adds:]: add_paths.add(new.path) self._changes.append(TreeChange(CHANGE_COPY, old, new)) self._prune(add_paths, delete_paths) def _should_find_content_renames(self): return len(self._adds) * len(self._deletes) <= self._max_files ** 2 def _rename_type(self, check_paths, delete, add): if check_paths and delete.old.path == add.new.path: # If the paths match, this must be a split modify, so make sure it # comes out as a modify. return CHANGE_MODIFY elif delete.type != CHANGE_DELETE: # If it's in deletes but not marked as a delete, it must have been # added due to find_copies_harder, and needs to be marked as a # copy. return CHANGE_COPY return CHANGE_RENAME def _find_content_rename_candidates(self): candidates = self._candidates = [] # TODO: Optimizations: # - Compare object sizes before counting blocks. # - Skip if delete's S_IFMT differs from all adds. # - Skip if adds or deletes is empty. # Match C git's behavior of not attempting to find content renames if # the matrix size exceeds the threshold. if not self._should_find_content_renames(): return block_cache = {} check_paths = self._rename_threshold is not None for delete in self._deletes: if S_ISGITLINK(delete.old.mode): continue # Git links don't exist in this repo. old_sha = delete.old.sha old_obj = self._store[old_sha] block_cache[old_sha] = _count_blocks(old_obj) for add in self._adds: if stat.S_IFMT(delete.old.mode) != stat.S_IFMT(add.new.mode): continue new_obj = self._store[add.new.sha] score = _similarity_score(old_obj, new_obj, block_cache=block_cache) if score > self._rename_threshold: new_type = self._rename_type(check_paths, delete, add) rename = TreeChange(new_type, delete.old, add.new) candidates.append((-score, rename)) def _choose_content_renames(self): # Sort scores from highest to lowest, but keep names in ascending # order. self._candidates.sort() delete_paths = set() add_paths = set() for _, change in self._candidates: new_path = change.new.path if new_path in add_paths: continue old_path = change.old.path orig_type = change.type if old_path in delete_paths: change = TreeChange(CHANGE_COPY, change.old, change.new) # If the candidate was originally a copy, that means it came from a # modified or unchanged path, so we don't want to prune it. if orig_type != CHANGE_COPY: delete_paths.add(old_path) add_paths.add(new_path) self._changes.append(change) self._prune(add_paths, delete_paths) def _join_modifies(self): if self._rewrite_threshold is None: return modifies = {} delete_map = dict((d.old.path, d) for d in self._deletes) for add in self._adds: path = add.new.path delete = delete_map.get(path) if (delete is not None and stat.S_IFMT(delete.old.mode) == stat.S_IFMT(add.new.mode)): modifies[path] = TreeChange(CHANGE_MODIFY, delete.old, add.new) self._adds = [a for a in self._adds if a.new.path not in modifies] self._deletes = [a for a in self._deletes if a.new.path not in modifies] self._changes += modifies.values() def _sorted_changes(self): result = [] result.extend(self._adds) result.extend(self._deletes) result.extend(self._changes) result.sort(key=_tree_change_key) return result def _prune_unchanged(self): if self._want_unchanged: return self._deletes = [ d for d in self._deletes if d.type != CHANGE_UNCHANGED] def changes_with_renames(self, tree1_id, tree2_id, want_unchanged=False, include_trees=False): """Iterate TreeChanges between two tree SHAs, with rename detection.""" self._reset() self._want_unchanged = want_unchanged self._include_trees = include_trees self._collect_changes(tree1_id, tree2_id) self._find_exact_renames() self._find_content_rename_candidates() self._choose_content_renames() self._join_modifies() self._prune_unchanged() return self._sorted_changes() # Hold on to the pure-python implementations for testing. _is_tree_py = _is_tree _merge_entries_py = _merge_entries _count_blocks_py = _count_blocks try: # Try to import C versions - from dulwich._diff_tree import _is_tree, _merge_entries, _count_blocks + from dulwich._diff_tree import ( # type: ignore + _is_tree, + _merge_entries, + _count_blocks, + ) except ImportError: pass diff --git a/dulwich/objects.py b/dulwich/objects.py index ed062390..e081eac7 100644 --- a/dulwich/objects.py +++ b/dulwich/objects.py @@ -1,1434 +1,1443 @@ # objects.py -- Access to base git objects # Copyright (C) 2007 James Westby # Copyright (C) 2008-2013 Jelmer Vernooij # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Access to base git objects.""" import binascii from io import BytesIO from collections import namedtuple import os import posixpath import stat +from typing import ( + Optional, + Dict, + Union, + Type, + ) import warnings import zlib from hashlib import sha1 from dulwich.errors import ( ChecksumMismatch, NotBlobError, NotCommitError, NotTagError, NotTreeError, ObjectFormatException, FileFormatException, ) from dulwich.file import GitFile ZERO_SHA = b'0' * 40 # Header fields for commits _TREE_HEADER = b'tree' _PARENT_HEADER = b'parent' _AUTHOR_HEADER = b'author' _COMMITTER_HEADER = b'committer' _ENCODING_HEADER = b'encoding' _MERGETAG_HEADER = b'mergetag' _GPGSIG_HEADER = b'gpgsig' # Header fields for objects _OBJECT_HEADER = b'object' _TYPE_HEADER = b'type' _TAG_HEADER = b'tag' _TAGGER_HEADER = b'tagger' S_IFGITLINK = 0o160000 MAX_TIME = 9223372036854775807 # (2**63) - 1 - signed long int max BEGIN_PGP_SIGNATURE = b"-----BEGIN PGP SIGNATURE-----" class EmptyFileException(FileFormatException): """An unexpectedly empty file was encountered.""" def S_ISGITLINK(m): """Check if a mode indicates a submodule. Args: m: Mode to check Returns: a ``boolean`` """ return (stat.S_IFMT(m) == S_IFGITLINK) def _decompress(string): dcomp = zlib.decompressobj() dcomped = dcomp.decompress(string) dcomped += dcomp.flush() return dcomped def sha_to_hex(sha): """Takes a string and returns the hex of the sha within""" hexsha = binascii.hexlify(sha) assert len(hexsha) == 40, "Incorrect length of sha1 string: %d" % hexsha return hexsha def hex_to_sha(hex): """Takes a hex sha and returns a binary sha""" assert len(hex) == 40, "Incorrect length of hexsha: %s" % hex try: return binascii.unhexlify(hex) except TypeError as exc: if not isinstance(hex, bytes): raise raise ValueError(exc.args[0]) def valid_hexsha(hex): if len(hex) != 40: return False try: binascii.unhexlify(hex) except (TypeError, binascii.Error): return False else: return True def hex_to_filename(path, hex): """Takes a hex sha and returns its filename relative to the given path.""" # os.path.join accepts bytes or unicode, but all args must be of the same # type. Make sure that hex which is expected to be bytes, is the same type # as path. if getattr(path, 'encode', None) is not None: hex = hex.decode('ascii') dir = hex[:2] file = hex[2:] # Check from object dir return os.path.join(path, dir, file) def filename_to_hex(filename): """Takes an object filename and returns its corresponding hex sha.""" # grab the last (up to) two path components names = filename.rsplit(os.path.sep, 2)[-2:] errmsg = "Invalid object filename: %s" % filename assert len(names) == 2, errmsg base, rest = names assert len(base) == 2 and len(rest) == 38, errmsg hex = (base + rest).encode('ascii') hex_to_sha(hex) return hex -def object_header(num_type, length): +def object_header(num_type: int, length: int) -> bytes: """Return an object header for the given numeric type and text length.""" return (object_class(num_type).type_name + b' ' + str(length).encode('ascii') + b'\0') -def serializable_property(name, docstring=None): +def serializable_property(name: str, docstring:Optional[str]=None): """A property that helps tracking whether serialization is necessary. """ def set(obj, value): setattr(obj, "_"+name, value) obj._needs_serialization = True def get(obj): return getattr(obj, "_"+name) return property(get, set, doc=docstring) def object_class(type): """Get the object class corresponding to the given type. Args: type: Either a type name string or a numeric type. Returns: The ShaFile subclass corresponding to the given type, or None if type is not a valid type name/number. """ return _TYPE_MAP.get(type, None) def check_hexsha(hex, error_msg): """Check if a string is a valid hex sha string. Args: hex: Hex string to check error_msg: Error message to use in exception Raises: ObjectFormatException: Raised when the string is not valid """ if not valid_hexsha(hex): raise ObjectFormatException("%s %s" % (error_msg, hex)) def check_identity(identity, error_msg): """Check if the specified identity is valid. This will raise an exception if the identity is not valid. Args: identity: Identity string error_msg: Error message to use in exception """ email_start = identity.find(b'<') email_end = identity.find(b'>') if (email_start < 0 or email_end < 0 or email_end <= email_start or identity.find(b'<', email_start + 1) >= 0 or identity.find(b'>', email_end + 1) >= 0 or not identity.endswith(b'>')): raise ObjectFormatException(error_msg) def check_time(time_seconds): """Check if the specified time is not prone to overflow error. This will raise an exception if the time is not valid. Args: time_info: author/committer/tagger info """ # Prevent overflow error if time_seconds > MAX_TIME: raise ObjectFormatException( 'Date field should not exceed %s' % MAX_TIME) def git_line(*items): """Formats items into a space separated line.""" return b' '.join(items) + b'\n' class FixedSha(object): """SHA object that behaves like hashlib's but is given a fixed value.""" __slots__ = ('_hexsha', '_sha') def __init__(self, hexsha): if getattr(hexsha, 'encode', None) is not None: hexsha = hexsha.encode('ascii') if not isinstance(hexsha, bytes): raise TypeError('Expected bytes for hexsha, got %r' % hexsha) self._hexsha = hexsha self._sha = hex_to_sha(hexsha) def digest(self): """Return the raw SHA digest.""" return self._sha def hexdigest(self): """Return the hex SHA digest.""" return self._hexsha.decode('ascii') class ShaFile(object): """A git SHA file.""" __slots__ = ('_chunked_text', '_sha', '_needs_serialization') + type_name:bytes + type_num:int + @staticmethod def _parse_legacy_object_header(magic, f): """Parse a legacy object, creating it but not reading the file.""" bufsize = 1024 decomp = zlib.decompressobj() header = decomp.decompress(magic) start = 0 end = -1 while end < 0: extra = f.read(bufsize) header += decomp.decompress(extra) magic += extra end = header.find(b'\0', start) start = len(header) header = header[:end] type_name, size = header.split(b' ', 1) try: int(size) # sanity check except ValueError as e: raise ObjectFormatException("Object size not an integer: %s" % e) obj_class = object_class(type_name) if not obj_class: raise ObjectFormatException("Not a known type: %s" % type_name) return obj_class() def _parse_legacy_object(self, map): """Parse a legacy object, setting the raw string.""" text = _decompress(map) header_end = text.find(b'\0') if header_end < 0: raise ObjectFormatException("Invalid object header, no \\0") self.set_raw_string(text[header_end+1:]) def as_legacy_object_chunks(self, compression_level=-1): """Return chunks representing the object in the experimental format. Returns: List of strings """ compobj = zlib.compressobj(compression_level) yield compobj.compress(self._header()) for chunk in self.as_raw_chunks(): yield compobj.compress(chunk) yield compobj.flush() def as_legacy_object(self, compression_level=-1): """Return string representing the object in the experimental format. """ return b''.join(self.as_legacy_object_chunks( compression_level=compression_level)) def as_raw_chunks(self): """Return chunks with serialization of the object. Returns: List of strings, not necessarily one per line """ if self._needs_serialization: self._sha = None self._chunked_text = self._serialize() self._needs_serialization = False return self._chunked_text def as_raw_string(self): """Return raw string with serialization of the object. Returns: String object """ return b''.join(self.as_raw_chunks()) def __bytes__(self): """Return raw string serialization of this object.""" return self.as_raw_string() def __hash__(self): """Return unique hash for this object.""" return hash(self.id) def as_pretty_string(self): """Return a string representing this object, fit for display.""" return self.as_raw_string() def set_raw_string(self, text, sha=None): """Set the contents of this object from a serialized string.""" if not isinstance(text, bytes): raise TypeError('Expected bytes for text, got %r' % text) self.set_raw_chunks([text], sha) def set_raw_chunks(self, chunks, sha=None): """Set the contents of this object from a list of chunks.""" self._chunked_text = chunks self._deserialize(chunks) if sha is None: self._sha = None else: self._sha = FixedSha(sha) self._needs_serialization = False @staticmethod def _parse_object_header(magic, f): """Parse a new style object, creating it but not reading the file.""" num_type = (ord(magic[0:1]) >> 4) & 7 obj_class = object_class(num_type) if not obj_class: raise ObjectFormatException("Not a known type %d" % num_type) return obj_class() def _parse_object(self, map): """Parse a new style object, setting self._text.""" # skip type and size; type must have already been determined, and # we trust zlib to fail if it's otherwise corrupted byte = ord(map[0:1]) used = 1 while (byte & 0x80) != 0: byte = ord(map[used:used+1]) used += 1 raw = map[used:] self.set_raw_string(_decompress(raw)) @classmethod def _is_legacy_object(cls, magic): b0 = ord(magic[0:1]) b1 = ord(magic[1:2]) word = (b0 << 8) + b1 return (b0 & 0x8F) == 0x08 and (word % 31) == 0 @classmethod def _parse_file(cls, f): map = f.read() if not map: raise EmptyFileException('Corrupted empty file detected') if cls._is_legacy_object(map): obj = cls._parse_legacy_object_header(map, f) obj._parse_legacy_object(map) else: obj = cls._parse_object_header(map, f) obj._parse_object(map) return obj def __init__(self): """Don't call this directly""" self._sha = None self._chunked_text = [] self._needs_serialization = True def _deserialize(self, chunks): raise NotImplementedError(self._deserialize) def _serialize(self): raise NotImplementedError(self._serialize) @classmethod def from_path(cls, path): """Open a SHA file from disk.""" with GitFile(path, 'rb') as f: return cls.from_file(f) @classmethod def from_file(cls, f): """Get the contents of a SHA file on disk.""" try: obj = cls._parse_file(f) obj._sha = None return obj except (IndexError, ValueError): raise ObjectFormatException("invalid object header") @staticmethod def from_raw_string(type_num, string, sha=None): """Creates an object of the indicated type from the raw string given. Args: type_num: The numeric type of the object. string: The raw uncompressed contents. sha: Optional known sha for the object """ obj = object_class(type_num)() obj.set_raw_string(string, sha) return obj @staticmethod def from_raw_chunks(type_num, chunks, sha=None): """Creates an object of the indicated type from the raw chunks given. Args: type_num: The numeric type of the object. chunks: An iterable of the raw uncompressed contents. sha: Optional known sha for the object """ obj = object_class(type_num)() obj.set_raw_chunks(chunks, sha) return obj @classmethod def from_string(cls, string): """Create a ShaFile from a string.""" obj = cls() obj.set_raw_string(string) return obj def _check_has_member(self, member, error_msg): """Check that the object has a given member variable. Args: member: the member variable to check for error_msg: the message for an error if the member is missing Raises: ObjectFormatException: with the given error_msg if member is missing or is None """ if getattr(self, member, None) is None: raise ObjectFormatException(error_msg) def check(self): """Check this object for internal consistency. Raises: ObjectFormatException: if the object is malformed in some way ChecksumMismatch: if the object was created with a SHA that does not match its contents """ # TODO: if we find that error-checking during object parsing is a # performance bottleneck, those checks should be moved to the class's # check() method during optimization so we can still check the object # when necessary. old_sha = self.id try: self._deserialize(self.as_raw_chunks()) self._sha = None new_sha = self.id except Exception as e: raise ObjectFormatException(e) if old_sha != new_sha: raise ChecksumMismatch(new_sha, old_sha) def _header(self): return object_header(self.type, self.raw_length()) def raw_length(self): """Returns the length of the raw string of this object.""" ret = 0 for chunk in self.as_raw_chunks(): ret += len(chunk) return ret def sha(self): """The SHA1 object that is the name of this object.""" if self._sha is None or self._needs_serialization: # this is a local because as_raw_chunks() overwrites self._sha new_sha = sha1() new_sha.update(self._header()) for chunk in self.as_raw_chunks(): new_sha.update(chunk) self._sha = new_sha return self._sha def copy(self): """Create a new copy of this SHA1 object from its raw string""" obj_class = object_class(self.get_type()) return obj_class.from_raw_string( self.get_type(), self.as_raw_string(), self.id) @property def id(self): """The hex SHA of this object.""" return self.sha().hexdigest().encode('ascii') def get_type(self): """Return the type number for this object class.""" return self.type_num def set_type(self, type): """Set the type number for this object class.""" self.type_num = type # DEPRECATED: use type_num or type_name as needed. type = property(get_type, set_type) def __repr__(self): return "<%s %s>" % (self.__class__.__name__, self.id) def __ne__(self, other): """Check whether this object does not match the other.""" return not isinstance(other, ShaFile) or self.id != other.id def __eq__(self, other): """Return True if the SHAs of the two objects match. """ return isinstance(other, ShaFile) and self.id == other.id def __lt__(self, other): """Return whether SHA of this object is less than the other. """ if not isinstance(other, ShaFile): raise TypeError return self.id < other.id def __le__(self, other): """Check whether SHA of this object is less than or equal to the other. """ if not isinstance(other, ShaFile): raise TypeError return self.id <= other.id def __cmp__(self, other): """Compare the SHA of this object with that of the other object. """ if not isinstance(other, ShaFile): raise TypeError return cmp(self.id, other.id) # noqa: F821 class Blob(ShaFile): """A Git Blob object.""" __slots__ = () type_name = b'blob' type_num = 3 def __init__(self): super(Blob, self).__init__() self._chunked_text = [] self._needs_serialization = False def _get_data(self): return self.as_raw_string() def _set_data(self, data): self.set_raw_string(data) data = property(_get_data, _set_data, doc="The text contained within the blob object.") def _get_chunked(self): return self._chunked_text def _set_chunked(self, chunks): self._chunked_text = chunks def _serialize(self): return self._chunked_text def _deserialize(self, chunks): self._chunked_text = chunks chunked = property( _get_chunked, _set_chunked, doc="The text in the blob object, as chunks (not necessarily lines)") @classmethod def from_path(cls, path): blob = ShaFile.from_path(path) if not isinstance(blob, cls): raise NotBlobError(path) return blob def check(self): """Check this object for internal consistency. Raises: ObjectFormatException: if the object is malformed in some way """ super(Blob, self).check() def splitlines(self): """Return list of lines in this blob. This preserves the original line endings. """ chunks = self.chunked if not chunks: return [] if len(chunks) == 1: return chunks[0].splitlines(True) remaining = None ret = [] for chunk in chunks: lines = chunk.splitlines(True) if len(lines) > 1: ret.append((remaining or b"") + lines[0]) ret.extend(lines[1:-1]) remaining = lines[-1] elif len(lines) == 1: if remaining is None: remaining = lines.pop() else: remaining += lines.pop() if remaining is not None: ret.append(remaining) return ret def _parse_message(chunks): """Parse a message with a list of fields and a body. Args: chunks: the raw chunks of the tag or commit object. Returns: iterator of tuples of (field, value), one per header line, in the order read from the text, possibly including duplicates. Includes a field named None for the freeform tag/commit text. """ f = BytesIO(b''.join(chunks)) k = None v = "" eof = False def _strip_last_newline(value): """Strip the last newline from value""" if value and value.endswith(b'\n'): return value[:-1] return value # Parse the headers # # Headers can contain newlines. The next line is indented with a space. # We store the latest key as 'k', and the accumulated value as 'v'. for line in f: if line.startswith(b' '): # Indented continuation of the previous line v += line[1:] else: if k is not None: # We parsed a new header, return its value yield (k, _strip_last_newline(v)) if line == b'\n': # Empty line indicates end of headers break (k, v) = line.split(b' ', 1) else: # We reached end of file before the headers ended. We still need to # return the previous header, then we need to return a None field for # the text. eof = True if k is not None: yield (k, _strip_last_newline(v)) yield (None, None) if not eof: # We didn't reach the end of file while parsing headers. We can return # the rest of the file as a message. yield (None, f.read()) f.close() class Tag(ShaFile): """A Git Tag object.""" type_name = b'tag' type_num = 4 __slots__ = ('_tag_timezone_neg_utc', '_name', '_object_sha', '_object_class', '_tag_time', '_tag_timezone', '_tagger', '_message', '_signature') def __init__(self): super(Tag, self).__init__() self._tagger = None self._tag_time = None self._tag_timezone = None self._tag_timezone_neg_utc = False self._signature = None @classmethod def from_path(cls, filename): tag = ShaFile.from_path(filename) if not isinstance(tag, cls): raise NotTagError(filename) return tag def check(self): """Check this object for internal consistency. Raises: ObjectFormatException: if the object is malformed in some way """ super(Tag, self).check() self._check_has_member("_object_sha", "missing object sha") self._check_has_member("_object_class", "missing object type") self._check_has_member("_name", "missing tag name") if not self._name: raise ObjectFormatException("empty tag name") check_hexsha(self._object_sha, "invalid object sha") if getattr(self, "_tagger", None): check_identity(self._tagger, "invalid tagger") self._check_has_member("_tag_time", "missing tag time") check_time(self._tag_time) last = None for field, _ in _parse_message(self._chunked_text): if field == _OBJECT_HEADER and last is not None: raise ObjectFormatException("unexpected object") elif field == _TYPE_HEADER and last != _OBJECT_HEADER: raise ObjectFormatException("unexpected type") elif field == _TAG_HEADER and last != _TYPE_HEADER: raise ObjectFormatException("unexpected tag name") elif field == _TAGGER_HEADER and last != _TAG_HEADER: raise ObjectFormatException("unexpected tagger") last = field def _serialize(self): chunks = [] chunks.append(git_line(_OBJECT_HEADER, self._object_sha)) chunks.append(git_line(_TYPE_HEADER, self._object_class.type_name)) chunks.append(git_line(_TAG_HEADER, self._name)) if self._tagger: if self._tag_time is None: chunks.append(git_line(_TAGGER_HEADER, self._tagger)) else: chunks.append(git_line( _TAGGER_HEADER, self._tagger, str(self._tag_time).encode('ascii'), format_timezone( self._tag_timezone, self._tag_timezone_neg_utc))) if self._message is not None: chunks.append(b'\n') # To close headers chunks.append(self._message) if self._signature is not None: chunks.append(self._signature) return chunks def _deserialize(self, chunks): """Grab the metadata attached to the tag""" self._tagger = None self._tag_time = None self._tag_timezone = None self._tag_timezone_neg_utc = False for field, value in _parse_message(chunks): if field == _OBJECT_HEADER: self._object_sha = value elif field == _TYPE_HEADER: obj_class = object_class(value) if not obj_class: raise ObjectFormatException("Not a known type: %s" % value) self._object_class = obj_class elif field == _TAG_HEADER: self._name = value elif field == _TAGGER_HEADER: (self._tagger, self._tag_time, (self._tag_timezone, self._tag_timezone_neg_utc)) = parse_time_entry(value) elif field is None: if value is None: self._message = None self._signature = None else: try: sig_idx = value.index(BEGIN_PGP_SIGNATURE) except ValueError: self._message = value self._signature = None else: self._message = value[:sig_idx] self._signature = value[sig_idx:] else: raise ObjectFormatException("Unknown field %s" % field) def _get_object(self): """Get the object pointed to by this tag. Returns: tuple of (object class, sha). """ return (self._object_class, self._object_sha) def _set_object(self, value): (self._object_class, self._object_sha) = value self._needs_serialization = True object = property(_get_object, _set_object) name = serializable_property("name", "The name of this tag") tagger = serializable_property( "tagger", "Returns the name of the person who created this tag") tag_time = serializable_property( "tag_time", "The creation timestamp of the tag. As the number of seconds " "since the epoch") tag_timezone = serializable_property( "tag_timezone", "The timezone that tag_time is in.") message = serializable_property( "message", "the message attached to this tag") signature = serializable_property( "signature", "Optional detached GPG signature") class TreeEntry(namedtuple('TreeEntry', ['path', 'mode', 'sha'])): """Named tuple encapsulating a single tree entry.""" def in_path(self, path): """Return a copy of this entry with the given path prepended.""" if not isinstance(self.path, bytes): raise TypeError('Expected bytes for path, got %r' % path) return TreeEntry(posixpath.join(path, self.path), self.mode, self.sha) def parse_tree(text, strict=False): """Parse a tree text. Args: text: Serialized text to parse Returns: iterator of tuples of (name, mode, sha) Raises: ObjectFormatException: if the object was malformed in some way """ count = 0 length = len(text) while count < length: mode_end = text.index(b' ', count) mode_text = text[count:mode_end] if strict and mode_text.startswith(b'0'): raise ObjectFormatException("Invalid mode '%s'" % mode_text) try: mode = int(mode_text, 8) except ValueError: raise ObjectFormatException("Invalid mode '%s'" % mode_text) name_end = text.index(b'\0', mode_end) name = text[mode_end+1:name_end] count = name_end+21 sha = text[name_end+1:count] if len(sha) != 20: raise ObjectFormatException("Sha has invalid length") hexsha = sha_to_hex(sha) yield (name, mode, hexsha) def serialize_tree(items): """Serialize the items in a tree to a text. Args: items: Sorted iterable over (name, mode, sha) tuples Returns: Serialized tree text as chunks """ for name, mode, hexsha in items: yield (("%04o" % mode).encode('ascii') + b' ' + name + b'\0' + hex_to_sha(hexsha)) def sorted_tree_items(entries, name_order): """Iterate over a tree entries dictionary. Args: name_order: If True, iterate entries in order of their name. If False, iterate entries in tree order, that is, treat subtree entries as having '/' appended. entries: Dictionary mapping names to (mode, sha) tuples Returns: Iterator over (name, mode, hexsha) """ key_func = name_order and key_entry_name_order or key_entry for name, entry in sorted(entries.items(), key=key_func): mode, hexsha = entry # Stricter type checks than normal to mirror checks in the C version. mode = int(mode) if not isinstance(hexsha, bytes): raise TypeError('Expected bytes for SHA, got %r' % hexsha) yield TreeEntry(name, mode, hexsha) def key_entry(entry): """Sort key for tree entry. Args: entry: (name, value) tuplee """ (name, value) = entry if stat.S_ISDIR(value[0]): name += b'/' return name def key_entry_name_order(entry): """Sort key for tree entry in name order.""" return entry[0] def pretty_format_tree_entry(name, mode, hexsha, encoding="utf-8"): """Pretty format tree entry. Args: name: Name of the directory entry mode: Mode of entry hexsha: Hexsha of the referenced object Returns: string describing the tree entry """ if mode & stat.S_IFDIR: kind = "tree" else: kind = "blob" return "%04o %s %s\t%s\n" % ( mode, kind, hexsha.decode('ascii'), name.decode(encoding, 'replace')) class Tree(ShaFile): """A Git tree object""" type_name = b'tree' type_num = 2 __slots__ = ('_entries') def __init__(self): super(Tree, self).__init__() self._entries = {} @classmethod def from_path(cls, filename): tree = ShaFile.from_path(filename) if not isinstance(tree, cls): raise NotTreeError(filename) return tree def __contains__(self, name): return name in self._entries def __getitem__(self, name): return self._entries[name] def __setitem__(self, name, value): """Set a tree entry by name. Args: name: The name of the entry, as a string. value: A tuple of (mode, hexsha), where mode is the mode of the entry as an integral type and hexsha is the hex SHA of the entry as a string. """ mode, hexsha = value self._entries[name] = (mode, hexsha) self._needs_serialization = True def __delitem__(self, name): del self._entries[name] self._needs_serialization = True def __len__(self): return len(self._entries) def __iter__(self): return iter(self._entries) def add(self, name, mode, hexsha): """Add an entry to the tree. Args: mode: The mode of the entry as an integral type. Not all possible modes are supported by git; see check() for details. name: The name of the entry, as a string. hexsha: The hex SHA of the entry as a string. """ if isinstance(name, int) and isinstance(mode, bytes): (name, mode) = (mode, name) warnings.warn( "Please use Tree.add(name, mode, hexsha)", category=DeprecationWarning, stacklevel=2) self._entries[name] = mode, hexsha self._needs_serialization = True def iteritems(self, name_order=False): """Iterate over entries. Args: name_order: If True, iterate in name order instead of tree order. Returns: Iterator over (name, mode, sha) tuples """ return sorted_tree_items(self._entries, name_order) def items(self): """Return the sorted entries in this tree. Returns: List with (name, mode, sha) tuples """ return list(self.iteritems()) def _deserialize(self, chunks): """Grab the entries in the tree""" try: parsed_entries = parse_tree(b''.join(chunks)) except ValueError as e: raise ObjectFormatException(e) # TODO: list comprehension is for efficiency in the common (small) # case; if memory efficiency in the large case is a concern, use a # genexp. self._entries = dict([(n, (m, s)) for n, m, s in parsed_entries]) def check(self): """Check this object for internal consistency. Raises: ObjectFormatException: if the object is malformed in some way """ super(Tree, self).check() last = None allowed_modes = (stat.S_IFREG | 0o755, stat.S_IFREG | 0o644, stat.S_IFLNK, stat.S_IFDIR, S_IFGITLINK, # TODO: optionally exclude as in git fsck --strict stat.S_IFREG | 0o664) for name, mode, sha in parse_tree(b''.join(self._chunked_text), True): check_hexsha(sha, 'invalid sha %s' % sha) if b'/' in name or name in (b'', b'.', b'..', b'.git'): raise ObjectFormatException( 'invalid name %s' % name.decode('utf-8', 'replace')) if mode not in allowed_modes: raise ObjectFormatException('invalid mode %06o' % mode) entry = (name, (mode, sha)) if last: if key_entry(last) > key_entry(entry): raise ObjectFormatException('entries not sorted') if name == last[0]: raise ObjectFormatException('duplicate entry %s' % name) last = entry def _serialize(self): return list(serialize_tree(self.iteritems())) def as_pretty_string(self): text = [] for name, mode, hexsha in self.iteritems(): text.append(pretty_format_tree_entry(name, mode, hexsha)) return "".join(text) def lookup_path(self, lookup_obj, path): """Look up an object in a Git tree. Args: lookup_obj: Callback for retrieving object by SHA1 path: Path to lookup Returns: A tuple of (mode, SHA) of the resulting path. """ parts = path.split(b'/') sha = self.id mode = None for p in parts: if not p: continue obj = lookup_obj(sha) if not isinstance(obj, Tree): raise NotTreeError(sha) mode, sha = obj[p] return mode, sha def parse_timezone(text): """Parse a timezone text fragment (e.g. '+0100'). Args: text: Text to parse. Returns: Tuple with timezone as seconds difference to UTC and a boolean indicating whether this was a UTC timezone prefixed with a negative sign (-0000). """ # cgit parses the first character as the sign, and the rest # as an integer (using strtol), which could also be negative. # We do the same for compatibility. See #697828. if not text[0] in b'+-': raise ValueError("Timezone must start with + or - (%(text)s)" % vars()) sign = text[:1] offset = int(text[1:]) if sign == b'-': offset = -offset unnecessary_negative_timezone = (offset >= 0 and sign == b'-') signum = (offset < 0) and -1 or 1 offset = abs(offset) hours = int(offset / 100) minutes = (offset % 100) return (signum * (hours * 3600 + minutes * 60), unnecessary_negative_timezone) def format_timezone(offset, unnecessary_negative_timezone=False): """Format a timezone for Git serialization. Args: offset: Timezone offset as seconds difference to UTC unnecessary_negative_timezone: Whether to use a minus sign for UTC or positive timezones (-0000 and --700 rather than +0000 / +0700). """ if offset % 60 != 0: raise ValueError("Unable to handle non-minute offset.") if offset < 0 or unnecessary_negative_timezone: sign = '-' offset = -offset else: sign = '+' return ('%c%02d%02d' % (sign, offset / 3600, (offset / 60) % 60)).encode('ascii') def parse_time_entry(value): """Parse time entry behavior Args: value: Bytes representing a git commit/tag line Raises: ObjectFormatException in case of parsing error (malformed field date) Returns: Tuple of (author, time, (timezone, timezone_neg_utc)) """ try: sep = value.rindex(b'> ') except ValueError: return (value, None, (None, False)) try: person = value[0:sep+1] rest = value[sep+2:] timetext, timezonetext = rest.rsplit(b' ', 1) time = int(timetext) timezone, timezone_neg_utc = parse_timezone(timezonetext) except ValueError as e: raise ObjectFormatException(e) return person, time, (timezone, timezone_neg_utc) def parse_commit(chunks): """Parse a commit object from chunks. Args: chunks: Chunks to parse Returns: Tuple of (tree, parents, author_info, commit_info, encoding, mergetag, gpgsig, message, extra) """ parents = [] extra = [] tree = None author_info = (None, None, (None, None)) commit_info = (None, None, (None, None)) encoding = None mergetag = [] message = None gpgsig = None for field, value in _parse_message(chunks): # TODO(jelmer): Enforce ordering if field == _TREE_HEADER: tree = value elif field == _PARENT_HEADER: parents.append(value) elif field == _AUTHOR_HEADER: author_info = parse_time_entry(value) elif field == _COMMITTER_HEADER: commit_info = parse_time_entry(value) elif field == _ENCODING_HEADER: encoding = value elif field == _MERGETAG_HEADER: mergetag.append(Tag.from_string(value + b'\n')) elif field == _GPGSIG_HEADER: gpgsig = value elif field is None: message = value else: extra.append((field, value)) return (tree, parents, author_info, commit_info, encoding, mergetag, gpgsig, message, extra) class Commit(ShaFile): """A git commit object""" type_name = b'commit' type_num = 1 __slots__ = ('_parents', '_encoding', '_extra', '_author_timezone_neg_utc', '_commit_timezone_neg_utc', '_commit_time', '_author_time', '_author_timezone', '_commit_timezone', '_author', '_committer', '_tree', '_message', '_mergetag', '_gpgsig') def __init__(self): super(Commit, self).__init__() self._parents = [] self._encoding = None self._mergetag = [] self._gpgsig = None self._extra = [] self._author_timezone_neg_utc = False self._commit_timezone_neg_utc = False @classmethod def from_path(cls, path): commit = ShaFile.from_path(path) if not isinstance(commit, cls): raise NotCommitError(path) return commit def _deserialize(self, chunks): (self._tree, self._parents, author_info, commit_info, self._encoding, self._mergetag, self._gpgsig, self._message, self._extra) = ( parse_commit(chunks)) (self._author, self._author_time, (self._author_timezone, self._author_timezone_neg_utc)) = author_info (self._committer, self._commit_time, (self._commit_timezone, self._commit_timezone_neg_utc)) = commit_info def check(self): """Check this object for internal consistency. Raises: ObjectFormatException: if the object is malformed in some way """ super(Commit, self).check() self._check_has_member("_tree", "missing tree") self._check_has_member("_author", "missing author") self._check_has_member("_committer", "missing committer") self._check_has_member("_author_time", "missing author time") self._check_has_member("_commit_time", "missing commit time") for parent in self._parents: check_hexsha(parent, "invalid parent sha") check_hexsha(self._tree, "invalid tree sha") check_identity(self._author, "invalid author") check_identity(self._committer, "invalid committer") check_time(self._author_time) check_time(self._commit_time) last = None for field, _ in _parse_message(self._chunked_text): if field == _TREE_HEADER and last is not None: raise ObjectFormatException("unexpected tree") elif field == _PARENT_HEADER and last not in (_PARENT_HEADER, _TREE_HEADER): raise ObjectFormatException("unexpected parent") elif field == _AUTHOR_HEADER and last not in (_TREE_HEADER, _PARENT_HEADER): raise ObjectFormatException("unexpected author") elif field == _COMMITTER_HEADER and last != _AUTHOR_HEADER: raise ObjectFormatException("unexpected committer") elif field == _ENCODING_HEADER and last != _COMMITTER_HEADER: raise ObjectFormatException("unexpected encoding") last = field # TODO: optionally check for duplicate parents def _serialize(self): chunks = [] tree_bytes = ( self._tree.id if isinstance(self._tree, Tree) else self._tree) chunks.append(git_line(_TREE_HEADER, tree_bytes)) for p in self._parents: chunks.append(git_line(_PARENT_HEADER, p)) chunks.append(git_line( _AUTHOR_HEADER, self._author, str(self._author_time).encode('ascii'), format_timezone( self._author_timezone, self._author_timezone_neg_utc))) chunks.append(git_line( _COMMITTER_HEADER, self._committer, str(self._commit_time).encode('ascii'), format_timezone(self._commit_timezone, self._commit_timezone_neg_utc))) if self.encoding: chunks.append(git_line(_ENCODING_HEADER, self.encoding)) for mergetag in self.mergetag: mergetag_chunks = mergetag.as_raw_string().split(b'\n') chunks.append(git_line(_MERGETAG_HEADER, mergetag_chunks[0])) # Embedded extra header needs leading space for chunk in mergetag_chunks[1:]: chunks.append(b' ' + chunk + b'\n') # No trailing empty line if chunks[-1].endswith(b' \n'): chunks[-1] = chunks[-1][:-2] for k, v in self.extra: if b'\n' in k or b'\n' in v: raise AssertionError( "newline in extra data: %r -> %r" % (k, v)) chunks.append(git_line(k, v)) if self.gpgsig: sig_chunks = self.gpgsig.split(b'\n') chunks.append(git_line(_GPGSIG_HEADER, sig_chunks[0])) for chunk in sig_chunks[1:]: chunks.append(git_line(b'', chunk)) chunks.append(b'\n') # There must be a new line after the headers chunks.append(self._message) return chunks tree = serializable_property( "tree", "Tree that is the state of this commit") def _get_parents(self): """Return a list of parents of this commit.""" return self._parents def _set_parents(self, value): """Set a list of parents of this commit.""" self._needs_serialization = True self._parents = value parents = property(_get_parents, _set_parents, doc="Parents of this commit, by their SHA1.") def _get_extra(self): """Return extra settings of this commit.""" return self._extra extra = property( _get_extra, doc="Extra header fields not understood (presumably added in a " "newer version of git). Kept verbatim so the object can " "be correctly reserialized. For private commit metadata, use " "pseudo-headers in Commit.message, rather than this field.") author = serializable_property( "author", "The name of the author of the commit") committer = serializable_property( "committer", "The name of the committer of the commit") message = serializable_property( "message", "The commit message") commit_time = serializable_property( "commit_time", "The timestamp of the commit. As the number of seconds since the " "epoch.") commit_timezone = serializable_property( "commit_timezone", "The zone the commit time is in") author_time = serializable_property( "author_time", "The timestamp the commit was written. As the number of " "seconds since the epoch.") author_timezone = serializable_property( "author_timezone", "Returns the zone the author time is in.") encoding = serializable_property( "encoding", "Encoding of the commit message.") mergetag = serializable_property( "mergetag", "Associated signed tag.") gpgsig = serializable_property( "gpgsig", "GPG Signature.") OBJECT_CLASSES = ( Commit, Tree, Blob, Tag, ) -_TYPE_MAP = {} +_TYPE_MAP:Dict[Union[bytes,int],Type[ShaFile]] = {} for cls in OBJECT_CLASSES: _TYPE_MAP[cls.type_name] = cls _TYPE_MAP[cls.type_num] = cls # Hold on to the pure-python implementations for testing _parse_tree_py = parse_tree _sorted_tree_items_py = sorted_tree_items try: # Try to import C versions - from dulwich._objects import parse_tree, sorted_tree_items + from dulwich._objects import parse_tree, sorted_tree_items # type: ignore except ImportError: pass diff --git a/dulwich/pack.py b/dulwich/pack.py index ce1600d8..6571d9c3 100644 --- a/dulwich/pack.py +++ b/dulwich/pack.py @@ -1,2099 +1,2096 @@ # pack.py -- For dealing with packed git objects. # Copyright (C) 2007 James Westby # Copyright (C) 2008-2013 Jelmer Vernooij # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Classes for dealing with packed git objects. A pack is a compact representation of a bunch of objects, stored using deltas where possible. They have two parts, the pack file, which stores the data, and an index that tells you where the data is. To find an object you look in all of the index files 'til you find a match for the object name. You then use the pointer got from this as a pointer in to the corresponding packfile. """ from collections import defaultdict import binascii from io import BytesIO, UnsupportedOperation from collections import ( deque, ) import difflib import struct from itertools import chain -try: - from itertools import imap, izip -except ImportError: - # Python3 - imap = map - izip = zip import os import sys from hashlib import sha1 from os import ( SEEK_CUR, SEEK_END, ) from struct import unpack_from import zlib try: import mmap except ImportError: has_mmap = False else: has_mmap = True # For some reason the above try, except fails to set has_mmap = False for plan9 if sys.platform == 'Plan9': has_mmap = False from dulwich.errors import ( # noqa: E402 ApplyDeltaError, ChecksumMismatch, ) from dulwich.file import GitFile # noqa: E402 from dulwich.lru_cache import ( # noqa: E402 LRUSizeCache, ) from dulwich.objects import ( # noqa: E402 ShaFile, hex_to_sha, sha_to_hex, object_header, ) OFS_DELTA = 6 REF_DELTA = 7 DELTA_TYPES = (OFS_DELTA, REF_DELTA) DEFAULT_PACK_DELTA_WINDOW_SIZE = 10 def take_msb_bytes(read, crc32=None): """Read bytes marked with most significant bit. Args: read: Read function """ ret = [] while len(ret) == 0 or ret[-1] & 0x80: b = read(1) if crc32 is not None: crc32 = binascii.crc32(b, crc32) ret.append(ord(b[:1])) return ret, crc32 class PackFileDisappeared(Exception): def __init__(self, obj): self.obj = obj class UnpackedObject(object): """Class encapsulating an object unpacked from a pack file. These objects should only be created from within unpack_object. Most members start out as empty and are filled in at various points by read_zlib_chunks, unpack_object, DeltaChainIterator, etc. End users of this object should take care that the function they're getting this object from is guaranteed to set the members they need. """ __slots__ = [ 'offset', # Offset in its pack. '_sha', # Cached binary SHA. 'obj_type_num', # Type of this object. 'obj_chunks', # Decompressed and delta-resolved chunks. 'pack_type_num', # Type of this object in the pack (may be a delta). 'delta_base', # Delta base offset or SHA. 'comp_chunks', # Compressed object chunks. 'decomp_chunks', # Decompressed object chunks. 'decomp_len', # Decompressed length of this object. 'crc32', # CRC32. ] # TODO(dborowitz): read_zlib_chunks and unpack_object could very well be # methods of this object. def __init__(self, pack_type_num, delta_base, decomp_len, crc32): self.offset = None self._sha = None self.pack_type_num = pack_type_num self.delta_base = delta_base self.comp_chunks = None self.decomp_chunks = [] self.decomp_len = decomp_len self.crc32 = crc32 if pack_type_num in DELTA_TYPES: self.obj_type_num = None self.obj_chunks = None else: self.obj_type_num = pack_type_num self.obj_chunks = self.decomp_chunks self.delta_base = delta_base def sha(self): """Return the binary SHA of this object.""" if self._sha is None: self._sha = obj_sha(self.obj_type_num, self.obj_chunks) return self._sha def sha_file(self): """Return a ShaFile from this object.""" return ShaFile.from_raw_chunks(self.obj_type_num, self.obj_chunks) # Only provided for backwards compatibility with code that expects either # chunks or a delta tuple. def _obj(self): """Return the decompressed chunks, or (delta base, delta chunks).""" if self.pack_type_num in DELTA_TYPES: return (self.delta_base, self.decomp_chunks) else: return self.decomp_chunks def __eq__(self, other): if not isinstance(other, UnpackedObject): return False for slot in self.__slots__: if getattr(self, slot) != getattr(other, slot): return False return True def __ne__(self, other): return not (self == other) def __repr__(self): data = ['%s=%r' % (s, getattr(self, s)) for s in self.__slots__] return '%s(%s)' % (self.__class__.__name__, ', '.join(data)) _ZLIB_BUFSIZE = 4096 def read_zlib_chunks(read_some, unpacked, include_comp=False, buffer_size=_ZLIB_BUFSIZE): """Read zlib data from a buffer. This function requires that the buffer have additional data following the compressed data, which is guaranteed to be the case for git pack files. Args: read_some: Read function that returns at least one byte, but may return less than the requested size. unpacked: An UnpackedObject to write result data to. If its crc32 attr is not None, the CRC32 of the compressed bytes will be computed using this starting CRC32. After this function, will have the following attrs set: * comp_chunks (if include_comp is True) * decomp_chunks * decomp_len * crc32 include_comp: If True, include compressed data in the result. buffer_size: Size of the read buffer. Returns: Leftover unused data from the decompression. Raises: zlib.error: if a decompression error occurred. """ if unpacked.decomp_len <= -1: raise ValueError('non-negative zlib data stream size expected') decomp_obj = zlib.decompressobj() comp_chunks = [] decomp_chunks = unpacked.decomp_chunks decomp_len = 0 crc32 = unpacked.crc32 while True: add = read_some(buffer_size) if not add: raise zlib.error('EOF before end of zlib stream') comp_chunks.append(add) decomp = decomp_obj.decompress(add) decomp_len += len(decomp) decomp_chunks.append(decomp) unused = decomp_obj.unused_data if unused: left = len(unused) if crc32 is not None: crc32 = binascii.crc32(add[:-left], crc32) if include_comp: comp_chunks[-1] = add[:-left] break elif crc32 is not None: crc32 = binascii.crc32(add, crc32) if crc32 is not None: crc32 &= 0xffffffff if decomp_len != unpacked.decomp_len: raise zlib.error('decompressed data does not match expected size') unpacked.crc32 = crc32 if include_comp: unpacked.comp_chunks = comp_chunks return unused def iter_sha1(iter): """Return the hexdigest of the SHA1 over a set of names. Args: iter: Iterator over string objects Returns: 40-byte hex sha1 digest """ sha = sha1() for name in iter: sha.update(name) return sha.hexdigest().encode('ascii') def load_pack_index(path): """Load an index file by path. Args: filename: Path to the index file Returns: A PackIndex loaded from the given path """ with GitFile(path, 'rb') as f: return load_pack_index_file(path, f) def _load_file_contents(f, size=None): try: fd = f.fileno() except (UnsupportedOperation, AttributeError): fd = None # Attempt to use mmap if possible if fd is not None: if size is None: size = os.fstat(fd).st_size if has_mmap: try: contents = mmap.mmap(fd, size, access=mmap.ACCESS_READ) except mmap.error: # Perhaps a socket? pass else: return contents, size contents = f.read() size = len(contents) return contents, size def load_pack_index_file(path, f): """Load an index file from a file-like object. Args: path: Path for the index file f: File-like object Returns: A PackIndex loaded from the given file """ contents, size = _load_file_contents(f) if contents[:4] == b'\377tOc': version = struct.unpack(b'>L', contents[4:8])[0] if version == 2: return PackIndex2( path, file=f, contents=contents, size=size) else: raise KeyError('Unknown pack index format %d' % version) else: return PackIndex1(path, file=f, contents=contents, size=size) def bisect_find_sha(start, end, sha, unpack_name): """Find a SHA in a data blob with sorted SHAs. Args: start: Start index of range to search end: End index of range to search sha: Sha to find unpack_name: Callback to retrieve SHA by index Returns: Index of the SHA, or None if it wasn't found """ assert start <= end while start <= end: i = (start + end) // 2 file_sha = unpack_name(i) if file_sha < sha: start = i + 1 elif file_sha > sha: end = i - 1 else: return i return None class PackIndex(object): """An index in to a packfile. Given a sha id of an object a pack index can tell you the location in the packfile of that object if it has it. """ def __eq__(self, other): if not isinstance(other, PackIndex): return False - for (name1, _, _), (name2, _, _) in izip(self.iterentries(), - other.iterentries()): + for (name1, _, _), (name2, _, _) in zip(self.iterentries(), + other.iterentries()): if name1 != name2: return False return True def __ne__(self, other): return not self.__eq__(other) def __len__(self): """Return the number of entries in this pack index.""" raise NotImplementedError(self.__len__) def __iter__(self): """Iterate over the SHAs in this pack.""" - return imap(sha_to_hex, self._itersha()) + return map(sha_to_hex, self._itersha()) def iterentries(self): """Iterate over the entries in this pack index. Returns: iterator over tuples with object name, offset in packfile and crc32 checksum. """ raise NotImplementedError(self.iterentries) def get_pack_checksum(self): """Return the SHA1 checksum stored for the corresponding packfile. Returns: 20-byte binary digest """ raise NotImplementedError(self.get_pack_checksum) def object_index(self, sha): """Return the index in to the corresponding packfile for the object. Given the name of an object it will return the offset that object lives at within the corresponding pack file. If the pack file doesn't have the object then None will be returned. """ if len(sha) == 40: sha = hex_to_sha(sha) try: return self._object_index(sha) except ValueError: closed = getattr(self._contents, 'closed', None) if closed in (None, True): raise PackFileDisappeared(self) raise def object_sha1(self, index): """Return the SHA1 corresponding to the index in the pack file. """ # PERFORMANCE/TODO(jelmer): Avoid scanning entire index for (name, offset, crc32) in self.iterentries(): if offset == index: return name else: raise KeyError(index) def _object_index(self, sha): """See object_index. Args: sha: A *binary* SHA string. (20 characters long)_ """ raise NotImplementedError(self._object_index) def objects_sha1(self): """Return the hex SHA1 over all the shas of all objects in this pack. Note: This is used for the filename of the pack. """ return iter_sha1(self._itersha()) def _itersha(self): """Yield all the SHA1's of the objects in the index, sorted.""" raise NotImplementedError(self._itersha) class MemoryPackIndex(PackIndex): """Pack index that is stored entirely in memory.""" def __init__(self, entries, pack_checksum=None): """Create a new MemoryPackIndex. Args: entries: Sequence of name, idx, crc32 (sorted) pack_checksum: Optional pack checksum """ self._by_sha = {} self._by_index = {} for name, idx, crc32 in entries: self._by_sha[name] = idx self._by_index[idx] = name self._entries = entries self._pack_checksum = pack_checksum def get_pack_checksum(self): return self._pack_checksum def __len__(self): return len(self._entries) def _object_index(self, sha): return self._by_sha[sha][0] def object_sha1(self, index): return self._by_index[index] def _itersha(self): return iter(self._by_sha) def iterentries(self): return iter(self._entries) class FilePackIndex(PackIndex): """Pack index that is based on a file. To do the loop it opens the file, and indexes first 256 4 byte groups with the first byte of the sha id. The value in the four byte group indexed is the end of the group that shares the same starting byte. Subtract one from the starting byte and index again to find the start of the group. The values are sorted by sha id within the group, so do the math to find the start and end offset and then bisect in to find if the value is present. """ def __init__(self, filename, file=None, contents=None, size=None): """Create a pack index object. Provide it with the name of the index file to consider, and it will map it whenever required. """ self._filename = filename # Take the size now, so it can be checked each time we map the file to # ensure that it hasn't changed. if file is None: self._file = GitFile(filename, 'rb') else: self._file = file if contents is None: self._contents, self._size = _load_file_contents(self._file, size) else: self._contents, self._size = (contents, size) @property def path(self): return self._filename def __eq__(self, other): # Quick optimization: if (isinstance(other, FilePackIndex) and self._fan_out_table != other._fan_out_table): return False return super(FilePackIndex, self).__eq__(other) def close(self): self._file.close() if getattr(self._contents, "close", None) is not None: self._contents.close() def __len__(self): """Return the number of entries in this pack index.""" return self._fan_out_table[-1] def _unpack_entry(self, i): """Unpack the i-th entry in the index file. Returns: Tuple with object name (SHA), offset in pack file and CRC32 checksum (if known). """ raise NotImplementedError(self._unpack_entry) def _unpack_name(self, i): """Unpack the i-th name from the index file.""" raise NotImplementedError(self._unpack_name) def _unpack_offset(self, i): """Unpack the i-th object offset from the index file.""" raise NotImplementedError(self._unpack_offset) def _unpack_crc32_checksum(self, i): """Unpack the crc32 checksum for the ith object from the index file. """ raise NotImplementedError(self._unpack_crc32_checksum) def _itersha(self): for i in range(len(self)): yield self._unpack_name(i) def iterentries(self): """Iterate over the entries in this pack index. Returns: iterator over tuples with object name, offset in packfile and crc32 checksum. """ for i in range(len(self)): yield self._unpack_entry(i) def _read_fan_out_table(self, start_offset): ret = [] for i in range(0x100): fanout_entry = self._contents[ start_offset+i*4:start_offset+(i+1)*4] ret.append(struct.unpack('>L', fanout_entry)[0]) return ret def check(self): """Check that the stored checksum matches the actual checksum.""" actual = self.calculate_checksum() stored = self.get_stored_checksum() if actual != stored: raise ChecksumMismatch(stored, actual) def calculate_checksum(self): """Calculate the SHA1 checksum over this pack index. Returns: This is a 20-byte binary digest """ return sha1(self._contents[:-20]).digest() def get_pack_checksum(self): """Return the SHA1 checksum stored for the corresponding packfile. Returns: 20-byte binary digest """ return bytes(self._contents[-40:-20]) def get_stored_checksum(self): """Return the SHA1 checksum stored for this index. Returns: 20-byte binary digest """ return bytes(self._contents[-20:]) def _object_index(self, sha): """See object_index. Args: sha: A *binary* SHA string. (20 characters long)_ """ assert len(sha) == 20 idx = ord(sha[:1]) if idx == 0: start = 0 else: start = self._fan_out_table[idx-1] end = self._fan_out_table[idx] i = bisect_find_sha(start, end, sha, self._unpack_name) if i is None: raise KeyError(sha) return self._unpack_offset(i) class PackIndex1(FilePackIndex): """Version 1 Pack Index file.""" def __init__(self, filename, file=None, contents=None, size=None): super(PackIndex1, self).__init__(filename, file, contents, size) self.version = 1 self._fan_out_table = self._read_fan_out_table(0) def _unpack_entry(self, i): (offset, name) = unpack_from('>L20s', self._contents, (0x100 * 4) + (i * 24)) return (name, offset, None) def _unpack_name(self, i): offset = (0x100 * 4) + (i * 24) + 4 return self._contents[offset:offset+20] def _unpack_offset(self, i): offset = (0x100 * 4) + (i * 24) return unpack_from('>L', self._contents, offset)[0] def _unpack_crc32_checksum(self, i): # Not stored in v1 index files return None class PackIndex2(FilePackIndex): """Version 2 Pack Index file.""" def __init__(self, filename, file=None, contents=None, size=None): super(PackIndex2, self).__init__(filename, file, contents, size) if self._contents[:4] != b'\377tOc': raise AssertionError('Not a v2 pack index file') (self.version, ) = unpack_from(b'>L', self._contents, 4) if self.version != 2: raise AssertionError('Version was %d' % self.version) self._fan_out_table = self._read_fan_out_table(8) self._name_table_offset = 8 + 0x100 * 4 self._crc32_table_offset = self._name_table_offset + 20 * len(self) self._pack_offset_table_offset = (self._crc32_table_offset + 4 * len(self)) self._pack_offset_largetable_offset = ( self._pack_offset_table_offset + 4 * len(self)) def _unpack_entry(self, i): return (self._unpack_name(i), self._unpack_offset(i), self._unpack_crc32_checksum(i)) def _unpack_name(self, i): offset = self._name_table_offset + i * 20 return self._contents[offset:offset+20] def _unpack_offset(self, i): offset = self._pack_offset_table_offset + i * 4 offset = unpack_from('>L', self._contents, offset)[0] if offset & (2**31): offset = ( self._pack_offset_largetable_offset + (offset & (2 ** 31 - 1)) * 8) offset = unpack_from('>Q', self._contents, offset)[0] return offset def _unpack_crc32_checksum(self, i): return unpack_from('>L', self._contents, self._crc32_table_offset + i * 4)[0] def read_pack_header(read): """Read the header of a pack file. Args: read: Read function Returns: Tuple of (pack version, number of objects). If no data is available to read, returns (None, None). """ header = read(12) if not header: return None, None if header[:4] != b'PACK': raise AssertionError('Invalid pack header %r' % header) (version,) = unpack_from(b'>L', header, 4) if version not in (2, 3): raise AssertionError('Version was %d' % version) (num_objects,) = unpack_from(b'>L', header, 8) return (version, num_objects) def chunks_length(chunks): if isinstance(chunks, bytes): return len(chunks) else: - return sum(imap(len, chunks)) + return sum(map(len, chunks)) def unpack_object(read_all, read_some=None, compute_crc32=False, include_comp=False, zlib_bufsize=_ZLIB_BUFSIZE): """Unpack a Git object. Args: read_all: Read function that blocks until the number of requested bytes are read. read_some: Read function that returns at least one byte, but may not return the number of bytes requested. compute_crc32: If True, compute the CRC32 of the compressed data. If False, the returned CRC32 will be None. include_comp: If True, include compressed data in the result. zlib_bufsize: An optional buffer size for zlib operations. Returns: A tuple of (unpacked, unused), where unused is the unused data leftover from decompression, and unpacked in an UnpackedObject with the following attrs set: * obj_chunks (for non-delta types) * pack_type_num * delta_base (for delta types) * comp_chunks (if include_comp is True) * decomp_chunks * decomp_len * crc32 (if compute_crc32 is True) """ if read_some is None: read_some = read_all if compute_crc32: crc32 = 0 else: crc32 = None bytes, crc32 = take_msb_bytes(read_all, crc32=crc32) type_num = (bytes[0] >> 4) & 0x07 size = bytes[0] & 0x0f for i, byte in enumerate(bytes[1:]): size += (byte & 0x7f) << ((i * 7) + 4) raw_base = len(bytes) if type_num == OFS_DELTA: bytes, crc32 = take_msb_bytes(read_all, crc32=crc32) raw_base += len(bytes) if bytes[-1] & 0x80: raise AssertionError delta_base_offset = bytes[0] & 0x7f for byte in bytes[1:]: delta_base_offset += 1 delta_base_offset <<= 7 delta_base_offset += (byte & 0x7f) delta_base = delta_base_offset elif type_num == REF_DELTA: delta_base = read_all(20) if compute_crc32: crc32 = binascii.crc32(delta_base, crc32) raw_base += 20 else: delta_base = None unpacked = UnpackedObject(type_num, delta_base, size, crc32) unused = read_zlib_chunks(read_some, unpacked, buffer_size=zlib_bufsize, include_comp=include_comp) return unpacked, unused def _compute_object_size(value): """Compute the size of a unresolved object for use with LRUSizeCache.""" (num, obj) = value if num in DELTA_TYPES: return chunks_length(obj[1]) return chunks_length(obj) class PackStreamReader(object): """Class to read a pack stream. The pack is read from a ReceivableProtocol using read() or recv() as appropriate. """ def __init__(self, read_all, read_some=None, zlib_bufsize=_ZLIB_BUFSIZE): self.read_all = read_all if read_some is None: self.read_some = read_all else: self.read_some = read_some self.sha = sha1() self._offset = 0 self._rbuf = BytesIO() # trailer is a deque to avoid memory allocation on small reads self._trailer = deque() self._zlib_bufsize = zlib_bufsize def _read(self, read, size): """Read up to size bytes using the given callback. As a side effect, update the verifier's hash (excluding the last 20 bytes read). Args: read: The read callback to read from. size: The maximum number of bytes to read; the particular behavior is callback-specific. """ data = read(size) # maintain a trailer of the last 20 bytes we've read n = len(data) self._offset += n tn = len(self._trailer) if n >= 20: to_pop = tn to_add = 20 else: to_pop = max(n + tn - 20, 0) to_add = n self.sha.update( bytes(bytearray([self._trailer.popleft() for _ in range(to_pop)]))) self._trailer.extend(data[-to_add:]) # hash everything but the trailer self.sha.update(data[:-to_add]) return data def _buf_len(self): buf = self._rbuf start = buf.tell() buf.seek(0, SEEK_END) end = buf.tell() buf.seek(start) return end - start @property def offset(self): return self._offset - self._buf_len() def read(self, size): """Read, blocking until size bytes are read.""" buf_len = self._buf_len() if buf_len >= size: return self._rbuf.read(size) buf_data = self._rbuf.read() self._rbuf = BytesIO() return buf_data + self._read(self.read_all, size - buf_len) def recv(self, size): """Read up to size bytes, blocking until one byte is read.""" buf_len = self._buf_len() if buf_len: data = self._rbuf.read(size) if size >= buf_len: self._rbuf = BytesIO() return data return self._read(self.read_some, size) def __len__(self): return self._num_objects def read_objects(self, compute_crc32=False): """Read the objects in this pack file. Args: compute_crc32: If True, compute the CRC32 of the compressed data. If False, the returned CRC32 will be None. Returns: Iterator over UnpackedObjects with the following members set: offset obj_type_num obj_chunks (for non-delta types) delta_base (for delta types) decomp_chunks decomp_len crc32 (if compute_crc32 is True) Raises: ChecksumMismatch: if the checksum of the pack contents does not match the checksum in the pack trailer. zlib.error: if an error occurred during zlib decompression. IOError: if an error occurred writing to the output file. """ pack_version, self._num_objects = read_pack_header(self.read) if pack_version is None: return for i in range(self._num_objects): offset = self.offset unpacked, unused = unpack_object( self.read, read_some=self.recv, compute_crc32=compute_crc32, zlib_bufsize=self._zlib_bufsize) unpacked.offset = offset # prepend any unused data to current read buffer buf = BytesIO() buf.write(unused) buf.write(self._rbuf.read()) buf.seek(0) self._rbuf = buf yield unpacked if self._buf_len() < 20: # If the read buffer is full, then the last read() got the whole # trailer off the wire. If not, it means there is still some of the # trailer to read. We need to read() all 20 bytes; N come from the # read buffer and (20 - N) come from the wire. self.read(20) pack_sha = bytearray(self._trailer) if pack_sha != self.sha.digest(): raise ChecksumMismatch(sha_to_hex(pack_sha), self.sha.hexdigest()) class PackStreamCopier(PackStreamReader): """Class to verify a pack stream as it is being read. The pack is read from a ReceivableProtocol using read() or recv() as appropriate and written out to the given file-like object. """ def __init__(self, read_all, read_some, outfile, delta_iter=None): """Initialize the copier. Args: read_all: Read function that blocks until the number of requested bytes are read. read_some: Read function that returns at least one byte, but may not return the number of bytes requested. outfile: File-like object to write output through. delta_iter: Optional DeltaChainIterator to record deltas as we read them. """ super(PackStreamCopier, self).__init__(read_all, read_some=read_some) self.outfile = outfile self._delta_iter = delta_iter def _read(self, read, size): """Read data from the read callback and write it to the file.""" data = super(PackStreamCopier, self)._read(read, size) self.outfile.write(data) return data def verify(self): """Verify a pack stream and write it to the output file. See PackStreamReader.iterobjects for a list of exceptions this may throw. """ if self._delta_iter: for unpacked in self.read_objects(): self._delta_iter.record(unpacked) else: for _ in self.read_objects(): pass def obj_sha(type, chunks): """Compute the SHA for a numeric type and object chunks.""" sha = sha1() sha.update(object_header(type, chunks_length(chunks))) if isinstance(chunks, bytes): sha.update(chunks) else: for chunk in chunks: sha.update(chunk) return sha.digest() def compute_file_sha(f, start_ofs=0, end_ofs=0, buffer_size=1 << 16): """Hash a portion of a file into a new SHA. Args: f: A file-like object to read from that supports seek(). start_ofs: The offset in the file to start reading at. end_ofs: The offset in the file to end reading at, relative to the end of the file. buffer_size: A buffer size for reading. Returns: A new SHA object updated with data read from the file. """ sha = sha1() f.seek(0, SEEK_END) length = f.tell() if (end_ofs < 0 and length + end_ofs < start_ofs) or end_ofs > length: raise AssertionError( "Attempt to read beyond file length. " "start_ofs: %d, end_ofs: %d, file length: %d" % ( start_ofs, end_ofs, length)) todo = length + end_ofs - start_ofs f.seek(start_ofs) while todo: data = f.read(min(todo, buffer_size)) sha.update(data) todo -= len(data) return sha class PackData(object): """The data contained in a packfile. Pack files can be accessed both sequentially for exploding a pack, and directly with the help of an index to retrieve a specific object. The objects within are either complete or a delta against another. The header is variable length. If the MSB of each byte is set then it indicates that the subsequent byte is still part of the header. For the first byte the next MS bits are the type, which tells you the type of object, and whether it is a delta. The LS byte is the lowest bits of the size. For each subsequent byte the LS 7 bits are the next MS bits of the size, i.e. the last byte of the header contains the MS bits of the size. For the complete objects the data is stored as zlib deflated data. The size in the header is the uncompressed object size, so to uncompress you need to just keep feeding data to zlib until you get an object back, or it errors on bad data. This is done here by just giving the complete buffer from the start of the deflated object on. This is bad, but until I get mmap sorted out it will have to do. Currently there are no integrity checks done. Also no attempt is made to try and detect the delta case, or a request for an object at the wrong position. It will all just throw a zlib or KeyError. """ def __init__(self, filename, file=None, size=None): """Create a PackData object representing the pack in the given filename. The file must exist and stay readable until the object is disposed of. It must also stay the same size. It will be mapped whenever needed. Currently there is a restriction on the size of the pack as the python mmap implementation is flawed. """ self._filename = filename self._size = size self._header_size = 12 if file is None: self._file = GitFile(self._filename, 'rb') else: self._file = file (version, self._num_objects) = read_pack_header(self._file.read) self._offset_cache = LRUSizeCache( 1024*1024*20, compute_size=_compute_object_size) self.pack = None @property def filename(self): return os.path.basename(self._filename) @property def path(self): return self._filename @classmethod def from_file(cls, file, size): return cls(str(file), file=file, size=size) @classmethod def from_path(cls, path): return cls(filename=path) def close(self): self._file.close() def __enter__(self): return self def __exit__(self, exc_type, exc_val, exc_tb): self.close() def _get_size(self): if self._size is not None: return self._size self._size = os.path.getsize(self._filename) if self._size < self._header_size: errmsg = ('%s is too small for a packfile (%d < %d)' % (self._filename, self._size, self._header_size)) raise AssertionError(errmsg) return self._size def __len__(self): """Returns the number of objects in this pack.""" return self._num_objects def calculate_checksum(self): """Calculate the checksum for this pack. Returns: 20-byte binary SHA1 digest """ return compute_file_sha(self._file, end_ofs=-20).digest() def get_ref(self, sha): """Get the object for a ref SHA, only looking in this pack.""" # TODO: cache these results if self.pack is None: raise KeyError(sha) try: offset = self.pack.index.object_index(sha) except KeyError: offset = None if offset: type, obj = self.get_object_at(offset) elif self.pack is not None and self.pack.resolve_ext_ref: type, obj = self.pack.resolve_ext_ref(sha) else: raise KeyError(sha) return offset, type, obj def resolve_object(self, offset, type, obj, get_ref=None): """Resolve an object, possibly resolving deltas when necessary. Returns: Tuple with object type and contents. """ # Walk down the delta chain, building a stack of deltas to reach # the requested object. base_offset = offset base_type = type base_obj = obj delta_stack = [] while base_type in DELTA_TYPES: prev_offset = base_offset if get_ref is None: get_ref = self.get_ref if base_type == OFS_DELTA: (delta_offset, delta) = base_obj # TODO: clean up asserts and replace with nicer error messages base_offset = base_offset - delta_offset base_type, base_obj = self.get_object_at(base_offset) assert isinstance(base_type, int) elif base_type == REF_DELTA: (basename, delta) = base_obj assert isinstance(basename, bytes) and len(basename) == 20 base_offset, base_type, base_obj = get_ref(basename) assert isinstance(base_type, int) delta_stack.append((prev_offset, base_type, delta)) # Now grab the base object (mustn't be a delta) and apply the # deltas all the way up the stack. chunks = base_obj for prev_offset, delta_type, delta in reversed(delta_stack): chunks = apply_delta(chunks, delta) # TODO(dborowitz): This can result in poor performance if # large base objects are separated from deltas in the pack. # We should reorganize so that we apply deltas to all # objects in a chain one after the other to optimize cache # performance. if prev_offset is not None: self._offset_cache[prev_offset] = base_type, chunks return base_type, chunks def iterobjects(self, progress=None, compute_crc32=True): self._file.seek(self._header_size) for i in range(1, self._num_objects + 1): offset = self._file.tell() unpacked, unused = unpack_object( self._file.read, compute_crc32=compute_crc32) if progress is not None: progress(i, self._num_objects) yield (offset, unpacked.pack_type_num, unpacked._obj(), unpacked.crc32) # Back up over unused data. self._file.seek(-len(unused), SEEK_CUR) def _iter_unpacked(self): # TODO(dborowitz): Merge this with iterobjects, if we can change its # return type. self._file.seek(self._header_size) if self._num_objects is None: return for _ in range(self._num_objects): offset = self._file.tell() unpacked, unused = unpack_object( self._file.read, compute_crc32=False) unpacked.offset = offset yield unpacked # Back up over unused data. self._file.seek(-len(unused), SEEK_CUR) def iterentries(self, progress=None): """Yield entries summarizing the contents of this pack. Args: progress: Progress function, called with current and total object count. Returns: iterator of tuples with (sha, offset, crc32) """ num_objects = self._num_objects resolve_ext_ref = ( self.pack.resolve_ext_ref if self.pack is not None else None) indexer = PackIndexer.for_pack_data( self, resolve_ext_ref=resolve_ext_ref) for i, result in enumerate(indexer): if progress is not None: progress(i, num_objects) yield result def sorted_entries(self, progress=None): """Return entries in this pack, sorted by SHA. Args: progress: Progress function, called with current and total object count Returns: List of tuples with (sha, offset, crc32) """ ret = sorted(self.iterentries(progress=progress)) return ret def create_index_v1(self, filename, progress=None): """Create a version 1 file for this data file. Args: filename: Index filename. progress: Progress report function Returns: Checksum of index file """ entries = self.sorted_entries(progress=progress) with GitFile(filename, 'wb') as f: return write_pack_index_v1(f, entries, self.calculate_checksum()) def create_index_v2(self, filename, progress=None): """Create a version 2 index file for this data file. Args: filename: Index filename. progress: Progress report function Returns: Checksum of index file """ entries = self.sorted_entries(progress=progress) with GitFile(filename, 'wb') as f: return write_pack_index_v2(f, entries, self.calculate_checksum()) def create_index(self, filename, progress=None, version=2): """Create an index file for this data file. Args: filename: Index filename. progress: Progress report function Returns: Checksum of index file """ if version == 1: return self.create_index_v1(filename, progress) elif version == 2: return self.create_index_v2(filename, progress) else: raise ValueError('unknown index format %d' % version) def get_stored_checksum(self): """Return the expected checksum stored in this pack.""" self._file.seek(-20, SEEK_END) return self._file.read(20) def check(self): """Check the consistency of this pack.""" actual = self.calculate_checksum() stored = self.get_stored_checksum() if actual != stored: raise ChecksumMismatch(stored, actual) def get_compressed_data_at(self, offset): """Given offset in the packfile return compressed data that is there. Using the associated index the location of an object can be looked up, and then the packfile can be asked directly for that object using this function. """ assert offset >= self._header_size self._file.seek(offset) unpacked, _ = unpack_object(self._file.read, include_comp=True) return (unpacked.pack_type_num, unpacked.delta_base, unpacked.comp_chunks) def get_object_at(self, offset): """Given an offset in to the packfile return the object that is there. Using the associated index the location of an object can be looked up, and then the packfile can be asked directly for that object using this function. """ try: return self._offset_cache[offset] except KeyError: pass assert offset >= self._header_size self._file.seek(offset) unpacked, _ = unpack_object(self._file.read) return (unpacked.pack_type_num, unpacked._obj()) class DeltaChainIterator(object): """Abstract iterator over pack data based on delta chains. Each object in the pack is guaranteed to be inflated exactly once, regardless of how many objects reference it as a delta base. As a result, memory usage is proportional to the length of the longest delta chain. Subclasses can override _result to define the result type of the iterator. By default, results are UnpackedObjects with the following members set: * offset * obj_type_num * obj_chunks * pack_type_num * delta_base (for delta types) * comp_chunks (if _include_comp is True) * decomp_chunks * decomp_len * crc32 (if _compute_crc32 is True) """ _compute_crc32 = False _include_comp = False def __init__(self, file_obj, resolve_ext_ref=None): self._file = file_obj self._resolve_ext_ref = resolve_ext_ref self._pending_ofs = defaultdict(list) self._pending_ref = defaultdict(list) self._full_ofs = [] self._shas = {} self._ext_refs = [] @classmethod def for_pack_data(cls, pack_data, resolve_ext_ref=None): walker = cls(None, resolve_ext_ref=resolve_ext_ref) walker.set_pack_data(pack_data) for unpacked in pack_data._iter_unpacked(): walker.record(unpacked) return walker def record(self, unpacked): type_num = unpacked.pack_type_num offset = unpacked.offset if type_num == OFS_DELTA: base_offset = offset - unpacked.delta_base self._pending_ofs[base_offset].append(offset) elif type_num == REF_DELTA: self._pending_ref[unpacked.delta_base].append(offset) else: self._full_ofs.append((offset, type_num)) def set_pack_data(self, pack_data): self._file = pack_data._file def _walk_all_chains(self): for offset, type_num in self._full_ofs: for result in self._follow_chain(offset, type_num, None): yield result for result in self._walk_ref_chains(): yield result assert not self._pending_ofs def _ensure_no_pending(self): if self._pending_ref: raise KeyError([sha_to_hex(s) for s in self._pending_ref]) def _walk_ref_chains(self): if not self._resolve_ext_ref: self._ensure_no_pending() return for base_sha, pending in sorted(self._pending_ref.items()): if base_sha not in self._pending_ref: continue try: type_num, chunks = self._resolve_ext_ref(base_sha) except KeyError: # Not an external ref, but may depend on one. Either it will # get popped via a _follow_chain call, or we will raise an # error below. continue self._ext_refs.append(base_sha) self._pending_ref.pop(base_sha) for new_offset in pending: for result in self._follow_chain(new_offset, type_num, chunks): yield result self._ensure_no_pending() def _result(self, unpacked): return unpacked def _resolve_object(self, offset, obj_type_num, base_chunks): self._file.seek(offset) unpacked, _ = unpack_object( self._file.read, include_comp=self._include_comp, compute_crc32=self._compute_crc32) unpacked.offset = offset if base_chunks is None: assert unpacked.pack_type_num == obj_type_num else: assert unpacked.pack_type_num in DELTA_TYPES unpacked.obj_type_num = obj_type_num unpacked.obj_chunks = apply_delta(base_chunks, unpacked.decomp_chunks) return unpacked def _follow_chain(self, offset, obj_type_num, base_chunks): # Unlike PackData.get_object_at, there is no need to cache offsets as # this approach by design inflates each object exactly once. todo = [(offset, obj_type_num, base_chunks)] for offset, obj_type_num, base_chunks in todo: unpacked = self._resolve_object(offset, obj_type_num, base_chunks) yield self._result(unpacked) unblocked = chain(self._pending_ofs.pop(unpacked.offset, []), self._pending_ref.pop(unpacked.sha(), [])) todo.extend( (new_offset, unpacked.obj_type_num, unpacked.obj_chunks) for new_offset in unblocked) def __iter__(self): return self._walk_all_chains() def ext_refs(self): return self._ext_refs class PackIndexer(DeltaChainIterator): """Delta chain iterator that yields index entries.""" _compute_crc32 = True def _result(self, unpacked): return unpacked.sha(), unpacked.offset, unpacked.crc32 class PackInflater(DeltaChainIterator): """Delta chain iterator that yields ShaFile objects.""" def _result(self, unpacked): return unpacked.sha_file() class SHA1Reader(object): """Wrapper for file-like object that remembers the SHA1 of its data.""" def __init__(self, f): self.f = f self.sha1 = sha1(b'') def read(self, num=None): data = self.f.read(num) self.sha1.update(data) return data def check_sha(self): stored = self.f.read(20) if stored != self.sha1.digest(): raise ChecksumMismatch(self.sha1.hexdigest(), sha_to_hex(stored)) def close(self): return self.f.close() def tell(self): return self.f.tell() class SHA1Writer(object): """Wrapper for file-like object that remembers the SHA1 of its data.""" def __init__(self, f): self.f = f self.length = 0 self.sha1 = sha1(b'') def write(self, data): self.sha1.update(data) self.f.write(data) self.length += len(data) def write_sha(self): sha = self.sha1.digest() assert len(sha) == 20 self.f.write(sha) self.length += len(sha) return sha def close(self): sha = self.write_sha() self.f.close() return sha def offset(self): return self.length def tell(self): return self.f.tell() def pack_object_header(type_num, delta_base, size): """Create a pack object header for the given object info. Args: type_num: Numeric type of the object. delta_base: Delta base offset or ref, or None for whole objects. size: Uncompressed object size. Returns: A header for a packed object. """ header = [] c = (type_num << 4) | (size & 15) size >>= 4 while size: header.append(c | 0x80) c = size & 0x7f size >>= 7 header.append(c) if type_num == OFS_DELTA: ret = [delta_base & 0x7f] delta_base >>= 7 while delta_base: delta_base -= 1 ret.insert(0, 0x80 | (delta_base & 0x7f)) delta_base >>= 7 header.extend(ret) elif type_num == REF_DELTA: assert len(delta_base) == 20 header += delta_base return bytearray(header) def write_pack_object(f, type, object, sha=None, compression_level=-1): """Write pack object to a file. Args: f: File to write to type: Numeric type of the object object: Object to write compression_level: the zlib compression level Returns: Tuple with offset at which the object was written, and crc32 """ if type in DELTA_TYPES: delta_base, object = object else: delta_base = None header = bytes(pack_object_header(type, delta_base, len(object))) comp_data = zlib.compress(object, compression_level) crc32 = 0 for data in (header, comp_data): f.write(data) if sha is not None: sha.update(data) crc32 = binascii.crc32(data, crc32) return crc32 & 0xffffffff def write_pack(filename, objects, deltify=None, delta_window_size=None, compression_level=-1): """Write a new pack data file. Args: filename: Path to the new pack file (without .pack extension) objects: Iterable of (object, path) tuples to write. Should provide __len__ window_size: Delta window size deltify: Whether to deltify pack objects compression_level: the zlib compression level Returns: Tuple with checksum of pack file and index file """ with GitFile(filename + '.pack', 'wb') as f: entries, data_sum = write_pack_objects( f, objects, delta_window_size=delta_window_size, deltify=deltify, compression_level=compression_level) entries = sorted([(k, v[0], v[1]) for (k, v) in entries.items()]) with GitFile(filename + '.idx', 'wb') as f: return data_sum, write_pack_index_v2(f, entries, data_sum) def write_pack_header(f, num_objects): """Write a pack header for the given number of objects.""" f.write(b'PACK') # Pack header f.write(struct.pack(b'>L', 2)) # Pack version f.write(struct.pack(b'>L', num_objects)) # Number of objects in pack def deltify_pack_objects(objects, window_size=None): """Generate deltas for pack objects. Args: objects: An iterable of (object, path) tuples to deltify. window_size: Window size; None for default Returns: Iterator over type_num, object id, delta_base, content delta_base is None for full text entries """ # TODO(jelmer): Use threads if window_size is None: window_size = DEFAULT_PACK_DELTA_WINDOW_SIZE # Build a list of objects ordered by the magic Linus heuristic # This helps us find good objects to diff against us magic = [] for obj, path in objects: magic.append((obj.type_num, path, -obj.raw_length(), obj)) magic.sort() possible_bases = deque() for type_num, path, neg_length, o in magic: raw = o.as_raw_string() winner = raw winner_base = None for base in possible_bases: if base.type_num != type_num: continue delta = create_delta(base.as_raw_string(), raw) if len(delta) < len(winner): winner_base = base.sha().digest() winner = delta yield type_num, o.sha().digest(), winner_base, winner possible_bases.appendleft(o) while len(possible_bases) > window_size: possible_bases.pop() def pack_objects_to_data(objects): """Create pack data from objects Args: objects: Pack objects Returns: Tuples with (type_num, hexdigest, delta base, object chunks) """ count = len(objects) return (count, ((o.type_num, o.sha().digest(), None, o.as_raw_string()) for (o, path) in objects)) def write_pack_objects(f, objects, delta_window_size=None, deltify=None, compression_level=-1): """Write a new pack data file. Args: f: File to write to objects: Iterable of (object, path) tuples to write. Should provide __len__ window_size: Sliding window size for searching for deltas; Set to None for default window size. deltify: Whether to deltify objects compression_level: the zlib compression level to use Returns: Dict mapping id -> (offset, crc32 checksum), pack checksum """ if deltify is None: # PERFORMANCE/TODO(jelmer): This should be enabled but is *much* too # slow at the moment. deltify = False if deltify: pack_contents = deltify_pack_objects(objects, delta_window_size) pack_contents_count = len(objects) else: pack_contents_count, pack_contents = pack_objects_to_data(objects) return write_pack_data( f, pack_contents_count, pack_contents, compression_level=compression_level) def write_pack_data( f, num_records, records, progress=None, compression_level=-1): """Write a new pack data file. Args: f: File to write to num_records: Number of records records: Iterator over type_num, object_id, delta_base, raw progress: Function to report progress to compression_level: the zlib compression level Returns: Dict mapping id -> (offset, crc32 checksum), pack checksum """ # Write the pack entries = {} f = SHA1Writer(f) write_pack_header(f, num_records) for i, (type_num, object_id, delta_base, raw) in enumerate(records): if progress is not None: progress(( 'writing pack data: %d/%d\r' % (i, num_records)).encode('ascii')) offset = f.offset() if delta_base is not None: try: base_offset, base_crc32 = entries[delta_base] except KeyError: type_num = REF_DELTA raw = (delta_base, raw) else: type_num = OFS_DELTA raw = (offset - base_offset, raw) crc32 = write_pack_object( f, type_num, raw, compression_level=compression_level) entries[object_id] = (offset, crc32) return entries, f.write_sha() def write_pack_index_v1(f, entries, pack_checksum): """Write a new pack index file. Args: f: A file-like object to write to entries: List of tuples with object name (sha), offset_in_pack, and crc32_checksum. pack_checksum: Checksum of the pack file. Returns: The SHA of the written index file """ f = SHA1Writer(f) fan_out_table = defaultdict(lambda: 0) for (name, offset, entry_checksum) in entries: fan_out_table[ord(name[:1])] += 1 # Fan-out table for i in range(0x100): f.write(struct.pack('>L', fan_out_table[i])) fan_out_table[i+1] += fan_out_table[i] for (name, offset, entry_checksum) in entries: if not (offset <= 0xffffffff): raise TypeError("pack format 1 only supports offsets < 2Gb") f.write(struct.pack('>L20s', offset, name)) assert len(pack_checksum) == 20 f.write(pack_checksum) return f.write_sha() def _delta_encode_size(size): ret = bytearray() c = size & 0x7f size >>= 7 while size: ret.append(c | 0x80) c = size & 0x7f size >>= 7 ret.append(c) return ret # The length of delta compression copy operations in version 2 packs is limited # to 64K. To copy more, we use several copy operations. Version 3 packs allow # 24-bit lengths in copy operations, but we always make version 2 packs. _MAX_COPY_LEN = 0xffff def _encode_copy_operation(start, length): scratch = [] op = 0x80 for i in range(4): if start & 0xff << i*8: scratch.append((start >> i*8) & 0xff) op |= 1 << i for i in range(2): if length & 0xff << i*8: scratch.append((length >> i*8) & 0xff) op |= 1 << (4+i) return bytearray([op] + scratch) def create_delta(base_buf, target_buf): """Use python difflib to work out how to transform base_buf to target_buf. Args: base_buf: Base buffer target_buf: Target buffer """ assert isinstance(base_buf, bytes) assert isinstance(target_buf, bytes) out_buf = bytearray() # write delta header out_buf += _delta_encode_size(len(base_buf)) out_buf += _delta_encode_size(len(target_buf)) # write out delta opcodes seq = difflib.SequenceMatcher(a=base_buf, b=target_buf) for opcode, i1, i2, j1, j2 in seq.get_opcodes(): # Git patch opcodes don't care about deletes! # if opcode == 'replace' or opcode == 'delete': # pass if opcode == 'equal': # If they are equal, unpacker will use data from base_buf # Write out an opcode that says what range to use copy_start = i1 copy_len = i2 - i1 while copy_len > 0: to_copy = min(copy_len, _MAX_COPY_LEN) out_buf += _encode_copy_operation(copy_start, to_copy) copy_start += to_copy copy_len -= to_copy if opcode == 'replace' or opcode == 'insert': # If we are replacing a range or adding one, then we just # output it to the stream (prefixed by its size) s = j2 - j1 o = j1 while s > 127: out_buf.append(127) out_buf += bytearray(target_buf[o:o+127]) s -= 127 o += 127 out_buf.append(s) out_buf += bytearray(target_buf[o:o+s]) return bytes(out_buf) def apply_delta(src_buf, delta): """Based on the similar function in git's patch-delta.c. Args: src_buf: Source buffer delta: Delta instructions """ if not isinstance(src_buf, bytes): src_buf = b''.join(src_buf) if not isinstance(delta, bytes): delta = b''.join(delta) out = [] index = 0 delta_length = len(delta) def get_delta_header_size(delta, index): size = 0 i = 0 while delta: cmd = ord(delta[index:index+1]) index += 1 size |= (cmd & ~0x80) << i i += 7 if not cmd & 0x80: break return size, index src_size, index = get_delta_header_size(delta, index) dest_size, index = get_delta_header_size(delta, index) assert src_size == len(src_buf), '%d vs %d' % (src_size, len(src_buf)) while index < delta_length: cmd = ord(delta[index:index+1]) index += 1 if cmd & 0x80: cp_off = 0 for i in range(4): if cmd & (1 << i): x = ord(delta[index:index+1]) index += 1 cp_off |= x << (i * 8) cp_size = 0 # Version 3 packs can contain copy sizes larger than 64K. for i in range(3): if cmd & (1 << (4+i)): x = ord(delta[index:index+1]) index += 1 cp_size |= x << (i * 8) if cp_size == 0: cp_size = 0x10000 if (cp_off + cp_size < cp_size or cp_off + cp_size > src_size or cp_size > dest_size): break out.append(src_buf[cp_off:cp_off+cp_size]) elif cmd != 0: out.append(delta[index:index+cmd]) index += cmd else: raise ApplyDeltaError('Invalid opcode 0') if index != delta_length: raise ApplyDeltaError('delta not empty: %r' % delta[index:]) if dest_size != chunks_length(out): raise ApplyDeltaError('dest size incorrect') return out def write_pack_index_v2(f, entries, pack_checksum): """Write a new pack index file. Args: f: File-like object to write to entries: List of tuples with object name (sha), offset_in_pack, and crc32_checksum. pack_checksum: Checksum of the pack file. Returns: The SHA of the index file written """ f = SHA1Writer(f) f.write(b'\377tOc') # Magic! f.write(struct.pack('>L', 2)) fan_out_table = defaultdict(lambda: 0) for (name, offset, entry_checksum) in entries: fan_out_table[ord(name[:1])] += 1 # Fan-out table largetable = [] for i in range(0x100): f.write(struct.pack(b'>L', fan_out_table[i])) fan_out_table[i+1] += fan_out_table[i] for (name, offset, entry_checksum) in entries: f.write(name) for (name, offset, entry_checksum) in entries: f.write(struct.pack(b'>L', entry_checksum)) for (name, offset, entry_checksum) in entries: if offset < 2**31: f.write(struct.pack(b'>L', offset)) else: f.write(struct.pack(b'>L', 2**31 + len(largetable))) largetable.append(offset) for offset in largetable: f.write(struct.pack(b'>Q', offset)) assert len(pack_checksum) == 20 f.write(pack_checksum) return f.write_sha() write_pack_index = write_pack_index_v2 class Pack(object): """A Git pack object.""" def __init__(self, basename, resolve_ext_ref=None): self._basename = basename self._data = None self._idx = None self._idx_path = self._basename + '.idx' self._data_path = self._basename + '.pack' self._data_load = lambda: PackData(self._data_path) self._idx_load = lambda: load_pack_index(self._idx_path) self.resolve_ext_ref = resolve_ext_ref @classmethod def from_lazy_objects(self, data_fn, idx_fn): """Create a new pack object from callables to load pack data and index objects.""" ret = Pack('') ret._data_load = data_fn ret._idx_load = idx_fn return ret @classmethod def from_objects(self, data, idx): """Create a new pack object from pack data and index objects.""" ret = Pack('') ret._data_load = lambda: data ret._idx_load = lambda: idx return ret def name(self): """The SHA over the SHAs of the objects in this pack.""" return self.index.objects_sha1() @property def data(self): """The pack data object being used.""" if self._data is None: self._data = self._data_load() self._data.pack = self self.check_length_and_checksum() return self._data @property def index(self): """The index being used. Note: This may be an in-memory index """ if self._idx is None: self._idx = self._idx_load() return self._idx def close(self): if self._data is not None: self._data.close() if self._idx is not None: self._idx.close() def __enter__(self): return self def __exit__(self, exc_type, exc_val, exc_tb): self.close() def __eq__(self, other): return isinstance(self, type(other)) and self.index == other.index def __len__(self): """Number of entries in this pack.""" return len(self.index) def __repr__(self): return '%s(%r)' % (self.__class__.__name__, self._basename) def __iter__(self): """Iterate over all the sha1s of the objects in this pack.""" return iter(self.index) def check_length_and_checksum(self): """Sanity check the length and checksum of the pack index and data.""" assert len(self.index) == len(self.data) idx_stored_checksum = self.index.get_pack_checksum() data_stored_checksum = self.data.get_stored_checksum() if idx_stored_checksum != data_stored_checksum: raise ChecksumMismatch(sha_to_hex(idx_stored_checksum), sha_to_hex(data_stored_checksum)) def check(self): """Check the integrity of this pack. Raises: ChecksumMismatch: if a checksum for the index or data is wrong """ self.index.check() self.data.check() for obj in self.iterobjects(): obj.check() # TODO: object connectivity checks def get_stored_checksum(self): return self.data.get_stored_checksum() def __contains__(self, sha1): """Check whether this pack contains a particular SHA1.""" try: self.index.object_index(sha1) return True except KeyError: return False def get_raw_unresolved(self, sha1): """Get raw unresolved data for a SHA. Args: sha1: SHA to return data for Returns: Tuple with pack object type, delta base (if applicable), list of data chunks """ offset = self.index.object_index(sha1) (obj_type, delta_base, chunks) = self.data.get_compressed_data_at( offset) if obj_type == OFS_DELTA: delta_base = sha_to_hex( self.index.object_sha1(offset - delta_base)) obj_type = REF_DELTA return (obj_type, delta_base, chunks) def get_raw(self, sha1): offset = self.index.object_index(sha1) obj_type, obj = self.data.get_object_at(offset) type_num, chunks = self.data.resolve_object(offset, obj_type, obj) return type_num, b''.join(chunks) def __getitem__(self, sha1): """Retrieve the specified SHA1.""" type, uncomp = self.get_raw(sha1) return ShaFile.from_raw_string(type, uncomp, sha=sha1) def iterobjects(self): """Iterate over the objects in this pack.""" return iter(PackInflater.for_pack_data( self.data, resolve_ext_ref=self.resolve_ext_ref)) def pack_tuples(self): """Provide an iterable for use with write_pack_objects. Returns: Object that can iterate over (object, path) tuples and provides __len__ """ class PackTupleIterable(object): def __init__(self, pack): self.pack = pack def __len__(self): return len(self.pack) def __iter__(self): return ((o, None) for o in self.pack.iterobjects()) return PackTupleIterable(self) def keep(self, msg=None): """Add a .keep file for the pack, preventing git from garbage collecting it. Args: msg: A message written inside the .keep file; can be used later to determine whether or not a .keep file is obsolete. Returns: The path of the .keep file, as a string. """ keepfile_name = '%s.keep' % self._basename with GitFile(keepfile_name, 'wb') as keepfile: if msg: keepfile.write(msg) keepfile.write(b'\n') return keepfile_name try: - from dulwich._pack import apply_delta, bisect_find_sha # noqa: F811 + from dulwich._pack import ( # type: ignore # noqa: F811 + apply_delta, + bisect_find_sha, + ) except ImportError: pass diff --git a/dulwich/tests/compat/test_web.py b/dulwich/tests/compat/test_web.py index c4c3f187..393e9f9e 100644 --- a/dulwich/tests/compat/test_web.py +++ b/dulwich/tests/compat/test_web.py @@ -1,206 +1,207 @@ # test_web.py -- Compatibility tests for the git web server. # Copyright (C) 2010 Google, Inc. # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Compatibility tests between Dulwich and the cgit HTTP server. warning: these tests should be fairly stable, but when writing/debugging new tests, deadlocks may freeze the test process such that it cannot be Ctrl-C'ed. On POSIX systems, you can kill the tests with Ctrl-Z, "kill %". """ import threading from wsgiref import simple_server import sys +from typing import Tuple from dulwich.server import ( DictBackend, UploadPackHandler, ReceivePackHandler, ) from dulwich.tests import ( SkipTest, skipIf, ) from dulwich.web import ( make_wsgi_chain, HTTPGitApplication, WSGIRequestHandlerLogger, WSGIServerLogger, ) from dulwich.tests.compat.server_utils import ( ServerTests, NoSideBand64kReceivePackHandler, ) from dulwich.tests.compat.utils import ( CompatTestCase, ) @skipIf(sys.platform == 'win32', 'Broken on windows, with very long fail time.') class WebTests(ServerTests): """Base tests for web server tests. Contains utility and setUp/tearDown methods, but does non inherit from TestCase so tests are not automatically run. """ protocol = 'http' def _start_server(self, repo): backend = DictBackend({'/': repo}) app = self._make_app(backend) dul_server = simple_server.make_server( 'localhost', 0, app, server_class=WSGIServerLogger, handler_class=WSGIRequestHandlerLogger) self.addCleanup(dul_server.shutdown) self.addCleanup(dul_server.server_close) threading.Thread(target=dul_server.serve_forever).start() self._server = dul_server _, port = dul_server.socket.getsockname() return port @skipIf(sys.platform == 'win32', 'Broken on windows, with very long fail time.') class SmartWebTestCase(WebTests, CompatTestCase): """Test cases for smart HTTP server. This server test case does not use side-band-64k in git-receive-pack. """ - min_git_version = (1, 6, 6) + min_git_version:Tuple[int, ...] = (1, 6, 6) def _handlers(self): return {b'git-receive-pack': NoSideBand64kReceivePackHandler} def _check_app(self, app): receive_pack_handler_cls = app.handlers[b'git-receive-pack'] caps = receive_pack_handler_cls.capabilities() self.assertNotIn(b'side-band-64k', caps) def _make_app(self, backend): app = make_wsgi_chain(backend, handlers=self._handlers()) to_check = app # peel back layers until we're at the base application while not issubclass(to_check.__class__, HTTPGitApplication): to_check = to_check.app self._check_app(to_check) return app def patch_capabilities(handler, caps_removed): # Patch a handler's capabilities by specifying a list of them to be # removed, and return the original classmethod for restoration. original_capabilities = handler.capabilities filtered_capabilities = [ i for i in original_capabilities() if i not in caps_removed] def capabilities(cls): return filtered_capabilities handler.capabilities = classmethod(capabilities) return original_capabilities @skipIf(sys.platform == 'win32', 'Broken on windows, with very long fail time.') class SmartWebSideBand64kTestCase(SmartWebTestCase): """Test cases for smart HTTP server with side-band-64k support.""" # side-band-64k in git-receive-pack was introduced in git 1.7.0.2 min_git_version = (1, 7, 0, 2) def setUp(self): self.o_uph_cap = patch_capabilities(UploadPackHandler, (b"no-done",)) self.o_rph_cap = patch_capabilities(ReceivePackHandler, (b"no-done",)) super(SmartWebSideBand64kTestCase, self).setUp() def tearDown(self): super(SmartWebSideBand64kTestCase, self).tearDown() UploadPackHandler.capabilities = self.o_uph_cap ReceivePackHandler.capabilities = self.o_rph_cap def _handlers(self): return None # default handlers include side-band-64k def _check_app(self, app): receive_pack_handler_cls = app.handlers[b'git-receive-pack'] caps = receive_pack_handler_cls.capabilities() self.assertIn(b'side-band-64k', caps) self.assertNotIn(b'no-done', caps) class SmartWebSideBand64kNoDoneTestCase(SmartWebTestCase): """Test cases for smart HTTP server with side-band-64k and no-done support. """ # no-done was introduced in git 1.7.4 min_git_version = (1, 7, 4) def _handlers(self): return None # default handlers include side-band-64k def _check_app(self, app): receive_pack_handler_cls = app.handlers[b'git-receive-pack'] caps = receive_pack_handler_cls.capabilities() self.assertIn(b'side-band-64k', caps) self.assertIn(b'no-done', caps) @skipIf(sys.platform == 'win32', 'Broken on windows, with very long fail time.') class DumbWebTestCase(WebTests, CompatTestCase): """Test cases for dumb HTTP server.""" def _make_app(self, backend): return make_wsgi_chain(backend, dumb=True) def test_push_to_dulwich(self): # Note: remove this if dulwich implements dumb web pushing. raise SkipTest('Dumb web pushing not supported.') def test_push_to_dulwich_remove_branch(self): # Note: remove this if dumb pushing is supported raise SkipTest('Dumb web pushing not supported.') def test_new_shallow_clone_from_dulwich(self): # Note: remove this if C git and dulwich implement dumb web shallow # clones. raise SkipTest('Dumb web shallow cloning not supported.') def test_shallow_clone_from_git_is_identical(self): # Note: remove this if C git and dulwich implement dumb web shallow # clones. raise SkipTest('Dumb web shallow cloning not supported.') def test_fetch_same_depth_into_shallow_clone_from_dulwich(self): # Note: remove this if C git and dulwich implement dumb web shallow # clones. raise SkipTest('Dumb web shallow cloning not supported.') def test_fetch_full_depth_into_shallow_clone_from_dulwich(self): # Note: remove this if C git and dulwich implement dumb web shallow # clones. raise SkipTest('Dumb web shallow cloning not supported.') def test_push_to_dulwich_issue_88_standard(self): raise SkipTest('Dumb web pushing not supported.') diff --git a/dulwich/tests/compat/utils.py b/dulwich/tests/compat/utils.py index 1f36a219..4173dbb4 100644 --- a/dulwich/tests/compat/utils.py +++ b/dulwich/tests/compat/utils.py @@ -1,264 +1,265 @@ # utils.py -- Git compatibility utilities # Copyright (C) 2010 Google, Inc. # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Utilities for interacting with cgit.""" import errno import functools import os import shutil import socket import stat import subprocess import sys import tempfile import time +from typing import Tuple from dulwich.repo import Repo from dulwich.protocol import TCP_GIT_PORT from dulwich.tests import ( SkipTest, TestCase, ) _DEFAULT_GIT = 'git' _VERSION_LEN = 4 _REPOS_DATA_DIR = os.path.abspath(os.path.join( os.path.dirname(__file__), os.pardir, 'data', 'repos')) def git_version(git_path=_DEFAULT_GIT): """Attempt to determine the version of git currently installed. Args: git_path: Path to the git executable; defaults to the version in the system path. Returns: A tuple of ints of the form (major, minor, point, sub-point), or None if no git installation was found. """ try: output = run_git_or_fail(['--version'], git_path=git_path) except OSError: return None version_prefix = b'git version ' if not output.startswith(version_prefix): return None parts = output[len(version_prefix):].split(b'.') nums = [] for part in parts: try: nums.append(int(part)) except ValueError: break while len(nums) < _VERSION_LEN: nums.append(0) return tuple(nums[:_VERSION_LEN]) def require_git_version(required_version, git_path=_DEFAULT_GIT): """Require git version >= version, or skip the calling test. Args: required_version: A tuple of ints of the form (major, minor, point, sub-point); ommitted components default to 0. git_path: Path to the git executable; defaults to the version in the system path. Raises: ValueError: if the required version tuple has too many parts. SkipTest: if no suitable git version was found at the given path. """ found_version = git_version(git_path=git_path) if found_version is None: raise SkipTest('Test requires git >= %s, but c git not found' % (required_version, )) if len(required_version) > _VERSION_LEN: raise ValueError('Invalid version tuple %s, expected %i parts' % (required_version, _VERSION_LEN)) required_version = list(required_version) while len(found_version) < len(required_version): required_version.append(0) required_version = tuple(required_version) if found_version < required_version: required_version = '.'.join(map(str, required_version)) found_version = '.'.join(map(str, found_version)) raise SkipTest('Test requires git >= %s, found %s' % (required_version, found_version)) def run_git(args, git_path=_DEFAULT_GIT, input=None, capture_stdout=False, **popen_kwargs): """Run a git command. Input is piped from the input parameter and output is sent to the standard streams, unless capture_stdout is set. Args: args: A list of args to the git command. git_path: Path to to the git executable. input: Input data to be sent to stdin. capture_stdout: Whether to capture and return stdout. popen_kwargs: Additional kwargs for subprocess.Popen; stdin/stdout args are ignored. Returns: A tuple of (returncode, stdout contents). If capture_stdout is False, None will be returned as stdout contents. Raises: OSError: if the git executable was not found. """ env = popen_kwargs.pop('env', {}) env['LC_ALL'] = env['LANG'] = 'C' args = [git_path] + args popen_kwargs['stdin'] = subprocess.PIPE if capture_stdout: popen_kwargs['stdout'] = subprocess.PIPE else: popen_kwargs.pop('stdout', None) p = subprocess.Popen(args, env=env, **popen_kwargs) stdout, stderr = p.communicate(input=input) return (p.returncode, stdout) def run_git_or_fail(args, git_path=_DEFAULT_GIT, input=None, **popen_kwargs): """Run a git command, capture stdout/stderr, and fail if git fails.""" if 'stderr' not in popen_kwargs: popen_kwargs['stderr'] = subprocess.STDOUT returncode, stdout = run_git(args, git_path=git_path, input=input, capture_stdout=True, **popen_kwargs) if returncode != 0: raise AssertionError("git with args %r failed with %d: %r" % ( args, returncode, stdout)) return stdout def import_repo_to_dir(name): """Import a repo from a fast-export file in a temporary directory. These are used rather than binary repos for compat tests because they are more compact and human-editable, and we already depend on git. Args: name: The name of the repository export file, relative to dulwich/tests/data/repos. Returns: The path to the imported repository. """ temp_dir = tempfile.mkdtemp() export_path = os.path.join(_REPOS_DATA_DIR, name) temp_repo_dir = os.path.join(temp_dir, name) export_file = open(export_path, 'rb') run_git_or_fail(['init', '--quiet', '--bare', temp_repo_dir]) run_git_or_fail(['fast-import'], input=export_file.read(), cwd=temp_repo_dir) export_file.close() return temp_repo_dir def check_for_daemon(limit=10, delay=0.1, timeout=0.1, port=TCP_GIT_PORT): """Check for a running TCP daemon. Defaults to checking 10 times with a delay of 0.1 sec between tries. Args: limit: Number of attempts before deciding no daemon is running. delay: Delay between connection attempts. timeout: Socket timeout for connection attempts. port: Port on which we expect the daemon to appear. Returns: A boolean, true if a daemon is running on the specified port, false if not. """ for _ in range(limit): time.sleep(delay) s = socket.socket(socket.AF_INET, socket.SOCK_STREAM) s.settimeout(delay) try: s.connect(('localhost', port)) return True except socket.timeout: pass except socket.error as e: if getattr(e, 'errno', False) and e.errno != errno.ECONNREFUSED: raise elif e.args[0] != errno.ECONNREFUSED: raise finally: s.close() return False class CompatTestCase(TestCase): """Test case that requires git for compatibility checks. Subclasses can change the git version required by overriding min_git_version. """ - min_git_version = (1, 5, 0) + min_git_version:Tuple[int, ...] = (1, 5, 0) def setUp(self): super(CompatTestCase, self).setUp() require_git_version(self.min_git_version) def assertObjectStoreEqual(self, store1, store2): self.assertEqual(sorted(set(store1)), sorted(set(store2))) def assertReposEqual(self, repo1, repo2): self.assertEqual(repo1.get_refs(), repo2.get_refs()) self.assertObjectStoreEqual(repo1.object_store, repo2.object_store) def assertReposNotEqual(self, repo1, repo2): refs1 = repo1.get_refs() objs1 = set(repo1.object_store) refs2 = repo2.get_refs() objs2 = set(repo2.object_store) self.assertFalse(refs1 == refs2 and objs1 == objs2) def import_repo(self, name): """Import a repo from a fast-export file in a temporary directory. Args: name: The name of the repository export file, relative to dulwich/tests/data/repos. Returns: An initialized Repo object that lives in a temporary directory. """ path = import_repo_to_dir(name) repo = Repo(path) def cleanup(): repo.close() rmtree_ro(os.path.dirname(path.rstrip(os.sep))) self.addCleanup(cleanup) return repo if sys.platform == 'win32': def remove_ro(action, name, exc): os.chmod(name, stat.S_IWRITE) os.remove(name) rmtree_ro = functools.partial(shutil.rmtree, onerror=remove_ro) else: rmtree_ro = shutil.rmtree diff --git a/dulwich/tests/test_archive.py b/dulwich/tests/test_archive.py index 21d1bde4..5adb2eb5 100644 --- a/dulwich/tests/test_archive.py +++ b/dulwich/tests/test_archive.py @@ -1,102 +1,102 @@ # test_archive.py -- tests for archive # Copyright (C) 2015 Jelmer Vernooij # # Dulwich is dual-licensed under the Apache License, Version 2.0 and the GNU # General Public License as public by the Free Software Foundation; version 2.0 # or (at your option) any later version. You can redistribute it and/or # modify it under the terms of either of these two licenses. # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # You should have received a copy of the licenses; if not, see # for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Tests for archive support.""" from io import BytesIO import tarfile import struct from unittest import skipUnless from dulwich.archive import tar_stream from dulwich.object_store import ( MemoryObjectStore, ) from dulwich.objects import ( Blob, Tree, ) from dulwich.tests import ( TestCase, ) from dulwich.tests.utils import ( build_commit_graph, ) try: from mock import patch except ImportError: - patch = None + patch = None # type: ignore class ArchiveTests(TestCase): def test_empty(self): store = MemoryObjectStore() c1, c2, c3 = build_commit_graph(store, [[1], [2, 1], [3, 1, 2]]) tree = store[c3.tree] stream = b''.join(tar_stream(store, tree, 10)) out = BytesIO(stream) tf = tarfile.TarFile(fileobj=out) self.addCleanup(tf.close) self.assertEqual([], tf.getnames()) def _get_example_tar_stream(self, *tar_stream_args, **tar_stream_kwargs): store = MemoryObjectStore() b1 = Blob.from_string(b"somedata") store.add_object(b1) t1 = Tree() t1.add(b"somename", 0o100644, b1.id) store.add_object(t1) stream = b''.join( tar_stream(store, t1, *tar_stream_args, **tar_stream_kwargs)) return BytesIO(stream) def test_simple(self): stream = self._get_example_tar_stream(mtime=0) tf = tarfile.TarFile(fileobj=stream) self.addCleanup(tf.close) self.assertEqual(["somename"], tf.getnames()) def test_prefix(self): stream = self._get_example_tar_stream(mtime=0, prefix=b'blah') tf = tarfile.TarFile(fileobj=stream) self.addCleanup(tf.close) self.assertEqual(["blah/somename"], tf.getnames()) def test_gzip_mtime(self): stream = self._get_example_tar_stream(mtime=1234, format='gz') expected_mtime = struct.pack(' for a copy of the GNU General Public License # and for a copy of the Apache # License, Version 2.0. # """Tests for the Git HTTP server.""" from io import BytesIO import gzip import re import os +from typing import Type from dulwich.object_store import ( MemoryObjectStore, ) from dulwich.objects import ( Blob, ) from dulwich.repo import ( BaseRepo, MemoryRepo, ) from dulwich.server import ( DictBackend, ) from dulwich.tests import ( TestCase, ) from dulwich.web import ( HTTP_OK, HTTP_NOT_FOUND, HTTP_FORBIDDEN, HTTP_ERROR, GunzipFilter, send_file, get_text_file, get_loose_object, get_pack_file, get_idx_file, get_info_refs, get_info_packs, handle_service_request, _LengthLimitedFile, HTTPGitRequest, HTTPGitApplication, ) from dulwich.tests.utils import ( make_object, make_tag, ) class MinimalistWSGIInputStream(object): """WSGI input stream with no 'seek()' and 'tell()' methods.""" def __init__(self, data): self.data = data self.pos = 0 def read(self, howmuch): start = self.pos end = self.pos + howmuch if start >= len(self.data): return '' self.pos = end return self.data[start:end] class MinimalistWSGIInputStream2(MinimalistWSGIInputStream): """WSGI input stream with no *working* 'seek()' and 'tell()' methods.""" def seek(self, pos): raise NotImplementedError def tell(self): raise NotImplementedError class TestHTTPGitRequest(HTTPGitRequest): """HTTPGitRequest with overridden methods to help test caching.""" def __init__(self, *args, **kwargs): HTTPGitRequest.__init__(self, *args, **kwargs) self.cached = None def nocache(self): self.cached = False def cache_forever(self): self.cached = True class WebTestCase(TestCase): """Base TestCase with useful instance vars and utility functions.""" - _req_class = TestHTTPGitRequest + _req_class:Type[HTTPGitRequest] = TestHTTPGitRequest def setUp(self): super(WebTestCase, self).setUp() self._environ = {} self._req = self._req_class(self._environ, self._start_response, handlers=self._handlers()) self._status = None self._headers = [] self._output = BytesIO() def _start_response(self, status, headers): self._status = status self._headers = list(headers) return self._output.write def _handlers(self): return None def assertContentTypeEquals(self, expected): self.assertTrue(('Content-Type', expected) in self._headers) def _test_backend(objects, refs=None, named_files=None): if not refs: refs = {} if not named_files: named_files = {} repo = MemoryRepo.init_bare(objects, refs) for path, contents in named_files.items(): repo._put_named_file(path, contents) return DictBackend({'/': repo}) class DumbHandlersTestCase(WebTestCase): def test_send_file_not_found(self): list(send_file(self._req, None, 'text/plain')) self.assertEqual(HTTP_NOT_FOUND, self._status) def test_send_file(self): f = BytesIO(b'foobar') output = b''.join(send_file(self._req, f, 'some/thing')) self.assertEqual(b'foobar', output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('some/thing') self.assertTrue(f.closed) def test_send_file_buffered(self): bufsize = 10240 xs = b'x' * bufsize f = BytesIO(2 * xs) self.assertEqual([xs, xs], list(send_file(self._req, f, 'some/thing'))) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('some/thing') self.assertTrue(f.closed) def test_send_file_error(self): class TestFile(object): def __init__(self, exc_class): self.closed = False self._exc_class = exc_class def read(self, size=-1): raise self._exc_class() def close(self): self.closed = True f = TestFile(IOError) list(send_file(self._req, f, 'some/thing')) self.assertEqual(HTTP_ERROR, self._status) self.assertTrue(f.closed) self.assertFalse(self._req.cached) # non-IOErrors are reraised f = TestFile(AttributeError) self.assertRaises(AttributeError, list, send_file(self._req, f, 'some/thing')) self.assertTrue(f.closed) self.assertFalse(self._req.cached) def test_get_text_file(self): backend = _test_backend([], named_files={'description': b'foo'}) mat = re.search('.*', 'description') output = b''.join(get_text_file(self._req, backend, mat)) self.assertEqual(b'foo', output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('text/plain') self.assertFalse(self._req.cached) def test_get_loose_object(self): blob = make_object(Blob, data=b'foo') backend = _test_backend([blob]) mat = re.search('^(..)(.{38})$', blob.id.decode('ascii')) output = b''.join(get_loose_object(self._req, backend, mat)) self.assertEqual(blob.as_legacy_object(), output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('application/x-git-loose-object') self.assertTrue(self._req.cached) def test_get_loose_object_missing(self): mat = re.search('^(..)(.{38})$', '1' * 40) list(get_loose_object(self._req, _test_backend([]), mat)) self.assertEqual(HTTP_NOT_FOUND, self._status) def test_get_loose_object_error(self): blob = make_object(Blob, data=b'foo') backend = _test_backend([blob]) mat = re.search('^(..)(.{38})$', blob.id.decode('ascii')) def as_legacy_object_error(self): raise IOError self.addCleanup( setattr, Blob, 'as_legacy_object', Blob.as_legacy_object) Blob.as_legacy_object = as_legacy_object_error list(get_loose_object(self._req, backend, mat)) self.assertEqual(HTTP_ERROR, self._status) def test_get_pack_file(self): pack_name = os.path.join( 'objects', 'pack', 'pack-%s.pack' % ('1' * 40)) backend = _test_backend([], named_files={pack_name: b'pack contents'}) mat = re.search('.*', pack_name) output = b''.join(get_pack_file(self._req, backend, mat)) self.assertEqual(b'pack contents', output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('application/x-git-packed-objects') self.assertTrue(self._req.cached) def test_get_idx_file(self): idx_name = os.path.join('objects', 'pack', 'pack-%s.idx' % ('1' * 40)) backend = _test_backend([], named_files={idx_name: b'idx contents'}) mat = re.search('.*', idx_name) output = b''.join(get_idx_file(self._req, backend, mat)) self.assertEqual(b'idx contents', output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('application/x-git-packed-objects-toc') self.assertTrue(self._req.cached) def test_get_info_refs(self): self._environ['QUERY_STRING'] = '' blob1 = make_object(Blob, data=b'1') blob2 = make_object(Blob, data=b'2') blob3 = make_object(Blob, data=b'3') tag1 = make_tag(blob2, name=b'tag-tag') objects = [blob1, blob2, blob3, tag1] refs = { b'HEAD': b'000', b'refs/heads/master': blob1.id, b'refs/tags/tag-tag': tag1.id, b'refs/tags/blob-tag': blob3.id, } backend = _test_backend(objects, refs=refs) mat = re.search('.*', '//info/refs') self.assertEqual([blob1.id + b'\trefs/heads/master\n', blob3.id + b'\trefs/tags/blob-tag\n', tag1.id + b'\trefs/tags/tag-tag\n', blob2.id + b'\trefs/tags/tag-tag^{}\n'], list(get_info_refs(self._req, backend, mat))) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('text/plain') self.assertFalse(self._req.cached) def test_get_info_refs_not_found(self): self._environ['QUERY_STRING'] = '' objects = [] refs = {} backend = _test_backend(objects, refs=refs) mat = re.search('info/refs', '/foo/info/refs') self.assertEqual( [b'No git repository was found at /foo'], list(get_info_refs(self._req, backend, mat))) self.assertEqual(HTTP_NOT_FOUND, self._status) self.assertContentTypeEquals('text/plain') def test_get_info_packs(self): class TestPackData(object): def __init__(self, sha): self.filename = "pack-%s.pack" % sha class TestPack(object): def __init__(self, sha): self.data = TestPackData(sha) packs = [TestPack(str(i) * 40) for i in range(1, 4)] class TestObjectStore(MemoryObjectStore): # property must be overridden, can't be assigned @property def packs(self): return packs store = TestObjectStore() repo = BaseRepo(store, None) backend = DictBackend({'/': repo}) mat = re.search('.*', '//info/packs') output = b''.join(get_info_packs(self._req, backend, mat)) expected = b''.join( [(b'P pack-' + s + b'.pack\n') for s in [b'1' * 40, b'2' * 40, b'3' * 40]]) self.assertEqual(expected, output) self.assertEqual(HTTP_OK, self._status) self.assertContentTypeEquals('text/plain') self.assertFalse(self._req.cached) class SmartHandlersTestCase(WebTestCase): class _TestUploadPackHandler(object): def __init__(self, backend, args, proto, http_req=None, advertise_refs=False): self.args = args self.proto = proto self.http_req = http_req self.advertise_refs = advertise_refs def handle(self): self.proto.write(b'handled input: ' + self.proto.recv(1024)) def _make_handler(self, *args, **kwargs): self._handler = self._TestUploadPackHandler(*args, **kwargs) return self._handler def _handlers(self): return {b'git-upload-pack': self._make_handler} def test_handle_service_request_unknown(self): mat = re.search('.*', '/git-evil-handler') content = list(handle_service_request(self._req, 'backend', mat)) self.assertEqual(HTTP_FORBIDDEN, self._status) self.assertFalse(b'git-evil-handler' in b"".join(content)) self.assertFalse(self._req.cached) def _run_handle_service_request(self, content_length=None): self._environ['wsgi.input'] = BytesIO(b'foo') if content_length is not None: self._environ['CONTENT_LENGTH'] = content_length mat = re.search('.*', '/git-upload-pack') class Backend(object): def open_repository(self, path): return None handler_output = b''.join( handle_service_request(self._req, Backend(), mat)) write_output = self._output.getvalue() # Ensure all output was written via the write callback. self.assertEqual(b'', handler_output) self.assertEqual(b'handled input: foo', write_output) self.assertContentTypeEquals('application/x-git-upload-pack-result') self.assertFalse(self._handler.advertise_refs) self.assertTrue(self._handler.http_req) self.assertFalse(self._req.cached) def test_handle_service_request(self): self._run_handle_service_request() def test_handle_service_request_with_length(self): self._run_handle_service_request(content_length='3') def test_handle_service_request_empty_length(self): self._run_handle_service_request(content_length='') def test_get_info_refs_unknown(self): self._environ['QUERY_STRING'] = 'service=git-evil-handler' class Backend(object): def open_repository(self, url): return None mat = re.search('.*', '/git-evil-pack') content = list(get_info_refs(self._req, Backend(), mat)) self.assertFalse(b'git-evil-handler' in b"".join(content)) self.assertEqual(HTTP_FORBIDDEN, self._status) self.assertFalse(self._req.cached) def test_get_info_refs(self): self._environ['wsgi.input'] = BytesIO(b'foo') self._environ['QUERY_STRING'] = 'service=git-upload-pack' class Backend(object): def open_repository(self, url): return None mat = re.search('.*', '/git-upload-pack') handler_output = b''.join(get_info_refs(self._req, Backend(), mat)) write_output = self._output.getvalue() self.assertEqual((b'001e# service=git-upload-pack\n' b'0000' # input is ignored by the handler b'handled input: '), write_output) # Ensure all output was written via the write callback. self.assertEqual(b'', handler_output) self.assertTrue(self._handler.advertise_refs) self.assertTrue(self._handler.http_req) self.assertFalse(self._req.cached) class LengthLimitedFileTestCase(TestCase): def test_no_cutoff(self): f = _LengthLimitedFile(BytesIO(b'foobar'), 1024) self.assertEqual(b'foobar', f.read()) def test_cutoff(self): f = _LengthLimitedFile(BytesIO(b'foobar'), 3) self.assertEqual(b'foo', f.read()) self.assertEqual(b'', f.read()) def test_multiple_reads(self): f = _LengthLimitedFile(BytesIO(b'foobar'), 3) self.assertEqual(b'fo', f.read(2)) self.assertEqual(b'o', f.read(2)) self.assertEqual(b'', f.read()) class HTTPGitRequestTestCase(WebTestCase): # This class tests the contents of the actual cache headers _req_class = HTTPGitRequest def test_not_found(self): self._req.cache_forever() # cache headers should be discarded message = 'Something not found' self.assertEqual(message.encode('ascii'), self._req.not_found(message)) self.assertEqual(HTTP_NOT_FOUND, self._status) self.assertEqual(set([('Content-Type', 'text/plain')]), set(self._headers)) def test_forbidden(self): self._req.cache_forever() # cache headers should be discarded message = 'Something not found' self.assertEqual(message.encode('ascii'), self._req.forbidden(message)) self.assertEqual(HTTP_FORBIDDEN, self._status) self.assertEqual(set([('Content-Type', 'text/plain')]), set(self._headers)) def test_respond_ok(self): self._req.respond() self.assertEqual([], self._headers) self.assertEqual(HTTP_OK, self._status) def test_respond(self): self._req.nocache() self._req.respond(status=402, content_type='some/type', headers=[('X-Foo', 'foo'), ('X-Bar', 'bar')]) self.assertEqual(set([ ('X-Foo', 'foo'), ('X-Bar', 'bar'), ('Content-Type', 'some/type'), ('Expires', 'Fri, 01 Jan 1980 00:00:00 GMT'), ('Pragma', 'no-cache'), ('Cache-Control', 'no-cache, max-age=0, must-revalidate'), ]), set(self._headers)) self.assertEqual(402, self._status) class HTTPGitApplicationTestCase(TestCase): def setUp(self): super(HTTPGitApplicationTestCase, self).setUp() self._app = HTTPGitApplication('backend') self._environ = { 'PATH_INFO': '/foo', 'REQUEST_METHOD': 'GET', } def _test_handler(self, req, backend, mat): # tests interface used by all handlers self.assertEqual(self._environ, req.environ) self.assertEqual('backend', backend) self.assertEqual('/foo', mat.group(0)) return 'output' def _add_handler(self, app): req = self._environ['REQUEST_METHOD'] app.services = { (req, re.compile('/foo$')): self._test_handler, } def test_call(self): self._add_handler(self._app) self.assertEqual('output', self._app(self._environ, None)) def test_fallback_app(self): def test_app(environ, start_response): return 'output' app = HTTPGitApplication('backend', fallback_app=test_app) self.assertEqual('output', app(self._environ, None)) class GunzipTestCase(HTTPGitApplicationTestCase): __doc__ = """TestCase for testing the GunzipFilter, ensuring the wsgi.input is correctly decompressed and headers are corrected. """ example_text = __doc__.encode('ascii') def setUp(self): super(GunzipTestCase, self).setUp() self._app = GunzipFilter(self._app) self._environ['HTTP_CONTENT_ENCODING'] = 'gzip' self._environ['REQUEST_METHOD'] = 'POST' def _get_zstream(self, text): zstream = BytesIO() zfile = gzip.GzipFile(fileobj=zstream, mode='w') zfile.write(text) zfile.close() zlength = zstream.tell() zstream.seek(0) return zstream, zlength def _test_call(self, orig, zstream, zlength): self._add_handler(self._app.app) self.assertLess(zlength, len(orig)) self.assertEqual(self._environ['HTTP_CONTENT_ENCODING'], 'gzip') self._environ['CONTENT_LENGTH'] = zlength self._environ['wsgi.input'] = zstream self._app(self._environ, None) buf = self._environ['wsgi.input'] self.assertIsNot(buf, zstream) buf.seek(0) self.assertEqual(orig, buf.read()) self.assertIs(None, self._environ.get('CONTENT_LENGTH')) self.assertNotIn('HTTP_CONTENT_ENCODING', self._environ) def test_call(self): self._test_call( self.example_text, *self._get_zstream(self.example_text) ) def test_call_no_seek(self): """ This ensures that the gunzipping code doesn't require any methods on 'wsgi.input' except for '.read()'. (In particular, it shouldn't require '.seek()'. See https://github.com/jelmer/dulwich/issues/140.) """ zstream, zlength = self._get_zstream(self.example_text) self._test_call( self.example_text, MinimalistWSGIInputStream(zstream.read()), zlength) def test_call_no_working_seek(self): """ Similar to 'test_call_no_seek', but this time the methods are available (but defunct). See https://github.com/jonashaag/klaus/issues/154. """ zstream, zlength = self._get_zstream(self.example_text) self._test_call( self.example_text, MinimalistWSGIInputStream2(zstream.read()), zlength) diff --git a/setup.cfg b/setup.cfg index 6f815c04..9e02035d 100644 --- a/setup.cfg +++ b/setup.cfg @@ -1,2 +1,5 @@ [flake8] exclude = build,.git,build-pypy,.tox + +[mypy] +ignore_missing_imports = True