GitHub Viewer
# Copyright (C) 2008, 2009 Michael Trier (mtrier@gmail.com) and contributors
#
# This module is part of GitPython and is released under the
# 3-Clause BSD License: https://opensource.org/license/bsd-3-clause/
import copy
from datetime import datetime
from io import BytesIO
from pathlib import Path
import os.path as osp
import sys
import time
from unittest.mock import Mock
from gitdb import IStream
from git import Actor, Commit, Repo, _backend
from git.exc import UnsafeOptionError
from git.objects.util import tzoffset, utc
from test.lib.fixtures import fixture_repo, writable_fixture, read_only_fixture
from test.lib import (
StringProcessAdapter,
TestBase,
fixture_path,
with_rw_repo,
)
class TestCommitSerialization(TestBase):
def assert_commit_serialization(self, rwrepo, commit_id, print_performance_info=False):
"""Check selected root, ordinary and merge commits against stored bytes.
:param print_performance_info: If True, we will show how fast we are.
"""
ns = 0 # Number of serializations.
nds = 0 # Number of deserializations.
st = time.time()
for cm in [
rwrepo.commit(commit_id),
rwrepo.commit("4c39f9da792792d4e73fc3a5effde66576ae128c"),
rwrepo.commit("0.1.6"),
next(rwrepo.commit("0.1.6").iter_parents(max_count=1)),
]:
nds += 1
# Assert that we deserialize commits correctly, hence we get the same
# sha on serialization.
stream = BytesIO()
cm._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
istream = rwrepo.odb.store(IStream(Commit.type, streamlen, stream))
self.assertEqual(istream.hexsha, cm.hexsha.encode("ascii"))
nc = Commit(
rwrepo,
Commit.NULL_BIN_SHA,
cm.tree,
cm.author,
cm.authored_date,
cm.author_tz_offset,
cm.committer,
cm.committed_date,
cm.committer_tz_offset,
cm.message,
cm.parents,
cm.encoding,
)
self.assertEqual(nc.parents, cm.parents)
stream = BytesIO()
nc._serialize(stream)
ns += 1
streamlen = stream.tell()
stream.seek(0)
# Reuse istream.
istream.size = streamlen
istream.stream = stream
istream.binsha = None
nc.binsha = rwrepo.odb.store(istream).binsha
# If it worked, we have exactly the same contents!
self.assertEqual(nc.hexsha, cm.hexsha)
# END check commits
elapsed = time.time() - st
if print_performance_info:
print(
"Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s"
% (ns, nds, elapsed, ns / elapsed, nds / elapsed),
file=sys.stderr,
)
# END handle performance info
class TestCommit(TestCommitSerialization):
def test_bake(self):
commit = self.rorepo.commit("2454ae89983a4496a445ce347d7a41c0bb0ea7ae")
# Commits have no dict.
self.assertRaises(AttributeError, setattr, commit, "someattr", 1)
commit.author # bake
self.assertEqual("Sebastian Thiel", commit.author.name)
self.assertEqual("byronimo@gmail.com", commit.author.email)
self.assertEqual(commit.author, commit.committer)
assert isinstance(commit.authored_date, int) and isinstance(commit.committed_date, int)
assert isinstance(commit.author_tz_offset, int) and isinstance(commit.committer_tz_offset, int)
self.assertEqual(
commit.message,
"Added missing information to docstrings of commit and stats module\n",
)
@with_rw_repo("HEAD")
def test_replace_no_changes(self, rw_repo):
old_commit = rw_repo.commit("2454ae89983a4496a445ce347d7a41c0bb0ea7ae")
new_commit = old_commit.replace()
for attr in old_commit.__slots__:
assert getattr(new_commit, attr) == getattr(old_commit, attr)
@with_rw_repo("HEAD")
def test_replace_new_sha(self, rw_repo):
commit = rw_repo.commit("2454ae89983a4496a445ce347d7a41c0bb0ea7ae")
new_commit = commit.replace(message="Added replace method")
assert new_commit.hexsha == "fc84cbecac1bd4ba4deaac07c1044889edd536e6"
assert new_commit.message == "Added replace method"
@with_rw_repo("HEAD")
def test_replace_invalid_attribute(self, rw_repo):
commit = rw_repo.commit("2454ae89983a4496a445ce347d7a41c0bb0ea7ae")
with self.assertRaises(ValueError):
commit.replace(badattr="This will never work")
def test_stats(self):
commit = self.rorepo.commit("33ebe7acec14b25c5f84f35a664803fcab2f7781")
stats = commit.stats
def check_entries(d, has_change_type=False):
assert isinstance(d, dict)
keys = ("insertions", "deletions", "lines")
if has_change_type:
keys += ("change_type",)
for key in keys:
assert key in d
# END assertion helper
assert stats.files
assert stats.total
check_entries(stats.total)
assert "files" in stats.total
for _filepath, d in stats.files.items():
check_entries(d, True)
# END for each stated file
# Check that data is parsed properly.
michael = Actor._from_string("Michael Trier ")
self.assertEqual(commit.author, michael)
self.assertEqual(commit.committer, michael)
self.assertEqual(commit.authored_date, 1210193388)
self.assertEqual(commit.committed_date, 1210193388)
self.assertEqual(commit.author_tz_offset, 14400, commit.author_tz_offset)
self.assertEqual(commit.committer_tz_offset, 14400, commit.committer_tz_offset)
self.assertEqual(commit.message, "initial project\n")
def test_stats_preserve_filename_spaces(self):
paths = ("name", " name", "name ", " name ", " ")
with fixture_repo("commit.sh", "spaces") as repo:
for revision, change_type in (("update-1", "A"), ("update-2", "M")):
stats = repo.commit(revision).stats
assert stats.total == {"insertions": 15, "deletions": 0, "lines": 15, "files": 5}
assert stats.files == {
path: {"insertions": count, "deletions": 0, "lines": count, "change_type": change_type}
for count, path in enumerate(paths, 1)
}
empty_stats = repo.commit("empty").stats
assert empty_stats.files == {}
assert empty_stats.total == {"insertions": 0, "deletions": 0, "lines": 0, "files": 0}
removed_stats = repo.commit("removed").stats
assert removed_stats.files == {
path: {"insertions": 0, "deletions": 2 * count, "lines": 2 * count, "change_type": "D"}
for count, path in enumerate(paths, 1)
}
assert removed_stats.total == {"insertions": 0, "deletions": 30, "lines": 30, "files": 5}
def test_renames(self):
commit = self.rorepo.commit("185d847ec7647fd2642a82d9205fb3d07ea71715")
files = commit.stats.files
# When a file is renamed, the output of git diff is like "dir/{old => new}"
# unless we disable rename with --no-renames, which produces two lines,
# one with the old path deletes and another with the new added.
self.assertEqual(len(files), 2)
def check_entries(path, changes):
expected = {
".github/workflows/Future.yml": {
"insertions": 57,
"deletions": 0,
"lines": 57,
},
".github/workflows/test_pytest.yml": {
"insertions": 0,
"deletions": 55,
"lines": 55,
},
}
assert path in expected
assert isinstance(changes, dict)
for key in ("insertions", "deletions", "lines"):
assert changes[key] == expected[path][key]
for path, changes in files.items():
check_entries(path, changes)
# END for each stated file
def test_unicode_actor(self):
# Check that we can parse Unicode actors correctly.
name = "Üäöß ÄußÉ"
self.assertEqual(len(name), 9)
special = Actor._from_string("%s " % name)
self.assertEqual(special.name, name)
assert isinstance(special.name, str)
def test_traversal(self):
start = self.rorepo.commit("a4d06724202afccd2b5c54f81bcf2bf26dea7fff")
first = self.rorepo.commit("33ebe7acec14b25c5f84f35a664803fcab2f7781")
p0 = start.parents[0]
p1 = start.parents[1]
p00 = p0.parents[0]
p10 = p1.parents[0]
# Basic branch first, depth first.
dfirst = start.traverse(branch_first=False)
bfirst = start.traverse(branch_first=True)
self.assertEqual(next(dfirst), p0)
self.assertEqual(next(dfirst), p00)
self.assertEqual(next(bfirst), p0)
self.assertEqual(next(bfirst), p1)
self.assertEqual(next(bfirst), p00)
self.assertEqual(next(bfirst), p10)
# At some point, both iterations should stop.
self.assertEqual(list(bfirst)[-1], first)
stoptraverse = self.rorepo.commit("254d04aa3180eb8b8daf7b7ff25f010cd69b4e7d").traverse(
ignore_self=0, as_edge=True
)
stoptraverse_list = list(stoptraverse)
for itemtup in stoptraverse_list:
self.assertIsInstance(itemtup, (tuple)) and self.assertEqual(len(itemtup), 2) # as_edge=True -> tuple
src, item = itemtup
self.assertIsInstance(item, Commit)
if src:
self.assertIsInstance(src, Commit)
else:
self.assertIsNone(src) # ignore_self=0 -> first is (None, Commit)
stoptraverse = self.rorepo.commit("254d04aa3180eb8b8daf7b7ff25f010cd69b4e7d").traverse(as_edge=True)
self.assertEqual(len(next(stoptraverse)), 2)
# Ignore self
self.assertEqual(next(start.traverse(ignore_self=False)), start)
# Depth
self.assertEqual(len(list(start.traverse(ignore_self=False, depth=0))), 1)
# Prune
self.assertEqual(next(start.traverse(branch_first=1, prune=lambda i, d: i == p0)), p1)
# Predicate
self.assertEqual(next(start.traverse(branch_first=1, predicate=lambda i, d: i == p1)), p1)
# Traversal should stop when the beginning is reached.
self.assertRaises(StopIteration, next, first.traverse())
# Parents of the first commit should be empty (as the only parent has a null sha)
self.assertEqual(len(first.parents), 0)
def test_iteration(self):
# We can iterate commits.
all_commits = Commit.list_items(self.rorepo, self.rorepo.head)
assert all_commits
self.assertEqual(all_commits, list(self.rorepo.iter_commits()))
# This includes merge commits.
mcomit = self.rorepo.commit("d884adc80c80300b4cc05321494713904ef1df2d")
assert mcomit in all_commits
# We can limit the result to paths.
ltd_commits = list(self.rorepo.iter_commits(paths="CHANGES"))
assert ltd_commits and len(ltd_commits) < len(all_commits)
# Show commits of multiple paths, resulting in a union of commits.
less_ltd_commits = list(Commit.iter_items(self.rorepo, "HEAD", paths=("CHANGES", "AUTHORS")))
assert len(ltd_commits) < len(less_ltd_commits)
class Child(Commit):
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
child_commits = list(Child.iter_items(self.rorepo, "HEAD", paths=("CHANGES", "AUTHORS")))
assert type(child_commits[0]) is Child
def test_iter_items(self):
# pretty not allowed.
self.assertRaises(ValueError, Commit.iter_items, self.rorepo, "master", pretty="raw")
def test_iter_items_rejects_unsafe_revision(self):
with writable_fixture("empty.sh") as temporary:
tdir = str(temporary.path().resolve())
marker = osp.join(tdir, "pwn")
self.assertRaises(UnsafeOptionError, Commit.iter_items, self.rorepo, f"--output={marker}")
def test_iter_items_rejects_unsafe_options(self):
with writable_fixture("empty.sh") as temporary:
tdir = str(temporary.path().resolve())
marker = osp.join(tdir, "pwn")
with self.assertRaises(UnsafeOptionError):
list(Commit.iter_items(self.rorepo, "HEAD", output=marker))
def test_count_rejects_unsafe_options(self):
with writable_fixture("empty.sh") as temporary:
with self.assertRaises(UnsafeOptionError):
self.rorepo.head.commit.count(output=str(temporary.path() / "output"))
def test_rev_list_bisect_all(self):
"""
'git rev-list --bisect-all' returns additional information
in the commit header. This test ensures that we properly parse it.
"""
revs = self.rorepo.git.rev_list(
"933d23bf95a5bd1624fbcdf328d904e1fa173474",
first_parent=True,
bisect_all=True,
)
commits = Commit._iter_from_process_or_stream(self.rorepo, StringProcessAdapter(revs.encode("ascii")))
expected_ids = (
"7156cece3c49544abb6bf7a0c218eb36646fad6d",
"1f66cfbbce58b4b552b041707a12d437cc5f400a",
"33ebe7acec14b25c5f84f35a664803fcab2f7781",
"933d23bf95a5bd1624fbcdf328d904e1fa173474",
)
for sha1, commit in zip(expected_ids, commits):
self.assertEqual(sha1, commit.hexsha)
def test_iter_from_invalid_process_or_stream(self):
for source, error in ((Mock(wait=Mock(), stdout=None), ValueError), (object(), TypeError)):
with self.assertRaises(error):
list(Commit._iter_from_process_or_stream(self.rorepo, source))
def test_ambiguous_arg_iteration(self):
with fixture_repo("commit.sh", "ambiguous") as repo:
assert list(repo.iter_commits(repo.head.ref)) == [repo.head.commit]
def test_count(self):
self.assertEqual(self.rorepo.tag("refs/tags/0.1.5").commit.count(), 143)
def test_list(self):
# This doesn't work anymore, as we will either attempt getattr with bytes, or
# compare 20 byte string with actual 20 byte bytes. This usage makes no sense
# anyway.
assert isinstance(
Commit.list_items(self.rorepo, "0.1.5", max_count=5)["5117c9c8a4d3af19a9958677e45cda9269de1541"],
Commit,
)
def test_str(self):
commit = Commit(self.rorepo, Commit.NULL_BIN_SHA)
self.assertEqual(Commit.NULL_HEX_SHA, str(commit))
def test_repr(self):
commit = Commit(self.rorepo, Commit.NULL_BIN_SHA)
self.assertEqual('' % Commit.NULL_HEX_SHA, repr(commit))
def test_equality(self):
commit1 = Commit(self.rorepo, Commit.NULL_BIN_SHA)
commit2 = Commit(self.rorepo, Commit.NULL_BIN_SHA)
commit3 = Commit(self.rorepo, b"\1" * 20)
self.assertEqual(commit1, commit2)
self.assertNotEqual(commit2, commit3)
def test_iter_parents(self):
# Should return all but ourselves, even if skip is defined.
c = self.rorepo.commit("0.1.5")
for skip in (0, 1):
piter = c.iter_parents(skip=skip)
first_parent = next(piter)
assert first_parent != c
self.assertEqual(first_parent, c.parents[0])
# END for each
def test_name_rev(self):
name_rev = self.rorepo.head.commit.name_rev
assert isinstance(name_rev, str)
@with_rw_repo("HEAD", bare=True)
def test_serialization(self, rwrepo):
# Create all commits of our repo.
self.assert_commit_serialization(rwrepo, "0.1.6")
@with_rw_repo("HEAD")
def test_serialization_unicode_support(self, rw_repo):
self.assertEqual(Commit.default_encoding.lower(), "utf-8")
# Create a commit with Unicode in the message, and the author's name.
# Verify its serialization and deserialization.
cmt = rw_repo.commit("0.1.6")
assert isinstance(cmt.message, str) # It automatically decodes it as such.
assert isinstance(cmt.author.name, str) # Same here.
cmt.message = "üäêèß"
self.assertEqual(len(cmt.message), 5)
cmt.author.name = "äüß"
self.assertEqual(len(cmt.author.name), 3)
cstream = BytesIO()
cmt._serialize(cstream)
cstream.seek(0)
assert len(cstream.getvalue())
ncmt = Commit(rw_repo, cmt.binsha)
stored = rw_repo.odb.store(IStream("commit", len(cstream.getvalue()), cstream))
ncmt = Commit(rw_repo, stored.binsha)
self.assertEqual(cmt.author.name, ncmt.author.name)
self.assertEqual(cmt.message, ncmt.message)
# Actually, it can't be printed in a shell as repr wants to have ascii only it
# appears.
cmt.author.__repr__()
def test_invalid_commit(self):
fixture = read_only_fixture(
"commit.sh",
"raw",
fixture_path("commit_invalid_data"),
Path(fixture_path("commit_invalid_data")).read_bytes().hex(),
)
with Repo(fixture / "repo") as repo:
cmt = Commit(repo, bytes.fromhex((fixture / "object-id").read_text().strip()))
self.assertEqual(cmt.author.name, "E.Azer Ko�o�o�oculu", cmt.author.name)
self.assertEqual(cmt.author.email, "azer@kodfabrik.com", cmt.author.email)
def test_deserialization_declared_encoding(self):
# Keep literal text expectations for correctly declared UTF-8/Latin-1.
# Other codecs use the installed Git's whole-commit projection below.
for scenario, encoding, name, message in (
("utf8", "UTF-8", "André", "Résumé\n"),
("git-only-codec", "UTF-8", None, None),
("latin1", "ISO-8859-1", "André", "Résumé\n"),
("latin1-signature", "ISO-8859-1", "André", "Résumé\n"),
("utf16-signature", "utf-16-be", None, None),
("late-latin1", "ISO-8859-1", "André", "Résumé\n"),
("unknown", "UTF-8", None, None),
("nontext", "UTF-8", None, None),
("nonascii", "UTF-8", None, None),
("utf16-then-latin1", "utf-16-be", None, None),
("utf16be-then-latin1", "UTF-16BE", None, None),
("latin1-then-nonascii", "ISO-8859-1", "André", "Résumé\n"),
("latin1-then-utf8", "ISO-8859-1", "André", "Résumé\n"),
("unknown-then-latin1", "UTF-8", None, None),
("empty-then-latin1", "UTF-8", None, None),
):
with self.subTest(scenario=scenario), fixture_repo("commit.sh", "encoding", scenario) as repo:
oid = (Path(repo.working_dir).parent / "object-id").read_text().strip()
if scenario.startswith("utf16"):
# Successful UTF-16 recoding would erase Git's header delimiters.
self.assertEqual(int(repo.git.cat_file("-s", oid)) % 2, 1)
before = _backend.statistics()
cmt = Commit(repo, bytes.fromhex(oid))
self.assertEqual(cmt.encoding, encoding)
self.assertEqual((cmt.authored_date, cmt.committed_date), (1112591163, 1112591163))
self.assertEqual((cmt.author_tz_offset, cmt.committer_tz_offset), (-7200, -7200))
if name is not None:
self.assertEqual(cmt.author.name, name)
self.assertEqual(cmt.author.email, "andre@example.com")
self.assertEqual(cmt.message, message)
self.assertEqual(cmt.committer, cmt.author)
# Git selects the first header; codec aliases and empty labels vary by platform.
expected = (
repo.git.show(
"--no-patch",
"--no-notes",
"--no-show-signature",
"--no-use-mailmap",
"--encoding=UTF-8",
"--format=format:%an%x00%ae%x00%cn%x00%ce%x00%B",
oid,
"--",
stdout_as_string=False,
strip_newline_in_stdout=False,
)
.decode("UTF-8", "replace")
.split("\0", 4)
)
self.assertEqual(
[cmt.author.name, cmt.author.email, cmt.committer.name, cmt.committer.email, cmt.message],
expected,
)
if scenario.endswith("-signature"):
signature = "signé\ncontinuedé" if scenario == "latin1-signature" else "sign\ncontinued"
self.assertEqual(cmt.gpgsig, signature)
raw = repo.git.show(
"--no-patch",
"--no-notes",
"--no-show-signature",
"--encoding=UTF-8",
"--format=raw",
oid,
"--",
stdout_as_string=False,
strip_newline_in_stdout=False,
).decode("UTF-8", "ignore")
self.assertIn("gpgsig " + signature.replace("\n", "\n ") + "\n", raw)
if _backend.name == "gix":
outcome = "native" if scenario == "utf8" else "CLI: commit text encoding (GIX-29)"
if scenario == "late-latin1":
outcome = "CLI: commit encoding header order (GIX-30)"
key = ("commit.decode", outcome)
self.assertEqual(_backend.statistics().get(key, 0), before.get(key, 0) + 1)
def test_identity_cannot_alter_headers(self):
"""A name or email must not add header lines or present another identity."""
with fixture_repo("commit.sh", "tree", write=True) as rw_repo:
tree = rw_repo.tree((Path(rw_repo.working_tree_dir).parent / "tree-id").read_text().strip())
service = Actor("Service", "service@example.com")
forged = "committer Forged 0 +0000"
for name, email in (
# A line feed ends the header line, so the remainder would become headers
# of its own, which Git reads before the committer written after them.
("User 0 +0000\n" + forged, "user@example.com"),
("User", "user@example.com> 0 +0000\n" + forged),
# Angle brackets delimit the email, so these would present another one.
("Forged ", "user@example.com"),
("User", "forged@example.com> ", "user@example.com"),
("User", "