def test_email_mapping(self):
contributors = Contributor.Mapping()
Contributor.from_scm_log('Author: Jonathan Bedard <*****@*****.**>',
contributors)
contributor = Contributor.from_scm_log(
'r266751 | [email protected] | 2020-09-08 14:33:42 -0700 (Tue, 08 Sep 2020) | 10 lines',
contributors)
self.assertEqual(contributor.name, 'Jonathan Bedard')
self.assertEqual(contributor.emails, ['*****@*****.**'])
def test_author_mapping(self):
contributors = Contributor.Mapping()
Contributor.from_scm_log('Author: Jonathan Bedard <*****@*****.**>',
contributors)
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <*****@*****.**>', contributors)
self.assertEqual(contributor.name, 'Jonathan Bedard')
self.assertEqual(contributor.emails,
['*****@*****.**', '*****@*****.**'])
def test_git_svn_log(self):
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <[email protected]@268f45cc-cd09-0410-ab3c-d52691b4dbfc>'
)
self.assertEqual(contributor.name, 'Jonathan Bedard')
self.assertEqual(contributor.emails, ['*****@*****.**'])
def test_short_svn_log(self):
contributor = Contributor.from_scm_log(
'r266751 | [email protected] | 2020-09-08 14:33:42 -0700 (Tue, 08 Sep 2020) | 1 line'
)
self.assertEqual(contributor.name, '*****@*****.**')
self.assertEqual(contributor.emails, ['*****@*****.**'])
def test_json_encode(self):
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <*****@*****.**>')
self.assertDictEqual(
dict(
revision=1,
hash='c3bd784f8b88bd03f64467ddd3304ed8be28acbe',
branch='main',
identifier='1@main',
timestamp=1000,
order=0,
author=dict(
name='Jonathan Bedard',
emails=['*****@*****.**'],
),
message='Message',
),
json.loads(
json.dumps(Commit(
revision=1,
hash='c3bd784f8b88bd03f64467ddd3304ed8be28acbe',
identifier='1@main',
timestamp=1000,
author=contributor,
message='Message',
),
cls=Commit.Encoder)))
def _args_from_content(self, content, include_log=True):
author = None
timestamp = None
for line in content.splitlines()[:4]:
split = line.split(': ')
if split[0] == 'Author':
author = Contributor.from_scm_log(line.lstrip(),
self.contributors)
elif split[0] == 'CommitDate':
tz_diff = line.split(' ')[-1]
date = datetime.strptime(split[1].lstrip()[:-len(tz_diff)],
'%a %b %d %H:%M:%S %Y ')
date += timedelta(
hours=int(tz_diff[1:3]),
minutes=int(tz_diff[3:5]),
) * (1 if tz_diff[0] == '-' else -1)
timestamp = int(calendar.timegm(
date.timetuple())) - time.timezone
message = ''
for line in content.splitlines()[5:]:
message += line[4:] + '\n'
matches = self.GIT_SVN_REVISION.findall(message)
return dict(
revision=int(matches[-1].split('@')[0]) if matches else None,
author=author,
timestamp=timestamp,
message=message.rstrip() if include_log else None,
)
def test_contributor(self):
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <*****@*****.**>')
commit = Commit(revision=1, identifier=1, author=contributor)
self.assertEqual(commit.author, contributor)
commit = Commit(revision=1,
identifier=1,
author=Contributor.Encoder().default(contributor))
self.assertEqual(commit.author, contributor)
def test_json_decode(self):
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <*****@*****.**>')
commit_a = Commit(revision=1,
hash='c3bd784f8b88bd03f64467ddd3304ed8be28acbe',
identifier='1@main',
timestamp=1000,
author=Contributor.Encoder().default(contributor),
message='Message')
dictionary = json.loads(json.dumps(commit_a, cls=Commit.Encoder))
commit_b = Commit(**dictionary)
self.assertEqual(commit_a, commit_b)
def test_from_json_str(self):
contributor = Contributor.from_scm_log(
'Author: [email protected] <*****@*****.**>')
self.assertEqual(
Commit.from_json('''{
"revision": 1,
"hash": "c3bd784f8b88bd03f64467ddd3304ed8be28acbe",
"identifier": "1@main",
"timestamp": 1000,
"author": "*****@*****.**",
"message": "Message"
}'''),
Commit(revision=1,
hash='c3bd784f8b88bd03f64467ddd3304ed8be28acbe',
identifier='1@main',
timestamp=1000,
author=Contributor.Encoder().default(contributor),
message='Message'),
)
def _args_from_content(self, content, include_log=True):
leading = content.splitlines()[0]
match = Contributor.SVN_AUTHOR_RE.match(
leading) or Contributor.SVN_AUTHOR_Q_RE.match(leading)
if not match:
return {}
tz_diff = match.group('date').split(' ', 2)[-1]
date = datetime.strptime(
match.group('date')[:-len(tz_diff)], '%Y-%m-%d %H:%M:%S ')
date += timedelta(
hours=int(tz_diff[1:3]),
minutes=int(tz_diff[3:5]),
) * (1 if tz_diff[0] == '-' else -1)
return dict(
revision=int(match.group('revision')),
timestamp=int(calendar.timegm(date.timetuple())),
author=Contributor.from_scm_log(leading, self.contributors),
message='\n'.join(content.splitlines()[2:]).rstrip()
if include_log else None,
)
def test_git_no_author(self):
contributor = Contributor.from_scm_log(
'Author: Automated Checkin <devnull>')
self.assertIsNone(contributor)
def commit(self, hash=None, revision=None, identifier=None, branch=None, tag=None, include_log=True):
if hash:
raise ValueError('SVN does not support Git hashes')
parsed_branch_point = None
if identifier is not None:
if revision:
raise ValueError('Cannot define both revision and identifier')
if tag:
raise ValueError('Cannot define both tag and identifier')
parsed_branch_point, identifier, parsed_branch = Commit._parse_identifier(identifier, do_assert=True)
if parsed_branch:
if branch and branch != parsed_branch:
raise ValueError(
"Caller passed both 'branch' and 'identifier', but specified different branches ({} and {})".format(
branch, parsed_branch,
),
)
branch = parsed_branch
branch = branch or self.branch
if branch == self.default_branch and parsed_branch_point:
raise self.Exception('Cannot provide a branch point for a commit on the default branch')
if not self._metadata_cache.get(branch, []) or identifier >= len(self._metadata_cache.get(branch, [])):
if branch != self.default_branch:
self._cache_revisions(branch=self.default_branch)
self._cache_revisions(branch=branch)
if identifier > len(self._metadata_cache.get(branch, [])):
raise self.Exception('Identifier {} cannot be found on the specified branch in the current checkout'.format(identifier))
if identifier <= 0:
if branch == self.default_branch:
raise self.Exception('Illegal negative identifier on the default branch')
identifier = self._commit_count(branch=branch) + identifier
if identifier < 0:
raise self.Exception('Identifier does not exist on the specified branch')
branch = self.default_branch
revision = self._metadata_cache[branch][identifier]
info = self.info(cached=True, branch=branch, revision=revision)
branch = self._branch_for(revision)
if not self._metadata_cache.get(branch, []) or identifier >= len(self._metadata_cache.get(branch, [])):
self._cache_revisions(branch=branch)
elif revision:
if branch:
raise ValueError('Cannot define both branch and revision')
if tag:
raise ValueError('Cannot define both tag and revision')
revision = Commit._parse_revision(revision, do_assert=True)
branch = self._branch_for(revision)
info = self.info(cached=True, revision=revision)
else:
if branch and tag:
raise ValueError('Cannot define both branch and tag')
branch = None if tag else branch or self.branch
info = self.info(tag=tag) if tag else self.info(branch=branch)
if not info:
raise self.Exception("'{}' is not a recognized {}".format(
tag or branch,
'tag' if tag else 'branch',
))
revision = int(info['Last Changed Rev'])
if branch != self.default_branch:
branch = self._branch_for(revision)
date = info['Last Changed Date'].split(' (')[0]
tz_diff = date.split(' ')[-1]
date = datetime.strptime(date[:-len(tz_diff)], '%Y-%m-%d %H:%M:%S ')
date += timedelta(
hours=int(tz_diff[1:3]),
minutes=int(tz_diff[3:5]),
) * (1 if tz_diff[0] == '-' else -1)
if not identifier:
if branch != self.default_branch and revision > self._metadata_cache.get(self.default_branch, [0])[-1]:
self._cache_revisions(branch=self.default_branch)
if revision not in self._metadata_cache.get(branch, []):
self._cache_revisions(branch=branch)
identifier = self._commit_count(revision=revision, branch=branch)
branch_point = None if branch == self.default_branch else self._commit_count(branch=branch)
if branch_point and parsed_branch_point and branch_point != parsed_branch_point:
raise ValueError("Provided 'branch_point' does not match branch point of specified branch")
if branch == self.default_branch or '/' in branch:
branch_arg = '^/{}'.format(branch)
else:
branch_arg = '^/branches/{}'.format(branch)
log = run(
[self.executable(), 'log', '-l', '1', '-r', str(revision), branch_arg], cwd=self.root_path,
capture_output=True, encoding='utf-8',
) if include_log else None
split_log = log.stdout.splitlines() if log else []
if log and (not log.returncode or len(split_log) >= 3):
author_line = split_log[1]
for line in split_log[2:8]:
if Contributor.SVN_PATCH_FROM_RE.match(line):
author_line = line
break
author = Contributor.from_scm_log(author_line, self.contributors)
message = '\n'.join(split_log[3:-1])
else:
if include_log:
self.log('Failed to connect to remote, cannot compute commit message')
email = info.get('Last Changed Author')
author = self.contributors.create(email, email) if '@' in email else self.contributors.create(email)
message = None
return Commit(
revision=int(revision),
branch=branch,
identifier=identifier,
branch_point=branch_point,
timestamp=int(calendar.timegm(date.timetuple())),
author=author,
message=message,
)
def commit(self,
hash=None,
revision=None,
identifier=None,
branch=None,
tag=None,
include_log=True,
include_identifier=True):
# Only git-svn checkouts can convert revisions to fully qualified commits
if revision and not self.is_svn:
raise self.Exception(
'This git checkout does not support SVN revisions')
# Determine the hash for a provided Subversion revision
elif revision:
if hash:
raise ValueError('Cannot define both hash and revision')
revision = Commit._parse_revision(revision, do_assert=True)
revision_log = run(
[self.executable(), 'svn', 'find-rev', 'r{}'.format(revision)],
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
timeout=3,
)
if revision_log.returncode:
raise self.Exception(
"Failed to retrieve commit information for 'r{}'".format(
revision))
hash = revision_log.stdout.rstrip()
if not hash:
raise self.Exception("Failed to find 'r{}'".format(revision))
default_branch = self.default_branch
parsed_branch_point = None
log_format = ['-1'] if include_log else ['-1', '--format=short']
# Determine the `git log` output and branch for a given identifier
if identifier is not None:
if revision:
raise ValueError('Cannot define both revision and identifier')
if hash:
raise ValueError('Cannot define both hash and identifier')
if tag:
raise ValueError('Cannot define both tag and identifier')
parsed_branch_point, identifier, parsed_branch = Commit._parse_identifier(
identifier, do_assert=True)
if parsed_branch:
if branch and branch != parsed_branch:
raise ValueError(
"Caller passed both 'branch' and 'identifier', but specified different branches ({} and {})"
.format(
branch,
parsed_branch,
), )
branch = parsed_branch
baseline = branch or 'HEAD'
is_default = baseline == default_branch
if baseline == 'HEAD':
is_default = default_branch in self._branches_for(baseline)
if is_default and parsed_branch_point:
raise self.Exception(
'Cannot provide a branch point for a commit on the default branch'
)
base_count = self._commit_count(
baseline if is_default else '{}..{}'.
format(default_branch, baseline))
if identifier > base_count:
raise self.Exception(
'Identifier {} cannot be found on the specified branch in the current checkout'
.format(identifier))
log = run(
[
self.executable(), 'log', '{}~{}'.format(
branch or 'HEAD', base_count - identifier)
] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
)
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for 'i{}@{}'".
format(identifier, branch or 'HEAD'))
# Negative identifiers are actually commits on the default branch, we will need to re-compute the identifier
if identifier < 0 and is_default:
raise self.Exception(
'Illegal negative identifier on the default branch')
if identifier < 0:
identifier = None
# Determine the `git log` output for a given branch or tag
elif branch or tag:
if hash:
raise ValueError('Cannot define both tag/branch and hash')
if branch and tag:
raise ValueError('Cannot define both tag and branch')
log = run([self.executable(), 'log', branch or tag] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8')
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for '{}'".format(
branch or tag))
# Determine the `git log` output for a given hash
else:
hash = Commit._parse_hash(hash, do_assert=True)
log = run([self.executable(), 'log', hash or 'HEAD'] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8')
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for '{}'".format(
hash or 'HEAD'))
# Fully define the hash from the `git log` output
match = self.GIT_COMMIT.match(log.stdout.splitlines()[0])
if not match:
raise self.Exception('Invalid commit hash in git log')
hash = match.group('hash')
# A commit is often on multiple branches, the canonical branch is the one with the highest priority
branch = self.prioritize_branches(self._branches_for(hash))
# Compute the identifier if the function did not receive one and we were asked to
if not identifier and include_identifier:
identifier = self._commit_count(
hash if branch ==
default_branch else '{}..{}'.format(default_branch, hash))
# Only compute the branch point we're on something other than the default branch
branch_point = None if not include_identifier or branch == default_branch else self._commit_count(
hash) - identifier
if branch_point and parsed_branch_point and branch_point != parsed_branch_point:
raise ValueError(
"Provided 'branch_point' does not match branch point of specified branch"
)
# Check the commit log for a git-svn revision
logcontent = '\n'.join(line[4:]
for line in log.stdout.splitlines()[4:])
matches = self.GIT_SVN_REVISION.findall(logcontent)
revision = int(matches[-1].split('@')[0]) if matches else None
# We only care about when a commit was commited
commit_time = run(
[self.executable(), 'show', '-s', '--format=%ct', hash],
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
)
if commit_time.returncode:
raise self.Exception(
'Failed to retrieve commit time for {}'.format(hash))
timestamp = int(commit_time.stdout.lstrip())
# Comparing commits in different repositories involves comparing timestamps. This is problematic because it git,
# it's possible for a series of commits to share a commit time. To handle this case, we assign each commit a
# zero-indexed "order" within it's timestamp.
order = 0
while not identifier or order + 1 < identifier + (branch_point or 0):
commit_time = run(
[
self.executable(), 'show', '-s', '--format=%ct',
'{}~{}'.format(hash, order + 1)
],
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
)
if commit_time.returncode:
break
if int(commit_time.stdout.lstrip()) != timestamp:
break
order += 1
return Commit(
repository_id=self.id,
hash=hash,
revision=revision,
identifier=identifier if include_identifier else None,
branch_point=branch_point,
branch=branch,
timestamp=timestamp,
order=order,
author=Contributor.from_scm_log(log.stdout.splitlines()[1],
self.contributors),
message=logcontent if include_log else None,
)
def commit(self,
hash=None,
revision=None,
identifier=None,
branch=None,
tag=None,
include_log=True):
if revision and not self.is_svn:
raise self.Exception(
'This git checkout does not support SVN revisions')
elif revision:
if hash:
raise ValueError('Cannot define both hash and revision')
revision = Commit._parse_revision(revision, do_assert=True)
revision_log = run(
[self.executable(), 'svn', 'find-rev', 'r{}'.format(revision)],
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
timeout=3,
)
if revision_log.returncode:
raise self.Exception(
"Failed to retrieve commit information for 'r{}'".format(
revision))
hash = revision_log.stdout.rstrip()
if not hash:
raise self.Exception("Failed to find 'r{}'".format(revision))
default_branch = self.default_branch
parsed_branch_point = None
log_format = ['-1'] if include_log else ['-1', '--format=short']
if identifier is not None:
if revision:
raise ValueError('Cannot define both revision and identifier')
if hash:
raise ValueError('Cannot define both hash and identifier')
if tag:
raise ValueError('Cannot define both tag and identifier')
parsed_branch_point, identifier, parsed_branch = Commit._parse_identifier(
identifier, do_assert=True)
if parsed_branch:
if branch and branch != parsed_branch:
raise ValueError(
"Caller passed both 'branch' and 'identifier', but specified different branches ({} and {})"
.format(
branch,
parsed_branch,
), )
branch = parsed_branch
baseline = branch or 'HEAD'
is_default = baseline == default_branch
if baseline == 'HEAD':
is_default = default_branch in self._branches_for(baseline)
if is_default and parsed_branch_point:
raise self.Exception(
'Cannot provide a branch point for a commit on the default branch'
)
base_count = self._commit_count(
baseline if is_default else '{}..{}'.
format(default_branch, baseline))
if identifier > base_count:
raise self.Exception(
'Identifier {} cannot be found on the specified branch in the current checkout'
.format(identifier))
log = run(
[
self.executable(), 'log', '{}~{}'.format(
branch or 'HEAD', base_count - identifier)
] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
)
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for 'i{}@{}'".
format(identifier, branch or 'HEAD'))
# Negative identifiers are actually commits on the default branch, we will need to re-compute the identifier
if identifier < 0 and is_default:
raise self.Exception(
'Illegal negative identifier on the default branch')
if identifier < 0:
identifier = None
elif branch or tag:
if hash:
raise ValueError('Cannot define both tag/branch and hash')
if branch and tag:
raise ValueError('Cannot define both tag and branch')
log = run([self.executable(), 'log', branch or tag] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8')
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for '{}'".format(
branch or tag))
else:
hash = Commit._parse_hash(hash, do_assert=True)
log = run([self.executable(), 'log', hash or 'HEAD'] + log_format,
cwd=self.root_path,
capture_output=True,
encoding='utf-8')
if log.returncode:
raise self.Exception(
"Failed to retrieve commit information for '{}'".format(
hash or 'HEAD'))
match = self.GIT_COMMIT.match(log.stdout.splitlines()[0])
if not match:
raise self.Exception('Invalid commit hash in git log')
hash = match.group('hash')
branch = self.prioritize_branches(self._branches_for(hash))
if not identifier:
identifier = self._commit_count(
hash if branch ==
default_branch else '{}..{}'.format(default_branch, hash))
branch_point = None if branch == default_branch else self._commit_count(
hash) - identifier
if branch_point and parsed_branch_point and branch_point != parsed_branch_point:
raise ValueError(
"Provided 'branch_point' does not match branch point of specified branch"
)
match = self.GIT_SVN_REVISION.search(log.stdout)
revision = int(match.group('revision')) if match else None
commit_time = run(
[self.executable(), 'show', '-s', '--format=%ct', hash],
cwd=self.root_path,
capture_output=True,
encoding='utf-8',
)
if commit_time.returncode:
raise self.Exception(
'Failed to retrieve commit time for {}'.format(hash))
return Commit(
hash=hash,
revision=revision,
identifier=identifier,
branch_point=branch_point,
branch=branch,
timestamp=int(commit_time.stdout.lstrip()),
author=Contributor.from_scm_log(log.stdout.splitlines()[1],
self.contributors),
message='\n'.join(line[4:] for line in log.stdout.splitlines()[4:])
if include_log else None,
)
def test_invalid_log(self):
with self.assertRaises(ValueError):
Contributor.from_scm_log('Jonathan Bedard <*****@*****.**>')
def test_svn_patch_by_log(self):
contributor = Contributor.from_scm_log(
'Patch by Jonathan Bedard <*****@*****.**> on 2020-09-10')
self.assertEqual(contributor.name, 'Jonathan Bedard')
self.assertEqual(contributor.emails, ['*****@*****.**'])
def test_git_log(self):
contributor = Contributor.from_scm_log(
'Author: Jonathan Bedard <*****@*****.**>')
self.assertEqual(contributor.name, 'Jonathan Bedard')
self.assertEqual(contributor.emails, ['*****@*****.**'])
def test_git_svn_no_author(self):
contributor = Contributor.from_scm_log(
'Author: (no author) <(no author)@268f45cc-cd09-0410-ab3c-d52691b4dbfc>'
)
self.assertIsNone(contributor)