def test_parse_hash(self):
self.assertEqual(
'1a2e41e3f7cdf51b1e1d02880cfb65eab9327ef2',
Commit._parse_hash('1a2e41e3f7cdf51b1e1d02880cfb65eab9327ef2'))
self.assertEqual(
'c3bd784f8b88bd03f64467ddd3304ed8be28acbe',
Commit._parse_hash('C3BD784F8B88BD03F64467DDD3304ED8BE28ACBE'))
self.assertEqual(None, Commit._parse_hash('invalid hash'))
self.assertEqual(
None,
Commit._parse_hash('c3bd784f8b88bd03f64467ddd3304ed8be28acbe1'))
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, include_identifier=True):
if revision:
raise self.Exception('Cannot map revisions to commits on GitHub')
# Determine the commit data 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
branch = branch or self.default_branch
is_default = branch == self.default_branch
if is_default and parsed_branch_point:
raise self.Exception('Cannot provide a branch point for a commit on the default branch')
if is_default:
base_count, base_ref = self._count_for_ref(ref=self.default_branch)
else:
_, base_ref = self._count_for_ref(ref=branch)
base_count = self._difference(self.default_branch, base_ref)
if identifier > base_count:
raise self.Exception('Identifier {} cannot be found on {}'.format(identifier, branch))
# 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')
commit_data = self.request('commits/{}~{}'.format(base_ref, base_count - identifier))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}@{}'".format(identifier, branch or 'HEAD'))
# If an identifier is negative, unset it so we re-compute before constructing the commit.
if identifier <= 0:
identifier = None
# Determine the commit data 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')
commit_data = self.request('commits/{}'.format(branch or tag))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(branch or tag))
# Determine the commit data for a given hash
else:
hash = Commit._parse_hash(hash, do_assert=True)
commit_data = self.request('commits/{}'.format(hash or self.default_branch))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(hash or 'HEAD'))
# A commit is often on multiple branches, the canonical branch is the one with the highest priority
branches = self._branches_for(commit_data['sha'])
if branches:
branch = self.prioritize_branches(branches)
else:
# A commit not on any branches cannot have an identifier
identifier = None
branch = None
# Define identifiers on default branch
branch_point = None
if include_identifier and branch and branch == self.default_branch:
if not identifier:
result = self._count_for_ref(ref=commit_data['sha'])
if not result:
raise Exception('{} {}'.format(result, commit_data['sha']))
identifier, _ = result
# Define identifiers on branches diverged from the default branch
elif include_identifier and branch:
if not identifier:
identifier = self._difference(self.default_branch, commit_data['sha'])
branch_point = self._count_for_ref(ref=commit_data['sha'])[0] - identifier
# Check the commit log for a git-svn revision
matches = self.GIT_SVN_REVISION.findall(commit_data['commit']['message'])
revision = int(matches[-1].split('@')[0]) if matches else None
email_match = self.EMAIL_RE.match(commit_data['commit']['author']['email'])
timestamp = int(calendar.timegm(datetime.strptime(
commit_data['commit']['committer']['date'], '%Y-%m-%dT%H:%M:%SZ',
).timetuple()))
# 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
lhash = commit_data['sha']
while lhash:
response = self.request('commits', paginate=False, params=dict(sha=lhash, per_page=20))
if len(response) <= 1:
break
for c in response:
if lhash == c['sha']:
continue
parent_timestamp = int(calendar.timegm(datetime.strptime(
c['commit']['committer']['date'], '%Y-%m-%dT%H:%M:%SZ',
).timetuple()))
if parent_timestamp != timestamp:
lhash = None
break
lhash = c['sha']
order += 1
return Commit(
repository_id=self.id,
hash=commit_data['sha'],
revision=revision,
branch_point=branch_point,
identifier=identifier if include_identifier else None,
branch=branch,
timestamp=timestamp,
order=order,
author=self.contributors.create(
commit_data['commit']['author']['name'],
email_match.group('email') if email_match else None,
), message=commit_data['commit']['message'] if include_log else None,
)
def commit(self, hash=None, revision=None, identifier=None, branch=None, tag=None, include_log=True, include_identifier=True):
if revision:
raise self.Exception('Cannot map revisions to commits on BitBucket')
# Determine the commit data 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
branch = branch or self.default_branch
is_default = branch == self.default_branch
if is_default and parsed_branch_point:
raise self.Exception('Cannot provide a branch point for a commit on the default branch')
commit_data = self.request('commits/{}'.format(branch), params=dict(limit=1))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(branch))
base_ref = commit_data['id']
if is_default:
base_count = self._distance(base_ref)
else:
base_count = self._distance(base_ref, magnitude=256, condition=lambda val: self.default_branch not in val)
if identifier > base_count:
raise self.Exception('Identifier {} cannot be found on {}'.format(identifier, branch))
# 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')
commit_data = self.request('commits/{}~{}'.format(base_ref, base_count - identifier))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}@{}'".format(identifier, branch or 'HEAD'))
# If an identifier is negative, unset it so we re-compute before constructing the commit.
if identifier <= 0:
identifier = None
# Determine the commit data 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')
commit_data = self.request('commits/{}'.format(branch or tag))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(branch or tag))
# Determine the commit data for a given hash
else:
hash = Commit._parse_hash(hash, do_assert=True)
commit_data = self.request('commits/{}'.format(hash or self.default_branch))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(hash or 'HEAD'))
# A commit is often on multiple branches, the canonical branch is the one with the highest priority
branches = self._branches_for(commit_data['id'])
if branches:
branch = self.prioritize_branches(branches)
else:
# A commit not on any branches cannot have an identifier
identifier = None
branch = None
# Define identifiers on default branch
branch_point = None
if include_identifier and branch and branch == self.default_branch:
if not identifier:
identifier = self._distance(commit_data['id'])
# Define identifiers on branches diverged from the default branch
elif include_identifier and branch:
if not identifier:
identifier = self._distance(commit_data['id'], magnitude=256, condition=lambda val: self.default_branch not in val)
branch_point = self._distance(commit_data['id']) - identifier
# Check the commit log for a git-svn revision
matches = self.GIT_SVN_REVISION.findall(commit_data['message'])
revision = int(matches[-1].split('@')[0]) if matches else None
# 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.
timestamp = int(commit_data['committerTimestamp'] / 1000)
order = 0
while not identifier or order + 1 < identifier + (branch_point or 0):
response = self.request('commits/{}'.format('{}~{}'.format(commit_data['id'], order + 1)))
if not response:
break
parent_timestamp = int(response['committerTimestamp'] / 1000)
if parent_timestamp != timestamp:
break
order += 1
return Commit(
repository_id=self.id,
hash=commit_data['id'],
revision=revision,
branch_point=branch_point,
identifier=identifier if include_identifier else None,
branch=branch,
timestamp=timestamp,
order=order,
author=self.contributors.create(
commit_data.get('committer', {}).get('displayName', None),
commit_data.get('committer', {}).get('emailAddress', None),
), message=commit_data['message'] if include_log else None,
)
def commit(self, hash=None, revision=None, identifier=None, branch=None, tag=None, include_log=True, include_identifier=True):
if revision:
raise self.Exception('Cannot map revisions to commits on GitHub')
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
branch = branch or self.default_branch
is_default = branch == self.default_branch
if is_default and parsed_branch_point:
raise self.Exception('Cannot provide a branch point for a commit on the default branch')
if is_default:
base_count, base_ref = self._count_for_ref(ref=self.default_branch)
else:
_, base_ref = self._count_for_ref(ref=branch)
base_count = self._difference(self.default_branch, base_ref)
if identifier > base_count:
raise self.Exception('Identifier {} cannot be found on {}'.format(identifier, branch))
# 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')
commit_data = self.request('commits/{}~{}'.format(base_ref, base_count - identifier))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}@{}'".format(identifier, branch or 'HEAD'))
# If an identifier is negative, unset it so we re-compute before constructing the commit.
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')
commit_data = self.request('commits/{}'.format(branch or tag))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(branch or tag))
else:
hash = Commit._parse_hash(hash, do_assert=True)
commit_data = self.request('commits/{}'.format(hash or self.default_branch))
if not commit_data:
raise self.Exception("Failed to retrieve commit information for '{}'".format(hash or 'HEAD'))
branches = self._branches_for(commit_data['sha'])
if branches:
branch = self.prioritize_branches(branches)
else:
# A commit not on any branches cannot have an identifier
identifier = None
branch = None
branch_point = None
if include_identifier and branch and branch == self.default_branch:
if not identifier:
result = self._count_for_ref(ref=commit_data['sha'])
if not result:
raise Exception('{} {}'.format(result, commit_data['sha']))
identifier, _ = result
elif include_identifier and branch:
if not identifier:
identifier = self._difference(self.default_branch, commit_data['sha'])
branch_point = self._count_for_ref(ref=commit_data['sha'])[0] - identifier
matches = self.GIT_SVN_REVISION.findall(commit_data['commit']['message'])
revision = int(matches[-1].split('@')[0]) if matches else None
email_match = self.EMAIL_RE.match(commit_data['commit']['author']['email'])
timestamp = int(calendar.timegm(datetime.strptime(
commit_data['commit']['committer']['date'], '%Y-%m-%dT%H:%M:%SZ',
).timetuple()))
order = 0
while not identifier or order + 1 < identifier + (branch_point or 0):
response = self.request('commits/{}'.format('{}~{}'.format(commit_data['sha'], order + 1)))
if not response:
break
parent_timestamp = int(calendar.timegm(datetime.strptime(
response['commit']['committer']['date'], '%Y-%m-%dT%H:%M:%SZ',
).timetuple()))
if parent_timestamp != timestamp:
break
order += 1
return Commit(
repository_id=self.id,
hash=commit_data['sha'],
revision=revision,
branch_point=branch_point,
identifier=identifier if include_identifier else None,
branch=branch,
timestamp=timestamp,
order=order,
author=self.contributors.create(
commit_data['commit']['author']['name'],
email_match.group('email') if email_match else None,
), message=commit_data['commit']['message'] 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,
)