def get_changed_files(self):
"""Return the list of added or modified files on a commit"""
if self.changed_files is None:
output = decode_str(
check_output([
git_exe_path,
'diff-tree',
'-r',
'--root', # Get the initial commit as additions
'--no-commit-id', # We already know the commit id.
'--break-rewrites', # Get rewrites as additions
'--no-renames', # Get renames as additions
'--diff-filter=AM', # Only additions and modifications
self.commit_id,
]))
changed_files = []
for line in output.splitlines():
line_split = line.split(None, 5)
assert len(line_split) == 6
assert line_split[0].startswith(':')
file_mode = line_split[1]
# sc add object_id
object_id = line_split[3]
file_path = line_split[5]
changed_files.append(
CommittedFile(file_path, self, file_mode, object_id))
self.changed_files = changed_files
return self.changed_files
def get_shebang(self):
"""Get the shebang from the file content"""
if not self.regular():
return None
content = self.get_content()
if not content.startswith(b'#!'):
return None
content = content[len(b'#!'):].strip()
return decode_str(content.split(None, 1)[0])
def get_shebang_exe(self):
"""Get the executable from the shebang"""
shebang = self.get_shebang()
if not shebang:
return None
if shebang == '/usr/bin/env':
rest = self.get_content().splitlines()[0][len(b'#!/usr/bin/env'):]
rest_split = rest.split(None, 1)
if rest_split:
return decode_str(rest_split[0])
return shebang.rsplit('/', 1)[-1]
def get_symlink_target(self):
"""Get the symlink target as same kind of instance
We just return None, if the target has no chance to be on
the repository."""
content = self.get_content()
if isabs(content):
return None
path = normpath(joinpath(self.path, '..', decode_str(content)))
if path.startswith('..'):
return None
return type(self)(path, self.commit)
def get_new_commit_list(self, branch_name):
"""Get the list of parent new commits in order"""
output = decode_str(
check_output([
git_exe_path,
'rev-list',
self.commit_id,
'--not',
'--all',
'--reverse',
]))
commit_list = CommitList([], branch_name)
for commit_id in output.splitlines():
commit = Commit(commit_id, commit_list)
commit_list.append(commit)
return commit_list
def _fetch_content(self):
content = check_output(
[git_exe_path, 'cat-file', '-p', self.commit_id])
self._parents = []
self._message_lines = []
# The commit message starts after the empty line. We iterate until
# we find one, and then consume the rest as the message.
lines = iter(content.splitlines())
for line in lines:
if not line:
break
if line.startswith(b'parent '):
self._parents.append(Commit(line[len(b'parent '):].rstrip()))
elif line.startswith(b'author '):
self._author = Contributor.parse(line[len(b'author '):])
elif line.startswith(b'committer '):
self._committer = Contributor.parse(line[len(b'committer '):])
for line in lines:
self._message_lines.append(decode_str(line))
self.content_fetched = True
def get_binary_files(self):
"""Return the binary files on a commit"""
if self.binary_files is None:
output = decode_str(
check_output([
git_exe_path,
'log',
'--pretty=format:%H -M100%', # pretty format
'--numstat', # number state of the file
'--no-commit-id', # We already know the commit id.
'--break-rewrites', # Get rewrites as additions
'--no-renames', # Get renames as additions
'--diff-filter=AM', # Only additions and modifications
"{}^!".format(self.commit_id),
]))
binary_files = []
for line in output.splitlines():
line_split = line.split('\t')
if len(line_split) == 3:
if "-" == line_split[0] and "-" == line_split[1]:
binary_files.append(line_split[2])
self.binary_files = binary_files
return self.binary_files
def parse(cls, line):
"""Parse the contribution line as bytes"""
name, line = line.split(b' <', 1)
email, line = line.split(b'> ', 1)
timestamp, line = line.split(b' ', 1)
return cls(decode_str(name), decode_str(email), int(timestamp))