def _parse_line(self, line):
if not line:
if self.commit is not None and self.state == SVNParser.COMMIT \
or self.state == SVNParser.FILES:
self.state = SVNParser.MESSAGE
elif self.state == SVNParser.MESSAGE:
self.__append_message_line()
return
# Message
if self.state == SVNParser.MESSAGE and self.msg_lines > 0:
self.__append_message_line(line)
return
# Invalid commit. Some svn repos like asterisk have commits like this:
# r176840 | (no author) | (no date) | 1 line
# without any canged path, so I think we can just ignore them
if self.patterns['invalid'].match(line):
printdbg("SVN Parser: skipping invalid commit: %s", (line, ))
self.state = SVNParser.COMMIT
self.commit = None
return
# Separator
if self.patterns['separator'].match(line):
if self.commit is None or self.state == SVNParser.COMMIT:
return
elif self.state == SVNParser.MESSAGE \
or self.state == SVNParser.FILES:
# We can go directly from FILES to COMMIT
# when there is an empty log message
if self.msg_lines > 0:
printout("Warning (%d): parsing svn log, missing " + \
"lines in commit message!", (self.n_line,))
self.__convert_commit_actions(self.commit)
self.handler.commit(self.commit)
self.state = SVNParser.COMMIT
self.commit = None
self.msg_lines = 0
else:
printout("Warning (%d): parsing svn log, unexpected separator",
(self.n_line, ))
return
# Commit
match = self.patterns['commit'].match(line)
if match and self.state == SVNParser.COMMIT:
commit = Commit()
commit.revision = match.group(1)
commit.committer = Person()
commit.committer.name = match.group(2)
commit.date = datetime.datetime(int(match.group(3)),
int(match.group(4)),
int(match.group(5)),
int(match.group(6)),
int(match.group(7)),
int(match.group(8)))
self.msg_lines = int(match.group(10))
self.commit = commit
self.handler.committer(commit.committer)
return
elif match and self.state == SVNParser.MESSAGE:
# It seems a piece of a log message has been copied as
# part of the commit message
self.commit.message += line + '\n'
return
elif match and self.state != SVNParser.COMMIT:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
# Files
if self.state == SVNParser.COMMIT:
if self.patterns['paths'].match(line):
self.state = SVNParser.FILES
else:
printout("Warning(%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
# File moved/copied/replaced
match = self.patterns['file-moved'].match(line)
if match:
if self.state != SVNParser.FILES:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
action = Action()
action.type = match.group(1)
action.f1 = match.group(2)
action.f2 = match.group(3)
action.rev = match.group(4)
action.branch_f1 = self.__guess_branch_from_path(action.f1)
action.branch_f2 = self.__guess_branch_from_path(action.f2)
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# File
match = self.patterns['file'].match(line)
if match:
if self.state != SVNParser.FILES:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
path = match.group(2)
if path != '/':
# path == '/' is probably a properties change in /
# not interesting for us, ignoring
action = Action()
action.type = match.group(1)
action.f1 = path
action.branch_f1 = self.__guess_branch_from_path(path)
self.commit.actions.append(action)
self.handler.file(path)
return
def _parse_line(self, line):
if not line:
if self.commit is not None and self.state == SVNParser.COMMIT \
or self.state == SVNParser.FILES:
self.state = SVNParser.MESSAGE
elif self.state == SVNParser.MESSAGE:
self.__append_message_line()
return
# Message
if self.state == SVNParser.MESSAGE and self.msg_lines > 0:
self.__append_message_line(line)
return
# Invalid commit. Some svn repos like asterisk have commits like this:
# r176840 | (no author) | (no date) | 1 line
# without any canged path, so I think we can just ignore them
if self.patterns['invalid'].match(line):
printdbg("SVN Parser: skipping invalid commit: %s", (line,))
self.state = SVNParser.COMMIT
self.commit = None
return
# Separator
if self.patterns['separator'].match(line):
if self.commit is None or self.state == SVNParser.COMMIT:
return
elif self.state == SVNParser.MESSAGE \
or self.state == SVNParser.FILES:
# We can go directly from FILES to COMMIT
# when there is an empty log message
if self.msg_lines > 0:
printout("Warning (%d): parsing svn log, missing " + \
"lines in commit message!", (self.n_line,))
self.__convert_commit_actions(self.commit)
self.handler.commit(self.commit)
self.state = SVNParser.COMMIT
self.commit = None
self.msg_lines = 0
else:
printout("Warning (%d): parsing svn log, unexpected separator",
(self.n_line,))
return
# Commit
match = self.patterns['commit'].match(line)
if match and self.state == SVNParser.COMMIT:
commit = Commit()
commit.revision = match.group(1)
commit.committer = Person()
commit.committer.name = match.group(2)
commit.commit_date = datetime.datetime(int(match.group(3)),
int(match.group(4)),
int(match.group(5)),
int(match.group(6)),
int(match.group(7)),
int(match.group(8)))
self.msg_lines = int(match.group(10))
self.commit = commit
self.handler.committer(commit.committer)
return
elif match and self.state == SVNParser.MESSAGE:
# It seems a piece of a log message has been copied as
# part of the commit message
self.commit.message += line + '\n'
return
elif match and self.state != SVNParser.COMMIT:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
# Files
if self.state == SVNParser.COMMIT:
if self.patterns['paths'].match(line):
self.state = SVNParser.FILES
else:
printout("Warning(%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
# File moved/copied/replaced
match = self.patterns['file-moved'].match(line)
if match:
if self.state != SVNParser.FILES:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
action = Action()
action.type = match.group(1)
action.f1 = match.group(2)
action.f2 = match.group(3)
action.rev = match.group(4)
action.branch_f1 = self.__guess_branch_from_path(action.f1)
action.branch_f2 = self.__guess_branch_from_path(action.f2)
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# File
match = self.patterns['file'].match(line)
if match:
if self.state != SVNParser.FILES:
printout("Warning (%d): parsing svn log, unexpected line %s",
(self.n_line, line))
return
path = match.group(2)
if path != '/':
# path == '/' is probably a properties change in /
# not interesting for us, ignoring
action = Action()
action.type = match.group(1)
action.f1 = path
action.branch_f1 = self.__guess_branch_from_path(path)
self.commit.actions.append(action)
self.handler.file(path)
return
def _parse_line(self, line):
if line is None or line == '':
return
# Ignore
for patt in self.patterns['ignore']:
if patt.match(line):
return
# Commit
match = self.patterns['commit'].match(line)
if match:
if self.commit is not None:
# Skip commits on svn tags
if self.branch.tail.svn_tag is None:
self.handler.commit(self.branch.tail.commit)
self.commit = Commit()
self.commit.revision = match.group(1)
parents = match.group(3)
if parents:
parents = parents.split()
git_commit = self.GitCommit(self.commit, parents)
# If a specific branch has been configured, there
# won't be any decoration, so a branch needs to be
# created
if Config().branch is not None:
self.branch = self.GitBranch(self.GitBranch.LOCAL,
Config().branch,
git_commit)
decorate = match.group(5)
branch = None
if decorate:
# Remote branch
m = re.search(self.patterns['branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.REMOTE, m.group(1),
git_commit)
printdbg("Branch '%s' head at acommit %s",
(branch.name, self.commit.revision))
else:
# Local Branch
m = re.search(self.patterns['local-branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.LOCAL,
m.group(1), git_commit)
printdbg("Commit %s on local branch '%s'",
(self.commit.revision, branch.name))
# If local branch was merged we just ignore this
# decoration
if self.branch and \
self.branch.is_my_parent(git_commit):
printdbg("Local branch '%s' was merged",
(branch.name,))
branch = None
else:
# Stash
m = re.search(self.patterns['stash'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.STASH,
"stash", git_commit)
printdbg("Commit %s on stash",
(self.commit.revision,))
# Tag
m = re.search(self.patterns['tag'], decorate)
if m:
self.commit.tags = [m.group(1)]
printdbg("Commit %s tagged as '%s'",
(self.commit.revision, self.commit.tags[0]))
if branch is not None and self.branch is not None:
# Detect empty branches. Ideally, the head of a branch
# can't have children. When this happens is because the
# branch is empty, so we just ignore such branch
if self.branch.is_my_parent(git_commit):
printout("Warning: Detected empty branch '%s', " + \
"it'll be ignored", (branch.name,))
branch = None
if len(self.branches) >= 2:
# If current commit is the start point of a new branch
# we have to look at all the current branches since
# we haven't inserted the new branch yet.
# If not, look at all other branches excluding the current one
for i, b in enumerate(self.branches):
if i == 0 and branch is None:
continue
if b.is_my_parent(git_commit):
# We assume current branch is always the last one
# AFAIK there's no way to make sure this is right
printdbg("Start point of branch '%s' at commit %s",
(self.branches[0].name, self.commit.revision))
self.branches.pop(0)
self.branch = b
if self.branch and self.branch.tail.svn_tag is not None and \
self.branch.is_my_parent(git_commit):
# There's a pending tag in previous commit
pending_tag = self.branch.tail.svn_tag
printdbg("Move pending tag '%s' from previous commit %s " + \
"to current %s", (pending_tag,
self.branch.tail.commit.revision,
self.commit.revision))
if self.commit.tags and pending_tag not in self.commit.tags:
self.commit.tags.append(pending_tag)
else:
self.commit.tags = [pending_tag]
self.branch.tail.svn_tag = None
if branch is not None:
self.branch = branch
# Insert master always at the end
if branch.is_remote() and branch.name == 'master':
self.branches.append(self.branch)
else:
self.branches.insert(0, self.branch)
else:
self.branch.set_tail(git_commit)
if parents and len(parents) > 1 and not Config().analyze_merges:
#Skip merge commits
self.commit = None
return
elif self.commit is None:
return
# Committer
match = self.patterns['committer'].match(line)
if match:
self.commit.committer = Person()
self.commit.committer.name = match.group(1)
self.commit.committer.email = match.group(2)
self.handler.committer(self.commit.committer)
return
# Author
match = self.patterns['author'].match(line)
if match:
self.commit.author = Person()
self.commit.author.name = match.group(1)
self.commit.author.email = match.group(2)
self.handler.author(self.commit.author)
return
# Commit Date
match = self.patterns['commit-date'].match(line)
if match:
self.commit.commit_date = datetime.datetime(*(time.strptime(\
match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
return
# Author Date
match = self.patterns['author-date'].match(line)
if match:
self.commit.author_date = datetime.datetime(*(time.strptime(\
match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
return
# File
match = self.patterns['file'].match(line)
if match:
action = Action()
action.type = match.group(1)
action.f1 = match.group(2)
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# File moved/copied
match = self.patterns['file-moved'].match(line)
if match:
action = Action()
type = match.group(1)
if type == 'R':
action.type = 'V'
else:
action.type = type
action.f1 = match.group(3)
action.f2 = match.group(2)
action.rev = self.commit.revision
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# Message
self.commit.message += line + '\n'
assert True, "Not match for line %s" % (line)
def _parse_line(self, line):
if line is None or line == '':
return
# Separator
match = self.patterns['separator'].match(line)
if match:
self.flush()
return
# Ignore details about merges
match = self.patterns['ignore'].match(line)
if match:
self.state = BzrParser.UNKNOWN
return
# Commit
match = self.patterns['commit'].match(line)
if match:
self.flush()
self.commit = Commit()
self.commit.revision = match.group(1)
self.state = BzrParser.COMMIT
return
# Committer
match = self.patterns['committer'].match(line)
if match:
self.commit.committer = Person()
self.commit.committer.name = match.group(1)
self.commit.committer.email = match.group(2)
self.handler.committer(self.commit.committer)
return
# Author
match = self.patterns['author'].match(line)
if match:
self.commit.author = Person()
self.commit.author.name = match.group(1)
self.commit.author.email = match.group(2)
self.handler.author(self.commit.author)
return
# Date
match = self.patterns['commit-date'].match(line)
if match:
self.commit.commit_date = datetime.datetime(*(time.strptime
(match.group(1).strip(" "),
"%Y-%m-%d %H:%M:%S")[0:6]))
# datetime.datetime.strptime not supported by Python2.4
#self.commit.commit_date = datetime.datetime.strptime(\
# match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")
return
# Message
match = self.patterns['message'].match(line)
if match:
self.state = BzrParser.MESSAGE
return
# Added files
match = self.patterns['added'].match(line)
if match:
self.state = BzrParser.ADDED
return
# Modified files
match = self.patterns['modified'].match(line)
if match:
self.state = BzrParser.MODIFIED
return
# Removed files
match = self.patterns['removed'].match(line)
if match:
self.state = BzrParser.REMOVED
return
# Renamed files
match = self.patterns['renamed'].match(line)
if match:
self.state = BzrParser.RENAMED
return
if self.state == BzrParser.MESSAGE:
self.commit.message += line.lstrip() + '\n'
elif self.state == BzrParser.ADDED or \
self.state == BzrParser.MODIFIED or \
self.state == BzrParser.REMOVED:
action = Action()
if self.state == BzrParser.ADDED:
action.type = 'A'
elif self.state == BzrParser.MODIFIED:
action.type = 'M'
elif self.state == BzrParser.REMOVED:
action.type = 'D'
action.f1 = line.strip()
self.commit.actions.append(action)
self.handler.file(action.f1)
elif self.state == BzrParser.RENAMED:
m = re.compile("^[ \t]+(.*) => (.*)$").match(line)
if not m:
return
action = Action()
action.type = 'V'
action.f1 = m.group(2)
action.f2 = m.group(1)
self.commit.actions.append(action)
self.handler.file(action.f1)
else:
self.state = BzrParser.UNKNOWN
def _parse_line (self, line):
if line is None or line == '':
return
# Ignore
for patt in self.patterns['ignore']:
if patt.match (line):
return
# Commit
match = self.patterns['commit'].match (line)
if match:
if self.commit is not None and self.branch.is_remote ():
if self.branch.tail.svn_tag is None: # Skip commits on svn tags
self.handler.commit (self.branch.tail.commit)
self.commit = Commit ()
self.commit.revision = match.group (1)
parents = match.group (3)
if parents:
parents = parents.split ()
git_commit = self.GitCommit (self.commit, parents)
decorate = match.group (5)
branch = None
if decorate:
# Remote branch
m = re.search (self.patterns['branch'], decorate)
if m:
branch = self.GitBranch (self.GitBranch.REMOTE, m.group (1), git_commit)
printdbg ("Branch '%s' head at acommit %s", (branch.name, self.commit.revision))
else:
# Local Branch
m = re.search (self.patterns['local-branch'], decorate)
if m:
branch = self.GitBranch (self.GitBranch.LOCAL, m.group (1), git_commit)
printdbg ("Commit %s on local branch '%s'", (self.commit.revision, branch.name))
# If local branch was merged we just ignore this decoration
if self.branch and self.branch.is_my_parent (git_commit):
printdbg ("Local branch '%s' was merged", (branch.name,))
branch = None
else:
# Stash
m = re.search (self.patterns['stash'], decorate)
if m:
branch = self.GitBranch (self.GitBranch.STASH, "stash", git_commit)
printdbg ("Commit %s on stash", (self.commit.revision,))
# Tag
m = re.search (self.patterns['tag'], decorate)
if m:
self.commit.tags = [m.group (1)]
printdbg ("Commit %s tagged as '%s'", (self.commit.revision, self.commit.tags[0]))
if branch is not None and self.branch is not None:
# Detect empty branches. Ideally, the head of a branch
# can't have children. When this happens is because the
# branch is empty, so we just ignore such branch
if self.branch.is_my_parent (git_commit):
printout ("Warning: Detected empty branch '%s', it'll be ignored", (branch.name,))
branch = None
if len (self.branches) >= 2:
# If current commit is the start point of a new branch
# we have to look at all the current branches since
# we haven't inserted the new branch yet.
# If not, look at all other branches excluding the current one
for i, b in enumerate (self.branches):
if i == 0 and branch is None:
continue
if b.is_my_parent (git_commit):
# We assume current branch is always the last one
# AFAIK there's no way to make sure this is right
printdbg ("Start point of branch '%s' at commit %s", (self.branches[0].name, self.commit.revision))
self.branches.pop (0)
self.branch = b
if self.branch and self.branch.tail.svn_tag is not None and self.branch.is_my_parent (git_commit):
# There's a pending tag in previous commit
pending_tag = self.branch.tail.svn_tag
printdbg ("Move pending tag '%s' from previous commit %s to current %s", (pending_tag,
self.branch.tail.commit.revision,
self.commit.revision))
if self.commit.tags and pending_tag not in self.commit.tags:
self.commit.tags.append (pending_tag)
else:
self.commit.tags = [pending_tag]
self.branch.tail.svn_tag = None
if branch is not None:
self.branch = branch
# Insert master always at the end
if branch.is_remote () and branch.name == 'master':
self.branches.append (self.branch)
else:
self.branches.insert (0, self.branch)
else:
self.branch.set_tail (git_commit)
return
# Committer
match = self.patterns['committer'].match (line)
if match:
self.commit.committer = Person ()
self.commit.committer.name = match.group (1)
self.commit.committer.email = match.group (2)
self.handler.committer (self.commit.committer)
return
# Author
match = self.patterns['author'].match (line)
if match:
self.commit.author = Person ()
self.commit.author.name = match.group (1)
self.commit.author.email = match.group (2)
self.handler.author (self.commit.author)
return
# Date
match = self.patterns['date'].match (line)
if match:
self.commit.date = datetime.datetime (* (time.strptime (match.group (1).strip (" "), "%a %b %d %H:%M:%S %Y")[0:6]))
# datetime.datetime.strptime not supported by Python2.4
#self.commit.date = datetime.datetime.strptime (match.group (1).strip (" "), "%a %b %d %H:%M:%S %Y")
return
# File
match = self.patterns['file'].match (line)
if match:
action = Action ()
action.type = match.group (1)
action.f1 = match.group (2)
self.commit.actions.append (action)
self.handler.file (action.f1)
return
# File moved/copied
match = self.patterns['file-moved'].match (line)
if match:
action = Action ()
type = match.group (1)
if type == 'R':
action.type = 'V'
else:
action.type = type
action.f1 = match.group (3)
action.f2 = match.group (2)
action.rev = self.commit.revision
self.commit.actions.append (action)
self.handler.file (action.f1)
return
# This is a workaround for a bug in the GNOME Git migration
# There are commits on tags not correctly detected like this one:
# http://git.gnome.org/cgit/evolution/commit/?id=b8e52acac2b9fc5414a7795a73c74f7ee4eeb71f
# We want to ignore commits on tags since it doesn't make any sense in Git
if self.is_gnome:
match = self.patterns['svn-tag'].match (line.strip ())
if match:
printout ("Warning: detected a commit on a svn tag: %s", (match.group (0),))
tag = match.group (1)
if self.commit.tags and tag in self.commit.tags:
# The commit will be ignored, so move the tag
# to the next (previous in history) commit
self.branch.tail.svn_tag = tag
# Message
self.commit.message += line + '\n'
assert True, "Not match for line %s" % (line)
def _parse_line(self, line):
if line is None or line == '':
return
# Separator
match = self.patterns['separator'].match(line)
if match:
self.flush()
return
# Ignore details about merges
match = self.patterns['ignore'].match(line)
if match:
self.state = BzrParser.UNKNOWN
return
# Commit
match = self.patterns['commit'].match(line)
if match:
self.flush()
self.commit = Commit()
self.commit.revision = match.group(1)
self.state = BzrParser.COMMIT
return
# Committer
match = self.patterns['committer'].match(line)
if match:
self.commit.committer = Person()
self.commit.committer.name = match.group(1)
self.commit.committer.email = match.group(2)
self.handler.committer(self.commit.committer)
return
# Author
match = self.patterns['author'].match(line)
if match:
self.commit.author = Person()
self.commit.author.name = match.group(1)
self.commit.author.email = match.group(2)
self.handler.author(self.commit.author)
return
# Date
match = self.patterns['date'].match(line)
if match:
self.commit.date = datetime.datetime(*(time.strptime
(match.group(1).strip(" "),
"%Y-%m-%d %H:%M:%S")[0:6]))
# datetime.datetime.strptime not supported by Python2.4
#self.commit.date = datetime.datetime.strptime(\
# match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")
return
# Message
match = self.patterns['message'].match(line)
if match:
self.state = BzrParser.MESSAGE
return
# Added files
match = self.patterns['added'].match(line)
if match:
self.state = BzrParser.ADDED
return
# Modified files
match = self.patterns['modified'].match(line)
if match:
self.state = BzrParser.MODIFIED
return
# Removed files
match = self.patterns['removed'].match(line)
if match:
self.state = BzrParser.REMOVED
return
# Renamed files
match = self.patterns['renamed'].match(line)
if match:
self.state = BzrParser.RENAMED
return
if self.state == BzrParser.MESSAGE:
self.commit.message += line.lstrip() + '\n'
elif self.state == BzrParser.ADDED or \
self.state == BzrParser.MODIFIED or \
self.state == BzrParser.REMOVED:
action = Action()
if self.state == BzrParser.ADDED:
action.type = 'A'
elif self.state == BzrParser.MODIFIED:
action.type = 'M'
elif self.state == BzrParser.REMOVED:
action.type = 'D'
action.f1 = line.strip()
self.commit.actions.append(action)
self.handler.file(action.f1)
elif self.state == BzrParser.RENAMED:
m = re.compile("^[ \t]+(.*) => (.*)$").match(line)
if not m:
return
action = Action()
action.type = 'V'
action.f1 = m.group(2)
action.f2 = m.group(1)
self.commit.actions.append(action)
self.handler.file(action.f1)
else:
self.state = BzrParser.UNKNOWN
def _parse_line(self, line):
if line is None or line == '':
return
# Ignore
for patt in self.patterns['ignore']:
if patt.match(line):
return
# Commit
match = self.patterns['commit'].match(line)
if match:
if self.commit is not None:
# Skip commits on svn tags
if self.branch.tail.svn_tag is None:
self.handler.commit(self.branch.tail.commit)
if self.patterns['replace-commit'].search(line):
printdbg("Skipping commit, because it's a replacement")
self.commit = None
return
self.commit = Commit()
self.commit.revision = match.group(1)
parents = match.group(3)
if parents:
parents = parents.split()
git_commit = self.GitCommit(self.commit, parents)
# If a specific branch has been configured, there
# won't be any decoration, so a branch needs to be
# created
if Config().branch is not None:
self.branch = self.GitBranch(self.GitBranch.LOCAL,
Config().branch, git_commit)
decorate = match.group(5)
branch = None
if decorate:
# Remote branch
m = re.search(self.patterns['branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.REMOTE, m.group(2),
git_commit)
printdbg("Branch '%s' head at acommit %s",
(branch.name, self.commit.revision))
else:
# Local Branch
m = re.search(self.patterns['local-branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.LOCAL,
m.group(1), git_commit)
printdbg("Commit %s on local branch '%s'",
(self.commit.revision, branch.name))
# If local branch was merged we just ignore this
# decoration
if self.branch and \
self.branch.is_my_parent(git_commit):
printdbg("Local branch '%s' was merged",
(branch.name, ))
branch = None
else:
# Stash
m = re.search(self.patterns['stash'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.STASH,
"stash", git_commit)
printdbg("Commit %s on stash",
(self.commit.revision, ))
# Tag
m = re.search(self.patterns['tag'], decorate)
if m:
self.commit.tags = [m.group(1)]
printdbg("Commit %s tagged as '%s'",
(self.commit.revision, self.commit.tags[0]))
if branch is not None and self.branch is not None:
# Detect empty branches. Ideally, the head of a branch
# can't have children. When this happens is because the
# branch is empty, so we just ignore such branch
if self.branch.is_my_parent(git_commit):
printout("Warning: Detected empty branch '%s', " + \
"it'll be ignored", (branch.name,))
branch = None
if len(self.branches) >= 2:
# If current commit is the start point of a new branch
# we have to look at all the current branches since
# we haven't inserted the new branch yet.
# If not, look at all other branches excluding the current one
for i, b in enumerate(self.branches):
if i == 0 and branch is None:
continue
if b.is_my_parent(git_commit):
# We assume current branch is always the last one
# AFAIK there's no way to make sure this is right
printdbg("Start point of branch '%s' at commit %s",
(self.branches[0].name, self.commit.revision))
self.branches.pop(0)
self.branch = b
if self.branch and self.branch.tail.svn_tag is not None and \
self.branch.is_my_parent(git_commit):
# There's a pending tag in previous commit
pending_tag = self.branch.tail.svn_tag
printdbg("Move pending tag '%s' from previous commit %s " + \
"to current %s", (pending_tag,
self.branch.tail.commit.revision,
self.commit.revision))
if self.commit.tags and pending_tag not in self.commit.tags:
self.commit.tags.append(pending_tag)
else:
self.commit.tags = [pending_tag]
self.branch.tail.svn_tag = None
if branch is not None:
self.branch = branch
# Insert master always at the end
if branch.is_remote() and branch.name == 'master':
self.branches.append(self.branch)
else:
self.branches.insert(0, self.branch)
else:
self.branch.set_tail(git_commit)
if parents and len(parents) > 1 and not Config().analyze_merges:
#Skip merge commits
self.commit = None
return
elif self.commit is None:
return
# Committer
match = self.patterns['committer'].match(line)
if match:
self.commit.committer = Person()
self.commit.committer.name = match.group(1)
self.commit.committer.email = match.group(2)
self.handler.committer(self.commit.committer)
return
# Author
match = self.patterns['author'].match(line)
if match:
self.commit.author = Person()
self.commit.author.name = match.group(1)
self.commit.author.email = match.group(2)
self.handler.author(self.commit.author)
return
# Commit Date
match = self.patterns['commit-date'].match(line)
if match:
self.commit.commit_date = datetime.datetime(*(time.strptime(\
match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
return
# Author Date
match = self.patterns['author-date'].match(line)
if match:
self.commit.author_date = datetime.datetime(*(time.strptime(\
match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
return
# File
match = self.patterns['file'].match(line)
if match:
action = Action()
action.type = match.group(1)
action.f1 = match.group(2)
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# File moved/copied
match = self.patterns['file-moved'].match(line)
if match:
action = Action()
type = match.group(1)
if type == 'R':
action.type = 'V'
else:
action.type = type
action.f1 = match.group(3)
action.f2 = match.group(2)
action.rev = self.commit.revision
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# Message
self.commit.message += line + '\n'
assert True, "Not match for line %s" % (line)
def _parse_line(self, line):
if not line:
if self.commit is None:
return
if self.rev_separator is not None:
self.rev_separator += '\n'
elif self.file_separator is not None:
self.file_separator += '\n'
elif self.commit.message is not None:
self.commit.message += '\n'
return
# Revision Separator
if self.patterns['rev-separator'].match(line):
# Ignore separators so that we don't
# include it in the commit message
if self.rev_separator is None:
self.rev_separator = line
else:
self.rev_separator += line + '\n'
return
# File Separator
if self.patterns['file-separator'].match(line):
# Ignore separators so that we don't
# include it in the commit message
if self.file_separator is None:
self.file_separator = line
else:
self.file_separator += line + '\n'
return
# File
match = self.patterns['file'].match(line)
if match:
self.flush()
path = match.group(1)
path = path[len(self.root_path):]
path = path[:path.rfind(',')]
self.file = path
self.branches = {}
self.tags = {}
self.commit = None
self.file_separator = None
return
# Branch
match = self.patterns['branch'].match(line)
if match:
self.branches[match.group(2) + match.group(4)] = match.group(1)
return
# Tag (Keep this always after Branch pattern)
match = self.patterns['tag'].match(line)
if match:
revision = match.group(2)
# We are ignoring 1.1.1.1 revisions,
# so in case there's a tag pointing to that
# revision we have to redirect it to 1.1 revision
if revision == '1.1.1.1':
revision = '1.1'
self.tags.setdefault(revision, []).append(match.group(1))
return
# Revision
match = self.patterns['revision'].match(line)
if match and self.rev_separator is not None:
self._handle_commit()
revision = match.group(1)
commit = Commit()
# composed rev: revision + | + file path
# to make sure revision is unique
commit.composed_rev = True
commit.revision = "%s|%s" % (revision, self.file)
commit.tags = self.tags.get(revision, None)
self.commit = commit
self.rev_separator = None
return
# Commit info (date, author, etc.)
match = self.patterns['info'].match(line)
if match and self.commit is not None:
commit = self.commit
revision = commit.revision.split('|')[0]
if revision == '1.1.1.1':
self.commit = None
return
commit.committer = Person()
commit.committer.name = match.group(8)
self.handler.committer(commit.committer)
commit.date = datetime.datetime(int(match.group(1)), int(match.group(2)), int(match.group(3)),
int(match.group(4)), int(match.group(5)), int(match.group(6)))
if match.group(10) is not None:
self.lines[commit.revision] = (int(match.group(11)), int(match.group(12)))
else:
self.lines[commit.revision] = (0, 0)
action = Action()
act = match.group(9)
if act == 'dead':
action.type = 'D'
self.file = self.file.replace('/Attic', '')
commit.revision = commit.revision.replace('/Attic', '')
elif revision == '1.1':
action.type = 'A'
else:
action.type = 'M'
action.f1 = self.file
# Branch
try:
last_dot = revision.rfind('.')
prefix = revision[:last_dot]
branch = self.branches[prefix]
if self.file_added_on_branch and \
self.file_added_on_branch == prefix and \
revision[last_dot + 1:] == '1':
action.type = 'A'
self.file_added_on_branch = None
except KeyError:
branch = 'trunk'
commit.branch = branch
commit.actions.append(action)
return
# Branches
match = self.patterns['branches'].match(line)
if match:
if self.commit is None:
return
action = self.commit.actions[0]
revision = self.commit.revision.split('|')[0]
if action.type == 'D' and revision == '1.1':
# File added on a branch
self.file_added_on_branch = match.group(1)
# Discard this commit
self.commit = None
return
# Message.
if self.commit is not None:
if self.rev_separator is not None:
# Previous separator was probably a
# false positive
self.commit.message += self.rev_separator + '\n'
self.rev_separator = None
if self.file_separator is not None:
# Previous separator was probably a
# false positive
self.commit.message += self.file_separator + '\n'
self.file_separator = None
self.commit.message += line + '\n'
def _parse_line(self, line):
if line is None or line == '':
return
# Ignore
for patt in self.patterns['ignore']:
if patt.match(line):
return
# Commit
match = self.patterns['commit'].match(line)
if match:
if self.commit is not None and self.branch is not None:
if self.branch.tail.svn_tag is None: # Skip commits on svn tags
self.handler.commit(self.branch.tail.commit)
self.commit = Commit()
self.commit.revision = match.group(1)
parents = match.group(3)
if parents:
parents = parents.split()
self.commit.parents = parents
git_commit = self.GitCommit(self.commit, parents)
decorate = match.group(5)
branch = None
if decorate:
# Remote branch
m = re.search(self.patterns['branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.REMOTE, m.group(1), git_commit)
printdbg("Branch '%s' head at acommit %s", (branch.name, self.commit.revision))
else:
# Local Branch
m = re.search(self.patterns['local-branch'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.LOCAL, m.group(1), git_commit)
printdbg("Commit %s on local branch '%s'", (self.commit.revision, branch.name))
# If local branch was merged we just ignore this decoration
if self.branch and self.branch.is_my_parent(git_commit):
printdbg("Local branch '%s' was merged", (branch.name,))
branch = None
else:
# Stash
m = re.search(self.patterns['stash'], decorate)
if m:
branch = self.GitBranch(self.GitBranch.STASH, "stash", git_commit)
printdbg("Commit %s on stash", (self.commit.revision,))
# Tag
m = re.search(self.patterns['tag'], decorate)
if m:
self.commit.tags = [m.group(1)]
printdbg("Commit %s tagged as '%s'", (self.commit.revision, self.commit.tags[0]))
if not branch and not self.branch:
branch = self.GitBranch(self.GitBranch.LOCAL, "(no-branch)", git_commit)
printdbg("Commit %s on unknown local branch '%s'", (self.commit.revision, branch.name))
# This part of code looks wired at first time so here is a small description what it does:
#
# * self.branch is the branch to which the last inspected commit belonged to
# * branch is the branch of the current parsed commit
#
# This check is only to find branches which are fully merged into a already analyzed branch
#
# For more detailed information see https://github.com/MetricsGrimoire/CVSAnalY/issues/64
if branch is not None and self.branch is not None:
# Detect empty branches.
# Ideally, the head of a branch can't have children.
# When this happens is because the branch is empty, so we just ignore such branch.
if self.branch.is_my_parent(git_commit):
printout(
"Info: Branch '%s' will be ignored, because it was already merged in an active one.",
(branch.name,)
)
branch = None
if len(self.branches) >= 2:
# If current commit is the start point of a new branch
# we have to look at all the current branches since
# we haven't inserted the new branch yet.
# If not, look at all other branches excluding the current one
for i, b in enumerate(self.branches):
if i == 0 and branch is None:
continue
if b.is_my_parent(git_commit):
# We assume current branch is always the last one
# AFAIK there's no way to make sure this is right
printdbg("Start point of branch '%s' at commit %s",
(self.branches[0].name, self.commit.revision))
self.branches.pop(0)
self.branch = b
if self.branch and self.branch.tail.svn_tag is not None and self.branch.is_my_parent(git_commit):
# There's a pending tag in previous commit
pending_tag = self.branch.tail.svn_tag
printdbg("Move pending tag '%s' from previous commit %s to current %s", (pending_tag,
self.branch.tail.commit.revision,
self.commit.revision))
if self.commit.tags and pending_tag not in self.commit.tags:
self.commit.tags.append(pending_tag)
else:
self.commit.tags = [pending_tag]
self.branch.tail.svn_tag = None
if branch is not None:
self.branch = branch
# Insert master always at the end
if branch.name == 'master':
self.branches.append(self.branch)
else:
self.branches.insert(0, self.branch)
else:
if self.branch is not None:
self.branch.set_tail(git_commit)
return
# Committer
match = self.patterns['committer'].match(line)
if match:
self.commit.committer = Person()
self.commit.committer.name = match.group(1)
self.commit.committer.email = match.group(2)
self.handler.committer(self.commit.committer)
return
# Author
match = self.patterns['author'].match(line)
if match:
self.commit.author = Person()
self.commit.author.name = match.group(1)
self.commit.author.email = match.group(2)
self.handler.author(self.commit.author)
return
# Commit date
match = self.patterns['date'].match(line)
if match:
self.commit.date = datetime.datetime(
*(time.strptime(match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
# datetime.datetime.strptime not supported by Python2.4
#self.commit.date = datetime.datetime.strptime (match.group (1).strip (" "), "%a %b %d %H:%M:%S %Y")
# match.group(2) represents the timezone. E.g. -0300, +0200, +0430 (Afghanistan)
# This string will be parsed to int and recalculated into seconds (60 * 60)
self.commit.date_tz = (((int(match.group(2))) * 60 * 60) / 100)
return
# Author date
match = self.patterns['author_date'].match(line)
if match:
self.commit.author_date = datetime.datetime(
*(time.strptime(match.group(1).strip(" "), "%a %b %d %H:%M:%S %Y")[0:6]))
# datetime.datetime.strptime not supported by Python2.4
#self.commit.author_date = datetime.datetime.strptime (match.group (1).strip (" "), "%a %b %d %H:%M:%S %Y")
# match.group(2) represents the timezone. E.g. -0300, +0200, +0430 (Afghanistan)
# This string will be parsed to int and recalculated into seconds (60 * 60)
self.commit.author_date_tz = (((int(match.group(2))) * 60 * 60) / 100)
return
# File
match = self.patterns['file'].match(line)
if match:
action = Action()
type = match.group(1)
if len(type) > 1:
# merge actions
if 'M' in type:
type = 'M'
else:
# ignore merge actions without 'M'
return
action.type = type
action.f1 = match.group(2)
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# File moved/copied
match = self.patterns['file-moved'].match(line)
if match:
action = Action()
type = match.group(1)
if type == 'R':
action.type = 'V'
else:
action.type = type
action.f1 = match.group(3)
action.f2 = match.group(2)
action.rev = self.commit.revision
self.commit.actions.append(action)
self.handler.file(action.f1)
return
# This is a workaround for a bug in the GNOME Git migration
# There are commits on tags not correctly detected like this one:
# http://git.gnome.org/cgit/evolution/commit/?id=b8e52acac2b9fc5414a7795a73c74f7ee4eeb71f
# We want to ignore commits on tags since it doesn't make any sense in Git
if self.is_gnome:
match = self.patterns['svn-tag'].match(line.strip())
if match:
printout("Warning: detected a commit on a svn tag: %s", (match.group(0),))
tag = match.group(1)
if self.commit.tags and tag in self.commit.tags:
# The commit will be ignored, so move the tag
# to the next (previous in history) commit
self.branch.tail.svn_tag = tag
# Message
self.commit.message += line + '\n'
assert True, "Not match for line %s" % (line)