def init(req):
print "args:", req.args, req.args.path
repo_path = os.path.join(req.args.path,'.git')
if not os.path.isdir(repo_path):
raise Abort("no repository found in '%s' (.git not found)!" % (os.path.abspath(req.args.path)))
repo = Repository(req.args.path)
print "repo references:", repo.listall_references()
xxl_path = os.path.join(req.args.path,'.gitxxl')
if os.path.exists(xxl_path):
print "reinitializing existing XXL repo"
else:
print "initialiling XXL repo"
os.mkdir(xxl_path)
#install commit hooks
dst = os.path.join(repo_path,"hooks","pre-commit")
if not os.path.isfile(dst):
src = os.path.join(sys.exec_prefix,"dist","hooks","pre-commit")
print "installing pre-commit hook from:", src
shutil.copyfile(src, dst)
perm = os.stat(dst)
os.chmod(dst,perm.st_mode | stat.S_IXUSR)
dst = os.path.join(repo_path,"hooks","post-commit")
if not os.path.isfile(dst):
src = os.path.join(sys.exec_prefix,"dist","hooks","post-commit")
print "installing post-commit hook from:", src
shutil.copyfile(src, dst)
perm = os.stat(dst)
os.chmod(dst,perm.st_mode | stat.S_IXUSR)
def git_is_clean(srcdir, project):
repo = Repository(os.path.join(srcdir, project.workspace_path, ".git"))
for _, b in iteritems(repo.status()):
if b != GIT_STATUS_IGNORED and b != GIT_STATUS_CURRENT:
return False, "has uncommitted changes"
if repo.head_is_detached:
return False, "has detached HEAD"
origin = get_origin(repo, project)
if not origin:
return False, "has no upstream remote"
remote_refs = []
local_refs = {}
for refname in repo.listall_references():
if refname.startswith("refs/remotes/%s/" % origin.name):
ref = repo.lookup_reference(refname)
if ref.type == GIT_REF_OID:
remote_refs.append(ref.target)
elif not refname.startswith("refs/remotes/"):
ref = repo.lookup_reference(refname)
if ref.type == GIT_REF_OID:
local_refs[ref.peel().id] = refname
if not remote_refs:
return False, "has no upstream remote branches"
if not local_refs:
return False, "has no local branches"
if not repo.lookup_branch("%s/%s" % (origin.name, project.master_branch), GIT_BRANCH_REMOTE):
return False, "has no upstream master branch"
for remote_ref in remote_refs:
for commit in repo.walk(remote_ref):
if commit.id in local_refs:
del local_refs[commit.id]
if local_refs:
return False, "has local commits: %s" % ", ".join(["'%s'" % name for _, name in iteritems(local_refs)])
return True, ""
def init_repo(repo_path, clone_from=None, clone_refs=False,
alternate_repo_paths=None, is_bare=True):
"""Initialise a new git repository or clone from existing."""
assert is_valid_new_path(repo_path)
init_repository(repo_path, is_bare)
if clone_from:
# The clone_from's objects and refs are in fact cloned into a
# subordinate tree that's then set as an alternate for the real
# repo. This lets git-receive-pack expose available commits as
# extra haves without polluting refs in the real repo.
sub_path = os.path.join(repo_path, 'turnip-subordinate')
clone_repository(clone_from, sub_path, True)
assert is_bare
alt_path = os.path.join(repo_path, 'objects/info/alternates')
with open(alt_path, 'w') as f:
f.write('../turnip-subordinate/objects\n')
if clone_refs:
# With the objects all accessible via the subordinate, we
# can just copy all refs from the origin. Unlike
# pygit2.clone_repository, this won't set up a remote.
# TODO: Filter out internal (eg. MP) refs.
from_repo = Repository(clone_from)
to_repo = Repository(repo_path)
for ref in from_repo.listall_references():
to_repo.create_reference(
ref, from_repo.lookup_reference(ref).target)
if alternate_repo_paths:
write_alternates(repo_path, alternate_repo_paths)
ensure_config(repo_path) # set repository configuration defaults
return repo_path
class Git(object):
r"""
Interact with a git repository.
"""
def __init__(self, gitdir):
r"""
Take a path to the git repository. Other methods interact with
this git repository.
"""
self.repo = Repository(gitdir)
def branches(self):
r"""
Return the list of a branch name and its last commit id.
"""
return self._refs('heads')
def tags(self):
r"""
Return the list of a tag name and its last commit id.
"""
return self._refs('tags')
def _refs(self, type):
refs = {}
pattern = re.compile(r'refs/%s/(.*)$' % type)
for ref in self.repo.listall_references():
m = pattern.match(ref)
if m:
reference = self.repo.lookup_reference(ref)
refs[m.group(1)] = reference.hex
return refs
def create_branch(self, name, target):
r"""
Create new branch.
"""
if not is_valid_value(name):
raise InvalidParamException("name is required")
if not is_valid_hex(target):
raise InvalidParamException("target is required")
target = sha_hex2bin(target)
try:
self.repo.create_reference('refs/heads/%s' % name, target)
except Exception, e:
raise InvalidParamException(str(e))
return True
def git_is_clean(srcdir, project):
repo = Repository(os.path.join(srcdir, project.workspace_path, ".git"))
for _, b in iteritems(repo.status()):
if b != GIT_STATUS_IGNORED and b != GIT_STATUS_CURRENT:
return False, "has uncommitted changes"
if repo.head_is_detached:
return False, "has detached HEAD"
origin = get_origin(repo, project)
if not origin:
return False, "has no upstream remote"
remote_refs = []
local_refs = {}
for refname in repo.listall_references():
if refname.startswith("refs/remotes/%s/" % origin.name):
ref = repo.lookup_reference(refname)
if ref.type == GIT_REF_OID:
remote_refs.append(ref.target)
elif not refname.startswith("refs/remotes/"):
ref = repo.lookup_reference(refname)
if ref.type == GIT_REF_OID:
local_refs[ref.peel().id] = refname
if not remote_refs:
return False, "has no upstream remote branches"
if not local_refs:
return False, "has no local branches"
if not repo.lookup_branch("%s/%s" % (origin.name, project.master_branch),
GIT_BRANCH_REMOTE):
return False, "has no upstream master branch"
for remote_ref in remote_refs:
for commit in repo.walk(remote_ref):
if commit.id in local_refs:
del local_refs[commit.id]
if local_refs:
return False, "has local commits: %s" % ", ".join(
["'%s'" % name for _, name in iteritems(local_refs)])
return True, ""
class GitMiner(BaseMiner):
Id_Name_Login = namedtuple("Id_Name_Login", ["id", "name", "login"])
Code_Change = namedtuple("Code_Change", ["commit_id", "filename"])
def __init__(self, args):
super().__init__(args)
self._initialise_db()
if args.dbms == "sqlite":
self._conn.execute("PRAGMA foreign_keys=ON")
self.email_map = {}
self.commit_id = {}
self.id_commit = {}
self.code_change_map = {}
self.__init_user_emails()
self._dump_repository()
self.aio = args.aio
if self.aio:
self._create_loop()
self.repo = Repository(args.path)
self._fetch_references()
self._dump_tags()
self._fetch_commit_ids()
def _create_loop(self):
self.loop = asyncio.new_event_loop()
def load_from_file(self, file):
pass
def dump_to_file(self, path):
pass
def __init_user_emails(self):
res = self.execute_query(
"""
SELECT email, id, login, name
FROM contributors
WHERE email IS NOT NULL
"""
).fetchall()
for row in res:
self.email_map[row[0]] = self.Id_Name_Login(id=row[1], name=row[2], login=row[3])
def __init_code_change(self):
res = self.execute_query(
"""
SELECT id, commit_id, filename
FROM code_change
"""
).fetchall()
for row in res:
self.code_change_map[self.Code_Change(commit_id=row[1], filename=row[2])] = row[0]
def _dump_repository(self):
logger.info("Dumping Repository...")
res = self.execute_query(
f"""
SELECT repo_id
FROM repository
WHERE name="{self.repo_name}" and owner="{self.repo_owner}"
"""
).fetchone()
if res:
self._set_repo_id(res[0])
else:
repo = RepositoryStruct(
name=self.repo_name,
owner=self.repo_owner
).process()
obj = self.db_schema.repository_object(
name=self.repo_name,
owner=self.repo_owner,
created_at=repo.created_at,
updated_at=repo.updated_at,
description=repo.description,
disk_usage=repo.disk_usage,
fork_count=repo.fork_count,
url=repo.url,
homepage_url=repo.homepage_url,
primary_language=repo.primary_language,
total_stargazers=repo.stargazer_count,
total_watchers=repo.watcher_count,
forked_from=repo.forked_from
)
self._insert(self.db_schema.repository.insert(), obj)
self._set_repo_id()
def _fetch_references(self):
self.tags, self.branches = [], {}
for reference in self.repo.listall_references():
if 'refs/tags' in reference:
self.tags.append(reference)
else:
self.branches[reference] = self.repo.lookup_reference(reference).peel().oid
def _dump_tags(self):
objects = []
for tag in self.tags:
ref = self.repo.lookup_reference(tag)
tag_obj = self.repo[ref.target.hex]
if isinstance(tag_obj, Tag):
name = tag_obj.name
msg = tag_obj.message
tagged_object = tag_obj.hex
tagger = self.__get_user_id(name=tag_obj.tagger.name, email=tag_obj.tagger.email, oid=tagged_object,
is_author=False, is_tagger=True)
else:
name = tag.split('/')[-1]
msg = tag_obj.message
tagged_object = tag_obj.hex
tagger = self.__get_user_id(name=tag_obj.author.name, email=tag_obj.author.email, oid=tagged_object,
is_author=True, is_tagger=False)
obj = self.db_schema.tags_object(
name=name,
tagged_object=tagged_object,
message=msg,
tagger=tagger
)
objects.append(obj)
self._insert(object_=self.db_schema.tags.insert(), param=objects)
@staticmethod
def __get_status(status):
if status == 1:
return 'ADDED'
elif status == 2:
return 'DELETED'
elif status == 3:
return 'MODIFIED'
elif status == 4:
return 'RENAMED'
elif status == 5:
return 'COPIED'
elif status == 6:
return 'IGNORED'
elif status == 7:
return 'UNTRACKED'
elif status == 8:
return 'TYPECHANGED'
else:
return None
def __init_commits(self, inverse=False):
if not inverse:
res = self.execute_query(
f"""
SELECT oid, id
FROM commits
WHERE repo_id={self.repo_id}
"""
).fetchall()
for row in res:
self.commit_id[row[0]] = row[1]
else:
res = self._conn.execute(
f"""
SELECT id, oid
FROM commits
WHERE repo_id={self.repo_id}
"""
).fetchall()
for row in res:
self.id_commit[row[0]] = row[1]
def __get_commit_id(self, oid, pk=None):
if not pk:
try:
return self.commit_id[oid]
except KeyError:
return None
else:
try:
return self.id_commit[pk]
except KeyError:
self.__init_commits(inverse=True)
res = self.__get_commit_id(oid=None, pk=pk)
if not res:
raise Exception(f"GitMiner => __get_commit_id: Pk {pk} does not exist!")
else:
return res
def __check_user_id(self, email):
try:
map_ = self.email_map[email]
return [map_.id, map_.login, map_.name]
except KeyError:
res = self.execute_query(
f"""
SELECT id, login, name
FROM contributors
WHERE email="{email}"
"""
).fetchone()
if res:
self.email_map[email] = self.Id_Name_Login(id=res[0], login=res[1], name=res[2])
return res
def __update_contributor(self, name, id_, login, email):
name = name.replace('"', '""')
self.execute_query(
f"""
UPDATE contributors
SET name="{name}"
WHERE id={id_}
"""
)
self.email_map[email] = self.Id_Name_Login(id=id_, login=login, name=name)
def __get_user_id(self, name, email, oid, is_author, is_tagger):
if not email:
email = None
if not name:
name = None
res = self.__check_user_id(email)
if not res:
user = CommitUserStruct(
oid=oid,
repo_name=self.repo_name,
repo_owner=self.repo_owner,
name=name,
email=email,
is_author=is_author,
is_tagger=is_tagger
).process()
if user is None:
self._dump_anon_user_object(name=name, email=email, object_=self.db_schema.contributors.insert(),
locked_insert=LOCKED)
else:
self._dump_user_object(login=None, user_object=user, object_=self.db_schema.contributors.insert(),
locked_insert=LOCKED)
return self.__get_user_id(name=name, email=email, oid=oid, is_author=is_author, is_tagger=is_tagger)
else:
if name == res[2]:
return res[0]
elif name == res[1]:
return res[0]
else:
self.__update_contributor(name=name, id_=res[0], login=res[1], email=email)
return res[0]
def _dump_code_change(self, oid):
commit = self.repo.get(oid)
commit_id = self.__get_commit_id(oid)
logger.debug(f"Dumping Code Change for commit_id -> {commit_id}...")
code_change = []
if commit:
if not commit.parents:
diffs = [self.repo.diff("4b825dc642cb6eb9a060e54bf8d69288fbee4904", commit)]
else:
diffs = [self.repo.diff(i, commit) for i in commit.parents]
total_diffs = len(diffs)
for diff in diffs:
logger.debug(f"Remaining: {total_diffs}")
total_diffs -= 1
for patch in diff:
obj = self.db_schema.code_change_object(
repo_id=self.repo_id,
commit_id=commit_id,
filename=patch.delta.new_file.path,
additions=patch.line_stats[1],
deletions=patch.line_stats[2],
changes=patch.line_stats[1] + patch.line_stats[2],
change_type=self.__get_status(patch.delta.status)
)
code_change.append(obj)
self._insert(object_=self.db_schema.code_change.insert(), param=code_change)
logger.debug(f"Successfully dumped code change for {oid}!")
def __get_code_change_id(self, commit_id, filename):
try:
return self.code_change_map[self.Code_Change(commit_id=commit_id, filename=filename)]
except KeyError:
return Exception(f"GitMiner => __get_code_change_id: Object does not exist! commit_id={commit_id}, "
f"filename:{filename}")
def _dump_patches(self, oid):
commit = self.repo.get(oid)
commit_id = self.__get_commit_id(oid)
logger.debug(f"Dumping Patch for commit_id -> {commit_id}...")
patches = []
if not commit.parents:
diffs = [self.repo.diff("4b825dc642cb6eb9a060e54bf8d69288fbee4904", commit)]
else:
diffs = [self.repo.diff(i, commit) for i in commit.parents]
total_diffs = len(diffs)
for diff in diffs:
logger.debug(f"Remaining: {total_diffs}")
total_diffs -= 1
for patch in diff:
obj = self.db_schema.patches_object(
code_change_id=self.__get_code_change_id(commit_id, patch.delta.new_file.path),
patch=patch.patch
)
patches.append(obj)
self._insert(object_=self.db_schema.patches.insert(), param=patches)
logger.debug(f"Successfully dumped patch for {oid}!")
def _dump_commit(self, oid):
logger.debug(f"Inserting for commit: {oid}...")
commit = self.repo.get(oid)
if not commit.parents:
diffs = [self.repo.diff("4b825dc642cb6eb9a060e54bf8d69288fbee4904", commit)]
else:
diffs = [self.repo.diff(i, commit) for i in commit.parents]
num_files_changed = 0
additions, deletions = 0, 0
for diff in diffs:
num_files_changed += diff.stats.files_changed
additions += diff.stats.insertions
deletions += diff.stats.deletions
author_name = commit.author.name
author_email = commit.author.email
author_id = self.__get_user_id(name=author_name, email=author_email, oid=oid.hex, is_author=True,
is_tagger=False) if \
author_email.strip() else None
authored_date = datetime.fromtimestamp(commit.author.time)
committer_name = commit.committer.name
committer_email = commit.committer.email
if committer_email == "*****@*****.**":
committer_id = author_id
else:
committer_id = self.__get_user_id(name=committer_name, email=committer_email, oid=oid.hex,
is_author=False, is_tagger=False) if committer_email.strip() else None
committed_date = datetime.fromtimestamp(commit.commit_time)
message = commit.message
if len(commit.parents) > 1:
is_merge = 1
else:
is_merge = 0
obj = self.db_schema.commits_object(
repo_id=self.repo_id,
oid=oid.hex,
additions=additions,
deletions=deletions,
author_id=author_id,
authored_date=authored_date,
committer_id=committer_id,
committer_date=committed_date,
message=message,
num_files_changed=num_files_changed,
is_merge=is_merge
)
self._insert(object_=self.db_schema.commits.insert(), param=obj)
logger.debug(f"Successfully dumped commit: {oid.hex}")
def __fetch_branch_commits(self, branch_target):
logger.info(f"Ongoing Branch {branch_target[0]}...")
for commit in self.repo.walk(branch_target[1], GIT_SORT_TOPOLOGICAL | GIT_SORT_TIME):
if commit.oid not in self.commits:
self.commits.add(commit.oid)
else:
break
def _fetch_commit_ids(self):
try:
with open(f"{ROOT}/.gras-cache/{self.repo_name}_commits.txt", "rb") as fp:
self.commits = pickle.load(fp)
self.commits = [Oid(hex=x) for x in self.commits]
logger.info(f"TOTAL COMMITS: {len(self.commits)}")
return self.commits
except FileNotFoundError:
logger.error("Commits file not present, dumping...")
self.commits = set()
with concurrent.futures.ThreadPoolExecutor(max_workers=THREADS) as executor:
process = {executor.submit(self.__fetch_branch_commits, branch_target): branch_target for branch_target
in self.branches.items()}
for future in concurrent.futures.as_completed(process):
branch_target = process[future]
logger.info(f"Fetched for {branch_target[0]}, Total: {len(self.commits)}")
logger.info(f"TOTAL COMMITS: {len(self.commits)}")
with open(f"{ROOT}/.gras-cache/{self.repo_name}_commits.txt", "wb") as fp:
temp = [x.hex for x in self.commits]
pickle.dump(temp, fp)
del temp
@timing(name="commits", is_stage=True)
def _parse_commits(self):
res = self.execute_query(
f"""
SELECT DISTINCT oid
FROM commits
"""
).fetchall()
dumped_commits = [x[0] for x in res]
del res
commits = list(self.commits)
for i in range(0, len(commits), THREADS):
proc = [mp.Process(target=self._dump_commit, args=(oid,)) for oid in commits[i:i + THREADS] if
oid.hex not in dumped_commits]
for p in proc:
p.start()
while any([p.is_alive() for p in proc]):
continue
@timing(name="code change", is_stage=True)
def _parse_code_change(self):
id_oid = self.execute_query(
f"""
SELECT id, oid
FROM commits
"""
).fetchall()
dumped_ids = self.execute_query(
f"""
SELECT DISTINCT commit_id
FROM code_change
"""
).fetchall()
dumped_ids = [x[0] for x in dumped_ids]
not_dumped_commits = [x[1] for x in id_oid if x[0] not in dumped_ids]
del dumped_ids
del id_oid
for i in range(0, len(not_dumped_commits), THREADS):
proc = [mp.Process(target=self._dump_code_change, args=(oid,)) for oid in
not_dumped_commits[i: i + THREADS]]
for p in proc:
p.start()
while any([x.is_alive() for x in proc]):
continue
@timing(name="patches", is_stage=True)
def _parse_patches(self):
self.__init_commits(inverse=True)
res = self.execute_query(
f"""
SELECT id, commit_id
FROM code_change
"""
).fetchall()
cc_commit = {}
for row in res:
cc_commit[row[0]] = row[1]
res = self.execute_query(
"""
SELECT code_change_id
FROM patches
"""
)
not_dumped_commits = set(cc_commit.values()).difference({cc_commit[x[0]] for x in res})
not_dumped_commits = sorted([self.id_commit[id_] for id_ in not_dumped_commits])
del cc_commit
for i in range(0, len(not_dumped_commits), THREADS):
proc = [mp.Process(target=self._dump_code_change, args=(oid,)) for oid in
not_dumped_commits[i: i + THREADS]]
for p in proc:
p.start()
while any([x.is_alive() for x in proc]):
continue
@timing(name="async -> commits", is_stage=True)
async def _async_parse_commits(self):
loop = asyncio.get_event_loop()
tasks = [loop.run_in_executor(self.executor, self._dump_commit, oid) for oid in self.commits]
completed, _ = await asyncio.wait(tasks)
for t in completed:
logger.info(f"Dumped commit: {t.result()}")
@timing(name="async -> code change", is_stage=True)
async def _async_parse_code_change(self):
loop = asyncio.get_event_loop()
tasks = [loop.run_in_executor(self.executor, self._dump_code_change, oid) for oid in self.commits]
completed, _ = await asyncio.wait(tasks)
for t in completed:
logger.info(f"Dumped Code Change for commit: {t.result()}")
def process(self):
if self.aio:
self.loop.run_until_complete(self._parse_commits())
self.loop.run_until_complete(self._parse_code_change())
else:
# self._parse_commits()
self.__init_commits()
self._parse_code_change()
# self._parse_patches()
def __del__(self):
if self.aio:
self.loop.close()
class GitStorage(BaseStorage):
_backend = None
def __init__(self, context, repo_path=None):
self.context = context
rp = IStorageInfo(context).path
try:
self.repo = Repository(discover_repository(rp))
except KeyError:
# discover_repository may have failed.
raise PathNotFoundError('repository does not exist at path')
self.checkout() # defaults to HEAD.
@property
def empty_root(self):
return {'': '_empty_root'}
def _get_empty_root(self):
return self.empty_root
def _get_obj(self, path, cls=None):
if path == '' and self._commit is None:
# special case
return self._get_empty_root()
if self._commit is None:
raise PathNotFoundError('repository is empty')
root = self._commit.tree
try:
breadcrumbs = []
fragments = list(reversed(path.split('/')))
node = root
oid = None
while fragments:
fragment = fragments.pop()
if not fragment == '':
# no empty string entries, also skips over '//' and
# leaves the final node (if directory) as the tree.
oid = node[fragment].oid
node = self.repo.get(oid)
breadcrumbs.append(fragment)
if node is None:
# strange. Looks like it's either submodules only
# have entry nodes or pygit2 doesn't fully support
# this. Try to manually resolve the .gitmodules
# file.
if cls is None:
# Only return this if a specific type was not
# expected.
submods = parse_gitmodules(self.repo.get(
root[GIT_MODULE_FILE].oid).data)
submod = submods.get('/'.join(breadcrumbs))
if submod:
fragments.reverse()
return {
'': '_subrepo',
'location': submod,
'path': '/'.join(fragments),
'rev': oid.hex,
}
if node and (cls is None or isinstance(node, cls)):
return node
except KeyError:
# can't find what is needed in repo, raised by pygit2
raise PathNotFoundError('path not found')
# not what we were looking for.
if cls == Tree:
raise PathNotDirError('path not dir')
elif cls == Blob:
raise PathNotFileError('path not file')
raise PathNotFoundError('path not found')
@property
def _commit(self):
return self.__commit
@property
def rev(self):
if self.__commit:
return self.__commit.hex
return None
@property
def shortrev(self):
# TODO this is an interim solution.
if self.rev:
return self.rev[:12]
def basename(self, name):
return name.split('/')[-1]
def checkout(self, rev=None):
# None maps to the default revision.
if rev is None:
rev = 'HEAD'
try:
self.__commit = self.repo.revparse_single(rev)
except KeyError:
if rev == 'HEAD':
# probably a new repo.
self.__commit = None
return
raise RevisionNotFoundError('revision %s not found' % rev)
# otherwise a RevisionNotFoundError should be raised.
def files(self):
def _files(tree, current_path=None):
results = []
for node in tree:
if current_path:
name = '/'.join([current_path, node.name])
else:
name = node.name
obj = self.repo.get(node.oid)
if isinstance(obj, Blob):
results.append(name)
elif isinstance(obj, Tree):
results.extend(_files(obj, name))
return results
if not self._commit:
return []
results = _files(self._commit.tree)
return results
def file(self, path):
return self._get_obj(path, Blob).data
def listdir(self, path):
if path:
tree = self._get_obj(path, Tree)
else:
if self._commit is None:
return []
tree = self._commit.tree
return [entry.name for entry in tree]
def format(self, **kw):
# XXX backwards compatibility??
return kw
def log(self, start, count, branch=None, shortlog=False):
"""
start and branch are literally the same thing.
"""
def _log(iterator):
for pos, commit in iterator:
if pos == count:
raise StopIteration
yield {
'author': commit.committer.name,
'email': self._commit.committer.email,
'date': self.strftime(committer_dt(commit.committer)),
'node': commit.hex,
'rev': commit.hex,
'desc': commit.message
}
if start is None:
# assumption.
start = 'HEAD'
try:
self.repo.revparse_single(start)
except KeyError:
return []
try:
rev = self.repo.revparse_single(start).hex
except KeyError:
raise RevisionNotFoundError('revision %s not found' % start)
iterator = enumerate(self.repo.walk(rev, GIT_SORT_TIME))
return list(_log(iterator))
def pathinfo(self, path):
obj = self._get_obj(path)
if isinstance(obj, Blob):
return self.format(**{
'type': 'file',
'basename': self.basename(path),
'size': obj.size,
'date': self.strftime(committer_dt(self._commit.committer)),
})
elif isinstance(obj, dict):
# special cases are represented as dict.
if obj[''] == '_subrepo':
return self.format(**{
'type': 'subrepo',
'date': '',
'size': 0,
'basename': self.basename(path),
# extra field.
'obj': obj,
})
elif obj[''] == '_empty_root':
return self.format(**{
'type': 'folder',
'date': '',
'size': 0,
'basename': self.basename(path),
})
# Assume this is a Tree.
return self.format(**{
'basename': self.basename(path),
'size': 0,
'type': 'folder',
'date': '',
})
def branches(self):
return tuple(
(b, self.repo.lookup_branch(b).target.hex)
for b in self.repo.listall_branches()
)
def tags(self):
return tuple(
(b[10:], self.repo.lookup_reference(b).target.hex)
for b in self.repo.listall_references()
if b.startswith('refs/tags')
)
class CollectGit(object):
"""
Small Helper class for small repositories.
This does not scale because we hold a lot of data in memory.
"""
_regex_comment = re.compile(
r"(//[^\"\n\r]*(?:\"[^\"\n\r]*\"[^\"\n\r]*)*[\r\n]|/\*([^*]|\*(?!/))*?\*/)(?=[^\"]*(?:\"[^\"]*\"[^\"]*)*$)"
)
_regex_jdoc_line = re.compile(r"(- |\+)\s*(\*|/\*).*")
def __init__(self, path):
if not path.endswith('.git'):
if not path.endswith('/'):
path += '/'
path += '.git'
self._log = logging.getLogger(self.__class__.__name__)
self._path = path
self._repo = Repository(self._path)
self._hunks = {}
self._file_actions = {}
self._bugfix = {}
self._msgs = {}
self._days = {}
self._cdays = {}
self._branches = {}
self._tags = {}
self._dopts = GIT_DIFF_FIND_RENAMES | GIT_DIFF_FIND_COPIES
self._SIMILARITY_THRESHOLD = 50
self._graph = nx.DiGraph()
@classmethod
def clone_repo(cls, uri, local_path):
project_name = uri.split('/')[-1].split('.git')[0]
repo_path = local_path + '/' + project_name + '/'
if os.path.isdir(repo_path):
c = subprocess.run(['git', 'fetch'],
cwd=repo_path,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if c.returncode != 0:
err = 'Error pulling repository {} to {}'.format(
uri, repo_path)
raise Exception(err)
else:
os.mkdir(repo_path)
c = subprocess.run(['git', 'clone', uri, repo_path],
cwd=repo_path,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if c.returncode != 0:
err = 'Error cloning repository {} to {}'.format(
uri, repo_path)
raise Exception(err)
return repo_path
def _changed_lines(self, hunk):
added_lines = []
deleted_lines = []
del_line = hunk['old_start']
add_line = hunk['new_start']
for line in hunk['content'].split('\n'):
tmp = line[1:].strip()
# is_comment = tmp.startswith('//') or tmp.startswith('/*') or tmp.startswith('*')
if line.startswith('+'):
added_lines.append((add_line, tmp))
del_line -= 1
if line.startswith('-'):
deleted_lines.append((del_line, tmp))
add_line -= 1
del_line += 1
add_line += 1
return added_lines, deleted_lines
def _comment_only_change(self, content):
content = content + '\n' # required for regex to drop comments
content = re.sub(self._regex_comment, "", content)
removed = ''
added = ''
for line in content.split('\n'):
line = re.sub(
r"\s+", " ", line, flags=re.UNICODE
) # replace all kinds of whitespaces (also multiple) with sińgle whitespace
if not re.match(self._regex_jdoc_line, line):
if line.startswith('-'):
removed += line[1:].strip()
elif line.startswith('+'):
added += line[1:].strip()
return removed == added
def _blame_lines(self,
revision_hash,
filepath,
strategy,
ignore_lines=False,
validated_bugfix_lines=False):
"""We want to find changed lines for one file in one commit (from the previous commit).
For this we are iterating over the diff and counting the lines that are deleted (changed) from the original file.
We ignore all added lines.
ignore_lines is already specific to all changed hunks of the file for which blame_lines is called
"""
c = self._repo.revparse_single('{}'.format(revision_hash))
self._hunks[revision_hash] = self._get_hunks(c)
changed_lines = []
if revision_hash not in self._hunks.keys(
) or not self._hunks[revision_hash]:
return changed_lines
for h in self._hunks[revision_hash]:
if h['new_file'] != filepath:
continue
# only whitespace or comment changes in the hunk, ignore
if strategy == 'code_only' and self._comment_only_change(
h['content']):
self._log.debug(
'detected whitepace or comment only change in {} for {}'.
format(revision_hash, filepath))
continue
added, deleted = self._changed_lines(h)
for dt in deleted:
if dt not in changed_lines and dt[1]:
if strategy == 'code_only' and dt[1].startswith(
('//', '/*', '*')):
continue
# we may only want validated lines
if validated_bugfix_lines is not False:
if dt[0] not in validated_bugfix_lines:
continue
# we may ignore lines, e.g., refactorings
if ignore_lines:
ignore = False
for start_line, end_line in ignore_lines:
if start_line <= dt[0] <= end_line:
ignore = True
break
# if we hit the line in our ignore list we continue to the next
if ignore:
# self._log.warn('ignore line {} in file {} in commit {} because of refactoring detection'.format(dt[0], filepath, revision_hash))
continue
changed_lines.append(dt)
return changed_lines
def blame(self,
revision_hash,
filepath,
strategy='code_only',
ignore_lines=False,
validated_bugfix_lines=False):
"""Collect a list of commits where the given revision and file were last changed.
Uses git blame.
:param str revision_hash: Commit for which we want to collect blame commits.
:param str filepath: File for which we want to collect blame commits.
:rtype: list
:returns: A list of tuples of blame commits and the original file for the given parameters.
"""
commits = []
# - ignore if commit is not in graph
if revision_hash not in self._graph:
return []
# # - ignore package-info.java
# if strategy == 'code_only' and filepath.lower().endswith('package-info.java'):
# self._log.debug('skipping blame on revision: {} for file {} because it is package-info.java'.format(revision_hash, filepath))
# return []
# # - ignore test/ /test/ example/ examples/
# if strategy == 'code_only' and re.match(self._regex_test_example, filepath):
# self._log.debug('skipping blame on revision: {} for file {} because it is a test or an example'.format(revision_hash, filepath))
# return []
# bail on multiple parents
parents = list(self._graph.predecessors(revision_hash))
if len(parents) > 1:
self._log.debug(
'skipping blame on revision: {} because it is a merge commit'.
format(revision_hash))
return []
changed_lines = self._blame_lines(revision_hash, filepath, strategy,
ignore_lines, validated_bugfix_lines)
parent_commit = self._repo.revparse_single('{}^'.format(revision_hash))
blame = self._repo.blame(filepath,
flags=GIT_BLAME_TRACK_COPIES_SAME_FILE,
newest_commit=parent_commit.hex)
for lineno, line in changed_lines:
# returns blamehunk for specific line
try:
bh = blame.for_line(lineno)
except IndexError as e:
# this happens when we have the wrong parent node
bla = 'tried to get file: {}, line: {}, revision: {}, blame commit: {}'.format(
filepath, lineno, revision_hash, str(bh.orig_commit_id))
self._log.error(bla)
raise # this is critical
inducing_commit = self._repo.revparse_single(str(
bh.orig_commit_id))
# start = bh.orig_start_line_number
# lines = bh.lines_in_hunk
# final_start = bh.final_start_line_number
# print(revision_hash, '->', inducing_commit.hex)
# print('original: {}: {}'.format(lineno, line))
# print('{},{}: {},{}'.format(start, lines, final_start, lines))
# blame_lines = []
# for hunk in self._hunks[inducing_commit.hex]:
# if hunk['new_file'] != bh.orig_path:
# continue
# ls = final_start
# for i, blame_line in enumerate(hunk['content'].split('\n')):
# if blame_line[1:].strip() and line[1:].strip() and blame_line[1:] == line[1:]:
# print('blame: {}:{}'.format(ls, blame_line))
# ls += 1
commits.append((inducing_commit.hex, bh.orig_path))
# make unique
return list(set(commits))
def commit_information(self, revision_hash):
obj = self._repo.get(revision_hash)
return {
'author_name':
obj.author.name,
'author_email':
obj.author.email,
'committer_name':
obj.committer.name,
'committer_email':
obj.committer.email,
'committer_date_utc':
datetime.fromtimestamp(obj.commit_time, tz=timezone.utc),
'committer_date':
obj.commit_time,
'committer_date_offset':
obj.commit_time_offset,
'message':
obj.message,
'file_actions':
self._file_actions[revision_hash]
}
def file_actions(self, revision_hash):
return self._file_actions[revision_hash]
def all_files(self, revision_hash):
# 1. checkout repo
self._checkout_revision(revision_hash)
# 2. list files
return self._list_files()
def first_occurence(self, filename):
# file rename tracking is not possible currently in libgit, see:
# https://github.com/libgit2/libgit2/issues/3041
# find first occurence of file with git cli
# git log --follow --diff-filter=A --find-renames=40% foo.js
path = self._path.replace('.git', '')
c = subprocess.run([
'git', 'log', '--all', '--pretty=tformat:"%H %ci"', '--follow',
'--diff-filter=A', '--find-renames=80%', '--', filename
],
cwd=path,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
if c.returncode != 0:
err = 'Error finding first occurrence of file: {}'.format(filename)
self._log.error(err)
self._log.error(c.stderr)
raise Exception(err)
full = c.stdout.decode('utf-8')
try:
first_line = full.split('\n')[-2]
except IndexError:
if not full:
print('no git log for file {}'.format(filename))
print(full)
raise
first_date = ' '.join(first_line.split(' ')[1:]).replace('"', '')
dt = datetime.strptime(
first_date, '%Y-%m-%d %H:%M:%S %z'
) # we can do this here because we control the input format, %z does not cover +01:00 just +100 (at least in 3.6)
return dt
def tags(self):
regex = re.compile('^refs/tags')
ret = []
for tagref in filter(lambda r: regex.match(r),
self._repo.listall_references()):
tag = self._repo.lookup_reference(tagref)
target = self._repo.lookup_reference(tagref).peel()
ret.append({
'name': tag.name.replace('refs/tags/', ''),
'revision_hash': target.id
})
return ret
def _checkout_revision(self, revision):
"""Checkout via shell, we ignore stdout output."""
path = self._path.replace('.git', '')
c = subprocess.run(['git', 'checkout', '-q', '-f', revision],
cwd=path,
stdout=subprocess.PIPE)
return c.returncode == 0
def _list_files(self):
"""The slower list_files"""
path = self._path.replace('.git', '')
ret = []
for root, dirs, files in os.walk(path):
for file in files:
filepath = os.path.join(root, file)
relative_filepath = filepath.replace(path, '')
ret.append(relative_filepath)
return ret
def _list_files2(self):
"""The faster list_files (relies on find command)"""
path = self._path.replace('.git', '')
lines = subprocess.check_output(['find', '.', '-iname', '*.java'],
cwd=path)
files = []
for f in lines.decode('utf-8').split('\n'):
if f.lower().endswith('.java'):
files.append(f.replace('./', ''))
return files
def _get_hunks(self, commit):
diffs = []
hunks = []
# for initial commit (or orphan commits) pygit2 needs some special attention
initial = False
if not commit.parents:
initial = True
diffs.append((None,
commit.tree.diff_to_tree(context_lines=0,
interhunk_lines=1)))
# we may have multiple parents (merge commit)
for parent in commit.parents:
# we need all information from each parent because in a merge each parent may add different files
tmp = self._repo.diff(parent,
commit,
context_lines=0,
interhunk_lines=1)
tmp.find_similar(self._dopts, self._SIMILARITY_THRESHOLD,
self._SIMILARITY_THRESHOLD)
diffs.append((parent.hex, tmp))
for parent, diff in diffs:
checked_paths = set()
for patch in diff:
if patch.delta.new_file.path in checked_paths:
self._log.warn('already have {} in checked_paths'.format(
patch.delta.new_file.path))
continue
mode = 'X'
if patch.delta.status == 1:
mode = 'A'
elif patch.delta.status == 2:
mode = 'D'
elif patch.delta.status == 3:
mode = 'M'
elif patch.delta.status == 4:
mode = 'R'
elif patch.delta.status == 5:
mode = 'C'
elif patch.delta.status == 6:
mode = 'I'
elif patch.delta.status == 7:
mode = 'U'
elif patch.delta.status == 8:
mode = 'T'
# diff to tree gives D for inital commit otherwise
if initial:
mode = 'A'
# we may have hunks to add
if patch.hunks and commit.hex not in self._hunks.keys():
self._hunks[commit.hex] = []
# add hunks
for hunk in patch.hunks:
# initial is special case
if initial:
content = ''.join(
['+' + l.content for l in hunk.lines])
hunks.append({
'header': hunk.header,
'new_file': patch.delta.new_file.path,
'new_start': hunk.old_start,
'new_lines': hunk.old_lines,
'old_start': hunk.new_start,
'old_lines': hunk.new_lines,
'content': content
})
else:
content = ''.join(
[l.origin + l.content for l in hunk.lines])
hunks.append({
'header': hunk.header,
'new_file': patch.delta.new_file.path,
'new_start': hunk.new_start,
'new_lines': hunk.new_lines,
'old_start': hunk.old_start,
'old_lines': hunk.old_lines,
'content': content
})
return hunks
def _changed_files(self, commit):
changed_files = []
diffs = []
# for initial commit (or orphan commits) pygit2 needs some special attention
initial = False
if not commit.parents:
initial = True
diffs.append((None,
commit.tree.diff_to_tree(context_lines=0,
interhunk_lines=1)))
# we may have multiple parents (merge commit)
for parent in commit.parents:
# we need all information from each parent because in a merge each parent may add different files
tmp = self._repo.diff(parent,
commit,
context_lines=0,
interhunk_lines=1)
tmp.find_similar(self._dopts, self._SIMILARITY_THRESHOLD,
self._SIMILARITY_THRESHOLD)
diffs.append((parent.hex, tmp))
for parent, diff in diffs:
checked_paths = set()
for patch in diff:
if patch.delta.new_file.path in checked_paths:
self._log.warn('already have {} in checked_paths'.format(
patch.delta.new_file.path))
continue
mode = 'X'
if patch.delta.status == 1:
mode = 'A'
elif patch.delta.status == 2:
mode = 'D'
elif patch.delta.status == 3:
mode = 'M'
elif patch.delta.status == 4:
mode = 'R'
elif patch.delta.status == 5:
mode = 'C'
elif patch.delta.status == 6:
mode = 'I'
elif patch.delta.status == 7:
mode = 'U'
elif patch.delta.status == 8:
mode = 'T'
# diff to tree gives D for inital commit otherwise
if initial:
mode = 'A'
# we may have hunks to add
if patch.hunks and commit.hex not in self._hunks.keys():
self._hunks[commit.hex] = []
# add hunks
for hunk in patch.hunks:
# initial is special case
if initial:
content = ''.join(
['+' + l.content for l in hunk.lines])
self._hunks[commit.hex].append({
'header':
hunk.header,
'new_file':
patch.delta.new_file.path,
'new_start':
hunk.old_start,
'new_lines':
hunk.old_lines,
'old_start':
hunk.new_start,
'old_lines':
hunk.new_lines,
'content':
content
})
else:
content = ''.join(
[l.origin + l.content for l in hunk.lines])
self._hunks[commit.hex].append({
'header':
hunk.header,
'new_file':
patch.delta.new_file.path,
'new_start':
hunk.new_start,
'new_lines':
hunk.new_lines,
'old_start':
hunk.old_start,
'old_lines':
hunk.old_lines,
'content':
content
})
# collect line stats
if initial:
fa = {
'lines_added': patch.line_stats[2],
'lines_deleted': patch.line_stats[1],
'changeset_size': len(diff),
'parent': None
}
else:
fa = {
'lines_added': patch.line_stats[1],
'lines_deleted': patch.line_stats[2],
'changeset_size': len(diff),
'parent': parent
}
#if mode == 'R':
# print('R {} -> {}, sim: {}'.format(patch.delta.old_file.path, patch.delta.new_file.path, patch.delta.similarity))
if mode in ['C', 'R']:
changed_file = [
mode, patch.delta.new_file.path,
patch.delta.old_file.path, fa
]
else:
changed_file = [mode, patch.delta.new_file.path, None, fa]
checked_paths.add(patch.delta.new_file.path)
changed_files.append(changed_file)
return changed_files
def collect(self):
# list all branches
for branch in list(self._repo.branches):
self._collect_branch(branch)
# list all tags
for obj in self._repo:
tag = self._repo[obj]
if tag.type == GIT_OBJ_TAG:
self._collect_branch(tag, is_tag=True)
return self._graph
def _collect_branch(self, branch, is_tag=False):
if type(branch) == str:
branch = self._repo.branches[branch]
# add nodes to graph
try:
for c in self._repo.walk(branch.target):
self._graph.add_node(c.hex)
# branch stuff, used for traversing backwards for tags in svn->git conversions
# if c.hex not in self._branches.keys():
# self._branches[c.hex] = []
# what about tags which are also on branches?
# if is_tag:
# self._tags[c.hex] = branch.name
# else:
# self._branches[c.hex].append(branch.branch_name)
# add msg
# self._msgs[c.hex] = c.message
# add days, we use this later for lookup
# day = str(datetime.fromtimestamp(c.commit_time, tz=timezone.utc).date())
# if day not in self._days.keys():
# self._days[day] = []
# self._days[day].append(c.hex)
# add for convenience for OntdekBaanBfs
# self._cdays[c.hex] = day
# add changed files per node
# if c.hex not in self._file_actions.keys():
# self._file_actions[c.hex] = self._changed_files(c)
# still too expensive
# self._create_hunks(c)
# add edges to graph
for c in self._repo.walk(branch.target):
for p in c.parents:
self._graph.add_edge(p.hex, c.hex)
except ValueError as e:
pass
class GitRepo(object):
''' git repo class '''
def __init__(self, path):
try:
self.__repo = Repository(path)
except Exception as e:
self.__repo = None
print(e)
def get_info(self):
if not self.__repo:
return None
signature = self.__repo.default_signature
result = {
'path': self.__repo.path,
'workdir': self.__repo.workdir,
'bare': self.__repo.is_bare,
'empty': self.__repo.is_empty,
'name': signature.name,
'email': signature.email,
'time': signature.time,
'offset': signature.offset,
}
return result
def get_all_references(self):
return self.__repo.listall_references()
def get_reference(self, name):
if not self.__repo:
return None
ref = None
try:
ref = self.__repo.lookup_reference(name)
except Exception as e:
print(e)
return ref
def get_all_branches(self, branch_type=None):
if not self.__repo:
return None
if branch_type:
return self.__repo.listall_branches(branch_type)
r = self.__repo.listall_branches(GIT_BRANCH_LOCAL | GIT_BRANCH_REMOTE)
return r
def get_branch(self, name, branch_type=GIT_BRANCH_LOCAL):
if not self.__repo:
return None
return self.__repo.lookup_branch(name, branch_type)
def check_branch(self, name, branch_type=None):
if not branch_type:
if '/' in name:
branch_type = GIT_BRANCH_REMOTE
else:
branch_type = GIT_BRANCH_LOCAL
try:
result = self.get_branch(name, branch_type)
return result
except Exception as e:
print(e)
return False
def get_current_commit(self):
if not self.__repo:
return None
commit = self.__repo.revparse_single('HEAD')
return self.get_commit(commit)
def get_commit_by_branch(self, branch):
if not self.__repo:
return None
query = 'refs/'
if hasattr(branch, 'remote_name'):
query += 'remotes/'
else:
query += 'heads/'
query += branch.branch_name
try:
ref = self.get_reference(query)
commit = ref.target
return self.get_commit(commit)
except Exception as e:
print(e)
return None
def get_commit_by_tag(self, tag):
if self.__repo is None:
return None
if tag:
commit = tag.get_object()
return self.get_commit(commit)
return None
def get_commit(self, oid_or_commit):
''' return a commit w/ json '''
if not self.__repo or not oid_or_commit:
return None
try:
commit = oid_or_commit
if not isinstance(oid_or_commit, Commit):
commit = self.__repo.get(oid_or_commit)
if commit and commit.type == GIT_OBJ_COMMIT:
# t1 = self.__repo.revparse_single('HEAD^')
# t2 = self.__repo.revparse_single('HEAD^^')
# patches = self.__repo.diff(t1, t2)
# for p in patches:
# print(p.new_file_path)
result = {
'id': str(commit.id),
'author': commit.author.name,
'commiter': commit.committer.name,
'message': commit.message,
'message_encoding': commit.message_encoding,
'tree': str(commit.tree_id),
'parent': [str(pid) for pid in commit.parent_ids],
'time': str(commit.commit_time),
'time_offset': str(commit.commit_time_offset),
}
return result
except Exception as e:
print(e)
return None
def get_commits(self, depth=10, oid_or_commit=None):
result = []
if depth == 0:
return result
if oid_or_commit:
commit = self.get_commit(oid_or_commit)
else:
commit = self.get_current_commit()
if not commit:
return result
# TODO: starting from a commit or its parent
# TODO: author
result.append(commit)
depth -= 1
if commit and commit['parent']:
for parent in commit['parent']:
result.extend(self.get_commits(depth, parent))
return result
def get_commits_by_branch(self, name, path=None):
if not self.__repo:
return None
if self.check_branch(name):
ref = self.get_reference('refs/heads/' + name)
if ref:
commit = ref.target
commits = self.get_commits(commit)
result = {}
for key, val in commits.items():
if self.check_commit_by_path(val, path):
result[key] = val
return result
return None
def check_tag(self, name):
try:
ref = self.get_reference('refs/tags/' + name)
return ref
except Exception:
return False
def get_commits_by_tag(self, tag, path=None):
if not self.__repo:
return None
if tag:
commit = tag.target
commits = self.get_commits(commit)
result = {}
for key, val in commits.items():
if self.check_commit_by_path(val, path):
result[key] = val
return result
return None
def check_commit_by_path(self, commit, path):
if not commit:
return False
if path is None or len(path) == 0:
return True
result = self.get_tree(commit['tree'])
if not isinstance(path, list):
path = path.strip().split('/')
for name in path:
name = name.strip()
if name in result:
oid = result[name]
result = self.get_tree(oid)
if not result:
result = self.get_blob(oid)
return result is not None
def get_tree(self, oid, ppath=None):
if not self.__repo:
return None
try:
tree = self.__repo.get(oid)
if tree and tree.type == GIT_OBJ_TREE:
result = {}
for entry in tree:
item = {
'id': str(entry.id)
}
obj = self.__repo.get(entry.id)
if obj.type == GIT_OBJ_BLOB:
item['type'] = 'blob'
elif obj.type == GIT_OBJ_TREE:
item['type'] = 'tree'
item['ppath'] = ppath
result[entry.name] = item
return result
except Exception as e:
print(e)
return None
def get_tree_by_commit(self, commit, path=None):
if not commit:
return None
result = self.get_tree(commit['tree'])
if not path:
return result
# if not isinstance(path, list):
# path = path.strip().split('/')
try:
for name in path:
oid = result[name]['id']
p = result[name]['ppath']
p = name if not p else p + '/' + name
result = self.get_tree(oid, p)
if not result:
break
except Exception as e:
print(e)
result = None
return result
def get_current_root(self):
tree = self.get_current_commit()
if tree:
return self.get_tree(tree['tree'])
return None
def get_whole_tree(self, oid):
''' tree w/ json '''
if not self.__repo:
return None
result = tree_walker(self.__repo, oid)
return result
def get_blob(self, oid):
''' blob w/ json '''
if not self.__repo or not oid:
return None
try:
blob = self.__repo.get(oid)
if blob and blob.type == GIT_OBJ_BLOB:
content = blob.is_binary and None or blob.data.decode(
'utf8', 'ignore')
result = {
'id': str(blob.id),
'content': content,
'size': blob.size,
}
return result
except Exception as e:
print(e)
return None
def get_blob_by_commit(self, commit, path=None):
try:
tree = self.get_tree_by_commit(commit, path[:-1])
oid = tree[path[-1]]['id']
result = self.get_blob(oid)
return result
except Exception as e:
print(e)
return None
def get_tag(self, oid):
''' blob w/ json '''
if not self.__repo or not oid:
return None
try:
tag = self.__repo.get(oid)
if tag and tag.type == GIT_OBJ_TAG:
result = {
'id': str(oid),
'name': tag.name,
'target': str(tag.target.id),
'tagger': tag.tagger,
'message': tag.message,
}
return result
except Exception as e:
print(e)
return None
def get_patches(self, a=None, b=None):
try:
if not a:
a = 'HEAD'
if not b:
b = a + '^'
t1 = self.__repo.revparse_single(a)
t2 = self.__repo.revparse_single(b)
patches = self.__repo.diff(t1, t2)
result = []
for patch in patches:
p = {
'old_file_path': patch.old_file_path,
'new_file_path': patch.new_file_path,
'old_oid': str(patch.old_oid),
'new_oid': str(patch.new_oid),
'status': patch.status,
'similarity': patch.similarity,
'additions': patch.additions,
'deletions': patch.deletions,
'binary': patch.is_binary,
'hunks': [],
}
for hunk in patch.hunks:
h = {
'old_start': hunk.old_start,
'old_lines': hunk.old_lines,
'new_start': hunk.new_start,
'new_lines': hunk.new_lines,
'lines': hunk.lines,
}
p['hunks'].append(h)
result.append(p)
return result
except Exception as e:
print(e)
return None
class Document(object):
'''
Class representing a document, interacts with the git
database
'''
def __init__( self, name, create=False, rootPath=None ):
'''
Constructor
Args:
name The name of the document
create If true, will create a document
rootPath The rootPath to use (if not supplied, uses default)
Exceptions:
RepoNotFound if repository isn't found
'''
if not rootPath:
rootPath = DEFAULT_ROOT_PATH
targetDir = os.path.join( rootPath, name + '.git' )
if create:
# Create a bare repository
self.repo = init_repository( targetDir, True )
self._CreateMasterBranch()
else:
try:
self.repo = Repository( targetDir )
except KeyError:
raise RepoNotFound()
def _CreateMasterBranch( self ):
'''
Creates the master branch on the repo w/ default file.
For now this is just a file named layout
'''
commitId = CommitBlob(
self.repo, '', SECTION_INDEX_FILENAME, 'Initial commit'
)
self.repo.create_reference( MASTER_REF, commitId )
@staticmethod
def _IsSectionRef( refName ):
'''
Checks if a refererence name refers to a section
Args:
refName: The reference name
Returns:
A boolean
'''
return refName.startswith( SECTION_REF_PREFIX )
@staticmethod
def _RefNameToSectionName( refName ):
'''
Converts a reference name to a section name
Args:
ref: The reference name
'''
return refName[ len(SECTION_REF_PREFIX) : ]
def _SectionRefs( self ):
'''
Gets an iterator over the section refs
'''
return (
(
self._RefNameToSectionName( ref ),
self.repo.lookup_reference( ref ),
)
for ref in self.repo.listall_references()
if self._IsSectionRef( ref )
)
def Sections( self ):
'''
Gets an iterator over all the sections
'''
return (
Section( name, self.repo[ref.oid], self.repo )
for name, ref in self._SectionRefs()
)
def CurrentSections( self ):
'''
Gets the current sections with their positions
Returns:
A list of tuples ( position, section )
'''
return enumerate( self._CurrentSections() )
def _CurrentSections( self ):
'''
Internal method to get the current sections
in order, without position numbers
Returns:
An iterator over the sections
'''
index = SectionIndex(self.repo)
for s in index.CurrentSections():
yield self.FindSection(s.name)
def FindSection( self, name ):
'''
Finds a section by name
Args:
name The name of the section to find
Returns:
The section if found
Exceptions:
SectionNotFound if section not found
'''
try:
ref = self.repo.lookup_reference(
SECTION_REF_PREFIX + name
)
except KeyError:
raise SectionNotFound()
return Section( name, self.repo[ref.oid], self.repo )
def AddSection( self, name, content='' ):
'''
Creates a new section
Args:
name The name of the section
content The optional initial content of the
section
Returns:
The new Section object
'''
# TODO: Should probably make
# sure no such section exists
commitId = CommitBlob(
self.repo,
content,
name,
'Created section ' + name
)
ref = self.repo.create_reference(
SECTION_REF_PREFIX + name,
commitId
)
index = SectionIndex(self.repo)
index.AddSection( name )
index.Save( self.repo )
return Section( name, self.repo[ ref.oid ], self.repo )
def RemoveSection( self, name ):
'''
Removes a section.
This function does not actually delete the data associated
with a section, it just removes it from the index.
Args:
name The name of the section to remove
'''
index = SectionIndex( self.repo )
index.RemoveSection( name )
index.Save( self.repo )
testrepo = None
repo.addfile(file.name)
testrepo = Repository(dir.name)
index = testrepo.index
index.read()
repo = GitRepo(dir.name)
testrepo = Repository(dir.name)
# print('Commitloglänge', len(testrepo.walk(testrepo.head.target)), 0)
testrepo = None
# Commit 1
repo.commit()
testrepo = Repository(dir.name)
all_refs = testrepo.listall_references()
print('Refs', all_refs)
commits = testrepo.walk(testrepo.head.target)
print('Commitloglänge', len(list(testrepo.walk(testrepo.head.target))), 1)
testrepo = None
# Commit 2
testrepo = Repository(dir.name)
# print('Commitloglänge', len(testrepo.walk(testrepo.head.target)), 0)
testrepo = None
file.write(b'Test2\n')
file.write(b'Test3\n')
f = open(file.name, 'w')
f.write('Test2')
f.write('Test3')
class GitMixin(object):
tag_or_remote_regex = re.compile('^refs/(tags|remotes)/(.*)')
def __init__(self):
where = GitOperations.get_repository_location(self.user, self.name)
self.ondisk = Repository(where)
def refresh(self):
creds = GitOperations.get_credentials(self.git_user, self.user)
for remote in self.ondisk.remotes:
remote.credentials = creds
remote.fetch()
# update current reference
master_ref = self.ondisk.lookup_reference('refs/heads/master')
remote_ref = self.ondisk.lookup_reference('refs/remotes/origin/master')
master_ref.set_target(remote_ref.target)
def filter_references(self, regex):
return [ref for ref in self.ondisk.listall_references()
if regex.match(ref)]
def get_commit_time(self, name):
ref = self.ondisk.revparse_single(name)
if isinstance(ref, Tag):
return ref.get_object().commit_time
if isinstance(ref, Commit):
return ref.commit_time
raise GitException('invalid reference: commit time could not be found.') # pragma: no cover
def get_latest_refs(self, count=None):
info = self.filter_references(GitMixin.tag_or_remote_regex)
refs = list(zip(info, map(self.get_commit_time, info)))
refs.sort(key=itemgetter(1), reverse=True)
def ref_info(info):
(ref, commit_time) = info
what, name = GitMixin.tag_or_remote_regex.findall(ref)[0]
return (what, name, commit_time)
refs = map(ref_info, refs)
if not count:
return refs
return islice(refs, count)
def filter_commits(self, flags=0):
all_commits = self.ondisk.walk(self.ondisk.head.target, flags)
emails = [ue.email for ue in self.user.emails.all()]
return filter(lambda commit: commit.author.email in emails, all_commits)
def get_commits(self, count=None):
all_commits = self.filter_commits(GIT_SORT_TOPOLOGICAL)
if not count:
return all_commits
return islice(all_commits, count)
def get_commit_count(self):
return len(list(self.filter_commits()))
def get_shorthand_of_branch(self, branch):
commit = self.ondisk.lookup_branch(branch)
if commit:
return commit.shorthand
return '(none)'
def get_sha1_of_branch(self, branch):
commit = self.ondisk.lookup_branch(branch)
if commit:
return str(commit.get_object().id)[:6]
return '(none)'
def get_numstat(self, commit):
diff = None
try:
previous_commit = self.ondisk.revparse_single(str(commit.id) + '^')
diff = self.ondisk.diff(previous_commit, commit)
except KeyError:
# likely we hit the very first commit.
diff = commit.tree.diff_to_tree(swap=True)
additions, deletions = 0, 0
for patch in diff:
additions += patch.additions
deletions += patch.deletions
return (len(diff), additions, deletions)
def get_first_updated(self):
all_commits = self.ondisk.walk(self.ondisk.head.target,
GIT_SORT_TIME | GIT_SORT_REVERSE)
first_commit = next(all_commits)
return first_commit.commit_time
def get_last_updated(self):
all_commits = self.ondisk.walk(self.ondisk.head.target,
GIT_SORT_TIME)
last_commit = next(all_commits)
return last_commit.commit_time
def get_file_count(self):
diff = self.ondisk.head.get_object().tree.diff_to_tree()
return len([patch.old_file_path for patch in diff])
def get_line_count(self):
diff = self.ondisk.head.get_object().tree.diff_to_tree()
return sum([patch.deletions for patch in diff])
def get_author_count(self):
commits = self.filter_commits()
return len(set([commit.author.email for commit in commits]))
def commits_between(self, start, end):
all_commits = self.filter_commits(GIT_SORT_TIME | GIT_SORT_REVERSE)
starting = dropwhile(lambda obj: obj.commit_time < start, all_commits)
return takewhile(lambda obj: obj.commit_time <= end, starting)
@staticmethod
def by_day(obj):
# we want to group our commit times by the day. so convert
# timestamp -> date -> timestamp
new_date = date.fromtimestamp(obj.commit_time)
new_date += timedelta(days=1)
return timegm(new_date.timetuple())
@staticmethod
def group_by(series):
result = groupby(series, GitMixin.by_day)
return [{'date': commit_date,
'value': len(list(commits))}
for commit_date, commits in result]
def histogram(self, start, end):
series = self.commits_between(start, end)
return GitMixin.group_by(series)
