def construct_historage(self):
logger.info('convert a git repository to a historage...')
historage_repo = self.prepare_historage_repo()
committer = SyntaxTreesCommitter(Repo(self.org_repo.git_dir), historage_repo, self.syntax_trees_dir)
num_commits = self.num_commits if self.num_commits != 0 else '???'
for num, commit in izip(count(), get_reversed_topological_ordered_commits(self.org_repo, self.org_repo.refs)):
logger.info('[%d/%s] convert %s to: %s' % (num, num_commits, commit.hexsha, historage_repo.git_dir))
committer.apply_change(commit)
committer.create_heads()
committer.create_tags()
if not self.is_bare_repo:
historage_repo.head.reset(working_tree=True)
logger.info('completed!')
def construct_historage(self):
print 'create historage...'
base_repo = self.prepare_base_repo()
committer = SyntaxTreesCommitter(Repo(self.org_repo.git_dir), base_repo, self.syntax_trees_dir)
num_commits = self.num_commits if self.num_commits != 0 else '???'
for num, commit in izip(count(), get_reversed_topological_ordered_commits(self.org_repo, self.org_repo.refs)):
commit = self.org_repo.commit(commit)
print '[%d/%s] convert %s to: %s' % (num, num_commits, commit.hexsha, base_repo.git_dir)
committer.apply_change(commit)
committer.create_heads()
committer.create_tags()
if not self.is_bare_repo:
base_repo.head.reset(working_tree=True)
def construct_historage(self):
logger.info('convert a git repository to a historage...')
historage_repo = self.prepare_historage_repo()
committer = SyntaxTreesCommitter(Repo(self.org_repo.git_dir),
historage_repo, self.syntax_trees_dir)
num_commits = self.num_commits if self.num_commits != 0 else '???'
for num, commit in zip(
count(),
get_reversed_topological_ordered_commits(
self.org_repo, self.org_repo.refs)):
logger.info(
'[%d/%s] convert %s to: %s' %
(num, num_commits, commit.hexsha, historage_repo.git_dir))
committer.apply_change(commit)
committer.create_heads()
committer.create_tags()
if not self.is_bare_repo:
historage_repo.head.reset(working_tree=True)
logger.info('completed!')
def parse_all_target_files(self):
logger.info('create parser processes...')
blob_parser = BlobParser(extension_dict, self.syntax_trees_dir, self.org_repo.git_dir)
parsed_blob = set()
for commit in get_reversed_topological_ordered_commits(self.org_repo, self.org_repo.refs):
self.num_commits = self.num_commits + 1
if commit.parents:
for p in commit.parents:
for diff in p.diff(commit):
if is_target_blob(diff.b_blob):
if diff.b_blob.hexsha not in parsed_blob:
blob_parser.parse_blob(diff.b_blob)
parsed_blob.add(diff.b_blob.hexsha)
else:
for entry in commit.tree.traverse():
if isinstance(entry, Blob) and is_target_blob(entry):
if entry.hexsha not in parsed_blob:
blob_parser.parse_blob(entry)
parsed_blob.add(entry.hexsha)
logger.info('waiting parser processes')
blob_parser.join()
def parse_all_java_files(self):
print 'create paresr processes...'
parser_executor = ParserExecutor(self.syntax_trees_dir, self.parser_jar_path)
parsed_blob = set()
for commit in get_reversed_topological_ordered_commits(self.org_repo, self.org_repo.refs):
self.num_commits = self.num_commits + 1
commit = self.org_repo.commit(commit)
if commit.parents:
for p in commit.parents:
for diff in p.diff(commit):
if self.is_target_blob(diff.b_blob, '.java'):
if diff.b_blob.hexsha not in parsed_blob:
parser_executor.parse_blob(diff.b_blob)
parsed_blob.add(diff.b_blob.hexsha)
else:
for entry in commit.tree.traverse():
if isinstance(entry, Blob) and self.is_target_blob(entry, '.java'):
if entry.hexsha not in parsed_blob:
parser_executor.parse_blob(entry)
parsed_blob.add(entry.hexsha)
print 'waiting parser processes'
parser_executor.join()
def parse_all_target_files(self):
logger.info('create parser processes...')
blob_parser = BlobParser(extension_dict, self.syntax_trees_dir,
self.org_repo.git_dir)
parsed_blob = set()
for commit in get_reversed_topological_ordered_commits(
self.org_repo, self.org_repo.refs):
self.num_commits = self.num_commits + 1
if commit.parents:
for p in commit.parents:
for diff in p.diff(commit):
if is_target_blob(diff.b_blob):
if diff.b_blob.hexsha not in parsed_blob:
blob_parser.parse_blob(diff.b_blob)
parsed_blob.add(diff.b_blob.hexsha)
else:
for entry in commit.tree.traverse():
if isinstance(entry, Blob) and is_target_blob(entry):
if entry.hexsha not in parsed_blob:
blob_parser.parse_blob(entry)
parsed_blob.add(entry.hexsha)
logger.info('waiting parser processes')
blob_parser.join()
submodule_conf = "/Users/kenjif/test_gitmodules"
conf_mode, conf_binsha = write_blob(odb, submodule_conf)
tree_contents = []
tree_contents.append((conf_mode, conf_binsha, ".gitmodules"))
tree_contents.append(get_submodule_tree_content(submodule_commit_hexsha, "jEdit"))
tree_mode, binsha = mktree_from_iter(odb, tree_contents)
return bin_to_hex(binsha)
if __name__ == "__main__":
from kenja.git.util import get_reversed_topological_ordered_commits
repo = Repo("/Users/kenjif/msr_repos/git/jEdit")
commits = get_reversed_topological_ordered_commits(repo, repo.refs)
new_repo = Repo.init("/Users/kenjif/test_git_repo")
with open("/Users/kenjif/test_gitmodules", "wb") as f:
name = "jEdit"
path = "jEdit"
url = "/Users/kenjif/msr_repos/git/jEdit"
write_submodule_config(f, name, path, url)
committed = {}
tags = {}
heads = {}
for tag_ref in repo.tags:
tags[tag_ref.commit.hexsha] = tag_ref.name
