allow filepath to be used with other filters

pydriller/git_repository.py

@@ -266,25 +266,18 @@ def _useless_line(self, line: str):
  return not line or line.startswith('//') or line.startswith('#') or line.startswith("/*") or \
  line.startswith("'''") or line.startswith('"""') or line.startswith("*")
 
- def get_commits_modified_file(self, filepath: str) -> List[Commit]:
+ def get_commits_modified_file(self, filepath: str) -> List[str]:
  all_commits = self.get_list_commits()
 
  dict_commits = {}
  for commit in all_commits:
  dict_commits[commit.hash] = commit
 
- # getting the absolute path of the file
  path = str(Path(filepath).absolute())
  commits = []
  try:
  commits = self.git.log("--follow", "--format=%H", path).split('\n')
  except GitCommandError:
  logger.debug("Could not find information of file %s", path)
 
- list_commits = []
- for commit in commits:
- # I don't have a better idea than this:
- # unfortunately, this will call `git` for every commit
- list_commits.append(dict_commits[commit.strip()])
-
- return list_commits
+ return commits

pydriller/repository_mining.py

@@ -81,6 +81,7 @@ def __init__(self, path_to_repo: Union[str, List[str]],
  self._only_authors = only_authors
  self._only_commits = only_commits
  self._filepath = filepath
+ self._filepath_commits = None
 
  def _sanity_check_repos(self, path_to_repo):
  if not isinstance(path_to_repo, str) and not isinstance(path_to_repo, list):
@@ -151,11 +152,7 @@ def traverse_commits(self) -> Generator[Commit, None, None]:
  logger.info('Analyzing git repository in {}'.format(git_repo.path))
 
  if self._filepath is not None:
- commits = git_repo.get_commits_modified_file(self._filepath)
- for commit in commits:
- logger.info('Commit #{} in {} from {}'.format(commit.hash, commit.committer_date, commit.author.name))
- yield commit
- continue
+ self._filepath_commits = git_repo.get_commits_modified_file(self._filepath)
 
  for commit in git_repo.get_list_commits(self._only_in_branch, not self._reversed_order):
  logger.info('Commit #{} in {} from {}'.format(commit.hash, commit.committer_date, commit.author.name))
@@ -187,6 +184,11 @@ def _is_commit_filtered(self, commit: Commit):
  if self._only_commits is not None and commit.hash not in self._only_commits:
  logger.debug("Commit filtered because it is not one of the specified commits")
  return True
+ if self._filepath is not None:
+ if self._filepath_commits is not None:
+ if commit.hash not in self._filepath_commits:
+ return True
+
  return False
 
  def _has_modification_with_file_type(self, commit):

tests/test_git_repository.py

@@ -403,4 +403,16 @@ def test_get_commits_modified_file():
  gr = GitRepository('test-repos/test1/')
 
  commits = gr.get_commits_modified_file('test-repos/test1/file2.java')
+
  assert len(commits) == 3
+ assert '09f6182cef737db02a085e1d018963c7a29bde5a' in commits
+ assert '6411e3096dd2070438a17b225f44475136e54e3a' in commits
+ assert 'a88c84ddf42066611e76e6cb690144e5357d132c' in commits
+
+
+def test_get_commits_modified_file_missing_file():
+ gr = GitRepository('test-repos/test1/')
+
+ commits = gr.get_commits_modified_file('test-repos/test1/non-existing-file.java')
+
+ assert len(commits) == 0

tests/test_repository_mining.py

@@ -67,3 +67,24 @@ def test_badly_formatted_url():
 
  with pytest.raises(Exception):
  list(RepositoryMining(path_to_repo='test').traverse_commits())
+
+
+def test_filepath():
+ dt = datetime(2018, 6, 6)
+ assert 4 == len(list(RepositoryMining(
+ path_to_repo='test-repos/test5',
+ filepath='test-repos/test5/A.java',
+ to=dt).traverse_commits()))
+
+def test_filepath_with_rename():
+ dt = datetime(2018, 6, 6)
+ commits = list(RepositoryMining(
+ path_to_repo='test-repos/test1',
+ filepath='test-repos/test1/file4.java',
+ to=dt).traverse_commits())
+ assert 2 == len(commits)
+
+ commit_hashes = [commit.hash for commit in commits]
+
+ assert 'da39b1326dbc2edfe518b90672734a08f3c13458' in commit_hashes
+ assert 'a88c84ddf42066611e76e6cb690144e5357d132c' in commit_hashes