- removed only_in_main_branch, since its behaviour was wrong (by defa…

…ult is True, and even if set to False nothing changed)

- only_in_branch created: by default is the master branch, but one can now set to a different branch

pydriller/git_repository.py

@@ -68,17 +68,17 @@ def get_head(self) -> Commit:
         head_commit = self.repo.head.commit
         return Commit(head_commit, self.path, self.main_branch)
 
-    def get_list_commits(self) -> List[Commit]:
+    def get_list_commits(self, branch: str = None) -> List[Commit]:
         """
         Return the list of all the commits in the repo.
 
         :return: List[Commit], the list of all the commits in the repo
         """
-        return self._get_all_commits()
+        return self._get_all_commits(branch)
 
-    def _get_all_commits(self) -> List[Commit]:
+    def _get_all_commits(self, branch: str = None) -> List[Commit]:
         all_commits = []
-        for commit in self.repo.iter_commits():
+        for commit in self.repo.iter_commits(branch):
             all_commits.append(self.get_commit_from_gitpython(commit))
         return all_commits
 

pydriller/repository_mining.py

@@ -36,8 +36,7 @@ def __init__(self, path_to_repo: Union[str, List[str]],
                  from_commit: str = None, to_commit: str = None,
                  from_tag: str = None, to_tag: str = None,
                  reversed_order: bool = False,
-                 only_in_main_branch: bool = False,
-                 only_in_branches: List[str] = None,
+                 only_in_branch: str = None,
                  only_modifications_with_file_types: List[str] = None,
                  only_no_merge: bool = False):
         """
@@ -57,8 +56,7 @@ def __init__(self, path_to_repo: Union[str, List[str]],
         :param str from_tag: starting the analysis from specified tag (only if `since` and `from_commit` are None)
         :param str to_tag: ending the analysis from specified tag (only if `to` and `to_commit` are None)
         :param bool reversed_order: whether the commits should be analyzed in reversed order
-        :param bool only_in_main_branch: whether only commits in main branch should be analyzed
-        :param List[str] only_in_branches: only commits in these branches will be analyzed
+        :param str only_in_branch: only commits in this branch will be analyzed
         :param List[str] only_modifications_with_file_types: only modifications with that file types will be analyzed
         :param bool only_no_merge: if True, merges will not be analyzed
         """
@@ -74,14 +72,13 @@ def __init__(self, path_to_repo: Union[str, List[str]],
         self._since = since
         self._to = to
         self._reversed_order = reversed_order
-        self._only_in_main_branch = only_in_main_branch
-        self._only_in_branches = only_in_branches
+        self._only_in_branch = only_in_branch
         self._only_modifications_with_file_types = only_modifications_with_file_types
         self._only_no_merge = only_no_merge
 
     def _sanity_check_repos(self, path_to_repo):
         if not isinstance(path_to_repo, str) and not isinstance(path_to_repo, list):
-            raise Exception('The path to the repo has to be of type \"string\" or \"list of strings\"!')
+            raise Exception("The path to the repo has to be of type 'string' or 'list of strings'!")
 
     def _sanity_check_filters(self, git_repo: GitRepository):
         # If single is defined, not other filters should be
@@ -151,7 +148,7 @@ def traverse_commits(self) -> Generator[Commit, None, None]:
 
             logger.info('Analyzing git repository in {}'.format(git_repo.path))
 
-            all_cs = self._apply_filters_on_commits(git_repo.get_list_commits())
+            all_cs = self._apply_filters_on_commits(git_repo.get_list_commits(self._only_in_branch))
 
             if not self._reversed_order:
                 all_cs.reverse()
@@ -167,13 +164,6 @@ def traverse_commits(self) -> Generator[Commit, None, None]:
                 yield commit
 
     def _is_commit_filtered(self, commit: Commit):
-        if self._only_in_main_branch is True and commit.in_main_branch is False:
-            logger.debug('Commit filtered for main branch')
-            return True
-        if self._only_in_branches is not None:
-            if not self._commit_branch_in_branches(commit):
-                logger.debug('Commit filtered for only in branches')
-                return True
         if self._only_modifications_with_file_types is not None:
             if not self._has_modification_with_file_type(commit):
                 logger.debug('Commit filtered for modification types')
@@ -183,12 +173,6 @@ def _is_commit_filtered(self, commit: Commit):
             return True
         return False
 
-    def _commit_branch_in_branches(self, commit: Commit):
-        for branch in commit.branches:
-            if branch in self._only_in_branches:
-                return True
-        return False
-
     def _has_modification_with_file_type(self, commit):
         for mod in commit.modifications:
             if mod.filename.endswith(tuple(self._only_modifications_with_file_types)):

tests/integration/test_commit_filters.py

@@ -30,7 +30,7 @@ def test_mod_with_file_types():
 
 
 def test_only_in_main_branch():
-    lc = list(RepositoryMining('test-repos/git-5/', only_in_main_branch=True).traverse_commits())
+    lc = list(RepositoryMining('test-repos/git-5/').traverse_commits())
 
     assert 5 == len(lc)
     assert '4a17f31c0d1285477a3a467d0bc3cb38e775097d' == lc[0].hash
@@ -41,7 +41,7 @@ def test_only_in_main_branch():
 
 
 def test_multiple_filters():
-    lc = list(RepositoryMining('test-repos/git-5/', only_in_main_branch=True, only_no_merge=True).traverse_commits())
+    lc = list(RepositoryMining('test-repos/git-5/', only_no_merge=True).traverse_commits())
 
     assert 4 == len(lc)
     assert '4a17f31c0d1285477a3a467d0bc3cb38e775097d' == lc[0].hash
@@ -58,7 +58,20 @@ def test_no_filters():
     assert '375de7a8275ecdc0b28dc8de2568f47241f443e9' == lc[1].hash
     assert 'b8c2be250786975f1c6f47e96922096f1bb25e39' == lc[2].hash
 
+
 def test_no_single_commit():
     with pytest.raises(Exception):
         for commit in RepositoryMining('test-repos/git-5', single="6fe83d9fbf9a63cc1c51e5fe6fd5230f7fbbce6f").traverse_commits():
             print(commit.hash)
+
+
+def test_only_in_branch():
+    lc = list(RepositoryMining('test-repos/git-5/', only_in_branch='branch2').traverse_commits())
+    assert 6 == len(lc)
+
+    assert '4a17f31c0d1285477a3a467d0bc3cb38e775097d' == lc[0].hash
+    assert 'ff663cf1931a67d5e47b75fc77dcea432c728052' == lc[1].hash
+    assert 'fa8217c324e7fb46c80e1ddf907f4e141449637e' == lc[2].hash
+    assert '5d9d79607d7e82b6f236aa29be4ba89a28fb4f15' == lc[3].hash
+    assert '377e0f474d70f6205784d0150ee0069a050c29ed' == lc[4].hash
+    assert '6fe83d9fbf9a63cc1c51e5fe6fd5230f7fbbce6f' == lc[5].hash