def test_only_commits():
lc = list(RepositoryMining('test-repos/complex_repo',
only_commits=["9e71dd5726d775fb4a5f08506a539216e878adbb"]).traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "9e71dd5726d775fb4a5f08506a539216e878adbb"
lc = list(RepositoryMining('test-repos/complex_repo',
only_commits=["953737b199de233896f00b4d87a0bc2794317253",
"ffccf1e7497eb8136fd66ed5e42bef29677c4b71"]).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == "ffccf1e7497eb8136fd66ed5e42bef29677c4b71"
assert lc[1].hash == "953737b199de233896f00b4d87a0bc2794317253"
lc = list(RepositoryMining('test-repos/complex_repo',
only_commits=["866e997a9e44cb4ddd9e00efe49361420aff2559",
"57dbd017d1a744b949e7ca0b1c1a3b3dd4c1cbc1",
"e7d13b0511f8a176284ce4f92ed8c6e8d09c77f2"]).traverse_commits())
assert len(lc) == 3
assert lc[0].hash == "866e997a9e44cb4ddd9e00efe49361420aff2559"
assert lc[1].hash == "57dbd017d1a744b949e7ca0b1c1a3b3dd4c1cbc1"
assert lc[2].hash == "e7d13b0511f8a176284ce4f92ed8c6e8d09c77f2"
lc = list(RepositoryMining('test-repos/complex_repo',
only_commits=["fake hash"]).traverse_commits())
assert len(lc) == 0
total_commits = len(list(RepositoryMining('test-repos/complex_repo').traverse_commits()))
assert total_commits == 13
def test_multiple_filters_exceptions():
from_commit = '6411e3096dd2070438a17b225f44475136e54e3a'
from_tag = 'v1.4'
with pytest.raises(Exception):
RepositoryMining('test-repos/test1/',
from_commit=from_commit,
from_tag=from_tag)
with pytest.raises(Exception):
RepositoryMining('test-repos/test1/',
since=dt2,
from_commit=from_commit)
with pytest.raises(Exception):
RepositoryMining('test-repos/test1/', since=dt2, from_tag=from_tag)
with pytest.raises(Exception):
RepositoryMining('test-repos/test1/', to=dt2, to_tag=from_tag)
with pytest.raises(Exception):
RepositoryMining('test-repos/test1/',
single=from_commit,
to=dt2,
to_tag=from_tag)
def test_only_in_branches():
# by default, only analyze master
assert 3 == len(list(RepositoryMining('test-repos/test8/').traverse_commits()))
# only analyze b2
assert 4 == len(list(RepositoryMining('test-repos/test8/', only_in_branch='b2').traverse_commits()))
# only analyze b1
assert 5 == len(list(RepositoryMining('test-repos/test8/', only_in_branch='b1').traverse_commits()))
def test_single_commit_head():
lc = list(RepositoryMining('test-repos/complex_repo',
single="e7d13b0511f8a176284ce4f92ed8c6e8d09c77f2").traverse_commits())
assert len(lc) == 1
lc_head = list(RepositoryMining('test-repos/complex_repo', single="HEAD").traverse_commits())
assert len(lc_head) == 1
assert lc[0].hash == lc_head[0].hash
def test_single_commit():
lc = list(RepositoryMining('test-repos/complex_repo',
single="866e997a9e44cb4ddd9e00efe49361420aff2559").traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "866e997a9e44cb4ddd9e00efe49361420aff2559"
lc = list(RepositoryMining('test-repos/complex_repo',
single="ffccf1e7497eb8136fd66ed5e42bef29677c4b71").traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "ffccf1e7497eb8136fd66ed5e42bef29677c4b71"
def test_only_authors():
lc = list(
RepositoryMining('test-repos/git-10/',
only_authors=["Maurício Aniche"]).traverse_commits())
assert len(lc) == 4
lc = list(
RepositoryMining('test-repos/git-10/',
only_authors=["ishepard"]).traverse_commits())
assert len(lc) == 1
def test_single_commit():
lc = list(RepositoryMining('test-repos/git-10/',
single="4e669cb4f69245dc669e116517d80d038d8e0434").traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "4e669cb4f69245dc669e116517d80d038d8e0434"
lc = list(RepositoryMining('test-repos/git-10/',
single="168b3aab057ed61a769acf336a4ef5e64f76c9fd").traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "168b3aab057ed61a769acf336a4ef5e64f76c9fd"
def test_only_in_branches():
# by default, only analyze master
assert len(list(RepositoryMining('test-repos/branches_not_merged')
.traverse_commits())) == 3
# only analyze b2
assert len(list(RepositoryMining('test-repos/branches_not_merged',
only_in_branch='b2')
.traverse_commits())) == 4
# only analyze b1
assert len(list(RepositoryMining('test-repos/branches_not_merged',
only_in_branch='b1')
.traverse_commits())) == 5
def test_mod_with_file_types():
lc = list(RepositoryMining('test-repos/different_files',
only_modifications_with_file_types=['.java']).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
assert lc[1].hash == 'b8c2be250786975f1c6f47e96922096f1bb25e39'
lc = list(RepositoryMining('test-repos/different_files1',
only_modifications_with_file_types=['.java'])
.traverse_commits())
assert len(lc) == 2
assert lc[0].hash == '5adbb71167e79ab6b974827e74c9da4d81977655'
assert lc[1].hash == '0577bec2387ee131e1ccf336adcc172224d3f6f9'
def mine(_type):
p = psutil.Process(os.getpid())
dt1 = datetime(2017, 1, 1)
dt2 = datetime(2017, 7, 1)
all_commits = []
start = datetime.now()
for commit in RepositoryMining('test-repos/hadoop',
since=dt1,
to=dt2).traverse_commits():
memory = p.memory_info()[0] / (2 ** 20)
all_commits.append(memory)
h = commit.author.name
if _type == 0:
continue
for mod in commit.modifications:
dd = mod.diff
if _type == 1:
continue
if mod.filename.endswith('.java'):
cc = mod.complexity
end = datetime.now()
diff = end - start
return diff, all_commits
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 discard_undesired_fixing_commits(self, commits: List[str]):
"""
Given a list of commits, discard commits that do not modify at least one Ansible file.
Note, the update occurs in-place. That is, the original list is updated.
Parameters
----------
commits : List[str]
List of commit hash
"""
# get a sorted list of commits in ascending order of date
self.sort_commits(commits)
for commit in RepositoryMining(self.path_to_repo,
from_commit=commits[0], # first commit in commits
to_commit=commits[-1], # last commit in commits
only_in_branch=self.branch).traverse_commits():
# if none of the modified files is a Ansible file, then discard the commit
if not any(modified_file.change_type == ModificationType.MODIFY and filters.is_ansible_file(
modified_file.new_path) for modified_file in commit.modifications):
if commit.hash in commits:
commits.remove(commit.hash)
def distinct_dev_count(self, path_to_repo: str, filepath: str,
from_commit: str = None, to_commit: str = None):
"""
Return the cumulative number of distinct developers contributed to
the file up to the indicated commit.
:path_to_repo: path to a single repo
:commit_hash: the SHA of the commit to stop counting. If None,
the SHA is the latest commit SHA
:filepath: the path to the file to count for. E.g. 'doc/README.md'
:return: int number of distinct developers contributing to the file
"""
filepath = str(Path(filepath))
developers = set()
for commit in RepositoryMining(path_to_repo, from_commit=from_commit,
to_commit=to_commit,
reversed_order=True).traverse_commits():
for modified_file in commit.modifications:
if filepath in (modified_file.new_path,
modified_file.old_path):
developers.add(commit.author.email.strip())
if modified_file.change_type == ModificationType.RENAME:
filepath = str(Path(modified_file.old_path))
break
return len(developers)
def commits_count(self, path_to_repo: str, filepath: str,
from_commit: str = None, to_commit: str = None):
"""
Return the number of commits made to a file from the first commit to
the one identified by commit_hash.
:path_to_repo: path to a single repo
:commit_hash: the SHA of the commit to stop counting. If None, the
analysis starts from the latest commit
:filepath: the path to the file to count for. E.g. 'doc/README.md'
:return: int number of commits made to the file
"""
filepath = str(Path(filepath))
count = 0
for commit in RepositoryMining(path_to_repo, from_commit=from_commit,
to_commit=to_commit,
reversed_order=True).traverse_commits():
for modified_file in commit.modifications:
if filepath in (modified_file.new_path,
modified_file.old_path):
count += 1
if modified_file.change_type == ModificationType.RENAME:
filepath = str(Path(modified_file.old_path))
break
return count
def mine(_type):
p = psutil.Process(os.getpid())
dt1 = datetime(2015, 1, 1)
dt2 = datetime(2015, 6, 1)
all_commits = []
start = datetime.now()
for commit in RepositoryMining('test-repos/hadoop', since=dt1,
to=dt2).traverse_commits():
memory = p.memory_info()[0] / (2**20)
all_commits.append(memory)
h = commit.author.name
if _type == 0:
continue
for mod in commit.modifications:
a = mod.old_path
if _type == 2:
dd = mod.diff
end = datetime.now()
diff = end - start
return diff, all_commits
def test_filepath_with_to():
dt = datetime(2018, 6, 6)
assert len(
list(
RepositoryMining(path_to_repo='test-repos/szz',
filepath='myfolder/A.java',
to=dt).traverse_commits())) == 4
def test_fixes_configuration_data():
for commit in RepositoryMining(
path_to_repo='https://github.com/iiab/iiab/',
only_commits=['25702f4e1d39965b54dec0e48bda18e8225e01d7'
]).traverse_commits():
assert AnsibleFixingCommitClassifier(
commit).fixes_configuration_data()
def test_no_filters():
lc = list(RepositoryMining('test-repos/git-4/').traverse_commits())
assert len(lc) == 3
assert lc[0].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
assert lc[1].hash == '375de7a8275ecdc0b28dc8de2568f47241f443e9'
assert lc[2].hash == 'b8c2be250786975f1c6f47e96922096f1bb25e39'
def test_only_in_main_branch():
lc = list(RepositoryMining('test-repos/branches_not_merged').traverse_commits())
assert len(lc) == 3
assert lc[0].hash == '04b0af7b53c2a0095e98951571aa41c2e0e0dbec'
assert lc[1].hash == 'e51421e0beae6a3c20bdcdfc21066e05db675e03'
assert lc[2].hash == 'b197ef4f0b4bc5b7d55c8949ecb1c861731f0b9d'
def test_mod_with_file_types_no_extension():
lc = list(
RepositoryMining(
'test-repos/git-4/',
only_modifications_with_file_types=['.py']).traverse_commits())
assert len(lc) == 0
def test_from_and_to_commit_with_merge_commit():
commits = RepositoryMining(
'test-repos/pydriller',
from_commit="015f7144641a418f6a9fae4d024286ec17fd7ce8",
to_commit="01d2f2fbeb6980cc5568825d008017ca8ca767d6").traverse_commits(
)
assert len(list(commits)) == 3
def test_filepath_with_since():
since = datetime(2018, 6, 6)
assert len(
list(
RepositoryMining(path_to_repo='test-repos/test5',
filepath='myfolder/A.java',
since=since).traverse_commits())) == 10
def test_fixes_service():
for commit in RepositoryMining(
path_to_repo='https://github.com/iiab/iiab/',
only_commits=['e7872a2a9da875e47e29c4bb21771c12104cd68e'
]).traverse_commits():
assert AnsibleFixingCommitClassifier(commit).fixes_service()
def label(self) -> Generator[FailureProneFile, None, None]:
"""
For each FixedFile object, yield a FailureProneFile object for each commit between the FixedFile's
bug-introducing-commit and its fixing-commit.
`Note:` make sure to run the method ``get_fixed_files`` before.
Yields
------
FailureProneFile
A FailureProneFile object.
"""
if not (self.fixing_commits or self.fixed_files):
return
labeling = dict()
for file in self.fixed_files:
labeling.setdefault(file.filepath, list()).append(file)
for commit in RepositoryMining(self.path_to_repo,
from_commit=self.fixing_commits[-1],
to_commit=self.commit_hashes[0],
order='reverse').traverse_commits():
for files in labeling.values():
for file in files:
idx_fic = self.commit_hashes.index(file.fic)
idx_bic = self.commit_hashes.index(file.bic)
idx_commit = self.commit_hashes.index(commit.hash)
if idx_fic > idx_commit >= idx_bic:
yield FailureProneFile(filepath=file.filepath,
commit=commit.hash,
fixing_commit=file.fic)
if idx_commit == idx_bic and file.filepath in labeling:
if file in labeling[file.filepath]:
labeling[file.filepath].remove(file)
# Handle file renaming
for modified_file in commit.modifications:
filepath = modified_file.new_path
for file in list(labeling.get(filepath, list())):
if self.commit_hashes.index(
file.fic) > self.commit_hashes.index(
commit.hash) >= self.commit_hashes.index(
file.bic):
if modified_file.change_type == ModificationType.ADD:
if filepath in labeling and file in labeling[
filepath]:
labeling[filepath].remove(file)
elif modified_file.change_type == ModificationType.RENAME:
file.filepath = modified_file.old_path
break
def test_between_dates():
list_commits = list(
RepositoryMining('test-repos/different_files', since=dt1,
to=dt2).traverse_commits())
assert len(list_commits) == 2
assert list_commits[0].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
assert list_commits[1].hash == '375de7a8275ecdc0b28dc8de2568f47241f443e9'
def test_multiple_repos_with_tags():
from_tag = 'tag2'
to_tag = 'tag3'
repos = ['test-repos/tags', 'test-repos/tags', 'test-repos/tags']
lc = list(
RepositoryMining(path_to_repo=repos, from_tag=from_tag,
to_tag=to_tag).traverse_commits())
assert len(lc) == 9
def test_data_changed():
for commit in RepositoryMining(
path_to_repo='https://github.com/iiab/iiab/',
only_commits=[
'9272b34b196d9010679157e493e775edca1daa13',
'25702f4e1d39965b54dec0e48bda18e8225e01d7'
]).traverse_commits():
assert AnsibleFixingCommitClassifier(commit).data_changed()
def setUpClass(cls) -> None:
# Create a fake pydriller.commit.Commit for test
cls.commit_obj = list(
RepositoryMining(
path_to_repo=
'https://github.com/stefanodallapalma/test-github-apis',
only_commits=['c9ada15de53d048f4d8e74d12bea62174bc0f957'
]).traverse_commits())[0]
def test_between_dates():
list_commits = list(
RepositoryMining('test-repos/git-4/', since=dt1,
to=dt2).traverse_commits())
assert 2 == len(list_commits)
assert 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5' == list_commits[0].hash
assert '375de7a8275ecdc0b28dc8de2568f47241f443e9' == list_commits[1].hash
def test_should_visit_ascendent_order():
lc = list(RepositoryMining('test-repos/small_repo').traverse_commits())
assert len(lc) == 5
assert lc[0].hash == 'a88c84ddf42066611e76e6cb690144e5357d132c'
assert lc[1].hash == '6411e3096dd2070438a17b225f44475136e54e3a'
assert lc[2].hash == '09f6182cef737db02a085e1d018963c7a29bde5a'
assert lc[3].hash == '1f99848edadfffa903b8ba1286a935f1b92b2845'
assert lc[4].hash == 'da39b1326dbc2edfe518b90672734a08f3c13458'
