def test_commit_in_master_branch():
gr = GitRepository('test-repos/git-2/')
assert gr.get_head().hash == '29e929fbc5dc6a2e9c620069b24e2a143af4285f'
gr.checkout('8986af2a679759e5a15794f6d56e6d46c3f302f1')
git_to_change_head = GitRepository('test-repos/git-2/')
commit = git_to_change_head.get_commit(
'8169f76a3d7add54b4fc7bca7160d1f1eede6eda')
assert commit.in_main_branch is False
commit = git_to_change_head.get_commit(
'168b3aab057ed61a769acf336a4ef5e64f76c9fd')
assert commit.in_main_branch is True
gr.reset()
assert gr.get_head().hash == '29e929fbc5dc6a2e9c620069b24e2a143af4285f'
def test_get_head():
gr = GitRepository('test-repos/test1/')
assert gr is not None
cs = gr.get_head()
assert cs is not None
assert cs.hash == 'da39b1326dbc2edfe518b90672734a08f3c13458'
assert cs.author_date.timestamp() == 1522164679
def test_checkout_consecutive_commits():
gr = GitRepository('test-repos/git-1/')
gr.checkout('a7053a4dcd627f5f4f213dc9aa002eb1caf926f8')
gr.checkout('f0dd1308bd904a9b108a6a40865166ee962af3d4')
gr.checkout('9e71dd5726d775fb4a5f08506a539216e878adbb')
files3 = gr.files()
assert len(files3) == 3
gr.reset()
def test_get_commit_from_tag():
gr = GitRepository('test-repos/test1/')
commit = gr.get_commit_from_tag('v1.4')
assert commit.hash == '09f6182cef737db02a085e1d018963c7a29bde5a'
with pytest.raises(IndexError):
gr.get_commit_from_tag('v1.5')
def test_get_commits_last_modified_lines_rename_simple():
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(
gr.get_commit('45ba0a61ccc448625bce0fea0301cf0c1ab32696'))
assert len(buggy_commits) == 1
assert 'e358878a00e78aca8366264d61a7319d00dd8186' in buggy_commits
def test_get_commits_last_modified_lines_simple():
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(
gr.get_commit('e6d3b38a9ef683e8184eac10a0471075c2808bbd'))
assert len(buggy_commits) == 1
assert '540c7f31c18664a38190fafb6721b5174ff4a166' in buggy_commits
def test_number_of_modifications():
gr = GitRepository('test-repos/git-1/')
commit = gr.get_commit('866e997a9e44cb4ddd9e00efe49361420aff2559')
assert commit.modifications[0].added == 62
assert commit.modifications[0].removed == 0
commit = gr.get_commit('d11dd6734ff4e60cac3a7b58d9267f138c9e05c7')
assert commit.modifications[0].added == 1
assert commit.modifications[0].removed == 1
def test_detail_rename():
gr = GitRepository('test-repos/git-1/')
commit = gr.get_commit('f0dd1308bd904a9b108a6a40865166ee962af3d4')
assert commit.author.name == "Maurício Aniche"
assert commit.author.email == "*****@*****.**"
assert commit.modifications[0].new_path == "Matricula.javax"
assert commit.modifications[0].old_path == "Matricula.java"
def test_get_commits_modified_file():
gr = GitRepository('test-repos/test1/')
commits = gr.get_commits_modified_file('file2.java')
assert len(commits) == 3
assert '09f6182cef737db02a085e1d018963c7a29bde5a' in commits
assert '6411e3096dd2070438a17b225f44475136e54e3a' in commits
assert 'a88c84ddf42066611e76e6cb690144e5357d132c' in commits
def test_get_commits_last_modified_lines_multiple():
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(gr.get_commit('9942ee9dcdd1103e5808d544a84e6bc8cade0e54'))
assert len(buggy_commits) == 3
assert '2eb905e5e7be414fd184d6b4f1571b142621f4de' in buggy_commits
assert '20a40688521c1802569e60f9d55342c3bfdd772c' in buggy_commits
assert '22505e97dca6f843549b3a484b3609be4e3acf17' in buggy_commits
def test_get_tagged_commits():
gr = GitRepository('test-repos/git-8/')
tagged_commits = gr.get_tagged_commits()
assert len(tagged_commits) == 3
assert '6bb9e2c6a8080e6b5b34e6e316c894b2ddbf7fcd' == tagged_commits[0]
assert '4638730126d40716e230c2040751a13153fb1556' == tagged_commits[1]
assert '627e1ad917a188a861c9fedf6e5858b79edbe439' == tagged_commits[2]
def test_filepahs():
gr = GitRepository('test-repos/test7')
c = gr.get_commit('f0f8aea2db50ed9f16332d86af3629ff7780583e')
mod0 = c.modifications[0]
assert mod0.filename == 'a.java'
assert mod0.new_path == str(Path('dir2/a.java'))
assert mod0.old_path == str(Path('dir2/a.java'))
def traverse_commits(self) -> Generator[Commit, None, None]:
"""
Analyze all the specified commits (all of them by default), returning
a generator of commits.
"""
for path_repo in self._conf.get('path_to_repos'):
# if it is a remote repo, clone it first in a temporary folder!
if self._is_remote(path_repo):
if self._conf.get('clone_repo_to'):
clone_folder = str(Path(self._conf.get('clone_repo_to')))
if not os.path.isdir(clone_folder):
raise Exception(
"Not a directory: {0}".format(clone_folder))
path_repo = self._clone_remote_repos(
clone_folder, path_repo)
else:
tmp_folder = tempfile.TemporaryDirectory()
path_repo = self._clone_remote_repos(
tmp_folder.name, path_repo)
git_repo = GitRepository(path_repo, self._conf)
self._conf.set_value("git_repo", git_repo)
self._conf.sanity_check_filters()
logger.info('Analyzing git repository in %s', git_repo.path)
# Get the commits that modified the filepath. In this case, we can not use
# git rev-list since it doesn't have the option --follow, necessary to follow
# the renames. Hence, we manually call git log instead
if self._conf.get('filepath') is not None:
self._conf.set_value(
'filepath_commits',
git_repo.get_commits_modified_file(
self._conf.get('filepath')))
# Gets only the commits that are tagged
if self._conf.get('only_releases'):
self._conf.set_value('tagged_commits',
git_repo.get_tagged_commits())
# Build the arguments to pass to git rev-list.
rev, kwargs = self._conf.build_args()
# Iterate over all the commits returned by git rev-list
for commit in git_repo.get_list_commits(rev, **kwargs):
logger.info('Commit #%s in %s from %s', commit.hash,
commit.committer_date, commit.author.name)
if self._conf.is_commit_filtered(commit):
logger.info('Commit #%s filtered', commit.hash)
continue
yield commit
# cleaning, this is necessary since GitPython issues on memory leaks
self._conf.set_value("git_repo", None)
git_repo.clear()
def test_modification_status():
gr = GitRepository('test-repos/git-1/')
commit = gr.get_commit('866e997a9e44cb4ddd9e00efe49361420aff2559')
assert ModificationType.ADD == commit.modifications[0].change_type
commit = gr.get_commit('57dbd017d1a744b949e7ca0b1c1a3b3dd4c1cbc1')
assert ModificationType.MODIFY == commit.modifications[0].change_type
commit = gr.get_commit('ffccf1e7497eb8136fd66ed5e42bef29677c4b71')
assert ModificationType.DELETE == commit.modifications[0].change_type
def test_tags():
gr = GitRepository('test-repos/git-8/')
commit = gr.get_commit_from_tag('tag1')
assert commit.hash == '6bb9e2c6a8080e6b5b34e6e316c894b2ddbf7fcd'
commit = gr.get_commit_from_tag('tag2')
assert commit.hash == '4638730126d40716e230c2040751a13153fb1556'
with pytest.raises(IndexError):
gr.get_commit_from_tag('tag4')
def test_merge_commits():
gr = GitRepository('test-repos/git-2/')
commit = gr.get_commit("168b3aab057ed61a769acf336a4ef5e64f76c9fd")
assert commit.merge is False
commit = gr.get_commit("8169f76a3d7add54b4fc7bca7160d1f1eede6eda")
assert commit.merge is False
commit = gr.get_commit("29e929fbc5dc6a2e9c620069b24e2a143af4285f")
assert commit.merge is True
def test_parent_commits():
gr = GitRepository('test-repos/git-5/')
merge_commit = gr.get_commit('5d9d79607d7e82b6f236aa29be4ba89a28fb4f15')
assert len(merge_commit.parents) == 2
assert 'fa8217c324e7fb46c80e1ddf907f4e141449637e' in merge_commit.parents
assert 'ff663cf1931a67d5e47b75fc77dcea432c728052' in merge_commit.parents
normal_commit = gr.get_commit('ff663cf1931a67d5e47b75fc77dcea432c728052')
assert len(normal_commit.parents) == 1
assert '4a17f31c0d1285477a3a467d0bc3cb38e775097d' in normal_commit.parents
def test_eq_commit():
gr = GitRepository('test-repos/git-11')
c1 = gr.get_commit('1734d6da01378bad3aade12b52bb4aa8954835dc')
c2 = gr.get_commit('2c1327f957ba3b2a5e86eaed097b0a425236719e')
c3 = gr.get_commit('1734d6da01378bad3aade12b52bb4aa8954835dc')
m1 = gr.get_commit('1734d6da01378bad3aade12b52bb4aa8954835dc'
'').modifications[0]
assert c1 == c3
assert c1 == c1
assert c1 != m1
assert c1 != c2
def traverse_commits(self) -> Generator[Commit, None, None]:
"""
Analyze all the specified commits (all of them by default), returning
a generator of commits.
"""
for path_repo in self._conf.get('path_to_repos'):
if self._is_remote(path_repo):
path_repo = self._clone_remote_repos(self._clone_folder(),
path_repo)
git_repo = GitRepository(path_repo, self._conf)
# saving the GitRepository object for further use
self._conf.set_value("git_repo", git_repo)
# when multiple repos are given in input, this variable will serve as a reminder
# of which one we are currently analyzing
self._conf.set_value('path_to_repo', path_repo)
# checking that the filters are set correctly
self._conf.sanity_check_filters()
logger.info('Analyzing git repository in %s', git_repo.path)
# Get the commits that modified the filepath. In this case, we can not use
# git rev-list since it doesn't have the option --follow, necessary to follow
# the renames. Hence, we manually call git log instead
if self._conf.get('filepath') is not None:
self._conf.set_value(
'filepath_commits',
git_repo.get_commits_modified_file(
self._conf.get('filepath')))
# Gets only the commits that are tagged
if self._conf.get('only_releases'):
self._conf.set_value('tagged_commits',
git_repo.get_tagged_commits())
# Build the arguments to pass to git rev-list.
rev, kwargs = self._conf.build_args()
# Iterate over all the commits returned by git rev-list
for commit in git_repo.get_list_commits(rev, **kwargs):
logger.info('Commit #%s in %s from %s', commit.hash,
commit.committer_date, commit.author.name)
if self._conf.is_commit_filtered(commit):
logger.info('Commit #%s filtered', commit.hash)
continue
yield commit
# cleaning, this is necessary since GitPython issues on memory leaks
self._conf.set_value("git_repo", None)
git_repo.clear()
def test_commit_in_master_branch(repo: GitRepository):
assert repo.get_head().hash == '29e929fbc5dc6a2e9c620069b24e2a143af4285f'
repo.checkout('8986af2a679759e5a15794f6d56e6d46c3f302f1')
git_to_change_head = GitRepository('test-repos/branches_merged')
commit = git_to_change_head.get_commit('8169f76a3d7add54b4fc7bca7160d1f1eede6eda')
assert commit.in_main_branch is False
repo.reset()
assert repo.get_head().hash == '29e929fbc5dc6a2e9c620069b24e2a143af4285f'
def test_get_commits_last_modified_lines_for_single_file():
gr = GitRepository('test-repos/test5/')
commit = gr.get_commit('0f726924f96621e4965039123098ba83e39ffba6')
buggy_commits = None
for mod in commit.modifications:
if mod.filename == 'A.java':
buggy_commits = gr.get_commits_last_modified_lines(commit, mod)
assert len(buggy_commits) == 1
assert 'e2ed043eb96c05ebde653a44ae733ded9ef90750' in buggy_commits
def test_get_commits_last_modified_lines_rename_simple_more_commits():
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(
gr.get_commit('04fadd3e68c58281db6cf15119f9769880ac1cbc'))
assert len(buggy_commits) == 2
assert '9b373199c270f9b24c37fee70f9e2b3ee9b816e3' in buggy_commits[
'A.java']
assert '9b373199c270f9b24c37fee70f9e2b3ee9b816e3' in buggy_commits[
'B.java']
def test_branches_from_commit():
gr = GitRepository('test-repos/git-1/')
commit = gr.get_commit('a997e9d400f742003dea601bb05a9315d14d1124')
assert len(commit.branches) == 1
assert 'b2' in commit.branches
commit = gr.get_commit('866e997a9e44cb4ddd9e00efe49361420aff2559')
assert len(commit.branches) == 2
assert 'master' in commit.branches
assert 'b2' in commit.branches
def test_get_commits_last_modified_lines_hyper_blame_unblamable(tmp_path):
p = tmp_path / "ignore.txt"
p.write_text("540c7f31c18664a38190fafb6721b5174ff4a166")
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(
gr.get_commit('e6d3b38a9ef683e8184eac10a0471075c2808bbd'),
hashes_to_ignore_path=str(p))
assert len(buggy_commits) == 0
def test_get_commits_last_modified_lines_hyper_blame_with_renaming():
gr = GitRepository('test-repos/test5/')
buggy_commits = gr.get_commits_last_modified_lines(
gr.get_commit('be0772cbaa2eba32bf97aae885199d1a357ddc93'))
assert len(buggy_commits) == 2
assert '9568d20856728304ab0b4d2d02fb9e81d0e5156d' in buggy_commits[
'A.java']
assert '9568d20856728304ab0b4d2d02fb9e81d0e5156d' in buggy_commits[
'H.java']
def test_get_commit():
gr = GitRepository('test-repos/test1/')
c = gr.get_commit('09f6182cef737db02a085e1d018963c7a29bde5a')
to_zone = timezone(timedelta(hours=1))
assert '09f6182cef737db02a085e1d018963c7a29bde5a' == c.hash
assert 'ishepard' == c.author.name
assert 'ishepard' == c.committer.name
assert datetime(2018, 3, 22, 10, 42, 3, tzinfo=to_zone).timestamp() == c.author_date.timestamp()
assert 1 == len(c.modifications)
assert 'Ooops file2' == c.msg
assert c.in_main_branch is True
def test_get_first_commit():
gr = GitRepository('test-repos/test1/')
c = gr.get_commit('a88c84ddf42066611e76e6cb690144e5357d132c')
to_zone = timezone(timedelta(hours=1))
assert 'a88c84ddf42066611e76e6cb690144e5357d132c' == c.hash
assert 'ishepard' == c.author.name
assert 'ishepard' == c.committer.name
assert datetime(2018,3,22,10,41,11,tzinfo=to_zone).timestamp() == c.author_date.timestamp()
assert 2 == len(c.modifications)
assert 'First commit adding 2 files' == c.msg
assert c.in_main_branch is True
def test_changed_methods():
gr = GitRepository("test-repos/diff")
# add a new method
mod = gr.get_commit(
'ea95227e0fd128aa69c7ab6a8ac485f72251b3ed').modifications[0]
assert len(mod.changed_methods) == 1
assert mod.changed_methods[
0].name == 'GitRepository::singleProjectThirdMethod'
# add 2 new methods
mod = gr.get_commit(
'd8eb8e80b671246a43c98d97b05f6d1c5ada14fb').modifications[0]
assert len(mod.changed_methods) == 2
# remove one method
mod = gr.get_commit(
'0c8f9fdec926785198b399a2c49adb5884aa952c').modifications[0]
assert len(mod.changed_methods) == 1
# add and remove one one method at different locations
mod = gr.get_commit(
'd8bb142c5616041b71cbfaa11eeb768d9a1a296e').modifications[0]
assert len(mod.changed_methods) == 2
# add and remove one one method at the same location
# this is equivalent to replacing a method - although we expect 2 methods
mod = gr.get_commit(
'9e9473d5ca310b7663e9df93c402302b6b7f24aa').modifications[0]
assert len(mod.changed_methods) == 2
# update a method
mod = gr.get_commit(
'b267a14e0503fdac36d280422f16360d1f661f12').modifications[0]
assert len(mod.changed_methods) == 1
# update and add a new method
mod = gr.get_commit(
'2489099dfd90edb99ddc2c82b62524b66c07c687').modifications[0]
assert len(mod.changed_methods) == 2
# update and delete methods
mod = gr.get_commit(
'5aebeb30e0238543a93e5bed806639481460cd9a').modifications[0]
assert len(mod.changed_methods) == 2
# delete 3 methods (test cleanup - revert the test file to its
# initial set of methods)
mod = gr.get_commit(
'9f6ddc2aac740a257af59a76860590cb8a84c77b').modifications[0]
assert len(mod.changed_methods) == 3
def test_checkout_with_commit_not_fully_merged_to_master():
gr = GitRepository('test-repos/git-9/')
gr.checkout('developing')
files1 = gr.files()
assert len(files1) == 2
gr.reset()
assert 4, "temp branch should be cleared." == len(gr.repo.branches)
files2 = gr.files()
assert len(files2) == 1
gr.checkout('developing')
files1 = gr.files()
assert len(files1) == 2
gr.reset()
def test_files():
gr = GitRepository('test-repos/test2')
all = gr.files()
assert len(all) == 8
assert str(Path('test-repos/test2/tmp1.py')) in all
assert str(Path('test-repos/test2/tmp2.py')) in all
assert str(Path('test-repos/test2/fold1/tmp3.py')) in all
assert str(Path('test-repos/test2/fold1/tmp4.py')) in all
assert str(Path('test-repos/test2/fold2/tmp5.py')) in all
assert str(Path('test-repos/test2/fold2/tmp6.py')) in all
assert str(Path('test-repos/test2/fold2/fold3/tmp7.py')) in all
assert str(Path('test-repos/test2/fold2/fold3/tmp8.py')) in all
