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_only_in_branches():
# by default, only analyze master
assert len(list(
RepositoryMining('test-repos/test8/').traverse_commits())) == 3
# only analyze b2
assert len(
list(
RepositoryMining('test-repos/test8/',
only_in_branch='b2').traverse_commits())) == 4
# only analyze b1
assert len(
list(
RepositoryMining('test-repos/test8/',
only_in_branch='b1').traverse_commits())) == 5
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_between_dates_reversed():
# logging.basicConfig(format='%(asctime)s - %(levelname)s - %(message)s', level=logging.INFO)
lc = list(RepositoryMining('test-repos/git-4/', since=dt1, to=dt2, reversed_order=True).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == '375de7a8275ecdc0b28dc8de2568f47241f443e9'
assert lc[1].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
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_between_dates_without_timezone():
dt1 = datetime(2016, 10, 8, 21, 0, 0)
dt2 = datetime(2016, 10, 8, 21, 59, 0)
list_commits = list(RepositoryMining('test-repos/git-4/', 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_only_in_main_branch():
lc = list(RepositoryMining('test-repos/git-5/').traverse_commits())
assert len(lc) == 5
assert lc[0].hash == '4a17f31c0d1285477a3a467d0bc3cb38e775097d'
assert lc[1].hash == 'ff663cf1931a67d5e47b75fc77dcea432c728052'
assert lc[2].hash == 'fa8217c324e7fb46c80e1ddf907f4e141449637e'
assert lc[3].hash == '5d9d79607d7e82b6f236aa29be4ba89a28fb4f15'
assert lc[4].hash == '377e0f474d70f6205784d0150ee0069a050c29ed'
def test_mod_with_file_types():
lc = list(
RepositoryMining(
'test-repos/git-4/',
only_modifications_with_file_types=['.java']).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
assert lc[1].hash == 'b8c2be250786975f1c6f47e96922096f1bb25e39'
def test_mod_with_file_types():
lc = list(
RepositoryMining(
'test-repos/git-7/',
only_modifications_with_file_types=['.java']).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == '5adbb71167e79ab6b974827e74c9da4d81977655'
assert lc[1].hash == '0577bec2387ee131e1ccf336adcc172224d3f6f9'
def test_between_dates_reversed():
lc = list(
RepositoryMining('test-repos/git-4/',
single='375de7a8275ecdc0b28dc8de2568f47241f443e9').
traverse_commits())
to_zone = timezone(timedelta(hours=-4))
dt = datetime(2016, 10, 8, 17, 57, 49, tzinfo=to_zone)
assert lc[0].author_date == dt
def test_one_timezone():
lc = list(
RepositoryMining('test-repos/git-2/',
single='29e929fbc5dc6a2e9c620069b24e2a143af4285f').
traverse_commits())
to_zone = timezone(timedelta(hours=2))
dt = datetime(2016, 4, 4, 13, 21, 25, tzinfo=to_zone)
assert lc[0].author_date == dt
def test_multiple_filters():
lc = list(
RepositoryMining('test-repos/git-5/',
only_no_merge=True).traverse_commits())
assert len(lc) == 4
assert lc[0].hash == '4a17f31c0d1285477a3a467d0bc3cb38e775097d'
assert lc[1].hash == 'ff663cf1931a67d5e47b75fc77dcea432c728052'
assert lc[2].hash == 'fa8217c324e7fb46c80e1ddf907f4e141449637e'
assert lc[3].hash == '377e0f474d70f6205784d0150ee0069a050c29ed'
def test_between_revisions():
from_tag = 'tag1'
to_tag = 'tag3'
lc = list(
RepositoryMining('test-repos/git-8/', from_tag=from_tag,
to_tag=to_tag).traverse_commits())
assert len(lc) == 5
assert '6bb9e2c6a8080e6b5b34e6e316c894b2ddbf7fcd' == lc[0].hash
assert 'f1a90b8d7b151ceefd3e3dfc0dc1d0e12b5f48d0' == lc[1].hash
assert '4638730126d40716e230c2040751a13153fb1556' == lc[2].hash
assert 'a26f1438bd85d6b22497c0e5dae003812becd0bc' == lc[3].hash
assert '627e1ad917a188a861c9fedf6e5858b79edbe439' == lc[4].hash
def test_only_commits():
# 4e669cb4f69245dc669e116517d80d038d8e0434
# 29e929fbc5dc6a2e9c620069b24e2a143af4285f
# 8986af2a679759e5a15794f6d56e6d46c3f302f1
# 8169f76a3d7add54b4fc7bca7160d1f1eede6eda
# 168b3aab057ed61a769acf336a4ef5e64f76c9fd
lc = list(
RepositoryMining('test-repos/git-10/',
only_commits=[
"4e669cb4f69245dc669e116517d80d038d8e0434"
]).traverse_commits())
assert len(lc) == 1
assert lc[0].hash == "4e669cb4f69245dc669e116517d80d038d8e0434"
lc = list(
RepositoryMining('test-repos/git-10/',
only_commits=[
"4e669cb4f69245dc669e116517d80d038d8e0434",
"8986af2a679759e5a15794f6d56e6d46c3f302f1"
]).traverse_commits())
assert len(lc) == 2
assert lc[0].hash == "8986af2a679759e5a15794f6d56e6d46c3f302f1"
assert lc[1].hash == "4e669cb4f69245dc669e116517d80d038d8e0434"
lc = list(
RepositoryMining('test-repos/git-10/',
only_commits=[
"4e669cb4f69245dc669e116517d80d038d8e0434",
"8986af2a679759e5a15794f6d56e6d46c3f302f1",
"29e929fbc5dc6a2e9c620069b24e2a143af4285f"
]).traverse_commits())
assert len(lc) == 3
assert lc[0].hash == "8986af2a679759e5a15794f6d56e6d46c3f302f1"
assert lc[1].hash == "29e929fbc5dc6a2e9c620069b24e2a143af4285f"
assert lc[2].hash == "4e669cb4f69245dc669e116517d80d038d8e0434"
lc = list(
RepositoryMining('test-repos/git-10/',
only_commits=["fake hash"]).traverse_commits())
assert len(lc) == 0
only_commits = len(
list(
RepositoryMining('test-repos/git-10/',
only_commits=[
"4e669cb4f69245dc669e116517d80d038d8e0434",
"29e929fbc5dc6a2e9c620069b24e2a143af4285f",
"8986af2a679759e5a15794f6d56e6d46c3f302f1",
"8169f76a3d7add54b4fc7bca7160d1f1eede6eda",
"168b3aab057ed61a769acf336a4ef5e64f76c9fd"
]).traverse_commits()))
total_commits = len(
list(RepositoryMining('test-repos/git-10/').traverse_commits()))
assert total_commits == only_commits
def test_multiple_filters_exceptions():
from_commit = '6411e3096dd2070438a17b225f44475136e54e3a'
from_tag = 'v1.4'
with pytest.raises(Exception):
for commit in RepositoryMining('test-repos/test1/',
from_commit=from_commit,
from_tag=from_tag).traverse_commits():
print(commit.hash)
with pytest.raises(Exception):
for commit in RepositoryMining(
'test-repos/test1/', since=dt2,
from_commit=from_commit).traverse_commits():
print(commit.hash)
with pytest.raises(Exception):
for commit in RepositoryMining('test-repos/test1/',
since=dt2,
from_tag=from_tag).traverse_commits():
print(commit.hash)
with pytest.raises(Exception):
for commit in RepositoryMining('test-repos/test1/',
to=dt2,
to_tag=from_tag).traverse_commits():
print(commit.hash)
with pytest.raises(Exception):
for commit in RepositoryMining('test-repos/test1/',
single=from_commit,
to=dt2,
to_tag=from_tag).traverse_commits():
print(commit.hash)
with pytest.raises(Exception):
for commit in RepositoryMining('test-repos/test1/',
to_commit=from_commit,
to=dt2).traverse_commits():
print(commit.hash)
def repository_mining_st(path, since, to):
return list(RepositoryMining(path, since=since, to=to).traverse_commits())
def lc(request):
reversed = request.param
yield list(
RepositoryMining('test-repos/git-4',
reversed_order=reversed).traverse_commits())
def test_only_in_branch_not_exist():
with pytest.raises(Exception):
list(
RepositoryMining('test-repos/git-5/',
only_in_branch='branch7').traverse_commits())
def repository_mining_cc(path, from_commit, to_commit):
return list(
RepositoryMining(path, from_commit=from_commit,
to_commit=to_commit).traverse_commits())
def repository_mining_tt(path, from_tag, to_tag):
return list(
RepositoryMining(path, from_tag=from_tag,
to_tag=to_tag).traverse_commits())
def test_between_dates():
list_commits = list(RepositoryMining('test-repos/git-4/', since=dt1, to=dt2).traverse_commits())
assert len(list_commits) == 2
assert list_commits[0].hash == 'a1b6136f978644ff1d89816bc0f2bd86f6d9d7f5'
assert list_commits[1].hash == '375de7a8275ecdc0b28dc8de2568f47241f443e9'