def _parse_commit_log(repo_path):
from vbench.git import parser, _convert_timezones
from pandas import Series
git_cmd = "git --git-dir=%s/.git --work-tree=%s " % (repo_path, repo_path)
githist = git_cmd + (
"log --graph --pretty=format:" '"::%h::%cd::%s::%an" > githist.txt'
)
os.system(githist)
githist = open("githist.txt").read()
os.remove("githist.txt")
shas = []
timestamps = []
messages = []
authors = []
for line in githist.split("\n"):
if "*" not in line.split("::")[0]: # skip non-commit lines
continue
_, sha, stamp, message, author = line.split("::", 4)
# parse timestamp into datetime object
stamp = parser.parse(stamp)
shas.append(sha)
timestamps.append(stamp)
messages.append(message)
authors.append(author)
# to UTC for now
timestamps = _convert_timezones(timestamps)
shas = Series(shas, timestamps)
messages = Series(messages, shas)
timestamps = Series(timestamps, shas)
authors = Series(authors, shas)
return shas[::-1], messages[::-1], timestamps[::-1], authors[::-1]
def _parse_commit_log(this,repo_path,base_commit=None):
from vbench.git import parser, _convert_timezones
from pandas import Series
git_cmd = 'git --git-dir=%s/.git --work-tree=%s ' % (repo_path, repo_path)
githist = git_cmd + ('log --graph --pretty=format:'+
'\"::%h::%cd::%s::%an\"'+
('%s..' % base_commit)+
'> githist.txt')
os.system(githist)
githist = open('githist.txt').read()
os.remove('githist.txt')
shas = []
timestamps = []
messages = []
authors = []
for line in githist.split('\n'):
if '*' not in line.split("::")[0]: # skip non-commit lines
continue
_, sha, stamp, message, author = line.split('::', 4)
# parse timestamp into datetime object
stamp = parser.parse(stamp)
shas.append(sha)
timestamps.append(stamp)
messages.append(message)
authors.append(author)
# to UTC for now
timestamps = _convert_timezones(timestamps)
shas = Series(shas, timestamps)
messages = Series(messages, shas)
timestamps = Series(timestamps, shas)
authors = Series(authors, shas)
return shas[::-1], messages[::-1], timestamps[::-1], authors[::-1]
def _parse_commit_log(repo_path):
from vbench.git import parser, _convert_timezones
from pandas import Series
git_cmd = "git --git-dir=%s/.git --work-tree=%s " % (repo_path, repo_path)
githist = git_cmd + ("log --graph --pretty=format:" '"::%h::%cd::%s::%an" > githist.txt')
os.system(githist)
githist = open("githist.txt").read()
os.remove("githist.txt")
shas = []
timestamps = []
messages = []
authors = []
for line in githist.split("\n"):
if "*" not in line.split("::")[0]: # skip non-commit lines
continue
_, sha, stamp, message, author = line.split("::", 4)
# parse timestamp into datetime object
stamp = parser.parse(stamp)
shas.append(sha)
timestamps.append(stamp)
messages.append(message)
authors.append(author)
# to UTC for now
timestamps = _convert_timezones(timestamps)
shas = Series(shas, timestamps)
messages = Series(messages, shas)
timestamps = Series(timestamps, shas)
authors = Series(authors, shas)
return shas[::-1], messages[::-1], timestamps[::-1], authors[::-1]