def clone_git_repos(repo_url_list: list):
for item in repo_url_list:
clone_url = item + ".git"
print(clone_url)
app_name = item.split('/')[-1]
local_repo_path = config.APM_REPO_PATH + app_name
shell_util.run_command("git clone {} '{}'".format(
clone_url, local_repo_path))
def get_methods_or_exprs_of_file(file_path: str, is_java_file):
xml_name = file_util.generate_random_file_name_with_extension('xml')
methods = []
xml_p = None
try:
shell_util.run_command("srcml '{}' -o {}".format(file_path, xml_name))
xml_p = Path(xml_name)
xml_bytes = xml_p.read_bytes()
methods = get_methods_exprs_of_xml_bytes(xml_bytes, is_java_file)
finally:
xml_p.unlink()
return methods
def get_java_kotlin_loc(path: str, commit_id=None):
if commit_id is None:
output = shell_util.run_command(
CLOC_COMMAND +
" --include-lang=Java,Kotlin --not-match-f='^[Mm]ock|[Mm]ock$|.*[Tt]est.*' '{}'"
.format(path))
else:
output = shell_util.run_command(
CLOC_COMMAND +
" --include-lang=Java,Kotlin --not-match-f='^[Mm]ock|[Mm]ock$|.*[Tt]est.*' '{}' {}"
.format(path, commit_id),
cwd=path)
pattern = '.*SUM.*'
m = re.search(pattern, output)
result = LOC()
if m is not None:
line = m.group(0)
result = _convert_cloc_line_to_object(line)
return result
def get_code_churn_between_commits(path: str, old_commit, new_commit):
output = shell_util.run_command(
"git diff --shortstat {} {} -- '*.java' | head -n 1".format(
old_commit, new_commit),
cwd=path)
added_count = 0
deleted_count = 0
added_match = insertion_pattern.search(output)
if added_match:
added_count = int(added_match.group(1))
deleted_match = deletion_pattern.search(output)
if deleted_match:
deleted_count = int(deleted_match.group(1))
return added_count + deleted_count
def get_file_loc_diff(old_path: str, new_path: str):
output = shell_util.run_command(
CLOC_COMMAND +
" --diff --diff-timeout 1000 '{}' '{}'".format(old_path, new_path))
if file_util.is_java_file(old_path):
pattern = '.*Java(\s.*){4}'
else:
pattern = '.*Kotlin(\s.*){4}'
m = re.search(pattern, output)
same = LOC()
modified = LOC()
added = LOC()
removed = LOC()
if m is not None:
lines = m.group(0).split('\n')
lines.pop(0)
for line in lines:
line_type = line.split()[0]
loc_detail = _convert_cloc_line_to_object(line)
if line_type == 'same':
same = loc_detail
elif line_type == 'modified':
modified = loc_detail
elif line_type == 'added':
added = loc_detail
elif line_type == 'removed':
removed = loc_detail
return {
'same': same,
'modified': modified,
'added': added,
'removed': removed
}
def get_repo_age_str(path: str):
output = shell_util.run_command(
"git log --reverse --pretty=oneline --format='%ar' | head -n 1 | LC_ALL=C sed 's/ago//' | tr -d ' '",
cwd=path)
return output
def get_commits_num(path: str):
output = shell_util.run_command(
"git log --oneline $commit | wc -l | tr -d ' '", cwd=path)
return int(output)
def get_files_num(path: str):
output = shell_util.run_command("git ls-files | wc -l | tr -d ' '",
cwd=path)
return int(output)
def get_default_branch(path: str):
output = shell_util.run_command(
"git symbolic-ref refs/remotes/origin/HEAD | sed 's@^refs/remotes/origin/@@' | tr -d ' '",
cwd=path)
return output.strip()
def get_authors_num(path: str):
output = shell_util.run_command(
"git log --format='%aN' | sort -u | wc -l | tr -d ' '", cwd=path)
return int(output)
def get_first_commit_date(path: str):
output = shell_util.run_command(
"git log --reverse --pretty='format: %ai' | head -n 1", cwd=path)
datetime_str = output.strip()
return datetime.strptime(datetime_str, '%Y-%m-%d %H:%M:%S %z')