def sort_bug_reports_by_commit_date(bug_reports):
commit_dates = []
for index, commit in enumerate(tqdm(bug_reports)):
sha = bug_reports[commit]['commit']['metadata']['sha'].replace('commit ', '').strip()
commit_date = convert_commit_date(
bug_reports[commit]['commit']['metadata']['date'].replace('Date:', '').strip())
commit_dates.append((sha, commit_date))
sorted_commit_dates = sorted(commit_dates, key=itemgetter(1))
sorted_commits = [commit_date[0] for commit_date in sorted_commit_dates]
return sorted_commits
def fix_commit_metadata(data):
"""Trim metadata fields, and add 'timestamp' from 'date'
Assumes that if 'timestamp' exists, then there is no need
for fixes.
Parameters
----------
data : dict | OrderedDict
The combined bug report and repository information from the
JSON file.
Returns
-------
dict | OrderedDict
Changed and augmented data.
Side effects
------------
Changes its input.
"""
n_skipped = 0
for commit in tqdm(data):
if 'timestamp' in data[commit]['commit']['metadata']:
n_skipped = n_skipped + 1
continue
trim_commit_info(data[commit]['commit']['metadata'])
data[commit]['commit']['metadata']['timestamp'] = \
datetime_to_timestamp(convert_commit_date(
data[commit]['commit']['metadata']['date']
))
print('%d / %d skipped: had already "timestamp" field in commit metadata' %
(n_skipped, len(data)),
file=sys.stderr)
return data