def _fmt_cmd_for_log(cmd):
"""
Join the git_cmd, quoting individual segments first so that
it's relatively easy to see if there were whitespace issues or
not.
"""
return ' '.join(['"%s"' % util.utf8(seg) for seg in cmd])
def _write_to_stdout(str_or_unicode):
"""
Write `str_or_unicode` to stdout, encoding to utf-8 if it's
unicode.
No extra newline is written.
"""
sys.stdout.write(util.utf8(str_or_unicode))
def _write_to_stdout(str_or_unicode):
"""
Write `str_or_unicode` to stdout, encoding to utf-8 if it's
unicode.
No extra newline is written.
"""
sys.stdout.write(util.utf8(str_or_unicode))
def test_git_cmd_returns_std_out():
with fleeting_repo() as test_repo:
commit_msg = 'I remain committed.'
test_repo.commit([('test.txt', 'testing\n')],
commit_msg=commit_msg)
with git_cmd(["log"], cwd=test_repo.repo_root) as output_f:
ok_(commit_msg in util.utf8(output_f.read()))
def _wrap(fname, log_entry):
with tempfile.TemporaryFile() as fil:
fil.write('\0'.join(
[util.utf8(x) for x in [fname] + list(log_entry)]))
fil.flush()
fil.seek(0)
with util.wrap_popen(list_or_str, stdin_fil=fil) as out_fil:
return out_fil.read()
def _wrap(fname, log_entry):
with tempfile.TemporaryFile() as fil:
fil.write('\0'.join([util.utf8(x)
for x
in [fname] + list(log_entry)]))
fil.flush()
fil.seek(0)
with util.wrap_popen(list_or_str, stdin_fil=fil) as out_fil:
return out_fil.read()
def _parse_log_entry(raw_log_entry):
"""
Parse a single git log entry into a LogEntry, or return None
if it can't be parsed.
"""
# A note on the encodings. Git doesn't give us a way to get at
# the encodings of the files / diffs (short of .gitattributes,
# which has to be set by the original producers of the repo). We
# assume that the encoding is UTF-8, and just replace everything
# else with that lovely question mark thing, to do the parsing /
# manipulation, and then convert that utf-8 to unicode for the
# returned LogEntry.
utf8_log_entry = util.utf8(raw_log_entry)
# attempt to split the header from the diff.
split_log_entry = _split_entry_header(utf8_log_entry)
if split_log_entry is None:
return None
header_lines, diff_lines = split_log_entry
diff = '\n'.join(diff_lines)
if not diff.strip():
log.debug("Diff appeared to be empty.")
return None
author = _parse_header('Author: ', header_lines)
if not author:
log.debug("Could not parse author.")
return None
parsed_author = parse_name_and_email(author)
if not parsed_author:
log.debug("Could not parse author name / email.")
return None
author_name, author_email = parsed_author
commit = _parse_header('commit ', header_lines)
if not commit:
log.debug("Could not parse commit.")
return None
log_msg = '\n'.join(_parse_log_msg(header_lines))
return LogEntry(author_name=util.uc(author_name),
author_email=util.uc(author_email),
commit=util.uc(commit),
log_msg=util.uc(log_msg),
diff=util.uc(diff),
raw_log=raw_log_entry)
def _parse_log_entry(raw_log_entry):
"""
Parse a single git log entry into a LogEntry, or return None
if it can't be parsed.
"""
# A note on the encodings. Git doesn't give us a way to get at
# the encodings of the files / diffs (short of .gitattributes,
# which has to be set by the original producers of the repo). We
# assume that the encoding is UTF-8, and just replace everything
# else with that lovely question mark thing, to do the parsing /
# manipulation, and then convert that utf-8 to unicode for the
# returned LogEntry.
utf8_log_entry = util.utf8(raw_log_entry)
# attempt to split the header from the diff.
split_log_entry = _split_entry_header(utf8_log_entry)
if split_log_entry is None:
return None
header_lines, diff_lines = split_log_entry
diff = '\n'.join(diff_lines)
if not diff.strip():
log.debug("Diff appeared to be empty.")
return None
author = _parse_header('Author: ', header_lines)
if not author:
log.debug("Could not parse author.")
return None
parsed_author = parse_name_and_email(author)
if not parsed_author:
log.debug("Could not parse author name / email.")
return None
author_name, author_email = parsed_author
commit = _parse_header('commit ', header_lines)
if not commit:
log.debug("Could not parse commit.")
return None
log_msg = '\n'.join(_parse_log_msg(header_lines))
return LogEntry(author_name=util.uc(author_name),
author_email=util.uc(author_email),
commit=util.uc(commit),
log_msg=util.uc(log_msg),
diff=util.uc(diff),
raw_log=raw_log_entry)
def _validate_parsed_entry(parsed_entry, commit_info):
ok_(parsed_entry)
ok_(util.uc(commit_info.commit_msg) in parsed_entry.log_msg)
ok_(commit_info.file_body in util.utf8(parsed_entry.diff))
eq_(util.uc(commit_info.author_name), parsed_entry.author_name)
eq_(util.uc(commit_info.author_email), parsed_entry.author_email)
TEST_COMMIT_ONE = TestCommitInfo(file_body="test file body\n",
commit_msg="test commit msg",
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_TWO = TestCommitInfo(file_body="test second file body\n",
commit_msg="test second commit msg",
author_name="Testy Second McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_BODY = TestCommitInfo(
file_body=util.utf8(u"frosty the \u2603 was a kind soul\n"),
commit_msg="test commit msg",
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_MSG = TestCommitInfo(
file_body="hi unicode msg\n",
commit_msg=util.utf8(u"test commit \u2603 msg"),
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_AUTHOR = TestCommitInfo(
file_body="hi unicode author\n",
commit_msg="test commit",
def _validate_parsed_entry(parsed_entry, commit_info):
ok_(parsed_entry)
ok_(util.uc(commit_info.commit_msg) in parsed_entry.log_msg)
ok_(commit_info.file_body in util.utf8(parsed_entry.diff))
eq_(util.uc(commit_info.author_name), parsed_entry.author_name)
eq_(util.uc(commit_info.author_email), parsed_entry.author_email)
# purposes of testing log retrieval / parsing
TestCommitInfo = namedtuple('TestCommitInfo',
'file_body commit_msg author_name author_email')
TEST_COMMIT_ONE = TestCommitInfo(file_body="test file body\n",
commit_msg="test commit msg",
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_TWO = TestCommitInfo(file_body="test second file body\n",
commit_msg="test second commit msg",
author_name="Testy Second McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_BODY = TestCommitInfo(
file_body=util.utf8(u"frosty the \u2603 was a kind soul\n"),
commit_msg="test commit msg",
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_MSG = TestCommitInfo(
file_body="hi unicode msg\n",
commit_msg=util.utf8(u"test commit \u2603 msg"),
author_name="Testy McTesterson",
author_email="*****@*****.**")
TEST_COMMIT_UNICODE_AUTHOR = TestCommitInfo(
file_body="hi unicode author\n",
commit_msg="test commit",
author_name=util.utf8(u"Testy \u2603 McTesterson"),
author_email="*****@*****.**")
