Files
@ 24e1099e4f29
Branch filter:
Location: kallithea/scripts/shortlog.py - annotation
24e1099e4f29
1.0 KiB
text/x-python
py3: make get_current_authuser handle missing tg context consistently and explicitly
tg context handling ends up using
tg.support.registry.StackedObjectProxy._current_obj for attribute access ...
which if no context has been pushed will end up in:
raise TypeError(
'No object (name: %s) has been registered for this '
'thread' % self.____name__)
utils2.get_current_authuser used code like:
if hasattr(tg.tmpl_context, 'authuser'):
Python 2 hasattr will call __getattr__ and return False if it throws any
exception. (It would thus catch the TypeError and silently fall through to use
the default user None.) This hasattr behavior is confusing and hard to use
correctly. Here, it was used incorrectly. It has been common practice to work
around by using something like:
getattr(x, y, None) is not None
Python 3 hasattr fixed this flaw and only catches AttributeError. The TypeError
would thus (rightfully) be propagated. That is a change that must be handled
when introducing py3 support.
The get_current_authuser code could more clearly and simple and py3-compatible
be written as:
return getattr(tmpl_context, 'authuser', None)
- but then we also have to handle the TypeError explicitly ... which we are
happy to do.
tg context handling ends up using
tg.support.registry.StackedObjectProxy._current_obj for attribute access ...
which if no context has been pushed will end up in:
raise TypeError(
'No object (name: %s) has been registered for this '
'thread' % self.____name__)
utils2.get_current_authuser used code like:
if hasattr(tg.tmpl_context, 'authuser'):
Python 2 hasattr will call __getattr__ and return False if it throws any
exception. (It would thus catch the TypeError and silently fall through to use
the default user None.) This hasattr behavior is confusing and hard to use
correctly. Here, it was used incorrectly. It has been common practice to work
around by using something like:
getattr(x, y, None) is not None
Python 3 hasattr fixed this flaw and only catches AttributeError. The TypeError
would thus (rightfully) be propagated. That is a change that must be handled
when introducing py3 support.
The get_current_authuser code could more clearly and simple and py3-compatible
be written as:
return getattr(tmpl_context, 'authuser', None)
- but then we also have to handle the TypeError explicitly ... which we are
happy to do.
30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 0a277465fddf 2786730e56e0 30e3d0a14f09 0a277465fddf 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 30e3d0a14f09 | #!/usr/bin/env python2
# -*- coding: utf-8 -*-
"""
Kallithea script for generating a quick overview of contributors and their
commit counts in a given revision set.
"""
import argparse
import os
from collections import Counter
from . import contributor_data
def main():
parser = argparse.ArgumentParser(description='Generate a list of committers and commit counts.')
parser.add_argument('revset',
help='revision set specifying the commits to count')
args = parser.parse_args()
repo_entries = [
(contributor_data.name_fixes.get(name) or contributor_data.name_fixes.get(name.rsplit('<', 1)[0].strip()) or name).rsplit('<', 1)[0].strip()
for name in (line.strip()
for line in os.popen("""hg log -r '%s' -T '{author}\n'""" % args.revset).readlines())
]
counter = Counter(repo_entries)
for name, count in counter.most_common():
if name == '':
continue
print('%4s %s' % (count, name))
if __name__ == '__main__':
main()
|