|
|
# stringutil.py - utility for generic string formatting, parsing, etc.
|
|
|
#
|
|
|
# Copyright 2005 K. Thananchayan <thananck@yahoo.com>
|
|
|
# Copyright 2005-2007 Olivia Mackall <olivia@selenic.com>
|
|
|
# Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com>
|
|
|
#
|
|
|
# This software may be used and distributed according to the terms of the
|
|
|
# GNU General Public License version 2 or any later version.
|
|
|
|
|
|
from __future__ import absolute_import
|
|
|
|
|
|
import ast
|
|
|
import codecs
|
|
|
import re as remod
|
|
|
import textwrap
|
|
|
import types
|
|
|
|
|
|
from ..i18n import _
|
|
|
from ..thirdparty import attr
|
|
|
|
|
|
from .. import (
|
|
|
encoding,
|
|
|
error,
|
|
|
pycompat,
|
|
|
)
|
|
|
|
|
|
# regex special chars pulled from https://bugs.python.org/issue29995
|
|
|
# which was part of Python 3.7.
|
|
|
_respecial = pycompat.bytestr(b'()[]{}?*+-|^$\\.&~# \t\n\r\v\f')
|
|
|
_regexescapemap = {ord(i): (b'\\' + i).decode('latin1') for i in _respecial}
|
|
|
regexbytesescapemap = {i: (b'\\' + i) for i in _respecial}
|
|
|
|
|
|
|
|
|
def reescape(pat):
|
|
|
"""Drop-in replacement for re.escape."""
|
|
|
# NOTE: it is intentional that this works on unicodes and not
|
|
|
# bytes, as it's only possible to do the escaping with
|
|
|
# unicode.translate, not bytes.translate. Sigh.
|
|
|
wantuni = True
|
|
|
if isinstance(pat, bytes):
|
|
|
wantuni = False
|
|
|
pat = pat.decode('latin1')
|
|
|
pat = pat.translate(_regexescapemap)
|
|
|
if wantuni:
|
|
|
return pat
|
|
|
return pat.encode('latin1')
|
|
|
|
|
|
|
|
|
def pprint(o, bprefix=False, indent=0, level=0):
|
|
|
"""Pretty print an object."""
|
|
|
return b''.join(pprintgen(o, bprefix=bprefix, indent=indent, level=level))
|
|
|
|
|
|
|
|
|
def pprintgen(o, bprefix=False, indent=0, level=0):
|
|
|
"""Pretty print an object to a generator of atoms.
|
|
|
|
|
|
``bprefix`` is a flag influencing whether bytestrings are preferred with
|
|
|
a ``b''`` prefix.
|
|
|
|
|
|
``indent`` controls whether collections and nested data structures
|
|
|
span multiple lines via the indentation amount in spaces. By default,
|
|
|
no newlines are emitted.
|
|
|
|
|
|
``level`` specifies the initial indent level. Used if ``indent > 0``.
|
|
|
"""
|
|
|
|
|
|
if isinstance(o, bytes):
|
|
|
if bprefix:
|
|
|
yield b"b'%s'" % escapestr(o)
|
|
|
else:
|
|
|
yield b"'%s'" % escapestr(o)
|
|
|
elif isinstance(o, bytearray):
|
|
|
# codecs.escape_encode() can't handle bytearray, so escapestr fails
|
|
|
# without coercion.
|
|
|
yield b"bytearray['%s']" % escapestr(bytes(o))
|
|
|
elif isinstance(o, list):
|
|
|
if not o:
|
|
|
yield b'[]'
|
|
|
return
|
|
|
|
|
|
yield b'['
|
|
|
|
|
|
if indent:
|
|
|
level += 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
for i, a in enumerate(o):
|
|
|
for chunk in pprintgen(
|
|
|
a, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
if i + 1 < len(o):
|
|
|
if indent:
|
|
|
yield b',\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
else:
|
|
|
yield b', '
|
|
|
|
|
|
if indent:
|
|
|
level -= 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
yield b']'
|
|
|
elif isinstance(o, dict):
|
|
|
if not o:
|
|
|
yield b'{}'
|
|
|
return
|
|
|
|
|
|
yield b'{'
|
|
|
|
|
|
if indent:
|
|
|
level += 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
for i, (k, v) in enumerate(sorted(o.items())):
|
|
|
for chunk in pprintgen(
|
|
|
k, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
yield b': '
|
|
|
|
|
|
for chunk in pprintgen(
|
|
|
v, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
if i + 1 < len(o):
|
|
|
if indent:
|
|
|
yield b',\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
else:
|
|
|
yield b', '
|
|
|
|
|
|
if indent:
|
|
|
level -= 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
yield b'}'
|
|
|
elif isinstance(o, set):
|
|
|
if not o:
|
|
|
yield b'set([])'
|
|
|
return
|
|
|
|
|
|
yield b'set(['
|
|
|
|
|
|
if indent:
|
|
|
level += 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
for i, k in enumerate(sorted(o)):
|
|
|
for chunk in pprintgen(
|
|
|
k, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
if i + 1 < len(o):
|
|
|
if indent:
|
|
|
yield b',\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
else:
|
|
|
yield b', '
|
|
|
|
|
|
if indent:
|
|
|
level -= 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
yield b'])'
|
|
|
elif isinstance(o, tuple):
|
|
|
if not o:
|
|
|
yield b'()'
|
|
|
return
|
|
|
|
|
|
yield b'('
|
|
|
|
|
|
if indent:
|
|
|
level += 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
for i, a in enumerate(o):
|
|
|
for chunk in pprintgen(
|
|
|
a, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
if i + 1 < len(o):
|
|
|
if indent:
|
|
|
yield b',\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
else:
|
|
|
yield b', '
|
|
|
|
|
|
if indent:
|
|
|
level -= 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
yield b')'
|
|
|
elif isinstance(o, types.GeneratorType):
|
|
|
# Special case of empty generator.
|
|
|
try:
|
|
|
nextitem = next(o)
|
|
|
except StopIteration:
|
|
|
yield b'gen[]'
|
|
|
return
|
|
|
|
|
|
yield b'gen['
|
|
|
|
|
|
if indent:
|
|
|
level += 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
last = False
|
|
|
|
|
|
while not last:
|
|
|
current = nextitem
|
|
|
|
|
|
try:
|
|
|
nextitem = next(o)
|
|
|
except StopIteration:
|
|
|
last = True
|
|
|
|
|
|
for chunk in pprintgen(
|
|
|
current, bprefix=bprefix, indent=indent, level=level
|
|
|
):
|
|
|
yield chunk
|
|
|
|
|
|
if not last:
|
|
|
if indent:
|
|
|
yield b',\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
else:
|
|
|
yield b', '
|
|
|
|
|
|
if indent:
|
|
|
level -= 1
|
|
|
yield b'\n'
|
|
|
yield b' ' * (level * indent)
|
|
|
|
|
|
yield b']'
|
|
|
else:
|
|
|
yield pycompat.byterepr(o)
|
|
|
|
|
|
|
|
|
def prettyrepr(o):
|
|
|
"""Pretty print a representation of a possibly-nested object"""
|
|
|
lines = []
|
|
|
rs = pycompat.byterepr(o)
|
|
|
p0 = p1 = 0
|
|
|
while p0 < len(rs):
|
|
|
# '... field=<type ... field=<type ...'
|
|
|
# ~~~~~~~~~~~~~~~~
|
|
|
# p0 p1 q0 q1
|
|
|
q0 = -1
|
|
|
q1 = rs.find(b'<', p1 + 1)
|
|
|
if q1 < 0:
|
|
|
q1 = len(rs)
|
|
|
elif q1 > p1 + 1 and rs.startswith(b'=', q1 - 1):
|
|
|
# backtrack for ' field=<'
|
|
|
q0 = rs.rfind(b' ', p1 + 1, q1 - 1)
|
|
|
if q0 < 0:
|
|
|
q0 = q1
|
|
|
else:
|
|
|
q0 += 1 # skip ' '
|
|
|
l = rs.count(b'<', 0, p0) - rs.count(b'>', 0, p0)
|
|
|
assert l >= 0
|
|
|
lines.append((l, rs[p0:q0].rstrip()))
|
|
|
p0, p1 = q0, q1
|
|
|
return b'\n'.join(b' ' * l + s for l, s in lines)
|
|
|
|
|
|
|
|
|
def buildrepr(r):
|
|
|
"""Format an optional printable representation from unexpanded bits
|
|
|
|
|
|
======== =================================
|
|
|
type(r) example
|
|
|
======== =================================
|
|
|
tuple ('<not %r>', other)
|
|
|
bytes '<branch closed>'
|
|
|
callable lambda: '<branch %r>' % sorted(b)
|
|
|
object other
|
|
|
======== =================================
|
|
|
"""
|
|
|
if r is None:
|
|
|
return b''
|
|
|
elif isinstance(r, tuple):
|
|
|
return r[0] % pycompat.rapply(pycompat.maybebytestr, r[1:])
|
|
|
elif isinstance(r, bytes):
|
|
|
return r
|
|
|
elif callable(r):
|
|
|
return r()
|
|
|
else:
|
|
|
return pprint(r)
|
|
|
|
|
|
|
|
|
def binary(s):
|
|
|
"""return true if a string is binary data"""
|
|
|
return bool(s and b'\0' in s)
|
|
|
|
|
|
|
|
|
def _splitpattern(pattern):
|
|
|
if pattern.startswith(b're:'):
|
|
|
return b're', pattern[3:]
|
|
|
elif pattern.startswith(b'literal:'):
|
|
|
return b'literal', pattern[8:]
|
|
|
return b'literal', pattern
|
|
|
|
|
|
|
|
|
def stringmatcher(pattern, casesensitive=True):
|
|
|
"""
|
|
|
accepts a string, possibly starting with 're:' or 'literal:' prefix.
|
|
|
returns the matcher name, pattern, and matcher function.
|
|
|
missing or unknown prefixes are treated as literal matches.
|
|
|
|
|
|
helper for tests:
|
|
|
>>> def test(pattern, *tests):
|
|
|
... kind, pattern, matcher = stringmatcher(pattern)
|
|
|
... return (kind, pattern, [bool(matcher(t)) for t in tests])
|
|
|
>>> def itest(pattern, *tests):
|
|
|
... kind, pattern, matcher = stringmatcher(pattern, casesensitive=False)
|
|
|
... return (kind, pattern, [bool(matcher(t)) for t in tests])
|
|
|
|
|
|
exact matching (no prefix):
|
|
|
>>> test(b'abcdefg', b'abc', b'def', b'abcdefg')
|
|
|
('literal', 'abcdefg', [False, False, True])
|
|
|
|
|
|
regex matching ('re:' prefix)
|
|
|
>>> test(b're:a.+b', b'nomatch', b'fooadef', b'fooadefbar')
|
|
|
('re', 'a.+b', [False, False, True])
|
|
|
|
|
|
force exact matches ('literal:' prefix)
|
|
|
>>> test(b'literal:re:foobar', b'foobar', b're:foobar')
|
|
|
('literal', 're:foobar', [False, True])
|
|
|
|
|
|
unknown prefixes are ignored and treated as literals
|
|
|
>>> test(b'foo:bar', b'foo', b'bar', b'foo:bar')
|
|
|
('literal', 'foo:bar', [False, False, True])
|
|
|
|
|
|
case insensitive regex matches
|
|
|
>>> itest(b're:A.+b', b'nomatch', b'fooadef', b'fooadefBar')
|
|
|
('re', 'A.+b', [False, False, True])
|
|
|
|
|
|
case insensitive literal matches
|
|
|
>>> itest(b'ABCDEFG', b'abc', b'def', b'abcdefg')
|
|
|
('literal', 'ABCDEFG', [False, False, True])
|
|
|
"""
|
|
|
kind, pattern = _splitpattern(pattern)
|
|
|
if kind == b're':
|
|
|
try:
|
|
|
flags = 0
|
|
|
if not casesensitive:
|
|
|
flags = remod.I
|
|
|
regex = remod.compile(pattern, flags)
|
|
|
except remod.error as e:
|
|
|
raise error.ParseError(
|
|
|
_(b'invalid regular expression: %s') % forcebytestr(e)
|
|
|
)
|
|
|
return kind, pattern, regex.search
|
|
|
elif kind == b'literal':
|
|
|
if casesensitive:
|
|
|
match = pattern.__eq__
|
|
|
else:
|
|
|
ipat = encoding.lower(pattern)
|
|
|
match = lambda s: ipat == encoding.lower(s)
|
|
|
return kind, pattern, match
|
|
|
|
|
|
raise error.ProgrammingError(b'unhandled pattern kind: %s' % kind)
|
|
|
|
|
|
|
|
|
def substringregexp(pattern, flags=0):
|
|
|
"""Build a regexp object from a string pattern possibly starting with
|
|
|
're:' or 'literal:' prefix.
|
|
|
|
|
|
helper for tests:
|
|
|
>>> def test(pattern, *tests):
|
|
|
... regexp = substringregexp(pattern)
|
|
|
... return [bool(regexp.search(t)) for t in tests]
|
|
|
>>> def itest(pattern, *tests):
|
|
|
... regexp = substringregexp(pattern, remod.I)
|
|
|
... return [bool(regexp.search(t)) for t in tests]
|
|
|
|
|
|
substring matching (no prefix):
|
|
|
>>> test(b'bcde', b'abc', b'def', b'abcdefg')
|
|
|
[False, False, True]
|
|
|
|
|
|
substring pattern should be escaped:
|
|
|
>>> substringregexp(b'.bc').pattern
|
|
|
'\\\\.bc'
|
|
|
>>> test(b'.bc', b'abc', b'def', b'abcdefg')
|
|
|
[False, False, False]
|
|
|
|
|
|
regex matching ('re:' prefix)
|
|
|
>>> test(b're:a.+b', b'nomatch', b'fooadef', b'fooadefbar')
|
|
|
[False, False, True]
|
|
|
|
|
|
force substring matches ('literal:' prefix)
|
|
|
>>> test(b'literal:re:foobar', b'foobar', b're:foobar')
|
|
|
[False, True]
|
|
|
|
|
|
case insensitive literal matches
|
|
|
>>> itest(b'BCDE', b'abc', b'def', b'abcdefg')
|
|
|
[False, False, True]
|
|
|
|
|
|
case insensitive regex matches
|
|
|
>>> itest(b're:A.+b', b'nomatch', b'fooadef', b'fooadefBar')
|
|
|
[False, False, True]
|
|
|
"""
|
|
|
kind, pattern = _splitpattern(pattern)
|
|
|
if kind == b're':
|
|
|
try:
|
|
|
return remod.compile(pattern, flags)
|
|
|
except remod.error as e:
|
|
|
raise error.ParseError(
|
|
|
_(b'invalid regular expression: %s') % forcebytestr(e)
|
|
|
)
|
|
|
elif kind == b'literal':
|
|
|
return remod.compile(remod.escape(pattern), flags)
|
|
|
|
|
|
raise error.ProgrammingError(b'unhandled pattern kind: %s' % kind)
|
|
|
|
|
|
|
|
|
def shortuser(user):
|
|
|
"""Return a short representation of a user name or email address."""
|
|
|
f = user.find(b'@')
|
|
|
if f >= 0:
|
|
|
user = user[:f]
|
|
|
f = user.find(b'<')
|
|
|
if f >= 0:
|
|
|
user = user[f + 1 :]
|
|
|
f = user.find(b' ')
|
|
|
if f >= 0:
|
|
|
user = user[:f]
|
|
|
f = user.find(b'.')
|
|
|
if f >= 0:
|
|
|
user = user[:f]
|
|
|
return user
|
|
|
|
|
|
|
|
|
def emailuser(user):
|
|
|
"""Return the user portion of an email address."""
|
|
|
f = user.find(b'@')
|
|
|
if f >= 0:
|
|
|
user = user[:f]
|
|
|
f = user.find(b'<')
|
|
|
if f >= 0:
|
|
|
user = user[f + 1 :]
|
|
|
return user
|
|
|
|
|
|
|
|
|
def email(author):
|
|
|
'''get email of author.'''
|
|
|
r = author.find(b'>')
|
|
|
if r == -1:
|
|
|
r = None
|
|
|
return author[author.find(b'<') + 1 : r]
|
|
|
|
|
|
|
|
|
def person(author):
|
|
|
"""Returns the name before an email address,
|
|
|
interpreting it as per RFC 5322
|
|
|
|
|
|
>>> person(b'foo@bar')
|
|
|
'foo'
|
|
|
>>> person(b'Foo Bar <foo@bar>')
|
|
|
'Foo Bar'
|
|
|
>>> person(b'"Foo Bar" <foo@bar>')
|
|
|
'Foo Bar'
|
|
|
>>> person(b'"Foo \"buz\" Bar" <foo@bar>')
|
|
|
'Foo "buz" Bar'
|
|
|
>>> # The following are invalid, but do exist in real-life
|
|
|
...
|
|
|
>>> person(b'Foo "buz" Bar <foo@bar>')
|
|
|
'Foo "buz" Bar'
|
|
|
>>> person(b'"Foo Bar <foo@bar>')
|
|
|
'Foo Bar'
|
|
|
"""
|
|
|
if b'@' not in author:
|
|
|
return author
|
|
|
f = author.find(b'<')
|
|
|
if f != -1:
|
|
|
return author[:f].strip(b' "').replace(b'\\"', b'"')
|
|
|
f = author.find(b'@')
|
|
|
return author[:f].replace(b'.', b' ')
|
|
|
|
|
|
|
|
|
@attr.s(hash=True)
|
|
|
class mailmapping(object):
|
|
|
"""Represents a username/email key or value in
|
|
|
a mailmap file"""
|
|
|
|
|
|
email = attr.ib()
|
|
|
name = attr.ib(default=None)
|
|
|
|
|
|
|
|
|
def _ismailmaplineinvalid(names, emails):
|
|
|
"""Returns True if the parsed names and emails
|
|
|
in a mailmap entry are invalid.
|
|
|
|
|
|
>>> # No names or emails fails
|
|
|
>>> names, emails = [], []
|
|
|
>>> _ismailmaplineinvalid(names, emails)
|
|
|
True
|
|
|
>>> # Only one email fails
|
|
|
>>> emails = [b'email@email.com']
|
|
|
>>> _ismailmaplineinvalid(names, emails)
|
|
|
True
|
|
|
>>> # One email and one name passes
|
|
|
>>> names = [b'Test Name']
|
|
|
>>> _ismailmaplineinvalid(names, emails)
|
|
|
False
|
|
|
>>> # No names but two emails passes
|
|
|
>>> names = []
|
|
|
>>> emails = [b'proper@email.com', b'commit@email.com']
|
|
|
>>> _ismailmaplineinvalid(names, emails)
|
|
|
False
|
|
|
"""
|
|
|
return not emails or not names and len(emails) < 2
|
|
|
|
|
|
|
|
|
def parsemailmap(mailmapcontent):
|
|
|
"""Parses data in the .mailmap format
|
|
|
|
|
|
>>> mmdata = b"\\n".join([
|
|
|
... b'# Comment',
|
|
|
... b'Name <commit1@email.xx>',
|
|
|
... b'<name@email.xx> <commit2@email.xx>',
|
|
|
... b'Name <proper@email.xx> <commit3@email.xx>',
|
|
|
... b'Name <proper@email.xx> Commit <commit4@email.xx>',
|
|
|
... ])
|
|
|
>>> mm = parsemailmap(mmdata)
|
|
|
>>> for key in sorted(mm.keys()):
|
|
|
... print(key)
|
|
|
mailmapping(email='commit1@email.xx', name=None)
|
|
|
mailmapping(email='commit2@email.xx', name=None)
|
|
|
mailmapping(email='commit3@email.xx', name=None)
|
|
|
mailmapping(email='commit4@email.xx', name='Commit')
|
|
|
>>> for val in sorted(mm.values()):
|
|
|
... print(val)
|
|
|
mailmapping(email='commit1@email.xx', name='Name')
|
|
|
mailmapping(email='name@email.xx', name=None)
|
|
|
mailmapping(email='proper@email.xx', name='Name')
|
|
|
mailmapping(email='proper@email.xx', name='Name')
|
|
|
"""
|
|
|
mailmap = {}
|
|
|
|
|
|
if mailmapcontent is None:
|
|
|
return mailmap
|
|
|
|
|
|
for line in mailmapcontent.splitlines():
|
|
|
|
|
|
# Don't bother checking the line if it is a comment or
|
|
|
# is an improperly formed author field
|
|
|
if line.lstrip().startswith(b'#'):
|
|
|
continue
|
|
|
|
|
|
# names, emails hold the parsed emails and names for each line
|
|
|
# name_builder holds the words in a persons name
|
|
|
names, emails = [], []
|
|
|
namebuilder = []
|
|
|
|
|
|
for element in line.split():
|
|
|
if element.startswith(b'#'):
|
|
|
# If we reach a comment in the mailmap file, move on
|
|
|
break
|
|
|
|
|
|
elif element.startswith(b'<') and element.endswith(b'>'):
|
|
|
# We have found an email.
|
|
|
# Parse it, and finalize any names from earlier
|
|
|
emails.append(element[1:-1]) # Slice off the "<>"
|
|
|
|
|
|
if namebuilder:
|
|
|
names.append(b' '.join(namebuilder))
|
|
|
namebuilder = []
|
|
|
|
|
|
# Break if we have found a second email, any other
|
|
|
# data does not fit the spec for .mailmap
|
|
|
if len(emails) > 1:
|
|
|
break
|
|
|
|
|
|
else:
|
|
|
# We have found another word in the committers name
|
|
|
namebuilder.append(element)
|
|
|
|
|
|
# Check to see if we have parsed the line into a valid form
|
|
|
# We require at least one email, and either at least one
|
|
|
# name or a second email
|
|
|
if _ismailmaplineinvalid(names, emails):
|
|
|
continue
|
|
|
|
|
|
mailmapkey = mailmapping(
|
|
|
email=emails[-1],
|
|
|
name=names[-1] if len(names) == 2 else None,
|
|
|
)
|
|
|
|
|
|
mailmap[mailmapkey] = mailmapping(
|
|
|
email=emails[0],
|
|
|
name=names[0] if names else None,
|
|
|
)
|
|
|
|
|
|
return mailmap
|
|
|
|
|
|
|
|
|
def mapname(mailmap, author):
|
|
|
"""Returns the author field according to the mailmap cache, or
|
|
|
the original author field.
|
|
|
|
|
|
>>> mmdata = b"\\n".join([
|
|
|
... b'# Comment',
|
|
|
... b'Name <commit1@email.xx>',
|
|
|
... b'<name@email.xx> <commit2@email.xx>',
|
|
|
... b'Name <proper@email.xx> <commit3@email.xx>',
|
|
|
... b'Name <proper@email.xx> Commit <commit4@email.xx>',
|
|
|
... ])
|
|
|
>>> m = parsemailmap(mmdata)
|
|
|
>>> mapname(m, b'Commit <commit1@email.xx>')
|
|
|
'Name <commit1@email.xx>'
|
|
|
>>> mapname(m, b'Name <commit2@email.xx>')
|
|
|
'Name <name@email.xx>'
|
|
|
>>> mapname(m, b'Commit <commit3@email.xx>')
|
|
|
'Name <proper@email.xx>'
|
|
|
>>> mapname(m, b'Commit <commit4@email.xx>')
|
|
|
'Name <proper@email.xx>'
|
|
|
>>> mapname(m, b'Unknown Name <unknown@email.com>')
|
|
|
'Unknown Name <unknown@email.com>'
|
|
|
"""
|
|
|
# If the author field coming in isn't in the correct format,
|
|
|
# or the mailmap is empty just return the original author field
|
|
|
if not isauthorwellformed(author) or not mailmap:
|
|
|
return author
|
|
|
|
|
|
# Turn the user name into a mailmapping
|
|
|
commit = mailmapping(name=person(author), email=email(author))
|
|
|
|
|
|
try:
|
|
|
# Try and use both the commit email and name as the key
|
|
|
proper = mailmap[commit]
|
|
|
|
|
|
except KeyError:
|
|
|
# If the lookup fails, use just the email as the key instead
|
|
|
# We call this commit2 as not to erase original commit fields
|
|
|
commit2 = mailmapping(email=commit.email)
|
|
|
proper = mailmap.get(commit2, mailmapping(None, None))
|
|
|
|
|
|
# Return the author field with proper values filled in
|
|
|
return b'%s <%s>' % (
|
|
|
proper.name if proper.name else commit.name,
|
|
|
proper.email if proper.email else commit.email,
|
|
|
)
|
|
|
|
|
|
|
|
|
_correctauthorformat = remod.compile(br'^[^<]+\s<[^<>]+@[^<>]+>$')
|
|
|
|
|
|
|
|
|
def isauthorwellformed(author):
|
|
|
"""Return True if the author field is well formed
|
|
|
(ie "Contributor Name <contrib@email.dom>")
|
|
|
|
|
|
>>> isauthorwellformed(b'Good Author <good@author.com>')
|
|
|
True
|
|
|
>>> isauthorwellformed(b'Author <good@author.com>')
|
|
|
True
|
|
|
>>> isauthorwellformed(b'Bad Author')
|
|
|
False
|
|
|
>>> isauthorwellformed(b'Bad Author <author@author.com')
|
|
|
False
|
|
|
>>> isauthorwellformed(b'Bad Author author@author.com')
|
|
|
False
|
|
|
>>> isauthorwellformed(b'<author@author.com>')
|
|
|
False
|
|
|
>>> isauthorwellformed(b'Bad Author <author>')
|
|
|
False
|
|
|
"""
|
|
|
return _correctauthorformat.match(author) is not None
|
|
|
|
|
|
|
|
|
def ellipsis(text, maxlength=400):
|
|
|
"""Trim string to at most maxlength (default: 400) columns in display."""
|
|
|
return encoding.trim(text, maxlength, ellipsis=b'...')
|
|
|
|
|
|
|
|
|
def escapestr(s):
|
|
|
if isinstance(s, memoryview):
|
|
|
s = bytes(s)
|
|
|
# call underlying function of s.encode('string_escape') directly for
|
|
|
# Python 3 compatibility
|
|
|
return codecs.escape_encode(s)[0]
|
|
|
|
|
|
|
|
|
def unescapestr(s):
|
|
|
return codecs.escape_decode(s)[0]
|
|
|
|
|
|
|
|
|
def forcebytestr(obj):
|
|
|
"""Portably format an arbitrary object (e.g. exception) into a byte
|
|
|
string."""
|
|
|
try:
|
|
|
return pycompat.bytestr(obj)
|
|
|
except UnicodeEncodeError:
|
|
|
# non-ascii string, may be lossy
|
|
|
return pycompat.bytestr(encoding.strtolocal(str(obj)))
|
|
|
|
|
|
|
|
|
def uirepr(s):
|
|
|
# Avoid double backslash in Windows path repr()
|
|
|
return pycompat.byterepr(pycompat.bytestr(s)).replace(b'\\\\', b'\\')
|
|
|
|
|
|
|
|
|
# delay import of textwrap
|
|
|
def _MBTextWrapper(**kwargs):
|
|
|
class tw(textwrap.TextWrapper):
|
|
|
"""
|
|
|
Extend TextWrapper for width-awareness.
|
|
|
|
|
|
Neither number of 'bytes' in any encoding nor 'characters' is
|
|
|
appropriate to calculate terminal columns for specified string.
|
|
|
|
|
|
Original TextWrapper implementation uses built-in 'len()' directly,
|
|
|
so overriding is needed to use width information of each characters.
|
|
|
|
|
|
In addition, characters classified into 'ambiguous' width are
|
|
|
treated as wide in East Asian area, but as narrow in other.
|
|
|
|
|
|
This requires use decision to determine width of such characters.
|
|
|
"""
|
|
|
|
|
|
def _cutdown(self, ucstr, space_left):
|
|
|
l = 0
|
|
|
colwidth = encoding.ucolwidth
|
|
|
for i in pycompat.xrange(len(ucstr)):
|
|
|
l += colwidth(ucstr[i])
|
|
|
if space_left < l:
|
|
|
return (ucstr[:i], ucstr[i:])
|
|
|
return ucstr, b''
|
|
|
|
|
|
# overriding of base class
|
|
|
def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
|
|
|
space_left = max(width - cur_len, 1)
|
|
|
|
|
|
if self.break_long_words:
|
|
|
cut, res = self._cutdown(reversed_chunks[-1], space_left)
|
|
|
cur_line.append(cut)
|
|
|
reversed_chunks[-1] = res
|
|
|
elif not cur_line:
|
|
|
cur_line.append(reversed_chunks.pop())
|
|
|
|
|
|
# this overriding code is imported from TextWrapper of Python 2.6
|
|
|
# to calculate columns of string by 'encoding.ucolwidth()'
|
|
|
def _wrap_chunks(self, chunks):
|
|
|
colwidth = encoding.ucolwidth
|
|
|
|
|
|
lines = []
|
|
|
if self.width <= 0:
|
|
|
raise ValueError(b"invalid width %r (must be > 0)" % self.width)
|
|
|
|
|
|
# Arrange in reverse order so items can be efficiently popped
|
|
|
# from a stack of chucks.
|
|
|
chunks.reverse()
|
|
|
|
|
|
while chunks:
|
|
|
|
|
|
# Start the list of chunks that will make up the current line.
|
|
|
# cur_len is just the length of all the chunks in cur_line.
|
|
|
cur_line = []
|
|
|
cur_len = 0
|
|
|
|
|
|
# Figure out which static string will prefix this line.
|
|
|
if lines:
|
|
|
indent = self.subsequent_indent
|
|
|
else:
|
|
|
indent = self.initial_indent
|
|
|
|
|
|
# Maximum width for this line.
|
|
|
width = self.width - len(indent)
|
|
|
|
|
|
# First chunk on line is whitespace -- drop it, unless this
|
|
|
# is the very beginning of the text (i.e. no lines started yet).
|
|
|
if self.drop_whitespace and chunks[-1].strip() == '' and lines:
|
|
|
del chunks[-1]
|
|
|
|
|
|
while chunks:
|
|
|
l = colwidth(chunks[-1])
|
|
|
|
|
|
# Can at least squeeze this chunk onto the current line.
|
|
|
if cur_len + l <= width:
|
|
|
cur_line.append(chunks.pop())
|
|
|
cur_len += l
|
|
|
|
|
|
# Nope, this line is full.
|
|
|
else:
|
|
|
break
|
|
|
|
|
|
# The current line is full, and the next chunk is too big to
|
|
|
# fit on *any* line (not just this one).
|
|
|
if chunks and colwidth(chunks[-1]) > width:
|
|
|
self._handle_long_word(chunks, cur_line, cur_len, width)
|
|
|
|
|
|
# If the last chunk on this line is all whitespace, drop it.
|
|
|
if (
|
|
|
self.drop_whitespace
|
|
|
and cur_line
|
|
|
and cur_line[-1].strip() == r''
|
|
|
):
|
|
|
del cur_line[-1]
|
|
|
|
|
|
# Convert current line back to a string and store it in list
|
|
|
# of all lines (return value).
|
|
|
if cur_line:
|
|
|
lines.append(indent + ''.join(cur_line))
|
|
|
|
|
|
return lines
|
|
|
|
|
|
global _MBTextWrapper
|
|
|
_MBTextWrapper = tw
|
|
|
return tw(**kwargs)
|
|
|
|
|
|
|
|
|
def wrap(line, width, initindent=b'', hangindent=b''):
|
|
|
maxindent = max(len(hangindent), len(initindent))
|
|
|
if width <= maxindent:
|
|
|
# adjust for weird terminal size
|
|
|
width = max(78, maxindent + 1)
|
|
|
line = line.decode(
|
|
|
pycompat.sysstr(encoding.encoding),
|
|
|
pycompat.sysstr(encoding.encodingmode),
|
|
|
)
|
|
|
initindent = initindent.decode(
|
|
|
pycompat.sysstr(encoding.encoding),
|
|
|
pycompat.sysstr(encoding.encodingmode),
|
|
|
)
|
|
|
hangindent = hangindent.decode(
|
|
|
pycompat.sysstr(encoding.encoding),
|
|
|
pycompat.sysstr(encoding.encodingmode),
|
|
|
)
|
|
|
wrapper = _MBTextWrapper(
|
|
|
width=width, initial_indent=initindent, subsequent_indent=hangindent
|
|
|
)
|
|
|
return wrapper.fill(line).encode(pycompat.sysstr(encoding.encoding))
|
|
|
|
|
|
|
|
|
_booleans = {
|
|
|
b'1': True,
|
|
|
b'yes': True,
|
|
|
b'true': True,
|
|
|
b'on': True,
|
|
|
b'always': True,
|
|
|
b'0': False,
|
|
|
b'no': False,
|
|
|
b'false': False,
|
|
|
b'off': False,
|
|
|
b'never': False,
|
|
|
}
|
|
|
|
|
|
|
|
|
def parsebool(s):
|
|
|
"""Parse s into a boolean.
|
|
|
|
|
|
If s is not a valid boolean, returns None.
|
|
|
"""
|
|
|
return _booleans.get(s.lower(), None)
|
|
|
|
|
|
|
|
|
def evalpythonliteral(s):
|
|
|
"""Evaluate a string containing a Python literal expression"""
|
|
|
# We could backport our tokenizer hack to rewrite '' to u'' if we want
|
|
|
if pycompat.ispy3:
|
|
|
return ast.literal_eval(s.decode('latin1'))
|
|
|
return ast.literal_eval(s)
|
|
|
|