annotate mercurial/utils/stringutil.py @ 51700:7f0cb9ee0534

Backout accidental publication of a large range of revisions I accidentally published 25e7f9dcad0f::bd1483fd7088, this is the inverse.
author Raphaël Gomès <rgomes@octobus.net>
date Tue, 23 Jul 2024 10:02:46 +0200
parents 493034cc3265
children ca7bde5dbafb
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
37083
f99d64e8a4e4 stringutil: move generic string helpers to new module
Yuya Nishihara <yuya@tcha.org>
parents: 37082
diff changeset
1 # stringutil.py - utility for generic string formatting, parsing, etc.
8226
8b2cd04a6e97 put license and copyright info into comment blocks
Martin Geisler <mg@lazybytes.net>
parents: 8225
diff changeset
2 #
8b2cd04a6e97 put license and copyright info into comment blocks
Martin Geisler <mg@lazybytes.net>
parents: 8225
diff changeset
3 # Copyright 2005 K. Thananchayan <thananck@yahoo.com>
46819
d4ba4d51f85f contributor: change mentions of mpm to olivia
Raphaël Gomès <rgomes@octobus.net>
parents: 45942
diff changeset
4 # Copyright 2005-2007 Olivia Mackall <olivia@selenic.com>
8226
8b2cd04a6e97 put license and copyright info into comment blocks
Martin Geisler <mg@lazybytes.net>
parents: 8225
diff changeset
5 # Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com>
8b2cd04a6e97 put license and copyright info into comment blocks
Martin Geisler <mg@lazybytes.net>
parents: 8225
diff changeset
6 #
8b2cd04a6e97 put license and copyright info into comment blocks
Martin Geisler <mg@lazybytes.net>
parents: 8225
diff changeset
7 # This software may be used and distributed according to the terms of the
10263
25e572394f5c Update license to GPLv2+
Matt Mackall <mpm@selenic.com>
parents: 9996
diff changeset
8 # GNU General Public License version 2 or any later version.
1082
ce96e316278a Update util.py docstrings, fix walk test
mpm@selenic.com
parents: 1081
diff changeset
9
ce96e316278a Update util.py docstrings, fix walk test
mpm@selenic.com
parents: 1081
diff changeset
10
37476
e9dea82ea1f3 wireproto: convert python literal to object without using unsafe eval()
Yuya Nishihara <yuya@tcha.org>
parents: 37322
diff changeset
11 import ast
31453
3b7a6941a6ef py3: call codecs.escape_encode() directly
Yuya Nishihara <yuya@tcha.org>
parents: 31451
diff changeset
12 import codecs
21907
7e5dfa00e3c2 util: rename 're' to 'remod'
Siddharth Agarwal <sid0@fb.com>
parents: 21857
diff changeset
13 import re as remod
27358
ac839ee45b6a util: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 27357
diff changeset
14 import textwrap
39296
ce145f8eface stringutil: teach pprint() to recognize generators
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39063
diff changeset
15 import types
3769
96095d9ff1f8 Add encoding detection
Matt Mackall <mpm@selenic.com>
parents: 3767
diff changeset
16
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
17 from typing import (
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
18 Optional,
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
19 overload,
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
20 )
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
21
37083
f99d64e8a4e4 stringutil: move generic string helpers to new module
Yuya Nishihara <yuya@tcha.org>
parents: 37082
diff changeset
22 from ..i18n import _
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
23 from ..thirdparty import attr
37083
f99d64e8a4e4 stringutil: move generic string helpers to new module
Yuya Nishihara <yuya@tcha.org>
parents: 37082
diff changeset
24
f99d64e8a4e4 stringutil: move generic string helpers to new module
Yuya Nishihara <yuya@tcha.org>
parents: 37082
diff changeset
25 from .. import (
27358
ac839ee45b6a util: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 27357
diff changeset
26 encoding,
ac839ee45b6a util: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 27357
diff changeset
27 error,
28818
6041fb8f2da8 pycompat: add empty and queue to handle py3 divergence
timeless <timeless@mozdev.org>
parents: 28497
diff changeset
28 pycompat,
27358
ac839ee45b6a util: use absolute_import
Gregory Szorc <gregory.szorc@gmail.com>
parents: 27357
diff changeset
29 )
37010
8453699a1f21 util: observable proxy objects for sockets
Gregory Szorc <gregory.szorc@gmail.com>
parents: 36991
diff changeset
30
38474
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
31 # regex special chars pulled from https://bugs.python.org/issue29995
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
32 # which was part of Python 3.7.
38477
de275ab362cb stringutil: update list of re-special characters to include &~
Augie Fackler <augie@google.com>
parents: 38474
diff changeset
33 _respecial = pycompat.bytestr(b'()[]{}?*+-|^$\\.&~# \t\n\r\v\f')
38474
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
34 _regexescapemap = {ord(i): (b'\\' + i).decode('latin1') for i in _respecial}
40684
e6c9ef5e11a0 match: provide and use a quick way to escape a single byte
Boris Feld <boris.feld@octobus.net>
parents: 40276
diff changeset
35 regexbytesescapemap = {i: (b'\\' + i) for i in _respecial}
38474
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
36
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
37
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
38 @overload
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
39 def reescape(pat: bytes) -> bytes:
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
40 ...
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
41
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
42
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
43 @overload
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
44 def reescape(pat: str) -> str:
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
45 ...
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
46
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
47
38474
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
48 def reescape(pat):
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
49 """Drop-in replacement for re.escape."""
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
50 # NOTE: it is intentional that this works on unicodes and not
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
51 # bytes, as it's only possible to do the escaping with
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
52 # unicode.translate, not bytes.translate. Sigh.
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
53 wantuni = True
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
54 if isinstance(pat, bytes):
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
55 wantuni = False
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
56 pat = pat.decode('latin1')
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
57 pat = pat.translate(_regexescapemap)
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
58 if wantuni:
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
59 return pat
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
60 return pat.encode('latin1')
96f65bdf0bf4 stringutil: add a new function to do minimal regex escaping
Augie Fackler <augie@google.com>
parents: 38264
diff changeset
61
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
62
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
63 def pprint(o, bprefix: bool = False, indent: int = 0, level: int = 0) -> bytes:
37300
2f859ad7ed8c stringutil: add function to pretty print an object
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37290
diff changeset
64 """Pretty print an object."""
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
65 return b''.join(pprintgen(o, bprefix=bprefix, indent=indent, level=level))
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
66
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
67
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
68 def pprintgen(o, bprefix: bool = False, indent: int = 0, level: int = 0):
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
69 """Pretty print an object to a generator of atoms.
39353
0d21b1f1722c stringutil: refactor core of pprint so it emits chunks
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39296
diff changeset
70
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
71 ``bprefix`` is a flag influencing whether bytestrings are preferred with
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
72 a ``b''`` prefix.
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
73
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
74 ``indent`` controls whether collections and nested data structures
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
75 span multiple lines via the indentation amount in spaces. By default,
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
76 no newlines are emitted.
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
77
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
78 ``level`` specifies the initial indent level. Used if ``indent > 0``.
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
79 """
39353
0d21b1f1722c stringutil: refactor core of pprint so it emits chunks
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39296
diff changeset
80
37619
68132a95df31 stringutil: support more types with pprint()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37476
diff changeset
81 if isinstance(o, bytes):
37750
f7194c925003 stringutil: make b prefixes on string output optional
Augie Fackler <augie@google.com>
parents: 37749
diff changeset
82 if bprefix:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
83 yield b"b'%s'" % escapestr(o)
39353
0d21b1f1722c stringutil: refactor core of pprint so it emits chunks
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39296
diff changeset
84 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
85 yield b"'%s'" % escapestr(o)
37619
68132a95df31 stringutil: support more types with pprint()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37476
diff changeset
86 elif isinstance(o, bytearray):
68132a95df31 stringutil: support more types with pprint()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37476
diff changeset
87 # codecs.escape_encode() can't handle bytearray, so escapestr fails
68132a95df31 stringutil: support more types with pprint()
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37476
diff changeset
88 # without coercion.
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
89 yield b"bytearray['%s']" % escapestr(bytes(o))
37300
2f859ad7ed8c stringutil: add function to pretty print an object
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37290
diff changeset
90 elif isinstance(o, list):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
91 if not o:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
92 yield b'[]'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
93 return
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
94
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
95 yield b'['
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
96
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
97 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
98 level += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
99 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
100 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
101
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
102 for i, a in enumerate(o):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
103 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
104 a, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
105 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
106 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
107
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
108 if i + 1 < len(o):
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
109 if indent:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
110 yield b',\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
111 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
112 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
113 yield b', '
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
114
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
115 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
116 level -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
117 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
118 yield b' ' * (level * indent)
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
119
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
120 yield b']'
37300
2f859ad7ed8c stringutil: add function to pretty print an object
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37290
diff changeset
121 elif isinstance(o, dict):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
122 if not o:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
123 yield b'{}'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
124 return
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
125
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
126 yield b'{'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
127
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
128 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
129 level += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
130 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
131 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
132
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
133 for i, (k, v) in enumerate(sorted(o.items())):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
134 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
135 k, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
136 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
137 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
138
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
139 yield b': '
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
140
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
141 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
142 v, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
143 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
144 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
145
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
146 if i + 1 < len(o):
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
147 if indent:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
148 yield b',\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
149 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
150 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
151 yield b', '
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
152
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
153 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
154 level -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
155 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
156 yield b' ' * (level * indent)
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
157
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
158 yield b'}'
39051
2aebe138ef6e stringutil: teach pprint about sets
Augie Fackler <augie@google.com>
parents: 38783
diff changeset
159 elif isinstance(o, set):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
160 if not o:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
161 yield b'set([])'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
162 return
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
163
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
164 yield b'set(['
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
165
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
166 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
167 level += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
168 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
169 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
170
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
171 for i, k in enumerate(sorted(o)):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
172 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
173 k, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
174 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
175 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
176
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
177 if i + 1 < len(o):
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
178 if indent:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
179 yield b',\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
180 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
181 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
182 yield b', '
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
183
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
184 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
185 level -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
186 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
187 yield b' ' * (level * indent)
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
188
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
189 yield b'])'
37932
bf6bb710b40f stringutil: teach pprint about tuples
Augie Fackler <augie@google.com>
parents: 37750
diff changeset
190 elif isinstance(o, tuple):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
191 if not o:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
192 yield b'()'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
193 return
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
194
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
195 yield b'('
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
196
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
197 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
198 level += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
199 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
200 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
201
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
202 for i, a in enumerate(o):
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
203 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
204 a, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
205 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
206 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
207
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
208 if i + 1 < len(o):
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
209 if indent:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
210 yield b',\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
211 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
212 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
213 yield b', '
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
214
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
215 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
216 level -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
217 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
218 yield b' ' * (level * indent)
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
219
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
220 yield b')'
39296
ce145f8eface stringutil: teach pprint() to recognize generators
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39063
diff changeset
221 elif isinstance(o, types.GeneratorType):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
222 # Special case of empty generator.
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
223 try:
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
224 nextitem = next(o)
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
225 except StopIteration:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
226 yield b'gen[]'
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
227 return
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
228
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
229 yield b'gen['
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
230
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
231 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
232 level += 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
233 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
234 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
235
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
236 last = False
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
237
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
238 while not last:
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
239 current = nextitem
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
240
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
241 try:
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
242 nextitem = next(o)
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
243 except StopIteration:
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
244 last = True
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
245
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
246 for chunk in pprintgen(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
247 current, bprefix=bprefix, indent=indent, level=level
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
248 ):
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
249 yield chunk
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
250
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
251 if not last:
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
252 if indent:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
253 yield b',\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
254 yield b' ' * (level * indent)
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
255 else:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
256 yield b', '
39378
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
257
0f549da54379 stringutil: teach pprint() to indent
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39354
diff changeset
258 if indent:
40276
be57c7019c70 stringutil: allow to specify initial indent level of pprint()
Yuya Nishihara <yuya@tcha.org>
parents: 40275
diff changeset
259 level -= 1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
260 yield b'\n'
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
261 yield b' ' * (level * indent)
39354
5ed7c6caf24d stringutil: emit multiple chunks when pretty printing
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39353
diff changeset
262
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
263 yield b']'
37300
2f859ad7ed8c stringutil: add function to pretty print an object
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37290
diff changeset
264 else:
39353
0d21b1f1722c stringutil: refactor core of pprint so it emits chunks
Gregory Szorc <gregory.szorc@gmail.com>
parents: 39296
diff changeset
265 yield pycompat.byterepr(o)
37300
2f859ad7ed8c stringutil: add function to pretty print an object
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37290
diff changeset
266
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
267
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
268 def prettyrepr(o) -> bytes:
38261
f3033692ccef stringutil: promote smartset.prettyformat() to utility function
Yuya Nishihara <yuya@tcha.org>
parents: 37942
diff changeset
269 """Pretty print a representation of a possibly-nested object"""
f3033692ccef stringutil: promote smartset.prettyformat() to utility function
Yuya Nishihara <yuya@tcha.org>
parents: 37942
diff changeset
270 lines = []
f3033692ccef stringutil: promote smartset.prettyformat() to utility function
Yuya Nishihara <yuya@tcha.org>
parents: 37942
diff changeset
271 rs = pycompat.byterepr(o)
38264
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
272 p0 = p1 = 0
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
273 while p0 < len(rs):
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
274 # '... field=<type ... field=<type ...'
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
275 # ~~~~~~~~~~~~~~~~
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
276 # p0 p1 q0 q1
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
277 q0 = -1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
278 q1 = rs.find(b'<', p1 + 1)
38264
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
279 if q1 < 0:
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
280 q1 = len(rs)
48479
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
281 # pytype: disable=wrong-arg-count
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
282 # TODO: figure out why pytype doesn't recognize the optional start
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
283 # arg
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
284 elif q1 > p1 + 1 and rs.startswith(b'=', q1 - 1):
48479
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
285 # pytype: enable=wrong-arg-count
38264
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
286 # backtrack for ' field=<'
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
287 q0 = rs.rfind(b' ', p1 + 1, q1 - 1)
38264
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
288 if q0 < 0:
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
289 q0 = q1
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
290 else:
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
291 q0 += 1 # skip ' '
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
292 l = rs.count(b'<', 0, p0) - rs.count(b'>', 0, p0)
38261
f3033692ccef stringutil: promote smartset.prettyformat() to utility function
Yuya Nishihara <yuya@tcha.org>
parents: 37942
diff changeset
293 assert l >= 0
38264
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
294 lines.append((l, rs[p0:q0].rstrip()))
fbb2eddea4d2 stringutil: fix prettyrepr() to not orphan foo=<...> line
Yuya Nishihara <yuya@tcha.org>
parents: 38261
diff changeset
295 p0, p1 = q0, q1
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
296 return b'\n'.join(b' ' * l + s for l, s in lines)
38261
f3033692ccef stringutil: promote smartset.prettyformat() to utility function
Yuya Nishihara <yuya@tcha.org>
parents: 37942
diff changeset
297
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
298
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
299 def buildrepr(r) -> bytes:
38576
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
300 """Format an optional printable representation from unexpanded bits
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
301
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
302 ======== =================================
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
303 type(r) example
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
304 ======== =================================
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
305 tuple ('<not %r>', other)
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
306 bytes '<branch closed>'
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
307 callable lambda: '<branch %r>' % sorted(b)
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
308 object other
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
309 ======== =================================
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
310 """
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
311 if r is None:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
312 return b''
38576
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
313 elif isinstance(r, tuple):
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
314 return r[0] % pycompat.rapply(pycompat.maybebytestr, r[1:])
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
315 elif isinstance(r, bytes):
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
316 return r
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
317 elif callable(r):
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
318 return r()
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
319 else:
39052
38409be2f521 stringutil: have buildrepr delegate to pprint for unknown types
Augie Fackler <augie@google.com>
parents: 39051
diff changeset
320 return pprint(r)
38576
a3130208db1c stringutil: move _formatsetrepr() from smartset
Yuya Nishihara <yuya@tcha.org>
parents: 38477
diff changeset
321
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
322
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
323 def binary(s: bytes) -> bool:
6507
9699864de219 Let util.binary check entire data for \0 (issue1066, issue1079)
Christian Ebert <blacktrash@gmx.net>
parents: 6501
diff changeset
324 """return true if a string is binary data"""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
325 return bool(s and b'\0' in s)
6762
f67d1468ac50 util: add sort helper
Matt Mackall <mpm@selenic.com>
parents: 6746
diff changeset
326
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
327
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
328 def _splitpattern(pattern: bytes):
45722
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
329 if pattern.startswith(b're:'):
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
330 return b're', pattern[3:]
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
331 elif pattern.startswith(b'literal:'):
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
332 return b'literal', pattern[8:]
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
333 return b'literal', pattern
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
334
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
335
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
336 def stringmatcher(pattern: bytes, casesensitive: bool = True):
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
337 """
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
338 accepts a string, possibly starting with 're:' or 'literal:' prefix.
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
339 returns the matcher name, pattern, and matcher function.
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
340 missing or unknown prefixes are treated as literal matches.
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
341
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
342 helper for tests:
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
343 >>> def test(pattern, *tests):
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
344 ... kind, pattern, matcher = stringmatcher(pattern)
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
345 ... return (kind, pattern, [bool(matcher(t)) for t in tests])
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
346 >>> def itest(pattern, *tests):
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
347 ... kind, pattern, matcher = stringmatcher(pattern, casesensitive=False)
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
348 ... return (kind, pattern, [bool(matcher(t)) for t in tests])
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
349
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
350 exact matching (no prefix):
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
351 >>> test(b'abcdefg', b'abc', b'def', b'abcdefg')
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
352 ('literal', 'abcdefg', [False, False, True])
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
353
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
354 regex matching ('re:' prefix)
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
355 >>> test(b're:a.+b', b'nomatch', b'fooadef', b'fooadefbar')
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
356 ('re', 'a.+b', [False, False, True])
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
357
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
358 force exact matches ('literal:' prefix)
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
359 >>> test(b'literal:re:foobar', b'foobar', b're:foobar')
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
360 ('literal', 're:foobar', [False, True])
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
361
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
362 unknown prefixes are ignored and treated as literals
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
363 >>> test(b'foo:bar', b'foo', b'bar', b'foo:bar')
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
364 ('literal', 'foo:bar', [False, False, True])
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
365
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
366 case insensitive regex matches
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
367 >>> itest(b're:A.+b', b'nomatch', b'fooadef', b'fooadefBar')
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
368 ('re', 'A.+b', [False, False, True])
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
369
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
370 case insensitive literal matches
34131
0fa781320203 doctest: bulk-replace string literals with b'' for Python 3
Yuya Nishihara <yuya@tcha.org>
parents: 34084
diff changeset
371 >>> itest(b'ABCDEFG', b'abc', b'def', b'abcdefg')
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
372 ('literal', 'ABCDEFG', [False, False, True])
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
373 """
45722
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
374 kind, pattern = _splitpattern(pattern)
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
375 if kind == b're':
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
376 try:
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
377 flags = 0
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
378 if not casesensitive:
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
379 flags = remod.I
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
380 regex = remod.compile(pattern, flags)
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
381 except remod.error as e:
45723
edfc5820aae7 py3: fix stringmatcher() to byte-stringify exception message
Yuya Nishihara <yuya@tcha.org>
parents: 45722
diff changeset
382 raise error.ParseError(
edfc5820aae7 py3: fix stringmatcher() to byte-stringify exception message
Yuya Nishihara <yuya@tcha.org>
parents: 45722
diff changeset
383 _(b'invalid regular expression: %s') % forcebytestr(e)
edfc5820aae7 py3: fix stringmatcher() to byte-stringify exception message
Yuya Nishihara <yuya@tcha.org>
parents: 45722
diff changeset
384 )
45722
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
385 return kind, pattern, regex.search
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
386 elif kind == b'literal':
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
387 if casesensitive:
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
388 match = pattern.__eq__
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
389 else:
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
390 ipat = encoding.lower(pattern)
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
391 match = lambda s: ipat == encoding.lower(s)
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
392 return kind, pattern, match
30773
c390b40fe1d7 util: teach stringmatcher to handle forced case insensitive matches
Matt Harbison <matt_harbison@yahoo.com>
parents: 30761
diff changeset
393
45722
d502caab76bc stringutil: extract helper function that splits stringmatcher() pattern
Yuya Nishihara <yuya@tcha.org>
parents: 44022
diff changeset
394 raise error.ProgrammingError(b'unhandled pattern kind: %s' % kind)
26481
7d132557e44a util: extract stringmatcher() from revset
Matt Harbison <matt_harbison@yahoo.com>
parents: 26480
diff changeset
395
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
396
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
397 def substringregexp(pattern: bytes, flags: int = 0):
45724
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
398 """Build a regexp object from a string pattern possibly starting with
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
399 're:' or 'literal:' prefix.
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
400
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
401 helper for tests:
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
402 >>> def test(pattern, *tests):
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
403 ... regexp = substringregexp(pattern)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
404 ... return [bool(regexp.search(t)) for t in tests]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
405 >>> def itest(pattern, *tests):
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
406 ... regexp = substringregexp(pattern, remod.I)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
407 ... return [bool(regexp.search(t)) for t in tests]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
408
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
409 substring matching (no prefix):
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
410 >>> test(b'bcde', b'abc', b'def', b'abcdefg')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
411 [False, False, True]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
412
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
413 substring pattern should be escaped:
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
414 >>> substringregexp(b'.bc').pattern
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
415 '\\\\.bc'
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
416 >>> test(b'.bc', b'abc', b'def', b'abcdefg')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
417 [False, False, False]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
418
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
419 regex matching ('re:' prefix)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
420 >>> test(b're:a.+b', b'nomatch', b'fooadef', b'fooadefbar')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
421 [False, False, True]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
422
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
423 force substring matches ('literal:' prefix)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
424 >>> test(b'literal:re:foobar', b'foobar', b're:foobar')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
425 [False, True]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
426
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
427 case insensitive literal matches
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
428 >>> itest(b'BCDE', b'abc', b'def', b'abcdefg')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
429 [False, False, True]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
430
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
431 case insensitive regex matches
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
432 >>> itest(b're:A.+b', b'nomatch', b'fooadef', b'fooadefBar')
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
433 [False, False, True]
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
434 """
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
435 kind, pattern = _splitpattern(pattern)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
436 if kind == b're':
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
437 try:
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
438 return remod.compile(pattern, flags)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
439 except remod.error as e:
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
440 raise error.ParseError(
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
441 _(b'invalid regular expression: %s') % forcebytestr(e)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
442 )
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
443 elif kind == b'literal':
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
444 return remod.compile(remod.escape(pattern), flags)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
445
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
446 raise error.ProgrammingError(b'unhandled pattern kind: %s' % kind)
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
447
ac39a8a214b1 stringutil: add function to compile stringmatcher pattern into regexp
Yuya Nishihara <yuya@tcha.org>
parents: 45723
diff changeset
448
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
449 def shortuser(user: bytes) -> bytes:
1903
e4abeafd6eb1 move shortuser into util module.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1635
diff changeset
450 """Return a short representation of a user name or email address."""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
451 f = user.find(b'@')
1903
e4abeafd6eb1 move shortuser into util module.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1635
diff changeset
452 if f >= 0:
e4abeafd6eb1 move shortuser into util module.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1635
diff changeset
453 user = user[:f]
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
454 f = user.find(b'<')
1903
e4abeafd6eb1 move shortuser into util module.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1635
diff changeset
455 if f >= 0:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
456 user = user[f + 1 :]
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
457 f = user.find(b' ')
3176
7492b33bdd9f shortuser should stop before the first space character.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3147
diff changeset
458 if f >= 0:
7492b33bdd9f shortuser should stop before the first space character.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3147
diff changeset
459 user = user[:f]
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
460 f = user.find(b'.')
3533
bb44489b901f shortname: truncate at '.' too
Matt Mackall <mpm@selenic.com>
parents: 3466
diff changeset
461 if f >= 0:
bb44489b901f shortname: truncate at '.' too
Matt Mackall <mpm@selenic.com>
parents: 3466
diff changeset
462 user = user[:f]
1903
e4abeafd6eb1 move shortuser into util module.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1635
diff changeset
463 return user
1920
b7cc0f323a4c merge with crew.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 1903 1882
diff changeset
464
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
465
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
466 def emailuser(user: bytes) -> bytes:
16360
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
467 """Return the user portion of an email address."""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
468 f = user.find(b'@')
16360
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
469 if f >= 0:
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
470 user = user[:f]
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
471 f = user.find(b'<')
16360
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
472 if f >= 0:
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
473 user = user[f + 1 :]
16360
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
474 return user
e5788269741a templates/filters: extracting the user portion of an email address
Matteo Capobianco <m.capobianco@gmail.com>
parents: 15720
diff changeset
475
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
476
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
477 def email(author: bytes) -> bytes:
5975
75d9fe70c654 templater: move email function to util
Matt Mackall <mpm@selenic.com>
parents: 5949
diff changeset
478 '''get email of author.'''
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
479 r = author.find(b'>')
10282
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
480 if r == -1:
08a0f04b56bd many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents: 10264
diff changeset
481 r = None
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
482 return author[author.find(b'<') + 1 : r]
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
483
5975
75d9fe70c654 templater: move email function to util
Matt Mackall <mpm@selenic.com>
parents: 5949
diff changeset
484
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
485 def person(author: bytes) -> bytes:
37155
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
486 """Returns the name before an email address,
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
487 interpreting it as per RFC 5322
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
488
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
489 >>> person(b'foo@bar')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
490 'foo'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
491 >>> person(b'Foo Bar <foo@bar>')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
492 'Foo Bar'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
493 >>> person(b'"Foo Bar" <foo@bar>')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
494 'Foo Bar'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
495 >>> person(b'"Foo \"buz\" Bar" <foo@bar>')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
496 'Foo "buz" Bar'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
497 >>> # The following are invalid, but do exist in real-life
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
498 ...
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
499 >>> person(b'Foo "buz" Bar <foo@bar>')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
500 'Foo "buz" Bar'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
501 >>> person(b'"Foo Bar <foo@bar>')
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
502 'Foo Bar'
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
503 """
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
504 if b'@' not in author:
37155
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
505 return author
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
506 f = author.find(b'<')
37155
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
507 if f != -1:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
508 return author[:f].strip(b' "').replace(b'\\"', b'"')
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
509 f = author.find(b'@')
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
510 return author[:f].replace(b'.', b' ')
37155
fb7140f1d09d stringutil: move person function from templatefilters
Connor Sheehan <sheehan@mozilla.com>
parents: 37154
diff changeset
511
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
512
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
513 @attr.s(hash=True)
48946
642e31cb55f0 py3: use class X: instead of class X(object):
Gregory Szorc <gregory.szorc@gmail.com>
parents: 48911
diff changeset
514 class mailmapping:
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
515 """Represents a username/email key or value in
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
516 a mailmap file"""
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
517
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
518 email = attr.ib()
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
519 name = attr.ib(default=None)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
520
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
521
37246
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
522 def _ismailmaplineinvalid(names, emails):
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
523 """Returns True if the parsed names and emails
37246
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
524 in a mailmap entry are invalid.
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
525
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
526 >>> # No names or emails fails
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
527 >>> names, emails = [], []
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
528 >>> _ismailmaplineinvalid(names, emails)
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
529 True
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
530 >>> # Only one email fails
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
531 >>> emails = [b'email@email.com']
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
532 >>> _ismailmaplineinvalid(names, emails)
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
533 True
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
534 >>> # One email and one name passes
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
535 >>> names = [b'Test Name']
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
536 >>> _ismailmaplineinvalid(names, emails)
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
537 False
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
538 >>> # No names but two emails passes
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
539 >>> names = []
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
540 >>> emails = [b'proper@email.com', b'commit@email.com']
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
541 >>> _ismailmaplineinvalid(names, emails)
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
542 False
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
543 """
37246
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
544 return not emails or not names and len(emails) < 2
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
545
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
546
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
547 def parsemailmap(mailmapcontent):
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
548 """Parses data in the .mailmap format
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
549
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
550 >>> mmdata = b"\\n".join([
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
551 ... b'# Comment',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
552 ... b'Name <commit1@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
553 ... b'<name@email.xx> <commit2@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
554 ... b'Name <proper@email.xx> <commit3@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
555 ... b'Name <proper@email.xx> Commit <commit4@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
556 ... ])
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
557 >>> mm = parsemailmap(mmdata)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
558 >>> for key in sorted(mm.keys()):
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
559 ... print(key)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
560 mailmapping(email='commit1@email.xx', name=None)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
561 mailmapping(email='commit2@email.xx', name=None)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
562 mailmapping(email='commit3@email.xx', name=None)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
563 mailmapping(email='commit4@email.xx', name='Commit')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
564 >>> for val in sorted(mm.values()):
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
565 ... print(val)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
566 mailmapping(email='commit1@email.xx', name='Name')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
567 mailmapping(email='name@email.xx', name=None)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
568 mailmapping(email='proper@email.xx', name='Name')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
569 mailmapping(email='proper@email.xx', name='Name')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
570 """
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
571 mailmap = {}
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
572
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
573 if mailmapcontent is None:
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
574 return mailmap
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
575
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
576 for line in mailmapcontent.splitlines():
51700
7f0cb9ee0534 Backout accidental publication of a large range of revisions
Raphaël Gomès <rgomes@octobus.net>
parents: 51690
diff changeset
577
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
578 # Don't bother checking the line if it is a comment or
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
579 # is an improperly formed author field
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
580 if line.lstrip().startswith(b'#'):
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
581 continue
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
582
37245
54b896f195d1 stringutil: rename local email/names variables to their plural forms
Connor Sheehan <sheehan@mozilla.com>
parents: 37210
diff changeset
583 # names, emails hold the parsed emails and names for each line
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
584 # name_builder holds the words in a persons name
37245
54b896f195d1 stringutil: rename local email/names variables to their plural forms
Connor Sheehan <sheehan@mozilla.com>
parents: 37210
diff changeset
585 names, emails = [], []
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
586 namebuilder = []
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
587
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
588 for element in line.split():
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
589 if element.startswith(b'#'):
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
590 # If we reach a comment in the mailmap file, move on
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
591 break
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
592
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
593 elif element.startswith(b'<') and element.endswith(b'>'):
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
594 # We have found an email.
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
595 # Parse it, and finalize any names from earlier
37245
54b896f195d1 stringutil: rename local email/names variables to their plural forms
Connor Sheehan <sheehan@mozilla.com>
parents: 37210
diff changeset
596 emails.append(element[1:-1]) # Slice off the "<>"
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
597
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
598 if namebuilder:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
599 names.append(b' '.join(namebuilder))
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
600 namebuilder = []
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
601
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
602 # Break if we have found a second email, any other
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
603 # data does not fit the spec for .mailmap
37245
54b896f195d1 stringutil: rename local email/names variables to their plural forms
Connor Sheehan <sheehan@mozilla.com>
parents: 37210
diff changeset
604 if len(emails) > 1:
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
605 break
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
606
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
607 else:
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
608 # We have found another word in the committers name
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
609 namebuilder.append(element)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
610
37246
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
611 # Check to see if we have parsed the line into a valid form
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
612 # We require at least one email, and either at least one
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
613 # name or a second email
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
614 if _ismailmaplineinvalid(names, emails):
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
615 continue
0e7550b0964c stringutil: improve check for failed mailmap line parsing
Connor Sheehan <sheehan@mozilla.com>
parents: 37245
diff changeset
616
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
617 mailmapkey = mailmapping(
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
618 email=emails[-1],
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
619 name=names[-1] if len(names) == 2 else None,
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
620 )
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
621
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
622 mailmap[mailmapkey] = mailmapping(
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
623 email=emails[0],
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
624 name=names[0] if names else None,
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
625 )
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
626
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
627 return mailmap
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
628
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
629
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
630 def mapname(mailmap, author: bytes) -> bytes:
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
631 """Returns the author field according to the mailmap cache, or
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
632 the original author field.
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
633
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
634 >>> mmdata = b"\\n".join([
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
635 ... b'# Comment',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
636 ... b'Name <commit1@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
637 ... b'<name@email.xx> <commit2@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
638 ... b'Name <proper@email.xx> <commit3@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
639 ... b'Name <proper@email.xx> Commit <commit4@email.xx>',
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
640 ... ])
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
641 >>> m = parsemailmap(mmdata)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
642 >>> mapname(m, b'Commit <commit1@email.xx>')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
643 'Name <commit1@email.xx>'
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
644 >>> mapname(m, b'Name <commit2@email.xx>')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
645 'Name <name@email.xx>'
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
646 >>> mapname(m, b'Commit <commit3@email.xx>')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
647 'Name <proper@email.xx>'
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
648 >>> mapname(m, b'Commit <commit4@email.xx>')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
649 'Name <proper@email.xx>'
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
650 >>> mapname(m, b'Unknown Name <unknown@email.com>')
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
651 'Unknown Name <unknown@email.com>'
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
652 """
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
653 # If the author field coming in isn't in the correct format,
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
654 # or the mailmap is empty just return the original author field
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
655 if not isauthorwellformed(author) or not mailmap:
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
656 return author
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
657
37247
2ed180117f76 stringutil: edit comment to reflect actual data type name
Connor Sheehan <sheehan@mozilla.com>
parents: 37246
diff changeset
658 # Turn the user name into a mailmapping
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
659 commit = mailmapping(name=person(author), email=email(author))
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
660
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
661 try:
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
662 # Try and use both the commit email and name as the key
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
663 proper = mailmap[commit]
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
664
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
665 except KeyError:
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
666 # If the lookup fails, use just the email as the key instead
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
667 # We call this commit2 as not to erase original commit fields
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
668 commit2 = mailmapping(email=commit.email)
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
669 proper = mailmap.get(commit2, mailmapping(None, None))
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
670
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
671 # Return the author field with proper values filled in
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
672 return b'%s <%s>' % (
37210
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
673 proper.name if proper.name else commit.name,
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
674 proper.email if proper.email else commit.email,
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
675 )
2a2ce93e12f4 templatefuncs: add mailmap template function
Connor Sheehan <sheehan@mozilla.com>
parents: 37155
diff changeset
676
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
677
44022
c1ccefb513e4 cleanup: drop redundant character escapes outside of `[]`
Matt Harbison <matt_harbison@yahoo.com>
parents: 43506
diff changeset
678 _correctauthorformat = remod.compile(br'^[^<]+\s<[^<>]+@[^<>]+>$')
37154
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
679
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
680
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
681 def isauthorwellformed(author: bytes) -> bool:
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
682 """Return True if the author field is well formed
37154
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
683 (ie "Contributor Name <contrib@email.dom>")
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
684
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
685 >>> isauthorwellformed(b'Good Author <good@author.com>')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
686 True
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
687 >>> isauthorwellformed(b'Author <good@author.com>')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
688 True
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
689 >>> isauthorwellformed(b'Bad Author')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
690 False
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
691 >>> isauthorwellformed(b'Bad Author <author@author.com')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
692 False
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
693 >>> isauthorwellformed(b'Bad Author author@author.com')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
694 False
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
695 >>> isauthorwellformed(b'<author@author.com>')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
696 False
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
697 >>> isauthorwellformed(b'Bad Author <author>')
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
698 False
45942
89a2afe31e82 formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents: 45724
diff changeset
699 """
37154
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
700 return _correctauthorformat.match(author) is not None
f8e1f48de118 stringutil: add isauthorwellformed function
Connor Sheehan <sheehan@mozilla.com>
parents: 37083
diff changeset
701
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
702
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
703 def firstline(text: bytes) -> bytes:
49021
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
704 """Return the first line of the input"""
49030
75794847ef62 stringutil: try to avoid running `splitlines()` only to get first line
Martin von Zweigbergk <martinvonz@google.com>
parents: 49021
diff changeset
705 # Try to avoid running splitlines() on the whole string
75794847ef62 stringutil: try to avoid running `splitlines()` only to get first line
Martin von Zweigbergk <martinvonz@google.com>
parents: 49021
diff changeset
706 i = text.find(b'\n')
75794847ef62 stringutil: try to avoid running `splitlines()` only to get first line
Martin von Zweigbergk <martinvonz@google.com>
parents: 49021
diff changeset
707 if i != -1:
75794847ef62 stringutil: try to avoid running `splitlines()` only to get first line
Martin von Zweigbergk <martinvonz@google.com>
parents: 49021
diff changeset
708 text = text[:i]
49021
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
709 try:
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
710 return text.splitlines()[0]
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
711 except IndexError:
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
712 return b''
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
713
51aed118f9dc templates: extract function to `stringutil` for getting first line of text
Martin von Zweigbergk <martinvonz@google.com>
parents: 48946
diff changeset
714
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
715 def ellipsis(text: bytes, maxlength: int = 400) -> bytes:
21857
86c2d792a4b7 util: replace 'ellipsis' implementation by 'encoding.trim'
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 21813
diff changeset
716 """Trim string to at most maxlength (default: 400) columns in display."""
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
717 return encoding.trim(text, maxlength, ellipsis=b'...')
3767
1861fa38a6a7 Move ellipsis code to util.ellipsis() and improve maxlength handling.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 3721
diff changeset
718
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
719
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
720 def escapestr(s: bytes) -> bytes:
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
721 # "bytes" is also a typing shortcut for bytes, bytearray, and memoryview
39063
1419ba5e3b56 stringutil: if we get a memoryview in escapestr, coerce it to bytes
Augie Fackler <augie@google.com>
parents: 39052
diff changeset
722 if isinstance(s, memoryview):
1419ba5e3b56 stringutil: if we get a memoryview in escapestr, coerce it to bytes
Augie Fackler <augie@google.com>
parents: 39052
diff changeset
723 s = bytes(s)
31453
3b7a6941a6ef py3: call codecs.escape_encode() directly
Yuya Nishihara <yuya@tcha.org>
parents: 31451
diff changeset
724 # call underlying function of s.encode('string_escape') directly for
3b7a6941a6ef py3: call codecs.escape_encode() directly
Yuya Nishihara <yuya@tcha.org>
parents: 31451
diff changeset
725 # Python 3 compatibility
49648
9be765b82a90 typing: minor tweaks to allow updating to pytype 2022.11.18
Matt Harbison <matt_harbison@yahoo.com>
parents: 49575
diff changeset
726 # pytype: disable=bad-return-type
48479
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
727 return codecs.escape_encode(s)[0] # pytype: disable=module-attr
49648
9be765b82a90 typing: minor tweaks to allow updating to pytype 2022.11.18
Matt Harbison <matt_harbison@yahoo.com>
parents: 49575
diff changeset
728 # pytype: enable=bad-return-type
31451
53865692a354 util: wrap s.encode('string_escape') call for future py3 compatibility
Yuya Nishihara <yuya@tcha.org>
parents: 31449
diff changeset
729
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
730
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
731 def unescapestr(s: bytes) -> bytes:
49648
9be765b82a90 typing: minor tweaks to allow updating to pytype 2022.11.18
Matt Harbison <matt_harbison@yahoo.com>
parents: 49575
diff changeset
732 # pytype: disable=bad-return-type
48479
dcdecec401ca pytype: stop excluding stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 47189
diff changeset
733 return codecs.escape_decode(s)[0] # pytype: disable=module-attr
49648
9be765b82a90 typing: minor tweaks to allow updating to pytype 2022.11.18
Matt Harbison <matt_harbison@yahoo.com>
parents: 49575
diff changeset
734 # pytype: enable=bad-return-type
31484
afb335353d28 util: wrap s.decode('string_escape') calls for future py3 compatibility
Yuya Nishihara <yuya@tcha.org>
parents: 31465
diff changeset
735
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
736
33682
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
737 def forcebytestr(obj):
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
738 """Portably format an arbitrary object (e.g. exception) into a byte
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
739 string."""
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
740 try:
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
741 return pycompat.bytestr(obj)
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
742 except UnicodeEncodeError:
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
743 # non-ascii string, may be lossy
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
744 return pycompat.bytestr(encoding.strtolocal(str(obj)))
1d5e497c08b3 py3: convert arbitrary exception object to byte string more reliably
Yuya Nishihara <yuya@tcha.org>
parents: 33619
diff changeset
745
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
746
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
747 def uirepr(s: bytes) -> bytes:
5291
23651848d638 extdiff: avoid repr() doubling paths backslashes under Windows
Patrick Mezard <pmezard@gmail.com>
parents: 5213
diff changeset
748 # Avoid double backslash in Windows path repr()
36266
1fa33bd848ee py3: fix bytes-unicode dance while building docstring of extdiff
Yuya Nishihara <yuya@tcha.org>
parents: 36235
diff changeset
749 return pycompat.byterepr(pycompat.bytestr(s)).replace(b'\\\\', b'\\')
7547
4949729ee9ee python implementation of diffstat
Alexander Solovyov <piranha@piranha.org.ua>
parents: 7537
diff changeset
750
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
751
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
752 # delay import of textwrap
37079
736024df4498 util: mark MBTextWrapper as private
Yuya Nishihara <yuya@tcha.org>
parents: 37078
diff changeset
753 def _MBTextWrapper(**kwargs):
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
754 class tw(textwrap.TextWrapper):
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
755 """
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
756 Extend TextWrapper for width-awareness.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
757
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
758 Neither number of 'bytes' in any encoding nor 'characters' is
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
759 appropriate to calculate terminal columns for specified string.
12957
9f2ac318b92e util: clarify purpose of MBTextWrapper class
Nicolas Dumazet <nicdumz.commits@gmail.com>
parents: 12938
diff changeset
760
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
761 Original TextWrapper implementation uses built-in 'len()' directly,
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
762 so overriding is needed to use width information of each characters.
12957
9f2ac318b92e util: clarify purpose of MBTextWrapper class
Nicolas Dumazet <nicdumz.commits@gmail.com>
parents: 12938
diff changeset
763
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
764 In addition, characters classified into 'ambiguous' width are
17424
e7cfe3587ea4 fix trivial spelling errors
Mads Kiilerich <mads@kiilerich.com>
parents: 17391
diff changeset
765 treated as wide in East Asian area, but as narrow in other.
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
766
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
767 This requires use decision to determine width of such characters.
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
768 """
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
769
15065
24a6c3f903bb util: wrap lines with multi-byte characters correctly (issue2943)
Mads Kiilerich <mads@kiilerich.com>
parents: 15024
diff changeset
770 def _cutdown(self, ucstr, space_left):
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
771 l = 0
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
772 colwidth = encoding.ucolwidth
49284
d44e3c45f0e4 py3: replace `pycompat.xrange` by `range`
Manuel Jacob <me@manueljacob.de>
parents: 49030
diff changeset
773 for i in range(len(ucstr)):
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
774 l += colwidth(ucstr[i])
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
775 if space_left < l:
15065
24a6c3f903bb util: wrap lines with multi-byte characters correctly (issue2943)
Mads Kiilerich <mads@kiilerich.com>
parents: 15024
diff changeset
776 return (ucstr[:i], ucstr[i:])
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
777 return ucstr, b''
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
778
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
779 # overriding of base class
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
780 def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width):
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
781 space_left = max(width - cur_len, 1)
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
782
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
783 if self.break_long_words:
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
784 cut, res = self._cutdown(reversed_chunks[-1], space_left)
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
785 cur_line.append(cut)
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
786 reversed_chunks[-1] = res
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
787 elif not cur_line:
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
788 cur_line.append(reversed_chunks.pop())
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
789
26201
c5b2074ae8c0 util: capitalize Python in MBTextWrapper._wrap_chunks comment
timeless@mozdev.org
parents: 26126
diff changeset
790 # this overriding code is imported from TextWrapper of Python 2.6
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
791 # to calculate columns of string by 'encoding.ucolwidth()'
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
792 def _wrap_chunks(self, chunks):
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
793 colwidth = encoding.ucolwidth
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
794
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
795 lines = []
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
796 if self.width <= 0:
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
797 raise ValueError(b"invalid width %r (must be > 0)" % self.width)
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
798
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
799 # Arrange in reverse order so items can be efficiently popped
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
800 # from a stack of chucks.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
801 chunks.reverse()
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
802
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
803 while chunks:
51700
7f0cb9ee0534 Backout accidental publication of a large range of revisions
Raphaël Gomès <rgomes@octobus.net>
parents: 51690
diff changeset
804
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
805 # Start the list of chunks that will make up the current line.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
806 # cur_len is just the length of all the chunks in cur_line.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
807 cur_line = []
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
808 cur_len = 0
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
809
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
810 # Figure out which static string will prefix this line.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
811 if lines:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
812 indent = self.subsequent_indent
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
813 else:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
814 indent = self.initial_indent
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
815
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
816 # Maximum width for this line.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
817 width = self.width - len(indent)
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
818
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
819 # First chunk on line is whitespace -- drop it, unless this
17424
e7cfe3587ea4 fix trivial spelling errors
Mads Kiilerich <mads@kiilerich.com>
parents: 17391
diff changeset
820 # is the very beginning of the text (i.e. no lines started yet).
43506
9f70512ae2cf cleanup: remove pointless r-prefixes on single-quoted strings
Augie Fackler <augie@google.com>
parents: 43077
diff changeset
821 if self.drop_whitespace and chunks[-1].strip() == '' and lines:
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
822 del chunks[-1]
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
823
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
824 while chunks:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
825 l = colwidth(chunks[-1])
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
826
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
827 # Can at least squeeze this chunk onto the current line.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
828 if cur_len + l <= width:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
829 cur_line.append(chunks.pop())
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
830 cur_len += l
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
831
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
832 # Nope, this line is full.
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
833 else:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
834 break
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
835
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
836 # The current line is full, and the next chunk is too big to
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
837 # fit on *any* line (not just this one).
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
838 if chunks and colwidth(chunks[-1]) > width:
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
839 self._handle_long_word(chunks, cur_line, cur_len, width)
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
840
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
841 # If the last chunk on this line is all whitespace, drop it.
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
842 if (
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
843 self.drop_whitespace
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
844 and cur_line
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
845 and cur_line[-1].strip() == r''
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
846 ):
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
847 del cur_line[-1]
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
848
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
849 # Convert current line back to a string and store it in list
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
850 # of all lines (return value).
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
851 if cur_line:
43506
9f70512ae2cf cleanup: remove pointless r-prefixes on single-quoted strings
Augie Fackler <augie@google.com>
parents: 43077
diff changeset
852 lines.append(indent + ''.join(cur_line))
15066
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
853
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
854 return lines
24efa83d81cb i18n: calculate terminal columns by width information of each characters
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 15065
diff changeset
855
37079
736024df4498 util: mark MBTextWrapper as private
Yuya Nishihara <yuya@tcha.org>
parents: 37078
diff changeset
856 global _MBTextWrapper
736024df4498 util: mark MBTextWrapper as private
Yuya Nishihara <yuya@tcha.org>
parents: 37078
diff changeset
857 _MBTextWrapper = tw
13316
d119403fd266 util: delay loading of textwrap
Matt Mackall <mpm@selenic.com>
parents: 13313
diff changeset
858 return tw(**kwargs)
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
859
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
860
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
861 def wrap(
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
862 line: bytes, width: int, initindent: bytes = b'', hangindent: bytes = b''
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
863 ) -> bytes:
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
864 maxindent = max(len(hangindent), len(initindent))
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
865 if width <= maxindent:
9417
4c3fb45123e5 util, minirst: do not crash with COLUMNS=0
Martin Geisler <mg@lazybytes.net>
parents: 9397
diff changeset
866 # adjust for weird terminal size
11297
d320e70442a5 replace Python standard textwrap by MBCS sensitive one for i18n text
FUJIWARA Katsunori <foozy@lares.dti.ne.jp>
parents: 11256
diff changeset
867 width = max(78, maxindent + 1)
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
868 line = line.decode(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
869 pycompat.sysstr(encoding.encoding),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
870 pycompat.sysstr(encoding.encodingmode),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
871 )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
872 initindent = initindent.decode(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
873 pycompat.sysstr(encoding.encoding),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
874 pycompat.sysstr(encoding.encodingmode),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
875 )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
876 hangindent = hangindent.decode(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
877 pycompat.sysstr(encoding.encoding),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
878 pycompat.sysstr(encoding.encodingmode),
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
879 )
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
880 wrapper = _MBTextWrapper(
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
881 width=width, initial_indent=initindent, subsequent_indent=hangindent
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
882 )
31338
a9a28ca17615 util: pass encoding.[encoding|encodingmode] as unicodes
Pulkit Goyal <7895pulkit@gmail.com>
parents: 31315
diff changeset
883 return wrapper.fill(line).encode(pycompat.sysstr(encoding.encoding))
8938
9b8c9266c59d commands: wrap short descriptions in 'hg help'
Martin Geisler <mg@lazybytes.net>
parents: 8785
diff changeset
884
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
885
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
886 _booleans = {
43077
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
887 b'1': True,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
888 b'yes': True,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
889 b'true': True,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
890 b'on': True,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
891 b'always': True,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
892 b'0': False,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
893 b'no': False,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
894 b'false': False,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
895 b'off': False,
687b865b95ad formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents: 43076
diff changeset
896 b'never': False,
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
897 }
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
898
12087
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
899
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
900 def parsebool(s: bytes) -> Optional[bool]:
12087
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
901 """Parse s into a boolean.
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
902
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
903 If s is not a valid boolean, returns None.
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
904 """
a88a4720c2f0 parsebool: create new function and use it for config parsing
Augie Fackler <durin42@gmail.com>
parents: 12086
diff changeset
905 return _booleans.get(s.lower(), None)
37290
cc5a040fe150 wireproto: syntax for encoding CBOR into frames
Gregory Szorc <gregory.szorc@gmail.com>
parents: 37247
diff changeset
906
43076
2372284d9457 formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents: 40684
diff changeset
907
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
908 # TODO: make arg mandatory (and fix code below?)
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
909 def parselist(value: Optional[bytes]):
47189
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
910 """parse a configuration value as a list of comma/space separated strings
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
911
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
912 >>> parselist(b'this,is "a small" ,test')
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
913 ['this', 'is', 'a small', 'test']
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
914 """
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
915
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
916 def _parse_plain(parts, s, offset):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
917 whitespace = False
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
918 while offset < len(s) and (
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
919 s[offset : offset + 1].isspace() or s[offset : offset + 1] == b','
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
920 ):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
921 whitespace = True
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
922 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
923 if offset >= len(s):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
924 return None, parts, offset
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
925 if whitespace:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
926 parts.append(b'')
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
927 if s[offset : offset + 1] == b'"' and not parts[-1]:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
928 return _parse_quote, parts, offset + 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
929 elif s[offset : offset + 1] == b'"' and parts[-1][-1:] == b'\\':
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
930 parts[-1] = parts[-1][:-1] + s[offset : offset + 1]
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
931 return _parse_plain, parts, offset + 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
932 parts[-1] += s[offset : offset + 1]
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
933 return _parse_plain, parts, offset + 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
934
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
935 def _parse_quote(parts, s, offset):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
936 if offset < len(s) and s[offset : offset + 1] == b'"': # ""
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
937 parts.append(b'')
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
938 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
939 while offset < len(s) and (
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
940 s[offset : offset + 1].isspace()
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
941 or s[offset : offset + 1] == b','
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
942 ):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
943 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
944 return _parse_plain, parts, offset
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
945
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
946 while offset < len(s) and s[offset : offset + 1] != b'"':
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
947 if (
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
948 s[offset : offset + 1] == b'\\'
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
949 and offset + 1 < len(s)
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
950 and s[offset + 1 : offset + 2] == b'"'
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
951 ):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
952 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
953 parts[-1] += b'"'
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
954 else:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
955 parts[-1] += s[offset : offset + 1]
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
956 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
957
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
958 if offset >= len(s):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
959 real_parts = _configlist(parts[-1])
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
960 if not real_parts:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
961 parts[-1] = b'"'
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
962 else:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
963 real_parts[0] = b'"' + real_parts[0]
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
964 parts = parts[:-1]
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
965 parts.extend(real_parts)
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
966 return None, parts, offset
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
967
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
968 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
969 while offset < len(s) and s[offset : offset + 1] in [b' ', b',']:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
970 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
971
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
972 if offset < len(s):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
973 if offset + 1 == len(s) and s[offset : offset + 1] == b'"':
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
974 parts[-1] += b'"'
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
975 offset += 1
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
976 else:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
977 parts.append(b'')
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
978 else:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
979 return None, parts, offset
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
980
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
981 return _parse_plain, parts, offset
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
982
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
983 def _configlist(s):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
984 s = s.rstrip(b' ,')
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
985 if not s:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
986 return []
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
987 parser, parts, offset = _parse_plain, [b''], 0
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
988 while parser:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
989 parser, parts, offset = parser(parts, s, offset)
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
990 return parts
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
991
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
992 if value is not None and isinstance(value, bytes):
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
993 result = _configlist(value.lstrip(b' ,\n'))
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
994 else:
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
995 result = value
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
996 return result or []
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
997
b0e92313107e parselist: move the function from config to stringutil
Pierre-Yves David <pierre-yves.david@octobus.net>
parents: 46819
diff changeset
998
49575
bbbb5213d043 typing: add basic type hints to stringutil.py
Matt Harbison <matt_harbison@yahoo.com>
parents: 49284
diff changeset
999 def evalpythonliteral(s: bytes):
37476
e9dea82ea1f3 wireproto: convert python literal to object without using unsafe eval()
Yuya Nishihara <yuya@tcha.org>
parents: 37322
diff changeset
1000 """Evaluate a string containing a Python literal expression"""
e9dea82ea1f3 wireproto: convert python literal to object without using unsafe eval()
Yuya Nishihara <yuya@tcha.org>
parents: 37322
diff changeset
1001 # We could backport our tokenizer hack to rewrite '' to u'' if we want
48911
46b3ecfb16e2 stringutil: remove Python 2 support code
Gregory Szorc <gregory.szorc@gmail.com>
parents: 48875
diff changeset
1002 return ast.literal_eval(s.decode('latin1'))