author | Yuya Nishihara <yuya@tcha.org> |
Mon, 07 Dec 2020 20:12:36 +0900 | |
changeset 46065 | 56a5ea353182 |
parent 45942 | 89a2afe31e82 |
child 46819 | d4ba4d51f85f |
permissions | -rw-r--r-- |
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
1 |
# charencode.py - miscellaneous character encoding |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
2 |
# |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
3 |
# Copyright 2005-2009 Matt Mackall <mpm@selenic.com> and others |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
4 |
# |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
5 |
# This software may be used and distributed according to the terms of the |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
6 |
# GNU General Public License version 2 or any later version. |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
7 |
|
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
8 |
from __future__ import absolute_import |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
9 |
|
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
10 |
import array |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
11 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
12 |
from .. import pycompat |
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
13 |
|
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
14 |
|
33926
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
15 |
def isasciistr(s): |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
16 |
try: |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
17 |
s.decode('ascii') |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
18 |
return True |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
19 |
except UnicodeDecodeError: |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
20 |
return False |
f4433f2713d0
encoding: add function to test if a str consists of ASCII characters
Yuya Nishihara <yuya@tcha.org>
parents:
33924
diff
changeset
|
21 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
22 |
|
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
23 |
def asciilower(s): |
45942
89a2afe31e82
formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents:
43506
diff
changeset
|
24 |
"""convert a string to lowercase if ASCII |
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
25 |
|
45942
89a2afe31e82
formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents:
43506
diff
changeset
|
26 |
Raises UnicodeDecodeError if non-ASCII characters are found.""" |
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
27 |
s.decode('ascii') |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
28 |
return s.lower() |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
29 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
30 |
|
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
31 |
def asciiupper(s): |
45942
89a2afe31e82
formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents:
43506
diff
changeset
|
32 |
"""convert a string to uppercase if ASCII |
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
33 |
|
45942
89a2afe31e82
formating: upgrade to black 20.8b1
Augie Fackler <raf@durin42.com>
parents:
43506
diff
changeset
|
34 |
Raises UnicodeDecodeError if non-ASCII characters are found.""" |
33761
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
35 |
s.decode('ascii') |
f5fc54e7e467
encoding: drop circular import by proxying through '<policy>.charencode'
Yuya Nishihara <yuya@tcha.org>
parents:
diff
changeset
|
36 |
return s.upper() |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
37 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
38 |
|
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
39 |
_jsonmap = [] |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
40 |
_jsonmap.extend(b"\\u%04x" % x for x in range(32)) |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
41 |
_jsonmap.extend(pycompat.bytechr(x) for x in range(32, 127)) |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
42 |
_jsonmap.append(b'\\u007f') |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
43 |
_jsonmap[0x09] = b'\\t' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
44 |
_jsonmap[0x0A] = b'\\n' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
45 |
_jsonmap[0x22] = b'\\"' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
46 |
_jsonmap[0x5C] = b'\\\\' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
47 |
_jsonmap[0x08] = b'\\b' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
48 |
_jsonmap[0x0C] = b'\\f' |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
49 |
_jsonmap[0x0D] = b'\\r' |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
50 |
_paranoidjsonmap = _jsonmap[:] |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
51 |
_paranoidjsonmap[0x3C] = b'\\u003c' # '<' (e.g. escape "</script>") |
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
52 |
_paranoidjsonmap[0x3E] = b'\\u003e' # '>' |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
53 |
_jsonmap.extend(pycompat.bytechr(x) for x in range(128, 256)) |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
54 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
55 |
|
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
56 |
def jsonescapeu8fast(u8chars, paranoid): |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
57 |
"""Convert a UTF-8 byte string to JSON-escaped form (fast path) |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
58 |
|
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
59 |
Raises ValueError if non-ASCII characters have to be escaped. |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
60 |
""" |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
61 |
if paranoid: |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
62 |
jm = _paranoidjsonmap |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
63 |
else: |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
64 |
jm = _jsonmap |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
65 |
try: |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
66 |
return b''.join(jm[x] for x in bytearray(u8chars)) |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
67 |
except IndexError: |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
68 |
raise ValueError |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
69 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
70 |
|
34218
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
71 |
if pycompat.ispy3: |
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
72 |
_utf8strict = r'surrogatepass' |
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
73 |
else: |
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
74 |
_utf8strict = r'strict' |
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
75 |
|
43076
2372284d9457
formatting: blacken the codebase
Augie Fackler <augie@google.com>
parents:
34218
diff
changeset
|
76 |
|
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
77 |
def jsonescapeu8fallback(u8chars, paranoid): |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
78 |
"""Convert a UTF-8 byte string to JSON-escaped form (slow path) |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
79 |
|
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
80 |
Escapes all non-ASCII characters no matter if paranoid is False. |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
81 |
""" |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
82 |
if paranoid: |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
83 |
jm = _paranoidjsonmap |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
84 |
else: |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
85 |
jm = _jsonmap |
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
86 |
# non-BMP char is represented as UTF-16 surrogate pair |
34218
aa877860d4d7
py3: use 'surrogatepass' error handler to process U+DCxx transparently
Yuya Nishihara <yuya@tcha.org>
parents:
34217
diff
changeset
|
87 |
u16b = u8chars.decode('utf-8', _utf8strict).encode('utf-16', _utf8strict) |
43506
9f70512ae2cf
cleanup: remove pointless r-prefixes on single-quoted strings
Augie Fackler <augie@google.com>
parents:
43077
diff
changeset
|
88 |
u16codes = array.array('H', u16b) |
33924
b9101467d88b
encoding: extract stub for fast JSON escape
Yuya Nishihara <yuya@tcha.org>
parents:
33761
diff
changeset
|
89 |
u16codes.pop(0) # drop BOM |
43077
687b865b95ad
formatting: byteify all mercurial/ and hgext/ string literals
Augie Fackler <augie@google.com>
parents:
43076
diff
changeset
|
90 |
return b''.join(jm[x] if x < 128 else b'\\u%04x' % x for x in u16codes) |