annotate mercurial/encoding.py @ 8225:46293a0c7e9f

updated license to be explicit about GPL version 2
author Martin Geisler <mg@lazybytes.net>
date Sun, 26 Apr 2009 01:08:54 +0200
parents de377b1a9a84
children 8b2cd04a6e97
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
7948
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
1 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
2 encoding.py - character transcoding support for Mercurial
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
3
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
4 Copyright 2005-2009 Matt Mackall <mpm@selenic.com> and others
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
5
8225
46293a0c7e9f updated license to be explicit about GPL version 2
Martin Geisler <mg@lazybytes.net>
parents: 7948
diff changeset
6 This software may be used and distributed according to the terms of the
46293a0c7e9f updated license to be explicit about GPL version 2
Martin Geisler <mg@lazybytes.net>
parents: 7948
diff changeset
7 GNU General Public License version 2, incorporated herein by reference.
7948
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
8 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
9
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
10 import sys, unicodedata, locale, os, error
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
11
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
12 _encodingfixup = {'646': 'ascii', 'ANSI_X3.4-1968': 'ascii'}
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
13
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
14 try:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
15 encoding = os.environ.get("HGENCODING")
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
16 if sys.platform == 'darwin' and not encoding:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
17 # On darwin, getpreferredencoding ignores the locale environment and
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
18 # always returns mac-roman. We override this if the environment is
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
19 # not C (has been customized by the user).
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
20 locale.setlocale(locale.LC_CTYPE, '')
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
21 encoding = locale.getlocale()[1]
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
22 if not encoding:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
23 encoding = locale.getpreferredencoding() or 'ascii'
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
24 encoding = _encodingfixup.get(encoding, encoding)
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
25 except locale.Error:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
26 encoding = 'ascii'
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
27 encodingmode = os.environ.get("HGENCODINGMODE", "strict")
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
28 fallbackencoding = 'ISO-8859-1'
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
29
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
30 def tolocal(s):
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
31 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
32 Convert a string from internal UTF-8 to local encoding
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
33
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
34 All internal strings should be UTF-8 but some repos before the
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
35 implementation of locale support may contain latin1 or possibly
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
36 other character sets. We attempt to decode everything strictly
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
37 using UTF-8, then Latin-1, and failing that, we use UTF-8 and
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
38 replace unknown characters.
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
39 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
40 for e in ('UTF-8', fallbackencoding):
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
41 try:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
42 u = s.decode(e) # attempt strict decoding
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
43 return u.encode(encoding, "replace")
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
44 except LookupError, k:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
45 raise error.Abort("%s, please check your locale settings" % k)
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
46 except UnicodeDecodeError:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
47 pass
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
48 u = s.decode("utf-8", "replace") # last ditch
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
49 return u.encode(encoding, "replace")
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
50
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
51 def fromlocal(s):
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
52 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
53 Convert a string from the local character encoding to UTF-8
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
54
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
55 We attempt to decode strings using the encoding mode set by
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
56 HGENCODINGMODE, which defaults to 'strict'. In this mode, unknown
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
57 characters will cause an error message. Other modes include
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
58 'replace', which replaces unknown characters with a special
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
59 Unicode character, and 'ignore', which drops the character.
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
60 """
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
61 try:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
62 return s.decode(encoding, encodingmode).encode("utf-8")
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
63 except UnicodeDecodeError, inst:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
64 sub = s[max(0, inst.start-10):inst.start+10]
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
65 raise error.Abort("decoding near '%s': %s!" % (sub, inst))
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
66 except LookupError, k:
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
67 raise error.Abort("%s, please check your locale settings" % k)
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
68
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
69 def colwidth(s):
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
70 "Find the column width of a UTF-8 string for display"
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
71 d = s.decode(encoding, 'replace')
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
72 if hasattr(unicodedata, 'east_asian_width'):
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
73 w = unicodedata.east_asian_width
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
74 return sum([w(c) in 'WF' and 2 or 1 for c in d])
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
75 return len(d)
de377b1a9a84 move encoding bits from util to encoding
Matt Mackall <mpm@selenic.com>
parents:
diff changeset
76