##// END OF EJS Templates
win32mbcs: wrapped docstrings at 78 characters
Martin Geisler -
r9077:22a33e20 default
parent child Browse files
Show More
@@ -1,120 +1,117 b''
1 # win32mbcs.py -- MBCS filename support for Mercurial
1 # win32mbcs.py -- MBCS filename support for Mercurial
2 #
2 #
3 # Copyright (c) 2008 Shun-ichi Goto <shunichi.goto@gmail.com>
3 # Copyright (c) 2008 Shun-ichi Goto <shunichi.goto@gmail.com>
4 #
4 #
5 # Version: 0.2
5 # Version: 0.2
6 # Author: Shun-ichi Goto <shunichi.goto@gmail.com>
6 # Author: Shun-ichi Goto <shunichi.goto@gmail.com>
7 #
7 #
8 # This software may be used and distributed according to the terms of the
8 # This software may be used and distributed according to the terms of the
9 # GNU General Public License version 2, incorporated herein by reference.
9 # GNU General Public License version 2, incorporated herein by reference.
10 #
10 #
11
11
12 '''allow the use of MBCS paths with problematic encodings
12 '''allow the use of MBCS paths with problematic encodings
13
13
14 Some MBCS encodings are not good for some path operations (i.e.
14 Some MBCS encodings are not good for some path operations (i.e. splitting
15 splitting path, case conversion, etc.) with its encoded bytes. We call
15 path, case conversion, etc.) with its encoded bytes. We call such a encoding
16 such a encoding (i.e. shift_jis and big5) as "problematic encoding".
16 (i.e. shift_jis and big5) as "problematic encoding". This extension can be
17 This extension can be used to fix the issue with those encodings by
17 used to fix the issue with those encodings by wrapping some functions to
18 wrapping some functions to convert to Unicode string before path
18 convert to Unicode string before path operation.
19 operation.
20
19
21 This extension is useful for:
20 This extension is useful for:
22 * Japanese Windows users using shift_jis encoding.
21 * Japanese Windows users using shift_jis encoding.
23 * Chinese Windows users using big5 encoding.
22 * Chinese Windows users using big5 encoding.
24 * All users who use a repository with one of problematic encodings on
23 * All users who use a repository with one of problematic encodings on
25 case-insensitive file system.
24 case-insensitive file system.
26
25
27 This extension is not needed for:
26 This extension is not needed for:
28 * Any user who use only ASCII chars in path.
27 * Any user who use only ASCII chars in path.
29 * Any user who do not use any of problematic encodings.
28 * Any user who do not use any of problematic encodings.
30
29
31 Note that there are some limitations on using this extension:
30 Note that there are some limitations on using this extension:
32 * You should use single encoding in one repository.
31 * You should use single encoding in one repository.
33 * You should set same encoding for the repository by locale or
32 * You should set same encoding for the repository by locale or HGENCODING.
34 HGENCODING.
35
33
36 Path encoding conversion are done between Unicode and
34 Path encoding conversion are done between Unicode and encoding.encoding which
37 encoding.encoding which is decided by Mercurial from current locale
35 is decided by Mercurial from current locale setting or HGENCODING.
38 setting or HGENCODING.
39 '''
36 '''
40
37
41 import os
38 import os
42 from mercurial.i18n import _
39 from mercurial.i18n import _
43 from mercurial import util, encoding
40 from mercurial import util, encoding
44
41
45 def decode(arg):
42 def decode(arg):
46 if isinstance(arg, str):
43 if isinstance(arg, str):
47 uarg = arg.decode(encoding.encoding)
44 uarg = arg.decode(encoding.encoding)
48 if arg == uarg.encode(encoding.encoding):
45 if arg == uarg.encode(encoding.encoding):
49 return uarg
46 return uarg
50 raise UnicodeError("Not local encoding")
47 raise UnicodeError("Not local encoding")
51 elif isinstance(arg, tuple):
48 elif isinstance(arg, tuple):
52 return tuple(map(decode, arg))
49 return tuple(map(decode, arg))
53 elif isinstance(arg, list):
50 elif isinstance(arg, list):
54 return map(decode, arg)
51 return map(decode, arg)
55 return arg
52 return arg
56
53
57 def encode(arg):
54 def encode(arg):
58 if isinstance(arg, unicode):
55 if isinstance(arg, unicode):
59 return arg.encode(encoding.encoding)
56 return arg.encode(encoding.encoding)
60 elif isinstance(arg, tuple):
57 elif isinstance(arg, tuple):
61 return tuple(map(encode, arg))
58 return tuple(map(encode, arg))
62 elif isinstance(arg, list):
59 elif isinstance(arg, list):
63 return map(encode, arg)
60 return map(encode, arg)
64 return arg
61 return arg
65
62
66 def wrapper(func, args):
63 def wrapper(func, args):
67 # check argument is unicode, then call original
64 # check argument is unicode, then call original
68 for arg in args:
65 for arg in args:
69 if isinstance(arg, unicode):
66 if isinstance(arg, unicode):
70 return func(*args)
67 return func(*args)
71
68
72 try:
69 try:
73 # convert arguments to unicode, call func, then convert back
70 # convert arguments to unicode, call func, then convert back
74 return encode(func(*decode(args)))
71 return encode(func(*decode(args)))
75 except UnicodeError:
72 except UnicodeError:
76 # If not encoded with encoding.encoding, report it then
73 # If not encoded with encoding.encoding, report it then
77 # continue with calling original function.
74 # continue with calling original function.
78 raise util.Abort(_("[win32mbcs] filename conversion fail with"
75 raise util.Abort(_("[win32mbcs] filename conversion fail with"
79 " %s encoding\n") % (encoding.encoding))
76 " %s encoding\n") % (encoding.encoding))
80
77
81 def wrapname(name):
78 def wrapname(name):
82 idx = name.rfind('.')
79 idx = name.rfind('.')
83 module = name[:idx]
80 module = name[:idx]
84 name = name[idx+1:]
81 name = name[idx+1:]
85 module = globals()[module]
82 module = globals()[module]
86 func = getattr(module, name)
83 func = getattr(module, name)
87 def f(*args):
84 def f(*args):
88 return wrapper(func, args)
85 return wrapper(func, args)
89 try:
86 try:
90 f.__name__ = func.__name__ # fail with python23
87 f.__name__ = func.__name__ # fail with python23
91 except Exception:
88 except Exception:
92 pass
89 pass
93 setattr(module, name, f)
90 setattr(module, name, f)
94
91
95 # List of functions to be wrapped.
92 # List of functions to be wrapped.
96 # NOTE: os.path.dirname() and os.path.basename() are safe because
93 # NOTE: os.path.dirname() and os.path.basename() are safe because
97 # they use result of os.path.split()
94 # they use result of os.path.split()
98 funcs = '''os.path.join os.path.split os.path.splitext
95 funcs = '''os.path.join os.path.split os.path.splitext
99 os.path.splitunc os.path.normpath os.path.normcase os.makedirs
96 os.path.splitunc os.path.normpath os.path.normcase os.makedirs
100 util.endswithsep util.splitpath util.checkcase util.fspath'''
97 util.endswithsep util.splitpath util.checkcase util.fspath'''
101
98
102 # codec and alias names of sjis and big5 to be faked.
99 # codec and alias names of sjis and big5 to be faked.
103 problematic_encodings = '''big5 big5-tw csbig5 big5hkscs big5-hkscs
100 problematic_encodings = '''big5 big5-tw csbig5 big5hkscs big5-hkscs
104 hkscs cp932 932 ms932 mskanji ms-kanji shift_jis csshiftjis shiftjis
101 hkscs cp932 932 ms932 mskanji ms-kanji shift_jis csshiftjis shiftjis
105 sjis s_jis shift_jis_2004 shiftjis2004 sjis_2004 sjis2004
102 sjis s_jis shift_jis_2004 shiftjis2004 sjis_2004 sjis2004
106 shift_jisx0213 shiftjisx0213 sjisx0213 s_jisx0213 950 cp950 ms950 '''
103 shift_jisx0213 shiftjisx0213 sjisx0213 s_jisx0213 950 cp950 ms950 '''
107
104
108 def reposetup(ui, repo):
105 def reposetup(ui, repo):
109 # TODO: decide use of config section for this extension
106 # TODO: decide use of config section for this extension
110 if not os.path.supports_unicode_filenames:
107 if not os.path.supports_unicode_filenames:
111 ui.warn(_("[win32mbcs] cannot activate on this platform.\n"))
108 ui.warn(_("[win32mbcs] cannot activate on this platform.\n"))
112 return
109 return
113
110
114 # fake is only for relevant environment.
111 # fake is only for relevant environment.
115 if encoding.encoding.lower() in problematic_encodings.split():
112 if encoding.encoding.lower() in problematic_encodings.split():
116 for f in funcs.split():
113 for f in funcs.split():
117 wrapname(f)
114 wrapname(f)
118 ui.debug(_("[win32mbcs] activated with encoding: %s\n")
115 ui.debug(_("[win32mbcs] activated with encoding: %s\n")
119 % encoding.encoding)
116 % encoding.encoding)
120
117
General Comments 0
You need to be logged in to leave comments. Login now