##// END OF EJS Templates
win32mbcs: util.normpath should be wrapped....
Shun-ichi GOTO -
r9463:bd017f35 default
parent child Browse files
Show More
@@ -1,147 +1,147 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.
15 splitting path, case conversion, etc.) with its encoded bytes. We call
15 splitting path, case conversion, etc.) with its encoded bytes. We call
16 such a encoding (i.e. shift_jis and big5) as "problematic encoding".
16 such a encoding (i.e. shift_jis and big5) as "problematic encoding".
17 This extension can be used to fix the issue with those encodings by
17 This extension can be used to fix the issue with those encodings by
18 wrapping some functions to convert to Unicode string before path
18 wrapping some functions to convert to Unicode string before path
19 operation.
19 operation.
20
20
21 This extension is useful for:
21 This extension is useful for:
22
22
23 - Japanese Windows users using shift_jis encoding.
23 - Japanese Windows users using shift_jis encoding.
24 - Chinese Windows users using big5 encoding.
24 - Chinese Windows users using big5 encoding.
25 - All users who use a repository with one of problematic encodings on
25 - All users who use a repository with one of problematic encodings on
26 case-insensitive file system.
26 case-insensitive file system.
27
27
28 This extension is not needed for:
28 This extension is not needed for:
29
29
30 - Any user who use only ASCII chars in path.
30 - Any user who use only ASCII chars in path.
31 - Any user who do not use any of problematic encodings.
31 - Any user who do not use any of problematic encodings.
32
32
33 Note that there are some limitations on using this extension:
33 Note that there are some limitations on using this extension:
34
34
35 - You should use single encoding in one repository.
35 - You should use single encoding in one repository.
36 - You should set same encoding for the repository by locale or
36 - You should set same encoding for the repository by locale or
37 HGENCODING.
37 HGENCODING.
38
38
39 Path encoding conversion are done between Unicode and
39 Path encoding conversion are done between Unicode and
40 encoding.encoding which is decided by Mercurial from current locale
40 encoding.encoding which is decided by Mercurial from current locale
41 setting or HGENCODING.
41 setting or HGENCODING.
42 '''
42 '''
43
43
44 import os, sys
44 import os, sys
45 from mercurial.i18n import _
45 from mercurial.i18n import _
46 from mercurial import util, encoding
46 from mercurial import util, encoding
47
47
48 def decode(arg):
48 def decode(arg):
49 if isinstance(arg, str):
49 if isinstance(arg, str):
50 uarg = arg.decode(encoding.encoding)
50 uarg = arg.decode(encoding.encoding)
51 if arg == uarg.encode(encoding.encoding):
51 if arg == uarg.encode(encoding.encoding):
52 return uarg
52 return uarg
53 raise UnicodeError("Not local encoding")
53 raise UnicodeError("Not local encoding")
54 elif isinstance(arg, tuple):
54 elif isinstance(arg, tuple):
55 return tuple(map(decode, arg))
55 return tuple(map(decode, arg))
56 elif isinstance(arg, list):
56 elif isinstance(arg, list):
57 return map(decode, arg)
57 return map(decode, arg)
58 elif isinstance(arg, dict):
58 elif isinstance(arg, dict):
59 for k, v in arg.items():
59 for k, v in arg.items():
60 arg[k] = decode(v)
60 arg[k] = decode(v)
61 return arg
61 return arg
62
62
63 def encode(arg):
63 def encode(arg):
64 if isinstance(arg, unicode):
64 if isinstance(arg, unicode):
65 return arg.encode(encoding.encoding)
65 return arg.encode(encoding.encoding)
66 elif isinstance(arg, tuple):
66 elif isinstance(arg, tuple):
67 return tuple(map(encode, arg))
67 return tuple(map(encode, arg))
68 elif isinstance(arg, list):
68 elif isinstance(arg, list):
69 return map(encode, arg)
69 return map(encode, arg)
70 elif isinstance(arg, dict):
70 elif isinstance(arg, dict):
71 for k, v in arg.items():
71 for k, v in arg.items():
72 arg[k] = encode(v)
72 arg[k] = encode(v)
73 return arg
73 return arg
74
74
75 def appendsep(s):
75 def appendsep(s):
76 # ensure the path ends with os.sep, appending it if necessary.
76 # ensure the path ends with os.sep, appending it if necessary.
77 try:
77 try:
78 us = decode(s)
78 us = decode(s)
79 except UnicodeError:
79 except UnicodeError:
80 us = s
80 us = s
81 if us and us[-1] not in ':/\\':
81 if us and us[-1] not in ':/\\':
82 s += os.sep
82 s += os.sep
83 return s
83 return s
84
84
85 def wrapper(func, args, kwds):
85 def wrapper(func, args, kwds):
86 # check argument is unicode, then call original
86 # check argument is unicode, then call original
87 for arg in args:
87 for arg in args:
88 if isinstance(arg, unicode):
88 if isinstance(arg, unicode):
89 return func(*args, **kwds)
89 return func(*args, **kwds)
90
90
91 try:
91 try:
92 # convert arguments to unicode, call func, then convert back
92 # convert arguments to unicode, call func, then convert back
93 return encode(func(*decode(args), **decode(kwds)))
93 return encode(func(*decode(args), **decode(kwds)))
94 except UnicodeError:
94 except UnicodeError:
95 raise util.Abort(_("[win32mbcs] filename conversion failed with"
95 raise util.Abort(_("[win32mbcs] filename conversion failed with"
96 " %s encoding\n") % (encoding.encoding))
96 " %s encoding\n") % (encoding.encoding))
97
97
98 def wrapperforlistdir(func, args, kwds):
98 def wrapperforlistdir(func, args, kwds):
99 # Ensure 'path' argument ends with os.sep to avoids
99 # Ensure 'path' argument ends with os.sep to avoids
100 # misinterpreting last 0x5c of MBCS 2nd byte as path separator.
100 # misinterpreting last 0x5c of MBCS 2nd byte as path separator.
101 if args:
101 if args:
102 args = list(args)
102 args = list(args)
103 args[0] = appendsep(args[0])
103 args[0] = appendsep(args[0])
104 if 'path' in kwds:
104 if 'path' in kwds:
105 kwds['path'] = appendsep(kwds['path'])
105 kwds['path'] = appendsep(kwds['path'])
106 return func(*args, **kwds)
106 return func(*args, **kwds)
107
107
108 def wrapname(name, wrapper):
108 def wrapname(name, wrapper):
109 module, name = name.rsplit('.', 1)
109 module, name = name.rsplit('.', 1)
110 module = sys.modules[module]
110 module = sys.modules[module]
111 func = getattr(module, name)
111 func = getattr(module, name)
112 def f(*args, **kwds):
112 def f(*args, **kwds):
113 return wrapper(func, args, kwds)
113 return wrapper(func, args, kwds)
114 try:
114 try:
115 f.__name__ = func.__name__ # fail with python23
115 f.__name__ = func.__name__ # fail with python23
116 except Exception:
116 except Exception:
117 pass
117 pass
118 setattr(module, name, f)
118 setattr(module, name, f)
119
119
120 # List of functions to be wrapped.
120 # List of functions to be wrapped.
121 # NOTE: os.path.dirname() and os.path.basename() are safe because
121 # NOTE: os.path.dirname() and os.path.basename() are safe because
122 # they use result of os.path.split()
122 # they use result of os.path.split()
123 funcs = '''os.path.join os.path.split os.path.splitext
123 funcs = '''os.path.join os.path.split os.path.splitext
124 os.path.splitunc os.path.normpath os.path.normcase os.makedirs
124 os.path.splitunc os.path.normpath os.path.normcase os.makedirs
125 mercurial.util.endswithsep mercurial.util.splitpath mercurial.util.checkcase
125 mercurial.util.endswithsep mercurial.util.splitpath mercurial.util.checkcase
126 mercurial.util.fspath mercurial.util.pconvert'''
126 mercurial.util.fspath mercurial.util.pconvert mercurial.util.normpath'''
127
127
128 # codec and alias names of sjis and big5 to be faked.
128 # codec and alias names of sjis and big5 to be faked.
129 problematic_encodings = '''big5 big5-tw csbig5 big5hkscs big5-hkscs
129 problematic_encodings = '''big5 big5-tw csbig5 big5hkscs big5-hkscs
130 hkscs cp932 932 ms932 mskanji ms-kanji shift_jis csshiftjis shiftjis
130 hkscs cp932 932 ms932 mskanji ms-kanji shift_jis csshiftjis shiftjis
131 sjis s_jis shift_jis_2004 shiftjis2004 sjis_2004 sjis2004
131 sjis s_jis shift_jis_2004 shiftjis2004 sjis_2004 sjis2004
132 shift_jisx0213 shiftjisx0213 sjisx0213 s_jisx0213 950 cp950 ms950 '''
132 shift_jisx0213 shiftjisx0213 sjisx0213 s_jisx0213 950 cp950 ms950 '''
133
133
134 def reposetup(ui, repo):
134 def reposetup(ui, repo):
135 # TODO: decide use of config section for this extension
135 # TODO: decide use of config section for this extension
136 if not os.path.supports_unicode_filenames:
136 if not os.path.supports_unicode_filenames:
137 ui.warn(_("[win32mbcs] cannot activate on this platform.\n"))
137 ui.warn(_("[win32mbcs] cannot activate on this platform.\n"))
138 return
138 return
139
139
140 # fake is only for relevant environment.
140 # fake is only for relevant environment.
141 if encoding.encoding.lower() in problematic_encodings.split():
141 if encoding.encoding.lower() in problematic_encodings.split():
142 for f in funcs.split():
142 for f in funcs.split():
143 wrapname(f, wrapper)
143 wrapname(f, wrapper)
144 wrapname("mercurial.osutil.listdir", wrapperforlistdir)
144 wrapname("mercurial.osutil.listdir", wrapperforlistdir)
145 ui.debug(_("[win32mbcs] activated with encoding: %s\n")
145 ui.debug(_("[win32mbcs] activated with encoding: %s\n")
146 % encoding.encoding)
146 % encoding.encoding)
147
147
General Comments 0
You need to be logged in to leave comments. Login now