]>
Commit | Line | Data |
---|---|---|
11774a5d AF |
1 | #!/usr/bin/python |
2 | ||
3 | import os, os.path, re, sys, locale | |
4 | ||
5 | langs={ | |
6 | 'bg':'windows-1251', | |
7 | 'br':'iso8859-1', | |
8 | 'ca':'iso8859-1', | |
9 | 'cs':'iso8859-2', | |
10 | 'da':'iso8859-1', | |
11 | 'de':'iso8859-1', | |
12 | 'en':'iso8859-1', | |
13 | 'eo':'iso8859-3', | |
14 | 'es':'iso8859-1', | |
15 | 'fi':'iso8859-1', | |
16 | 'fo':'iso8859-1', | |
17 | 'fr':'iso8859-1', | |
18 | 'gl':'iso8859-1', | |
19 | 'he':'iso8859-8', | |
20 | 'id':'iso8859-1', | |
21 | 'is':'iso8859-1', | |
22 | 'it':'iso8859-1', | |
23 | 'ja':'euc-jp', | |
24 | 'ko':'euc-kr', | |
25 | 'nb':'iso8859-1', | |
26 | 'nl':'iso8859-1', | |
27 | 'pl':'iso8859-2', | |
28 | 'pt':'iso8859-1', | |
29 | 'pt_BR':'iso8859-1', | |
30 | 'ro':'iso8859-2', | |
e4fe63e5 | 31 | 'ru':'KOI8-R', |
11774a5d AF |
32 | 'se':'UTF-8', |
33 | 'sk':'iso8859-2', | |
34 | 'sl':'iso8859-2', | |
35 | 'sv':'iso8859-1', | |
36 | 'tr':'iso8859-9', | |
37 | 'uk':'KOI8-U', | |
38 | 'wa':'iso8859-1', | |
39 | 'zh_CN':'GB2312', | |
40 | 'zh_HK':'BIG5-HKSCS', | |
41 | 'zh_TW':'BIG5', | |
42 | 0:0} | |
43 | ||
fb3b78fc AF |
44 | def find_encoding(lang): |
45 | r = re.match("^([^.]+)(\.[^@]+)?$", lang) | |
46 | pure_lang = r.group(1) | |
47 | if r.group(2) == None: | |
48 | try: | |
49 | enc = langs[lang] | |
50 | except KeyError: | |
51 | enc = None | |
52 | else: | |
53 | # strip dot | |
54 | enc = r.group(2)[1:] | |
55 | return (enc, pure_lang) | |
56 | ||
11774a5d AF |
57 | def parse_spec(infile, outfile): |
58 | re_summary = re.compile("^Summary\(([^\)]+)\):\t+(.*)$") | |
fb3b78fc | 59 | re_utf = re.compile("^utf-8$", re.I) |
11774a5d AF |
60 | re_desc = re.compile("^(%description.*\s)-l\s+([\S]+)($|\s.*$)") |
61 | re_proc = re.compile("^%") | |
62 | in_desc = False | |
63 | ||
64 | for l in infile: | |
65 | outline = l | |
66 | r = re_summary.match(l) | |
67 | if r: | |
fb3b78fc AF |
68 | (enc, pure_lang) = find_encoding(r.group(1)) |
69 | if enc == None: | |
70 | outfile.write("#unknow lang code Summary(%s)\n" % (lang)) | |
71 | elif not re_utf.search(enc): | |
11774a5d | 72 | try: |
fb3b78fc AF |
73 | desc = unicode(r.group(2), enc).encode("UTF-8") |
74 | l = "Summary(%s.UTF-8): %s\n" % (pure_lang, desc) | |
11774a5d AF |
75 | except UnicodeDecodeError: |
76 | outfile.write("#transcoding error Summary(%s)\n" % (lang)) | |
11774a5d AF |
77 | if in_desc: |
78 | if re_proc.search(l): | |
79 | in_desc = False | |
80 | else: | |
fb3b78fc | 81 | if not re_utf.search(enc): |
11774a5d | 82 | try: |
fb3b78fc | 83 | l = unicode(l, enc).encode("UTF-8") |
11774a5d AF |
84 | except UnicodeDecodeError: |
85 | outfile.write("#transcoding error %%description -l %s\n" % (lang)) | |
86 | ||
87 | r = re_desc.match(l) | |
88 | if r: | |
fb3b78fc AF |
89 | (enc, pure_lang) = find_encoding(r.group(2)) |
90 | if enc == None: | |
983ae5dd | 91 | outfile.write("#unknown lang code in %%description -l %s\n" % (lang)) |
fb3b78fc AF |
92 | elif not re_utf.search(enc): |
93 | in_desc = True | |
94 | l = "%s-l %s.UTF-8%s\n" % (r.group(1), pure_lang, r.group(3)) | |
11774a5d AF |
95 | |
96 | outfile.write(l) | |
97 | ||
11774a5d AF |
98 | def main(): |
99 | parse_spec(sys.stdin, sys.stdout) | |
100 | ||
101 | if __name__ == "__main__": | |
102 | main() | |
103 |