| rev |
line source |
|
bos@149
|
1 #!/usr/bin/env python
|
|
bos@251
|
2 #
|
|
bos@251
|
3 # This script attempts to work around some of the more bizarre and
|
|
bos@251
|
4 # quirky behaviours of htlatex.
|
|
bos@251
|
5 #
|
|
bos@251
|
6 # - We've persuaded htlatex to produce UTF-8, which unfortunately
|
|
bos@251
|
7 # causes it to use huge character sequences to represent even the
|
|
bos@251
|
8 # safe 7-bit ASCII subset of UTF-8. We fix that up.
|
|
bos@251
|
9 #
|
|
bos@251
|
10 # - BUT we have to treat angle brackets (for example, redirections in
|
|
bos@251
|
11 # shell script snippets) specially, otherwise they'll break the
|
|
bos@251
|
12 # generated HTML. (Reported by Johannes Hoff.)
|
|
bos@251
|
13 #
|
|
bos@251
|
14 # - For some reason, htlatex gives a unique ID to each fancyvrb
|
|
bos@251
|
15 # environment, which makes writing a sane, small CSS stylesheet
|
|
bos@251
|
16 # impossible. We squish all those IDs down to nothing.
|
|
bos@149
|
17
|
|
bos@149
|
18 import os
|
|
bos@149
|
19 import sys
|
|
bos@149
|
20 import re
|
|
bos@149
|
21
|
|
bos@251
|
22 angle_re = re.compile(r'([CE];)')
|
|
bos@251
|
23 unicode_re = re.compile(r'�([0-7][0-9A-F]);')
|
|
bos@149
|
24 fancyvrb_re = re.compile(r'id="fancyvrb\d+"', re.I)
|
|
bos@149
|
25
|
|
bos@149
|
26 tmpsuffix = '.tmp.' + str(os.getpid())
|
|
bos@149
|
27
|
|
bos@251
|
28 def hide_angle(m):
|
|
bos@251
|
29 return m.group(1).lower()
|
|
bos@251
|
30
|
|
bos@149
|
31 def fix_ascii(m):
|
|
bos@149
|
32 return chr(int(m.group(1), 16))
|
|
bos@149
|
33
|
|
bos@149
|
34 for name in sys.argv[1:]:
|
|
bos@149
|
35 tmpname = name + tmpsuffix
|
|
bos@149
|
36 ofp = file(tmpname, 'w')
|
|
bos@149
|
37 for line in file(name):
|
|
bos@251
|
38 line = angle_re.sub(hide_angle, line)
|
|
bos@149
|
39 line = unicode_re.sub(fix_ascii, line)
|
|
bos@149
|
40 line = fancyvrb_re.sub('id="fancyvrb"', line)
|
|
bos@149
|
41 ofp.write(line)
|
|
bos@149
|
42 ofp.close()
|
|
bos@149
|
43 os.rename(tmpname, name)
|
