buildscripts/wiki-slurp.py

   1 #!@PYTHON@
   2
   3 #
   4 # translate an entire Wiki site into local html.
   5 #
   6
   7 import re
   8 import urllib
   9 import sys
  10 import getopt
  11 import os
  12 program_version = '@TOPLEVEL_VERSION@'
  13 if program_version == '@' + 'TOPLEVEL_VERSION' + '@':
  14         program_version = '1.5.69'
  15
  16
  17 def help ():
  18         print """Usage: wiki-slurp.py [OPTION]... ENTRY-PATTERN...
  19
  20 Download a WikiWiki site and convert to local html.
  21
  22 Example: wiki-slurp.py -d /tmp/output 'http://c2.com/cgi-bin/wiki?'
  23
  24 Options:
  25   -h,--help          this help
  26   -m,--mangle        mangle file names to be shorter
  27   -d,--outdir=DIR    set output directory to DIR
  28   -v,--version       version information
  29
  30 Warning: downloading an entire Wiki site generates a huge amount of
  31 traffic and server load.  Consider asking for a copy of the database.
  32 See also http://c2.com/cgi-bin/wiki?WikiSlurp
  33
  34 Report bugs to bug-lilypond@gnu.org.
  35
  36 Written by Han-Wen Nienhuys <hanwen@cs.uu.nl>
  37 """
  38
  39 def print_version ():
  40         print r"""wiki-slurp.py %s
  41
  42 This is free software.  It is covered by the GNU General Public License,
  43 and you are welcome to change it and/or distribute copies of it under
  44 certain conditions.  Invoke as `midi2ly --warranty' for more information.
  45
  46 Copyright (c) 2000-2002 by Han-Wen Nienhuys <hanwen@cs.uu.nl>
  47 """ % program_version
  48
  49 (options, files) = getopt.getopt (sys.argv[1:], 'vd:hm', ['help','mangle','version', 'outdir='])
  50
  51
  52 def identity (name):
  53         return name
  54
  55 def mangle (name):
  56         return '%d' % hash (name)
  57
  58 mangler = identity
  59
  60 outdir = '/tmp/'
  61
  62 for opt in options:
  63         o = opt[0]
  64         a = opt[1]
  65         if o== '--help' or o == '-h':
  66                 help ()
  67                 sys.exit (0)
  68         elif o == '--version' or o == '-v':
  69                 print_version ()
  70                 sys.exit(0)
  71         elif  o == '--mangle' or o == '-m':
  72                 mangler = mangle
  73         elif o == '--outdir' or o == '-d':
  74                 outdir = a
  75         else:
  76                 print o
  77                 raise getopt.error
  78
  79
  80 patterns = files
  81
  82 if not patterns:
  83         help ()
  84         sys.stderr.write ("\n")
  85         sys.exit (2)
  86
  87 re_patterns = []
  88 for pattern in patterns:
  89         re_patterns.append (re.sub ('([?.])', '\\\\\\1', pattern))
  90
  91 todo = ["FrontPage"]
  92
  93 print 'here!'
  94 done = {
  95         'EditText': 1,
  96         }
  97
  98 def unwiki (str, pat, mangler):
  99         local = '<a href="%s([A-Za-z]+)">([A-Za-z]+)</a>' % pat
 100
 101         newurls = []
 102         def do_replace (match, us = newurls, mangler = mangler):
 103                 newurl = match.group (1)
 104                 local = mangler (newurl)
 105
 106                 replacement = '<a href="%s.html">%s</a>' % (local,newurl)
 107                 us.append (newurl)
 108                 return replacement
 109
 110         str = re.sub (local, do_replace, str)
 111         otherurl = '<a href="%s[^>]*">([?A-Za-z]+)</a>' % pat
 112         str = re.sub (otherurl, '\\1', str)
 113
 114         imagesrc = '<a href="%s[^>]*">(<img[^>]*>)</a>' % pat
 115         str = re.sub (imagesrc, '\\1', str)
 116
 117         return (str, newurls)
 118
 119
 120 while todo:
 121         f = todo[-1]
 122         todo = todo[:-1]
 123
 124         if done.has_key (f):
 125                 continue
 126         done [f] = 1
 127
 128         mangled = mangler (f)
 129
 130         sys.stderr.write ("reading `%s' ... " % f)
 131         sys.stderr.flush ()
 132
 133         page = urllib.urlopen (patterns[0] + f).read ()
 134         sys.stderr.write ('done. ')
 135         sys.stderr.flush ()
 136
 137         for re_pattern in re_patterns:
 138                 (page, nus) = unwiki (page, re_pattern, mangler)
 139                 todo.extend (nus)
 140
 141         outname = os.path.join (outdir, mangled) + '.html'
 142         fo = open (outname, 'w')
 143
 144         sys.stderr.write ("Writing `%s'\n" % outname)
 145         fo.write (page)
 146         fo.close ()
 147
 148
 149 # test
 150 if 0:
 151         page = open ('/tmp/FrontPage.html').read()
 152         (str, us)=unwiki (page, re_patterns)
 153         print str
 154         print us
 155