htproc.py

   1 #!/usr/bin/env python\r
   2 #\r
   3 """htproc.py - HTML Processor"""\r
   4 \r
   5 import sys\r
   6 from BeautifulSoup import BeautifulSoup\r
   7 \r
   8 from rules import rules\r
   9 \r
  10 usage = """%s <src> <dest>""" % sys.argv[0]\r
  11 \r
  12 try:\r
  13     src = sys.argv[1]\r
  14     dest = sys.argv[2]\r
  15 except IndexError:\r
  16     sys.exit(usage)\r
  17 \r
  18 f_src = open(src, "r")\r
  19 f_dest = open(dest, "w")\r
  20 \r
  21 html = f_src.read()\r
  22 bsp = BeautifulSoup(html)\r
  23 \r
  24 for tag in rules:\r
  25     for result in bsp.findAll(tag):\r
  26         if rules[tag].has_key("begin"):\r
  27             result.insert(0, rules[tag]["begin"])\r
  28         if rules[tag].has_key("pre"):\r
  29             rawtag = result.prettify()\r
  30             if rules[tag].has_key("post"):\r
  31                 rawtag = rules[tag]["pre"] + rawtag + rules[tag]["post"]\r
  32             else:\r
  33                 rawtag = rules[tag]["pre"] + rawtag\r
  34             result.replaceWith(rawtag)\r
  35 \r
  36         elif rules[tag].has_key("pre"):\r
  37             rawtag = result.prettify() + rules[tag]["post"]\r
  38             result.replaceWith(rawtag)\r
  39 \r
  40 #f_dest.write(bsp.prettify())\r
  41 f_dest.write(bsp.renderContents())\r
  42 \r
  43 f_src.close()\r
  44 f_dest.close()\r
  45 \r