doc/markupSanitizer.py

   1 #!/usr/bin/env python
   2
   3 from bs4 import BeautifulSoup
   4 import sys
   5 import os
   6
   7 if sys.version < '3':
   8     import codecs
   9     def u(x):
  10         return codecs.unicode_escape_decode(x)[0]
  11 else:
  12     def u(x):
  13         return x
  14
  15 # Accept filename as user input
  16 argc = len( sys.argv )
  17 if (argc < 2): raise Exception
  18 fileName = sys.argv[1];
  19
  20 # Construct a DOM object
  21 soup = BeautifulSoup(open(fileName), "lxml")
  22
  23 # Assuming, tt tags are not spewed recklessly by latex2html,
  24 # replace them with code tags
  25 for t in soup('tt'):
  26     t.wrap( soup.new_tag('code') )
  27     t.unwrap()
  28
  29 # Rewrap all div class=alltt blocks in pre tags
  30 for d in soup('div','alltt'):
  31     d.wrap( soup.new_tag('pre') )
  32     d.unwrap()
  33
  34 # Remove br and span tags from within pre sections
  35 for p in soup('pre'):
  36     for b in p('br'):
  37         b.extract()
  38     for s in p('span'):
  39         s.unwrap()
  40
  41 # Remove all useless class 'arabic' spans
  42 for s in soup('span','arabic'):
  43     s.unwrap()
  44
  45 # Extract the navigation bar
  46 navmenu = soup.find('div', 'navigation')
  47 if navmenu:
  48     navmenu.extract()
  49
  50 # Wrap the remaining contents within a div
  51 if not soup.find('div', id='maincontainer'):
  52     soup.body['id'] = 'maincontainer'
  53     soup.body.name = 'div'
  54     soup.find('div', id='maincontainer').wrap( soup.new_tag('body') )
  55
  56 if navmenu:
  57     # If this navmenu doesn't already have a TOC, insert one
  58     if not navmenu.find('ul','manual-toc'):
  59         # Add a toc within the navmenu
  60         navmenuTOC = BeautifulSoup(open("tmp-navmenu.html"), "lxml")
  61         navmenuTOC = navmenuTOC.find('ul','manual-toc').extract()
  62         navmenuTOC.append( BeautifulSoup("".join([
  63         '<li><a href="http://charm.cs.illinois.edu">PPL Homepage</a></li>',
  64         '<li><a href="http://charm.cs.illinois.edu/help">Other Manuals</a></li>'])
  65         ) )
  66         navmenu.append(navmenuTOC)
  67
  68     # Insert navigation symbols to prev and next links
  69     prevsymbol = soup.new_tag('span')
  70     prevsymbol['class'] = 'navsymbol'
  71     prevsymbol.string = u('\xab')
  72     prv = navmenu.find('li',id='nav-prev')
  73     if prv:
  74         prv.find('a').insert(0, prevsymbol)
  75
  76     nextsymbol = soup.new_tag('span')
  77     nextsymbol['class'] = 'navsymbol'
  78     nextsymbol.string = u('\xbb')
  79     nxt = navmenu.find('li',id='nav-next')
  80     if nxt:
  81         nxt.find('a').append(nextsymbol)
  82
  83     # Reinsert the navigation bar at the end
  84     soup.body.append(navmenu)
  85
  86 # Extract the title
  87 titl = soup.find('title')
  88
  89 # Replace the head section with the user-supplied head markup
  90 soup.find('head').extract()
  91 newhead = BeautifulSoup(open("../assets/head.html"), "lxml")
  92 newhead = newhead.find('head').extract()
  93 newhead.append(titl)
  94 soup.html.body.insert_before(newhead)
  95
  96 # Print cleaned up markup to stdout
  97 print( soup.prettify(formatter="html") )
  98