tools/c10e-html.py

   1 #!/usr/bin/python3
   2 # canonicalize html dirs to ease comaring them
   3 #
   4 # run as:
   5 # ./tools/c10e-html html
   6
   7 import argparse
   8 import glob
   9 import re
  10 import os
  11 import sys
  12
  13 from bs4 import BeautifulSoup
  14
  15
  16 def prettify(filename, parser='lxml', fixup=False):
  17     with open(filename, 'r') as doc:
  18         soup = BeautifulSoup(doc.read(), parser)
  19     with open(filename, 'w') as doc:
  20         html = soup.prettify()
  21         if fixup:
  22             # strip things that mkhtml2 is not producing to reduce the diff
  23             html = html.replace('a class="link" href', 'a href')
  24             html = html.replace(' target="_top"', '')
  25             html = html.replace('summary="Navigation header" ', '')
  26             html = html.replace("""   <a name="idx">
  27    </a>
  28 """, '')
  29             html = re.sub("""  <div class="footer">
  30    <hr/>
  31    Generated by GTK-Doc V[.0-9]*
  32   </div>
  33 """, '', html)
  34             html = re.sub(r'\s*<p>\s*</p>', '', html)
  35             html = re.sub(r'\s*<a name="id-[.0-9]+">\s*</a>', '', html)
  36             html = re.sub(r'\s*<div class="titlepage">\s*</div>', '', html)
  37             html = re.sub(r'\s*<meta content="DocBook[^>]*>', '', html)
  38             html = re.sub(r'\s*<meta content="GTK-Doc[^>]*>', '', html)
  39         doc.write(html)
  40
  41
  42 def main(htmldir):
  43     for filename in glob.glob(os.path.join(htmldir, '*.devhelp2')):
  44         prettify(filename, parser='lxml-xml')
  45     for filename in glob.glob(os.path.join(htmldir, '*.html')):
  46         prettify(filename, fixup=True)
  47
  48
  49 if __name__ == '__main__':
  50     parser = argparse.ArgumentParser(
  51         description='c10e-html - canonicalize html files for diffing')
  52     parser.add_argument('args', nargs='*',  help='HTML_DIR')
  53
  54     options = parser.parse_args()
  55     if len(options.args) < 1:
  56         sys.exit('Too few arguments')
  57
  58     main(options.args[0])