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):
  17     with open(filename, 'r') as doc:
  18         soup = BeautifulSoup(doc.read(), 'lxml')
  19     with open(filename, 'w') as doc:
  20         html = soup.prettify()
  21         # strip things that mkhtml2 is not producing to reduce the diff
  22         html = html.replace('a class="link" href', 'a href')
  23         html = html.replace('summary="Navigation header" ', '')
  24         html = html.replace("""  <div class="footer">
  25    <hr/>
  26    Generated by GTK-Doc V1.28.1
  27   </div>
  28 """, '')
  29         html = re.sub(r'\s*<meta content="DocBook[^>]*>', '', html)
  30         html = re.sub(r'\s*<meta content="GTK-Doc[^>]*>', '', html)
  31         doc.write(html)
  32
  33
  34 def main(htmldir):
  35     for filename in glob.glob(os.path.join(htmldir, '*.devhelp2')):
  36         prettify(filename)
  37     for filename in glob.glob(os.path.join(htmldir, '*.html')):
  38         prettify(filename)
  39
  40
  41 if __name__ == '__main__':
  42     parser = argparse.ArgumentParser(
  43         description='c10e-html - canonicalize html files for diffing')
  44     parser.add_argument('args', nargs='*',  help='HTML_DIR')
  45
  46     options = parser.parse_args()
  47     if len(options.args) < 1:
  48         sys.exit('Too few arguments')
  49
  50     main(options.args[0])