scripts/verify-indexterm-range

   1 #!/usr/bin/python
   2 #
   3 # Verify the <indexterm> ranges of a docbook document, ensuring
   4 # endofrange come after startofrange, and the IDs used by startofrange
   5 # are unique.
   6
   7 from  lxml import etree
   8 def main():
   9     filename = 'freeculture.xml'
  10     doc = etree.parse(filename)
  11     ids = {}
  12     order = 0
  13     for it in doc.getroot().xpath('//indexterm'):
  14         order = order + 1
  15         indextermclass = None
  16         id = None
  17         if 'class' in it.attrib:
  18             indextermclass = it.attrib['class']
  19             if 'startofrange' == indextermclass:
  20                 id=it.attrib['id']
  21                 if id in ids:
  22                     print "error: non-unique indexterm id: %s" % id
  23                 ids[id] = order
  24             if  'endofrange' == indextermclass:
  25                 id = it.attrib['startref']
  26                 if id not in ids:
  27                     print "error: indexterm id=\"%s\" not listed before endofrange" % id
  28             print indextermclass, id
  29
  30 if __name__ == "main":
  31     main()