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 import subprocess
   9
  10 def main():
  11     filename = 'freeculture.xml'
  12
  13     # make sure entities are looked up / available
  14     # Based on idea from
  15     # http://stackoverflow.com/questions/14731633/how-to-resolve-external-entities-with-xml-etree-like-lxml-etree
  16     proc = subprocess.Popen(['xmllint','--noent',filename],stdout=subprocess.PIPE)
  17     output = proc.communicate()[0]
  18     doc = ElementTree.parse(StringIO.StringIO(output))
  19
  20     ids = {}
  21     order = 0
  22     for it in doc.getroot().xpath('//indexterm'):
  23         order = order + 1
  24         indextermclass = None
  25         id = None
  26         if 'class' in it.attrib:
  27             indextermclass = it.attrib['class']
  28             if 'startofrange' == indextermclass:
  29                 id=it.attrib['id']
  30                 if id in ids:
  31                     print "error: non-unique indexterm id: %s" % id
  32                 ids[id] = order
  33             if  'endofrange' == indextermclass:
  34                 id = it.attrib['startref']
  35                 if id not in ids:
  36                     print "error: indexterm id=\"%s\" not listed before endofrange" % id
  37             print indextermclass, id
  38
  39 if __name__ == "main":
  40     main()