#!/usr/bin/env python
# -*- coding: iso-8859-1 -*-
# Checks if all internal links in a html file (<a href="#foo">)
# can be resolved. This is not checked by tidy.
# (c) Thomas Güttler, June 2003
# Copying, Changing allowed
# Feedback welcome thomas@thomas-guettler.de

import sys
import os
from HTMLParser import HTMLParser

class LinkParser(HTMLParser):
    link_to={}
    links={}
    
    def handle_starttag(self, tag, attributes):
        if tag=="a":
            href=None
            name=None
            for (att, value) in attributes:
                if att=="href":
                    href=value
                elif att=="name":
                    name=value
            if href:
                if href[0]=="#": # Internal Link
                    self.link_to[href[1:]]=self.getpos()
                else:
                    if (href.startswith("http") or href.startswith("ftp")):
                        #External Link
                        pass
                    else:
                        if not os.path.isfile(href):
                            print "%s not found: %s" % (href, self.getpos())
            if name:
                if self.links.get(name):
                    print '<a name="%s"> appears twice: %s' % (name,
                                                               self.getpos())
                self.links[name]=1

    def handle_endtag(self, tag):
        if tag=="html":
            for (link, pos) in self.link_to.items():
                if not self.links.get(link):
                    print 'Link "%s" is unkown: %s' % (link, pos)

if __name__=="__main__":
    filename=sys.argv[1]
    fd=open(filename)
    content=fd.read()
    fd.close()

    parser=LinkParser()
    parser.feed(content)
    parser.close()