'''xml-parser.py

   Jeff Ondich, 5/29/09

   This program is a template program for parsing XML using the
   expat XML-parsing library.  This particular program just counts
   the XML tags in an input file.
'''

import sys
from xml.parsers import expat

class XMLFilter:
    ''' ... '''
    def __init__(self):
        self.parser = None
        self.tagCount = 0

    def report(self, out):
        out.write('There were %d tags.\n' % self.tagCount)

    def startElement(self, name, attributes):
        '''This method gets called for each opening tag.  name is the
           tag name, and attributes is a dictionary of names and values
           for the tag attributes.'''
        self.tagCount += 1

    def endElement(self, name):
        '''This method gets called for each closing tag.'''
        pass

    def characterData(self, data):
        '''This method gets called for each portion of text between a
           tag's opening and closing.  It's a bit tricky when the text
           is interspersed among other tags inside the tag the text
           belongs to.  For example, for <a>bbb<c>ddd</c>eee</a>,
           characterData will be called three times, with bbb, ddd,
           and eee.'''
        pass

    def parse(self, inFileName):
        '''Here, we instantiate a parser object and initialize its
           handler methods for the various types of XML-parsing events.'''
        self.parser = expat.ParserCreate()
        self.parser.StartElementHandler = self.startElement
        self.parser.EndElementHandler = self.endElement
        self.parser.CharacterDataHandler = self.characterData
        parserStatus = self.parser.Parse(open(inFileName).read(), 1)

if __name__ == '__main__':
    if len(sys.argv) != 2:
        sys.stderr.write('Usage: %s xmlFile\n' % sys.argv[0])
        sys.exit(1)

    inFileName = sys.argv[1]
    filter = XMLFilter()
    filter.parse(inFileName)
    filter.report(sys.stdout)