"""
This program is for processing the XML export from Bobby. Bobby is a software that can exam the 
accessiblity of a web page accoring to the rules defined by Web accessibility initiative and 
American Rahabilitation Act Section 508. The result of the programm list the url of the web page, 
the error id defined in Bobby's preference and the number of errors that Bobby has found. This can
be used to measure the general web accessiblity of a web site.

Author: Xiaoming Zeng
Date: 2/1/2003
"""

version = '0.1'

from xml.sax.handler import ContentHandler
import xml.sax
import sys


#===================================================
#
# Count Handler subclass of ContentHandler
#
#===================================================
class countHandler(ContentHandler):
    
    #Construct of the class
    def __init__(self):
        self.tags={}    #Page tag dictionary
        self.ids={}     #Error ID dictionary
        self.url=""     #Universal Resoruce Locator 
        self.id = ""    #Error ID  
        self.count = 0  #Count for Page Number

    #Overload startElement method
    def startElement(self, name, attrs):
        if name=="page":
            self.url = attrs.get('url', "")
            if not self.tags.has_key(self.url):
                self.tags[self.url] = 0
            self.tags[self.url] += 1
        elif name=="error":
            self.id = attrs.get("id", "")
            if not self.ids.has_key(self.id):
                self.ids[self.id] = 0
            self.ids[self.id] += 1
        elif name=="instance":
            self.ids[self.id]+=1
            
    #Overload endElement method
    def endElement(self, name):
        
        if name=="page": #Finish processing a page element
            self.count = self.count+1
            print "Processing page ", self.count
            self.tags[self.url] = self.ids
            self.ids = {} #Clear ids dictionary
        elif name == "error":
            pass
            
#======================================================
#
#Output Result
#
#======================================================
    

 
parser = xml.sax.make_parser()
handler = countHandler()
parser.setContentHandler(handler)
inputfile = raw_input('Enter input file name: ')
parser.parse(inputfile)
outputfilename = raw_input('Enter output file name: ')
outputfile = open(outputfilename,'w')
for eachKey in handler.tags.keys():
    outputfile.write(eachKey + '\n')
    for eachKey2 in handler.tags[eachKey]:
        outputfile.write("\t " + eachKey2 + '\t:\t'+ str(handler.tags[eachKey][eachKey2]) + '\n')
    #print eachKey, handler.tags[eachKey]
outputfile.close()