#!/usr/bin/env python
#Adds trivial "alt=pix" tag to <img src=...>, in order to
#satisfy "Bobby compliance". http://www.cast.org/bobby/
#example usage: bobby.py *.html 

from sys import * # with this import sys.argv is simply argv
import re #with this import re.sub is still re.sub

#Make two "regular expression objects".  Note the use of "raw strings" with
#a r" ", rather than " ".  Raw strings do not convert \\ to a single \.  So
#not really needed here. But raw strings are the safe habit for regular expressions. 
#Note the 5 groups, meaning the patterns grouped in ( ).
re_img=re.compile(r"(^.*?)(<\s*img\s+src)(.*?)(>)(.*$)")

#This one is simple, just to check if "alt=" already exists:
re_alt=re.compile(r"alt=")

#in Python, argv[0] is the calling program, the rest of argv is the *.html glob
for file in argv[1:]:
	print file #print name of file that is being modified
	infile=open(file)
	all=infile.readlines() #put the lines of the file into a list
	infile.close()
	outfile=open(file,'w') #reopen the file for writing
	for line in all:
		found_img=re_img.search(line) #search for html "<img src=...", and save the 5 pieces
		if found_img:
			z=found_img.groups() #copy the dictionary with the labelled pieces
			middle=z[2] #the middle piece is labelled 'cat'
			if not re_alt.search(middle): #check to see that "alt=anything" does not exist
				print line, #print this line to monitor
				line=z[0]+z[1]+z[2]+' alt="pix"'+z[3]+z[4] #put in "alt=pix"
				print line  #print the modified line to monitor
		outfile.write(line) #write the modified line to the file
	outfile.close() #close the file after inspecting and modifying all lines