cryptocd/documents/mod_html_files.py

#!/usr/bin/env python
'''gets a latex2html file and returns the file with a css tag
and stuff'''
import string
import sys
import re


def writeFile(data, filename):
	"""
	write data to the given filename
		@param filename String : name of file to write to
	"""
	try:
		f = open(filename,"w")#oeffnen und schliessen =>
		f.close()			#datei ist jetzt genullt
		f = open(filename,"a")	#anhaengend oeffnen
		f.write(data)
		f.close()
		return ""
	except:
		print "(WW)[%s]: \"%s\" is not writeable!"%(__name__, filename)
		return filename


### start of code
try:
	f=open(sys.argv[1],"r")
	content=f.read()
	f.close()

	# zu_manipulierende_variable=string.replace(zu_manipulierende_variable,alter_string,neuer_string)

	# recently switched to latex2html, now cutting of headers
	# dieser abschnitt muss ueberarbeitet werden!
	startoffset=string.find(content,"<HR>\n<ADDRESS>")
	endoffset=string.find(content,"</ADDRESS>",startoffset)+len("</ADDRESS>")
	if (startoffset >10) and (endoffset>startoffset):
		footer=content[startoffset:endoffset]
		content=string.replace(content,footer,"")
	
	#put whole content into div tags
	content = string.replace(content,"</BODY>",'</div>')
	content = string.replace(content,"</HTML>",'<!-- end of main -->')

	#remove empty image subtitles
	content = string.replace(content,"<STRONG>Abbildung:</STRONG>","")

	# replace "./filename.png" references with "filename.png"
	# (this avoids validation warnings)
	content = string.replace(content,'HREF="./','HREF="')
	
	# Linie vor Fussnoten wird entfernt
	# Fussnoten bekommen id, damit sie per css formatierbar sind
	content = string.replace(content,'<HR><H4>Fu&szlig;noten</H4>\n<DL>','<H4>Fu&szlig;noten</H4>\n<DL id="footnotes">')
	
	# Sonderzeichen in Fussnoten werden ersetzt (Hotfix fuer einen latex2html-Bug)
	content = string.replace(content,'<EFBFBD>','&auml;')
	content = string.replace(content,'<EFBFBD>','&Auml;')
	content = string.replace(content,'<EFBFBD>','&ouml;')
	content = string.replace(content,'<EFBFBD>','&Ouml;')
	content = string.replace(content,'<EFBFBD>','&uuml;')
	content = string.replace(content,'<EFBFBD>','&Uuml;')
	content = string.replace(content,'<EFBFBD>','&szlig;')
	
	# Links innerhalb von div="main" die zu externen Seiten verweisen, bekommen ein tag
	content = string.replace(content,'HREF="http:','class="externerLink" href="http:')

	#all done, writing tuned files
	writeFile(content,sys.argv[1])

except: #read/write/whatever failed, 
	# BOESE: unklare Fehlerbehandlung
	print "an defined error occured - but i'm too lame to fix that"
	print "usage:",sys.argv[0],"<file_to_modify>"
	sys.exit(1)
sys.exit(0)