#!/usr/bin/env python
import string
#import pdb
import pprint
import xmllib
import sys
SIPB = {
"start_sipb" : """""",
"end_sipb" : """""",
"start_footnote" : "",
"end_footnote" : "",
"start_title" : "",
"end_title" : ""
}
class Parser(xmllib.XMLParser):
#
def __init__(self, file=None):
xmllib.XMLParser.__init__(self)
self.__data = []
self.footnotes = []
self.__out = ""
if file:
self.load(file)
def load(self,file):
while 1:
s=file.read(512)
if not s:
break
self.feed(s)
self.close()
def start_sipb(self,attrs):
self.__data.append(SIPB["start_sipb"])
def start_title(self,attrs):
self.__data.append(SIPB["start_title"])
def end_title(self):
line = self.__data.pop()
if sys.argv[3] == 'title' :
print line
line = self.__data.pop() +line
self.__data.append(line +SIPB["end_title"])
def start_footnote(self,attrs):
self.__data.append(SIPB["start_footnote"])
def end_footnote(self):
ft = self.__data.pop()
ft2 = self.__data.pop()
self.footnotes.append(ft)
ftl = repr(len(self.footnotes))
self.__data.append(ft2+''+ftl+''+SIPB["end_footnote"] )
def unknown_starttag(self,tag,attrs):
self.__data.append("<"+tag+">")
def unknown_endtag(self,tag):
line = self.__data.pop()
if tag == 'title' :
print line
line = self.__data.pop() +line
self.__data.append(line + "")
def start_link(self,attrs) :
# Got to make several transformations
# here.
if attrs.get("href"):
self.__data.append('')
def end_link(self):
self.__data.append("")
def handle_data(self,data):
self.__data.append(data)
def end_sipb(self):
if self.footnotes:
self.__data.append("
")
for f in range(len(self.footnotes)):
self.__data.append(''+repr(f+1)+':'+self.footnotes[f]+'' )
self.__data.append(SIPB["end_sipb"])
self.outgo = string.join(self.__data,"")
try:
p = Parser()
p.load(open(sys.argv[1]))
open(sys.argv[2],'w').write(p.outgo)
except EOFError:
pass