#!/usr/bin/env python
# -*- coding: UTF-8 -*-

import string
import sys
import os
from pology.file.catalog import Catalog
import xml.parsers.expat
import xml.dom.minidom

if len(sys.argv) != 4:
    print "usage: %s <source_po> <source_html> <output_po>" % os.path.basename(sys.argv[0])
    sys.exit(1)

fname = sys.argv[1]
cat_input = Catalog(fname)
cat_output = Catalog(sys.argv[3], create=True)
cat_output.header = cat_input.header

message = None
tagstack = 0
hadcdata = False
output = ""

def start_element(name, attrs):
    global output, tagstack, hadcdata, message
    if name == 'msgid' or name == 'msgids':
        return
    
    replace = '<%s %s>' % (name, string.join(['%s="%s"' % t for t in attrs.items()]))
    message = message.replace('<a href="#%d">' % tagstack, replace)
    message = message.replace('<a href="#%d"/>' % tagstack, replace)
    tagstack = tagstack + 1

def end_element(name):
    global output, tagstack, hadcdata, message
    if name == 'msgid' or name == 'msgids':
        tagstack = 0
        return

    if hadcdata:
        message = message.replace('</a>', '</%s>' % name, 1)
    else:
        message = message[:-1] + '/>'
    
def char_data(data):
    global output, hadcdata
    hadcdata = True
    output += str(data)

p = xml.parsers.expat.ParserCreate()

p.StartElementHandler = start_element
p.EndElementHandler = end_element
p.CharacterDataHandler = char_data

i = 0
html = open(sys.argv[2], 'r').read()
dom = xml.dom.minidom.parseString(html.replace('&','&amp;'))
messages = [string.join([n.toxml().replace('&amp;','&') for n in m.childNodes], '')for m in dom.getElementsByTagName("p")]

p.Parse("<msgids>", False)

for msg in cat_input:
    
    message = messages[i]
    i += 1

    if msg.translated:
        text = msg.msgstr
    else:
        text = msg.msgid

    text = "<msgid>%s</msgid>" % text.replace('&', '&amp;')
    p.Parse(text, False)

    msg.msgstr[0] = message
    cat_output.add(msg)

p.Parse("</msgids>", True);

cat_output.sync()