morss/morss

#! /usr/bin/env python2.7
import sys
import os
from os.path import expanduser
from lxml import etree
import string
import urllib2
import urllib
from cookielib import CookieJar

def log(txt):
	if os.getenv('DEBUG', False):
		print txt

def xmlclean(xml):
	table = string.maketrans('', '')
	return xml.translate(table, table[:32])

node =	sys.argv[1] if len(sys.argv) > 1 else "//h1/.."

xml =	xmlclean(sys.stdin.read())
rss =	etree.XML(xml)
items =	rss.xpath('//item')

cache = expanduser("~") + "/.cache/morss"
if not os.path.exists(cache):
	os.makedirs(cache)

for item in items:
	link = 	item.findtext('link').encode('utf-8')
	desc =	item.xpath('description')[0]
	
	log(link)
	
	cached = cache + "/" + str(hash(link))
	log(cached)
	
	if os.path.exists(cached):
		log("cached")
		desc.text = open(cached, 'r').read()
	else:
		try:
			cj = CookieJar()
			opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))
			data =	opener.open(link).read()
			html =	etree.HTML(data)
			match =	html.xpath(node)
		
			if len(match):
				try:
					text =	etree.tostring(match[0])
					log("ok txt")
				except etree.SerialisationError:
					log('serialisation')
					continue
				try:
					desc.text = text
					open(cached, 'w').write(text)
				except ValueError:
					log('xml error')
			else:
				log("no match")
		
		except (urllib2.HTTPError, urllib2.URLError) as error:
			log(error)
			log("http error")

if not os.getenv('DEBUG', False):
	print etree.tostring(rss)