#!/usr/bin/python
import os, sys
sys.path.insert(0, '/home/mark/dev')
import timeoutsocket
timeoutsocket.setDefaultSocketTimeout(10)
import urllib, random, cgi
from sgmllib import SGMLParser

MAXWORDS = 100

class WordParser(SGMLParser):
    def reset(self):
        SGMLParser.reset(self)
        self.words = []
        self.capture=0
    def start_body(self, attrs):
        self.capture = 1
    def handle_data(self, data):
        if not self.capture: return
        if random.randint(0, 1): return
        self.words.extend(data.split())
        if len(self.words) > MAXWORDS:
            self.words = self.words[:MAXWORDS]
            self.setnomoretags()

def getWords(uri):
    try:
        usock = urllib.urlopen(uri)
        p = WordParser()
        p.feed(usock.read(100000))
        p.close()
        return p.words
    except:
        return ['Sorry', 'an', 'error', 'occurred', 'trying', 'to', 'download', uri]

def output(wordList):
    htmlIn = file('magnetic_poetry.html').read()
    htmlOut = "\n".join(["<span class='drag'>%s</span>" % cgi.escape(word) for word in wordList])
    return htmlIn.replace("<!--insert words here-->", htmlOut)

print 'Content-type: text/html'
print
# grab uri stuffed in by mod_rewrite
fs = cgi.FieldStorage()
remoteURI = fs["url"].value
if not remoteURI.startswith('http://'):
    remoteURI = 'http://%s' % remoteURI
print output(getWords(remoteURI))