threads-catalog.py

# -*- coding: utf-8 -*-
"""
Created on Sat Nov 21 14:29:50 2015

@author: User
"""

from bs4 import BeautifulSoup
import re
import os

# theres probably better way
def innerHtml(element):
    t = []
    for child in element.children:
        t.append('%s' % child)
    return '\n'.join(t)

home = 'C:/wamp/www/4chan'


threads = []

for item in os.listdir(home):
    threadNumber = item
    m = re.match('^[0-9]+$',item)
    if m == None:
        continue
    print threadNumber
    thread = home + "/" + threadNumber + "/index.html"

    f = open(thread,'r')
    html = f.read()
    f.close()
    soup = BeautifulSoup(html, 'html.parser')

    post = soup.find(class_="opContainer")

    postMessage = post.find(class_="postMessage")
    fileThumb = post.find(class_="fileThumb")
    postImage = fileThumb.find('img')

    imgStyle = postImage.get('style')

    w = re.search('width: ([0-9]+)px',imgStyle)
    h = re.search('height: ([0-9]+)px',imgStyle)
    w = float(w.group(1))
    h = float(h.group(1))
    #print float(w.group(1) + ' ' + h.group(1)

    asp = w / h

    w2 = 150
    h2 = w2 / asp

    newStyle = 'width: %dpx; height: %dpx' % (w2,h2)
    image = '<a href="' + threadNumber + '/index.html"><img src="' + threadNumber + '/' + postImage.get('src') + '" class="thumb" style="'+newStyle+'"/></a>'
    teaser = '<div class="teaser">'+innerHtml(postMessage)+'</div>'
    result = '<div class="thread">' + image + teaser + '</div>'
    threads.append(result)

    #result = topHtml + result + bottomHtml

dest = home + '/index.html'
topHtml = '<html><head><link href="https://s.4cdn.org/css/catalog_yotsuba_b_new.635.css" rel="stylesheet" id="base-css" type="text/css"></head><body><div id="threads" class="extended-small">'
bottomHtml = '</div></body></html>'

result = topHtml + "\n".join(threads) + bottomHtml

f = open(dest,'w')
f.write(result)
f.close()