#!/usr/bin/env python
"""
A small script to detect whether there are source code links
from the PDFs of conference papers
"""
import tarfile
import os

input_file = tarfile.open('nips2008_pdf.tgz','r')
files_with_link = []
for filename in input_file.getnames():
    if filename.endswith('.pdf') and not filename.endswith('_slide.pdf'):
        input_file.extract(filename,'.')
        os.system('pdftotext ' + filename)
        text_file = open(filename[:-4]+'.txt', 'r')
        found_link = False
        for text in text_file.readlines():
            # Find links, but filter out PDF generation stuff
            if (text.find('http://')>0 or text.find('ftp://')>0)\
                   and not (text.find('ns.adobe.com')>0 or
                            text.find('w3.org')>0 or
                            text.find('purl.org')>0 or
                            text.find('cairographics.org')>0 or
                            text.find('apple.com')>0 or
                            text.find('iec.ch')>0):
                print filename, text
                found_link = True
        if found_link:
            files_with_link.append(filename)
        text_file.close()
        os.system('rm ' + filename + ' ' + filename[:-4]+'.txt')
print files_with_link