Count of views of PyOhio 2012 videos at pyvideo.org

video_count.ipynb

{
 "metadata": {
  "name": "video_count"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import requests\n",
      "import bs4\n",
      "import re"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "mainpage = requests.get('http://pyvideo.org/category/22/pyohio-2012')\n",
      "mainsoup = bs4.BeautifulSoup(mainpage.content)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "linkpattern = re.compile('/video/')\n",
      "youtubepattern = re.compile('www.youtube.com/watch')\n",
      "video_urls = dict()\n",
      "for link in mainsoup.findAll('a', href=linkpattern):\n",
      "    video_urls['http://pyvideo.org' + link['href']] = None\n",
      "for video_url in video_urls:\n",
      "    video_page = requests.get(video_url)\n",
      "    video_soup = bs4.BeautifulSoup(video_page.content)\n",
      "    youtube_url = video_soup.find('a', href=youtubepattern)['href']\n",
      "    youtube_page = requests.get(youtube_url)\n",
      "    youtube_soup = bs4.BeautifulSoup(youtube_page.content)\n",
      "    view_span = youtube_soup.find('span', class_=\"watch-view-count \")\n",
      "    views = view_span.text.strip().split()[0]\n",
      "    video_urls[video_url] = views"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "video_urls\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 4,
       "text": [
        "{u'http://pyvideo.org/video/1352/lightning-talks-0-1': u'323',\n",
        " u'http://pyvideo.org/video/1353/you-used-python-for-what': u'549',\n",
        " u'http://pyvideo.org/video/1354/starting-your-project-right-setup-and-automation': u'582',\n",
        " u'http://pyvideo.org/video/1355/video-production-strategies-using-image-sequences': u'144',\n",
        " u'http://pyvideo.org/video/1356/python-101102': u'541',\n",
        " u'http://pyvideo.org/video/1357/devops-or-how-i-learned-to-stop-worrying-and-lov': u'175',\n",
        " u'http://pyvideo.org/video/1358/drive-in-double-header-datetimes-and-log-analysi': u'73',\n",
        " u'http://pyvideo.org/video/1359/coding-for-everyone': u'285',\n",
        " u'http://pyvideo.org/video/1360/really-good-logging': u'570',\n",
        " u'http://pyvideo.org/video/1361/develop-games-with-panda3d-and-python': u'2,816',\n",
        " u'http://pyvideo.org/video/1362/leo-a-paradigm-shifting-ide': u'641',\n",
        " u'http://pyvideo.org/video/1363/does-python-have-the-secret-sauce-to-be-web-scale': u'291',\n",
        " u'http://pyvideo.org/video/1364/an-introduction-to-tkinter': u'2,046',\n",
        " u'http://pyvideo.org/video/1365/family-project-three-keys-and-a-boss': u'111',\n",
        " u'http://pyvideo.org/video/1366/python-testing-fundamentals': u'1,205',\n",
        " u'http://pyvideo.org/video/1367/an-introduction-to-the-zodb': u'284',\n",
        " u'http://pyvideo.org/video/1368/using-python-on-android': u'1,417',\n",
        " u'http://pyvideo.org/video/1369/python-design-patterns-1': u'3,254',\n",
        " u'http://pyvideo.org/video/1370/python-for-educators': u'175',\n",
        " u'http://pyvideo.org/video/1371/learn-python-tdd-style-with-python-koans': u'1,537',\n",
        " u'http://pyvideo.org/video/1372/exploring-python-code-objects': u'471',\n",
        " u'http://pyvideo.org/video/1373/fun-observational-science-with-python-and-a-webca': u'217',\n",
        " u'http://pyvideo.org/video/1374/python-for-humans': u'731',\n",
        " u'http://pyvideo.org/video/1375/beyond-the-pil-alternative-solutions-for-working': u'665',\n",
        " u'http://pyvideo.org/video/1376/effective-django': u'1,172',\n",
        " u'http://pyvideo.org/video/1377/django-cms-friends-dont-let-friends-use-drupal': u'4,984',\n",
        " u'http://pyvideo.org/video/1378/deleting-code-is-hard-and-you-should-do-it': u'1,630'}"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "for url in video_urls:\n",
      "    views = int(video_urls[url].replace(',',''))\n",
      "    video_urls[url] = views"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "sum(v for v in video_urls.values())"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "pyout",
       "prompt_number": 11,
       "text": [
        "26889"
       ]
      }
     ],
     "prompt_number": 11
    }
   ],
   "metadata": {}
  }
 ]
}