[channel9] Initial implementation (#1885)
authordst <dstftw@gmail.com>
Thu, 12 Dec 2013 08:13:45 +0000 (15:13 +0700)
committerdst <dstftw@gmail.com>
Thu, 12 Dec 2013 08:13:45 +0000 (15:13 +0700)
youtube_dl/extractor/channel9.py [new file with mode: 0644]

diff --git a/youtube_dl/extractor/channel9.py b/youtube_dl/extractor/channel9.py
new file mode 100644 (file)
index 0000000..016c449
--- /dev/null
@@ -0,0 +1,351 @@
+# encoding: utf-8
+
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    format_bytes,
+    ExtractorError,
+)
+
+class Channel9IE(InfoExtractor):
+    '''
+    Common extractor for channel9.msdn.com.
+
+    The type of provided URL (video or playlist) is determined according to
+    meta Search.PageType from web page HTML rather than URL itself, as it is
+    not always possible to do.    
+    '''
+    IE_DESC = u'Channel 9'
+    IE_NAME = u'channel9'
+    _VALID_URL = r'^https?://(?:www\.)?channel9\.msdn\.com/(?P<contentpath>.+)/?'
+
+    _TESTS = [
+        {
+            u'url': u'http://channel9.msdn.com/Events/TechEd/Australia/2013/KOS002',
+            u'file': u'Events_TechEd_Australia_2013_KOS002.mp4',
+            u'md5': u'bbd75296ba47916b754e73c3a4bbdf10',
+            u'info_dict': {
+                u'title': u'Developer Kick-Off Session: Stuff We Love',
+                u'description': u'md5:c08d72240b7c87fcecafe2692f80e35f',
+                u'duration': 4576,
+                u'thumbnail': u'http://media.ch9.ms/ch9/9d51/03902f2d-fc97-4d3c-b195-0bfe15a19d51/KOS002_220.jpg',
+                u'session_code': u'KOS002',
+                u'session_day': u'Day 1',
+                u'session_room': u'Arena 1A',
+                u'session_speakers': [ u'Ed Blankenship', u'Andrew Coates', u'Brady Gaster', u'Patrick Klug', u'Mads Kristensen' ],
+            },
+        },
+        {
+            u'url': u'http://channel9.msdn.com/posts/Self-service-BI-with-Power-BI-nuclear-testing',
+            u'file': u'posts_Self-service-BI-with-Power-BI-nuclear-testing.mp4',
+            u'md5': u'b43ee4529d111bc37ba7ee4f34813e68',
+            u'info_dict': {
+                u'title': u'Self-service BI with Power BI - nuclear testing',
+                u'description': u'md5:a6d5cfd9ee46d1851cf6e40ea61cfc10',
+                u'duration': 1540,
+                u'thumbnail': u'http://media.ch9.ms/ch9/87e1/0300391f-a455-4c72-bec3-4422f19287e1/selfservicenuk_512.jpg',
+                u'authors': [ u'Mike Wilmot' ],
+            },
+        }
+    ]
+
+    _RSS_URL = 'http://channel9.msdn.com/%s/RSS'
+    _EXTRACT_ENTRY_ITEMS_FROM_RSS = False
+
+    # Sorted by quality
+    _known_formats = ['MP3', 'MP4', 'Mid Quality WMV', 'Mid Quality MP4', 'High Quality WMV', 'High Quality MP4']
+
+    def _restore_bytes(self, formatted_size):
+        if not formatted_size:
+            return 0
+        m = re.match(r'^(?P<size>\d+(?:\.\d+)?)\s+(?P<units>[a-zA-Z]+)', formatted_size)
+        if not m:
+            return 0
+        units = m.group('units')
+        try:
+            exponent = [u'B', u'KB', u'MB', u'GB', u'TB', u'PB', u'EB', u'ZB', u'YB'].index(units.upper())
+        except ValueError:
+            return 0
+        size = float(m.group('size'))
+        return int(size * (1024 ** exponent))
+
+    def _formats_from_html(self, html):
+        FORMAT_REGEX = r'''
+            (?x)
+            <a\s+href="(?P<url>[^"]+)">(?P<quality>[^<]+)</a>\s*
+            <span\s+class="usage">\((?P<note>[^\)]+)\)</span>\s*
+            (?:<div\s+class="popup\s+rounded">\s*
+            <h3>File\s+size</h3>\s*(?P<filesize>.*?)\s*
+            </div>)?                                                # File size part may be missing
+        '''
+        # Extract known formats
+        formats = [{'url': x.group('url'),
+                 'format_id': x.group('quality'),
+                 'format_note': x.group('note'),
+                 'format': '%s (%s)' % (x.group('quality'), x.group('note')), 
+                 'filesize': self._restore_bytes(x.group('filesize')), # File size is approximate
+                 } for x in list(re.finditer(FORMAT_REGEX, html)) if x.group('quality') in self._known_formats]
+        # Sort according to known formats list
+        formats.sort(key=lambda fmt: self._known_formats.index(fmt['format_id']))
+        return formats
+
+    def _formats_from_rss_item(self, item):
+
+        def process_formats(elem):
+            formats = []
+            for media_content in elem.findall('./{http://search.yahoo.com/mrss/}content'):
+                url = media_content.attrib['url']
+                # Ignore unrelated media
+                if url.endswith('.ism/manifest'):
+                    continue
+                format_note = media_content.attrib['type']
+                filesize = int(media_content.attrib['fileSize'])
+                formats.append({'url': url,
+                                'format_note': format_note,
+                                'format': '%s %s' % (format_note, format_bytes(filesize)),
+                                'filesize': filesize,
+                                })
+            return formats
+
+        formats = []
+
+        for media_group in item.findall('./{http://search.yahoo.com/mrss/}group'):
+            formats.extend(process_formats(media_group))
+
+        # Sometimes there are no media:groups in item, but there is media:content
+        # right in item (usually when there is the only media source)
+        formats.extend(process_formats(item))        
+
+        # Sort by file size
+        formats.sort(key=lambda fmt: fmt['filesize'])
+        return formats
+
+    def _extract_title(self, html):
+        title = self._html_search_meta(u'title', html, u'title')
+        if title is None:           
+            title = self._og_search_title(html)
+            TITLE_SUFFIX = u' (Channel 9)'
+            if title is not None and title.endswith(TITLE_SUFFIX):
+                title = title[:-len(TITLE_SUFFIX)]
+        return title
+
+    def _extract_description(self, html):
+        DESCRIPTION_REGEX = r'''(?sx)
+            <div\s+class="entry-content">\s*
+            <div\s+id="entry-body">\s*
+            (?P<description>.+?)\s*
+            </div>\s*
+            </div>
+        '''
+        m = re.search(DESCRIPTION_REGEX, html)
+        if m is not None:
+            return m.group('description')
+        return self._html_search_meta(u'description', html, u'description')
+
+    def _extract_duration(self, html):
+        m = re.search(r'data-video_duration="(?P<hours>\d{2}):(?P<minutes>\d{2}):(?P<seconds>\d{2})"', html)
+        return ((int(m.group('hours')) * 60 * 60) + (int(m.group('minutes')) * 60) + int(m.group('seconds'))) if m else None
+
+    def _extract_slides(self, html):
+        m = re.search(r'<a href="(?P<slidesurl>[^"]+)" class="slides">Slides</a>', html)
+        return m.group('slidesurl') if m is not None else None
+
+    def _extract_zip(self, html):
+        m = re.search(r'<a href="(?P<zipurl>[^"]+)" class="zip">Zip</a>', html)
+        return m.group('zipurl') if m is not None else None
+
+    def _extract_avg_rating(self, html):
+        m = re.search(r'<p class="avg-rating">Avg Rating: <span>(?P<avgrating>[^<]+)</span></p>', html)
+        return float(m.group('avgrating')) if m is not None else 0
+
+    def _extract_rating_count(self, html):
+        m = re.search(r'<div class="rating-count">\((?P<ratingcount>[^<]+)\)</div>', html)
+        return int(self._fix_count(m.group('ratingcount'))) if m is not None else 0
+
+    def _extract_view_count(self, html):
+        m = re.search(r'<li class="views">\s*<span class="count">(?P<viewcount>[^<]+)</span> Views\s*</li>', html)
+        return int(self._fix_count(m.group('viewcount'))) if m is not None else 0
+
+    def _extract_comment_count(self, html):
+        m = re.search(r'<li class="comments">\s*<a href="#comments">\s*<span class="count">(?P<commentcount>[^<]+)</span> Comments\s*</a>\s*</li>', html)
+        return int(self._fix_count(m.group('commentcount'))) if m is not None else 0
+
+    def _fix_count(self, count):
+        return int(str(count).replace(',', '')) if count is not None else None
+
+    def _extract_authors(self, html):
+        m = re.search(r'(?s)<li class="author">(.*?)</li>', html)
+        if m is None:
+            return None
+        return re.findall(r'<a href="/Niners/[^"]+">([^<]+)</a>', m.group(1))
+
+    def _extract_session_code(self, html):
+        m = re.search(r'<li class="code">\s*(?P<code>.+?)\s*</li>', html)
+        return m.group('code') if m is not None else None
+
+    def _extract_session_day(self, html):
+        m = re.search(r'<li class="day">\s*<a href="/Events/[^"]+">(?P<day>[^<]+)</a>\s*</li>', html)
+        return m.group('day') if m is not None else None
+
+    def _extract_session_room(self, html):
+        m = re.search(r'<li class="room">\s*(?P<room>.+?)\s*</li>', html)
+        return m.group('room') if m is not None else None
+
+    def _extract_session_speakers(self, html):
+        return re.findall(r'<a href="/Events/Speakers/[^"]+">([^<]+)</a>', html)
+
+    def _extract_content(self, html, content_path):
+        # Look for downloadable content        
+        formats = self._formats_from_html(html)
+        slides = self._extract_slides(html)
+        zip_ = self._extract_zip(html)
+
+        # Nothing to download
+        if len(formats) == 0 and slides is None and zip_ is None:
+            self._downloader.report_warning(u'None of recording, slides or zip are available for %s' % content_path)
+            return
+
+        # Extract meta
+        title = self._extract_title(html)
+        description = self._extract_description(html)
+        thumbnail = self._og_search_thumbnail(html)
+        duration = self._extract_duration(html)
+        avg_rating = self._extract_avg_rating(html)
+        rating_count = self._extract_rating_count(html)
+        view_count = self._extract_view_count(html)
+        comment_count = self._extract_comment_count(html)
+
+        common = {'_type': 'video',
+                  'id': content_path,
+                  'description': description,
+                  'thumbnail': thumbnail,
+                  'duration': duration,
+                  'avg_rating': avg_rating,
+                  'rating_count': rating_count,
+                  'view_count': view_count,
+                  'comment_count': comment_count,
+                }
+
+        result = []
+
+        if slides is not None:
+            d = common.copy()
+            d.update({ 'title': title + '-Slides', 'url': slides })
+            result.append(d)
+
+        if zip_ is not None:
+            d = common.copy()
+            d.update({ 'title': title + '-Zip', 'url': zip_ })
+            result.append(d)
+
+        if len(formats) > 0:
+            d = common.copy()
+            d.update({ 'title': title, 'formats': formats })
+            result.append(d)
+
+        return result
+
+    def _extract_entry_item(self, html, content_path):
+        contents = self._extract_content(html, content_path)
+        if contents is None:
+            return contents
+
+        authors = self._extract_authors(html)
+
+        for content in contents:
+            content['authors'] = authors
+
+        return contents
+
+    def _extract_session(self, html, content_path):
+        contents = self._extract_content(html, content_path)
+        if contents is None:
+            return contents
+
+        session_meta = {'session_code': self._extract_session_code(html),
+                        'session_day': self._extract_session_day(html),
+                        'session_room': self._extract_session_room(html),
+                        'session_speakers': self._extract_session_speakers(html),
+                        }
+
+        for content in contents:
+            content.update(session_meta)
+
+        return contents
+
+    def _extract_content_rss(self, rss):
+        '''
+        Extracts links to entry items right out of RSS feed.
+        This approach is faster than extracting from web pages
+        one by one, but suffers from some problems.
+        Pros:
+         - no need to download additional pages
+         - provides more media links
+         - accurate file size
+        Cons:
+         - fewer meta data provided
+         - links to media files have no appropriate data that may be used as format_id
+         - RSS does not contain links to presentation materials (slides, zip)
+        '''
+        entries = []
+        for item in rss.findall('./channel/item'):
+            url = item.find('./link').text
+            video_id = url.split('/')[-1]
+            formats = self._formats_from_rss_item(item)
+
+            if len(formats) == 0:
+                self._downloader.report_warning(u'The recording for session %s is not yet available' % video_id)
+                continue
+
+            title = item.find('./title').text
+            description = item.find('./description').text
+
+            thumbnail = item.find('./{http://search.yahoo.com/mrss/}thumbnail').text
+
+            duration_e = item.find('./{http://www.itunes.com/dtds/podcast-1.0.dtd}duration')
+            duration = duration_e.text if duration_e is not None else 0
+
+            speakers_e = item.find('./{http://purl.org/dc/elements/1.1/}creator')
+            speakers = speakers_e.text.split(', ') if speakers_e is not None and speakers_e.text else []
+
+            entries.append({'_type': 'video',
+                            'id': video_id,
+                            'formats': formats,
+                            'title': title,
+                            'description': description,
+                            'thumbnail': thumbnail,
+                            'duration': duration,
+                            'session_speakers': speakers,                            
+                            })
+        return entries
+
+    def _extract_list(self, content_path):
+        rss = self._download_xml(self._RSS_URL % content_path, content_path, u'Downloading RSS')
+        if self._EXTRACT_ENTRY_ITEMS_FROM_RSS:   
+            return self._extract_content_rss(rss)
+        else:
+            entries = [self.url_result(session_url.text, 'Channel9')
+                       for session_url in rss.findall('./channel/item/link')]
+            title_text = rss.find('./channel/title').text
+            return self.playlist_result(entries, content_path, title_text)
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        content_path = mobj.group('contentpath')
+
+        webpage = self._download_webpage(url, content_path, u'Downloading web page')
+
+        page_type_m = re.search(r'<meta name="Search.PageType" content="(?P<pagetype>[^"]+)"/>', webpage)
+        if page_type_m is None:
+            raise ExtractorError(u'Search.PageType not found, don\'t know how to process this page', expected=True)
+
+        page_type = page_type_m.group('pagetype')
+        if page_type == 'List':         # List page, may contain list of 'item'-like objects
+            return self._extract_list(content_path)
+        elif page_type == 'Entry.Item': # Any 'item'-like page, may contain downloadable content
+            return self._extract_entry_item(webpage, content_path)
+        elif page_type == 'Session':    # Event session page, may contain downloadable content
+            return self._extract_session(webpage, content_path)
+        else:
+            raise ExtractorError(u'Unexpected Search.PageType %s' % page_type, expected=True)
\ No newline at end of file