1 files changed, 127 insertions, 0 deletions
diff --git a/yt_dlp/extractor/dispeak.py b/yt_dlp/extractor/dispeak.py
new file mode 100644
index 0000000..37f89b9
--- /dev/null
+++ b/yt_dlp/extractor/dispeak.py
@@ -0,0 +1,127 @@
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    int_or_none,
+    parse_duration,
+    remove_end,
+    xpath_element,
+    xpath_text,
+)
+
+
+class DigitallySpeakingIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:s?evt\.dispeak|events\.digitallyspeaking)\.com/(?:[^/]+/)+xml/(?P<id>[^.]+)\.xml'
+
+    _TESTS = [{
+        # From http://gdcvault.com/play/1023460/Tenacious-Design-and-The-Interface
+        'url': 'http://evt.dispeak.com/ubm/gdc/sf16/xml/840376_BQRC.xml',
+        'md5': 'a8efb6c31ed06ca8739294960b2dbabd',
+        'info_dict': {
+            'id': '840376_BQRC',
+            'ext': 'mp4',
+            'title': 'Tenacious Design and The Interface of \'Destiny\'',
+        },
+    }, {
+        # From http://www.gdcvault.com/play/1014631/Classic-Game-Postmortem-PAC
+        'url': 'http://events.digitallyspeaking.com/gdc/sf11/xml/12396_1299111843500GMPX.xml',
+        'only_matching': True,
+    }, {
+        # From http://www.gdcvault.com/play/1013700/Advanced-Material
+        'url': 'http://sevt.dispeak.com/ubm/gdc/eur10/xml/11256_1282118587281VNIT.xml',
+        'only_matching': True,
+    }, {
+        # From https://gdcvault.com/play/1016624, empty speakerVideo
+        'url': 'https://sevt.dispeak.com/ubm/gdc/online12/xml/201210-822101_1349794556671DDDD.xml',
+        'info_dict': {
+            'id': '201210-822101_1349794556671DDDD',
+            'ext': 'flv',
+            'title': 'Pre-launch - Preparing to Take the Plunge',
+        },
+    }, {
+        # From http://www.gdcvault.com/play/1014846/Conference-Keynote-Shigeru, empty slideVideo
+        'url': 'http://events.digitallyspeaking.com/gdc/project25/xml/p25-miyamoto1999_1282467389849HSVB.xml',
+        'only_matching': True,
+    }]
+
+    def _parse_mp4(self, metadata):
+        video_formats = []
+        video_root = None
+
+        mp4_video = xpath_text(metadata, './mp4video', default=None)
+        if mp4_video is not None:
+            mobj = re.match(r'(?P<root>https?://.*?/).*', mp4_video)
+            video_root = mobj.group('root')
+        if video_root is None:
+            http_host = xpath_text(metadata, 'httpHost', default=None)
+            if http_host:
+                video_root = 'http://%s/' % http_host
+        if video_root is None:
+            # Hard-coded in http://evt.dispeak.com/ubm/gdc/sf16/custom/player2.js
+            # Works for GPUTechConf, too
+            video_root = 'http://s3-2u.digitallyspeaking.com/'
+
+        formats = metadata.findall('./MBRVideos/MBRVideo')
+        if not formats:
+            return None
+        for a_format in formats:
+            stream_name = xpath_text(a_format, 'streamName', fatal=True)
+            video_path = re.match(r'mp4\:(?P<path>.*)', stream_name).group('path')
+            url = video_root + video_path
+            bitrate = xpath_text(a_format, 'bitrate')
+            tbr = int_or_none(bitrate)
+            vbr = int_or_none(self._search_regex(
+                r'-(\d+)\.mp4', video_path, 'vbr', default=None))
+            video_formats.append({
+                'format_id': bitrate,
+                'url': url,
+                'tbr': tbr,
+                'vbr': vbr,
+            })
+        return video_formats
+
+    def _parse_flv(self, metadata):
+        formats = []
+        akamai_url = xpath_text(metadata, './akamaiHost', fatal=True)
+        audios = metadata.findall('./audios/audio')
+        for audio in audios:
+            formats.append({
+                'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
+                'play_path': remove_end(audio.get('url'), '.flv'),
+                'ext': 'flv',
+                'vcodec': 'none',
+                'quality': 1,
+                'format_id': audio.get('code'),
+            })
+        for video_key, format_id, preference in (
+                ('slide', 'slides', -2), ('speaker', 'speaker', -1)):
+            video_path = xpath_text(metadata, './%sVideo' % video_key)
+            if not video_path:
+                continue
+            formats.append({
+                'url': 'rtmp://%s/ondemand?ovpfv=1.1' % akamai_url,
+                'play_path': remove_end(video_path, '.flv'),
+                'ext': 'flv',
+                'format_note': '%s video' % video_key,
+                'quality': preference,
+                'format_id': format_id,
+            })
+        return formats
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        xml_description = self._download_xml(url, video_id)
+        metadata = xpath_element(xml_description, 'metadata')
+
+        video_formats = self._parse_mp4(metadata)
+        if video_formats is None:
+            video_formats = self._parse_flv(metadata)
+
+        return {
+            'id': video_id,
+            'formats': video_formats,
+            'title': xpath_text(metadata, 'title', fatal=True),
+            'duration': parse_duration(xpath_text(metadata, 'endTime')),
+            'creator': xpath_text(metadata, 'speaker'),
+        }