]>
Commit | Line | Data |
---|---|---|
0d97ef43 NJ |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import re | |
01b06aed | 5 | import base64 |
0d97ef43 NJ |
6 | |
7 | from .common import InfoExtractor | |
01b06aed | 8 | from ..compat import ( |
01b06aed | 9 | compat_urlparse, |
fac7e792 | 10 | compat_parse_qs, |
01b06aed | 11 | ) |
0d97ef43 | 12 | from ..utils import ( |
bdceea7a | 13 | clean_html, |
0d97ef43 NJ |
14 | ExtractorError, |
15 | int_or_none, | |
01b06aed | 16 | unsmuggle_url, |
81953d1a | 17 | smuggle_url, |
0d97ef43 NJ |
18 | ) |
19 | ||
20 | ||
21 | class KalturaIE(InfoExtractor): | |
22 | _VALID_URL = r'''(?x) | |
ee3ec091 | 23 | (?: |
fac7e792 | 24 | kaltura:(?P<partner_id>\d+):(?P<id>[0-9a-z_]+)| |
ee3ec091 | 25 | https?:// |
ff24261b | 26 | (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com(?::\d+)?/ |
ee3ec091 S |
27 | (?: |
28 | (?: | |
29 | # flash player | |
37084f66 | 30 | index\.php/(?:kwidget|extwidget/preview)| |
b184f944 | 31 | # html5 player |
fac7e792 | 32 | html5/html5lib/[^/]+/mwEmbedFrame\.php |
ae6a8021 | 33 | ) |
fac7e792 | 34 | )(?:/(?P<path>[^?]+))?(?:\?(?P<query>.*))? |
ee3ec091 S |
35 | ) |
36 | ''' | |
81953d1a RA |
37 | _SERVICE_URL = 'http://cdnapi.kaltura.com' |
38 | _SERVICE_BASE = '/api_v3/index.php' | |
92d4cfa3 S |
39 | # See https://github.com/kaltura/server/blob/master/plugins/content/caption/base/lib/model/enums/CaptionType.php |
40 | _CAPTION_TYPES = { | |
41 | 1: 'srt', | |
42 | 2: 'ttml', | |
43 | 3: 'vtt', | |
44 | } | |
0d97ef43 NJ |
45 | _TESTS = [ |
46 | { | |
47 | 'url': 'kaltura:269692:1_1jc2y3e4', | |
48 | 'md5': '3adcbdb3dcc02d647539e53f284ba171', | |
49 | 'info_dict': { | |
50 | 'id': '1_1jc2y3e4', | |
51 | 'ext': 'mp4', | |
bb4b8c57 | 52 | 'title': 'Straight from the Heart', |
0d97ef43 NJ |
53 | 'upload_date': '20131219', |
54 | 'uploader_id': 'mlundberg@wolfgangsvault.com', | |
55 | 'description': 'The Allman Brothers Band, 12/16/1981', | |
56 | 'thumbnail': 're:^https?://.*/thumbnail/.*', | |
57 | 'timestamp': int, | |
58 | }, | |
59 | }, | |
60 | { | |
61 | 'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4', | |
62 | 'only_matching': True, | |
63 | }, | |
bd3749ed S |
64 | { |
65 | 'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3', | |
66 | 'only_matching': True, | |
67 | }, | |
ef49b590 | 68 | { |
69 | 'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342', | |
70 | 'only_matching': True, | |
1094074c RA |
71 | }, |
72 | { | |
73 | # video with subtitles | |
74 | 'url': 'kaltura:111032:1_cw786r8q', | |
75 | 'only_matching': True, | |
2c6acdfd S |
76 | }, |
77 | { | |
78 | # video with ttml subtitles (no fileExt) | |
79 | 'url': 'kaltura:1926081:0_l5ye1133', | |
80 | 'info_dict': { | |
81 | 'id': '0_l5ye1133', | |
82 | 'ext': 'mp4', | |
83 | 'title': 'What Can You Do With Python?', | |
84 | 'upload_date': '20160221', | |
85 | 'uploader_id': 'stork', | |
86 | 'thumbnail': 're:^https?://.*/thumbnail/.*', | |
87 | 'timestamp': int, | |
88 | 'subtitles': { | |
89 | 'en': [{ | |
90 | 'ext': 'ttml', | |
91 | }], | |
92 | }, | |
93 | }, | |
a01825a5 | 94 | 'skip': 'Gone. Maybe https://www.safaribooksonline.com/library/tutorials/introduction-to-python-anon/3469/', |
2c6acdfd S |
95 | 'params': { |
96 | 'skip_download': True, | |
97 | }, | |
37084f66 RA |
98 | }, |
99 | { | |
100 | 'url': 'https://www.kaltura.com/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', | |
101 | 'only_matching': True, | |
ff24261b S |
102 | }, |
103 | { | |
104 | 'url': 'https://www.kaltura.com:443/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto', | |
105 | 'only_matching': True, | |
ef49b590 | 106 | } |
0d97ef43 NJ |
107 | ] |
108 | ||
9ea5c04c S |
109 | @staticmethod |
110 | def _extract_url(webpage): | |
78280352 | 111 | # Embed codes: https://knowledge.kaltura.com/embedding-kaltura-media-players-your-site |
9ea5c04c S |
112 | mobj = ( |
113 | re.search( | |
114 | r"""(?xs) | |
115 | kWidget\.(?:thumb)?[Ee]mbed\( | |
116 | \{.*? | |
78280352 S |
117 | (?P<q1>['"])wid(?P=q1)\s*:\s* |
118 | (?P<q2>['"])_?(?P<partner_id>(?:(?!(?P=q2)).)+)(?P=q2),.*? | |
119 | (?P<q3>['"])entry_?[Ii]d(?P=q3)\s*:\s* | |
120 | (?P<q4>['"])(?P<id>(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\}) | |
9ea5c04c S |
121 | """, webpage) or |
122 | re.search( | |
123 | r'''(?xs) | |
78280352 | 124 | (?P<q1>["']) |
ff24261b | 125 | (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com(?::\d+)?/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)(?:(?!(?P=q1)).)* |
9ea5c04c S |
126 | (?P=q1).*? |
127 | (?: | |
69d69da9 S |
128 | (?: |
129 | entry_?[Ii]d| | |
130 | (?P<q2>["'])entry_?[Ii]d(?P=q2) | |
131 | )\s*:\s*| | |
c2f2f8b1 | 132 | \[\s*(?P<q2_1>["'])entry_?[Ii]d(?P=q2_1)\s*\]\s*=\s* |
69d69da9 | 133 | ) |
78280352 | 134 | (?P<q3>["'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3) |
a01825a5 JH |
135 | ''', webpage) or |
136 | re.search( | |
137 | r'''(?xs) | |
e30991f9 | 138 | <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["']) |
c21692fa | 139 | (?:https?:)?//(?:(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+) |
a01825a5 | 140 | (?:(?!(?P=q1)).)* |
e30991f9 | 141 | [?&;]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+) |
c21692fa | 142 | (?:(?!(?P=q1)).)* |
a01825a5 JH |
143 | (?P=q1) |
144 | ''', webpage) | |
145 | ) | |
9ea5c04c | 146 | if mobj: |
81953d1a RA |
147 | embed_info = mobj.groupdict() |
148 | url = 'kaltura:%(partner_id)s:%(id)s' % embed_info | |
d9163ae3 | 149 | escaped_pid = re.escape(embed_info['partner_id']) |
81953d1a | 150 | service_url = re.search( |
d9163ae3 | 151 | r'<script[^>]+src=["\']((?:https?:)?//.+?)/p/%s/sp/%s00/embedIframeJs' % (escaped_pid, escaped_pid), |
81953d1a RA |
152 | webpage) |
153 | if service_url: | |
154 | url = smuggle_url(url, {'service_url': service_url.group(1)}) | |
155 | return url | |
9ea5c04c | 156 | |
d9163ae3 | 157 | def _kaltura_api_call(self, video_id, actions, service_url=None, *args, **kwargs): |
0d97ef43 NJ |
158 | params = actions[0] |
159 | if len(actions) > 1: | |
160 | for i, a in enumerate(actions[1:], start=1): | |
161 | for k, v in a.items(): | |
162 | params['%d:%s' % (i, k)] = v | |
163 | ||
81953d1a | 164 | data = self._download_json( |
d9163ae3 | 165 | (service_url or self._SERVICE_URL) + self._SERVICE_BASE, |
81953d1a | 166 | video_id, query=params, *args, **kwargs) |
0d97ef43 NJ |
167 | |
168 | status = data if len(actions) == 1 else data[0] | |
169 | if status.get('objectType') == 'KalturaAPIException': | |
170 | raise ExtractorError( | |
171 | '%s said: %s' % (self.IE_NAME, status['message'])) | |
172 | ||
173 | return data | |
174 | ||
d9163ae3 | 175 | def _get_video_info(self, video_id, partner_id, service_url=None): |
0d97ef43 NJ |
176 | actions = [ |
177 | { | |
178 | 'action': 'null', | |
179 | 'apiVersion': '3.1.5', | |
180 | 'clientTag': 'kdp:v3.8.5', | |
181 | 'format': 1, # JSON, 2 = XML, 3 = PHP | |
182 | 'service': 'multirequest', | |
1094074c RA |
183 | }, |
184 | { | |
185 | 'expiry': 86400, | |
186 | 'service': 'session', | |
187 | 'action': 'startWidgetSession', | |
188 | 'widgetId': '_%s' % partner_id, | |
0d97ef43 NJ |
189 | }, |
190 | { | |
191 | 'action': 'get', | |
192 | 'entryId': video_id, | |
193 | 'service': 'baseentry', | |
1094074c | 194 | 'ks': '{1:result:ks}', |
9b27a78a RA |
195 | 'responseProfile:fields': 'createdAt,dataUrl,duration,name,plays,thumbnailUrl,userId', |
196 | 'responseProfile:type': 1, | |
0d97ef43 NJ |
197 | }, |
198 | { | |
bb4b8c57 | 199 | 'action': 'getbyentryid', |
0d97ef43 | 200 | 'entryId': video_id, |
bb4b8c57 | 201 | 'service': 'flavorAsset', |
1094074c RA |
202 | 'ks': '{1:result:ks}', |
203 | }, | |
204 | { | |
205 | 'action': 'list', | |
206 | 'filter:entryIdEqual': video_id, | |
207 | 'service': 'caption_captionasset', | |
208 | 'ks': '{1:result:ks}', | |
0d97ef43 NJ |
209 | }, |
210 | ] | |
211 | return self._kaltura_api_call( | |
d9163ae3 | 212 | video_id, actions, service_url, note='Downloading video info JSON') |
0d97ef43 NJ |
213 | |
214 | def _real_extract(self, url): | |
01b06aed S |
215 | url, smuggled_data = unsmuggle_url(url, {}) |
216 | ||
0d97ef43 | 217 | mobj = re.match(self._VALID_URL, url) |
fac7e792 | 218 | partner_id, entry_id = mobj.group('partner_id', 'id') |
9dce3c09 | 219 | ks = None |
1094074c | 220 | captions = None |
fac7e792 | 221 | if partner_id and entry_id: |
1094074c | 222 | _, info, flavor_assets, captions = self._get_video_info(entry_id, partner_id, smuggled_data.get('service_url')) |
fac7e792 | 223 | else: |
224 | path, query = mobj.group('path', 'query') | |
225 | if not path and not query: | |
226 | raise ExtractorError('Invalid URL', expected=True) | |
227 | params = {} | |
228 | if query: | |
229 | params = compat_parse_qs(query) | |
230 | if path: | |
231 | splitted_path = path.split('/') | |
e5a2e17a | 232 | params.update(dict((zip(splitted_path[::2], [[v] for v in splitted_path[1::2]])))) |
fac7e792 | 233 | if 'wid' in params: |
234 | partner_id = params['wid'][0][1:] | |
235 | elif 'p' in params: | |
236 | partner_id = params['p'][0] | |
37084f66 RA |
237 | elif 'partner_id' in params: |
238 | partner_id = params['partner_id'][0] | |
fac7e792 | 239 | else: |
240 | raise ExtractorError('Invalid URL', expected=True) | |
241 | if 'entry_id' in params: | |
242 | entry_id = params['entry_id'][0] | |
1094074c | 243 | _, info, flavor_assets, captions = self._get_video_info(entry_id, partner_id) |
fac7e792 | 244 | elif 'uiconf_id' in params and 'flashvars[referenceId]' in params: |
245 | reference_id = params['flashvars[referenceId]'][0] | |
246 | webpage = self._download_webpage(url, reference_id) | |
247 | entry_data = self._parse_json(self._search_regex( | |
248 | r'window\.kalturaIframePackageData\s*=\s*({.*});', | |
249 | webpage, 'kalturaIframePackageData'), | |
250 | reference_id)['entryResult'] | |
251 | info, flavor_assets = entry_data['meta'], entry_data['contextData']['flavorAssets'] | |
252 | entry_id = info['id'] | |
a8094467 S |
253 | # Unfortunately, data returned in kalturaIframePackageData lacks |
254 | # captions so we will try requesting the complete data using | |
255 | # regular approach since we now know the entry_id | |
256 | try: | |
257 | _, info, flavor_assets, captions = self._get_video_info( | |
258 | entry_id, partner_id) | |
259 | except ExtractorError: | |
260 | # Regular scenario failed but we already have everything | |
261 | # extracted apart from captions and can process at least | |
262 | # with this | |
263 | pass | |
fac7e792 | 264 | else: |
265 | raise ExtractorError('Invalid URL', expected=True) | |
9dce3c09 | 266 | ks = params.get('flashvars[ks]', [None])[0] |
0d97ef43 | 267 | |
01b06aed S |
268 | source_url = smuggled_data.get('source_url') |
269 | if source_url: | |
270 | referrer = base64.b64encode( | |
271 | '://'.join(compat_urlparse.urlparse(source_url)[:2]) | |
272 | .encode('utf-8')).decode('utf-8') | |
273 | else: | |
274 | referrer = None | |
275 | ||
9dce3c09 S |
276 | def sign_url(unsigned_url): |
277 | if ks: | |
278 | unsigned_url += '/ks/%s' % ks | |
279 | if referrer: | |
280 | unsigned_url += '?referrer=%s' % referrer | |
281 | return unsigned_url | |
282 | ||
81953d1a RA |
283 | data_url = info['dataUrl'] |
284 | if '/flvclipper/' in data_url: | |
285 | data_url = re.sub(r'/flvclipper/.*', '/serveFlavor', data_url) | |
286 | ||
01b06aed | 287 | formats = [] |
bb4b8c57 | 288 | for f in flavor_assets: |
289 | # Continue if asset is not ready | |
1094074c | 290 | if f.get('status') != 2: |
bb4b8c57 | 291 | continue |
e8bcd982 S |
292 | # Original format that's not available (e.g. kaltura:1926081:0_c03e1b5g) |
293 | # skip for now. | |
294 | if f.get('fileExt') == 'chun': | |
295 | continue | |
c0bddd6d | 296 | # DRM-protected video, cannot be decrypted |
297 | if f.get('fileExt') == 'wvm': | |
298 | continue | |
ab6f6aee | 299 | if not f.get('fileExt'): |
8ab7e6c4 | 300 | # QT indicates QuickTime; some videos have broken fileExt |
ab6f6aee S |
301 | if f.get('containerFormat') == 'qt': |
302 | f['fileExt'] = 'mov' | |
303 | else: | |
304 | f['fileExt'] = 'mp4' | |
81953d1a RA |
305 | video_url = sign_url( |
306 | '%s/flavorId/%s' % (data_url, f['id'])) | |
1d16035b S |
307 | # audio-only has no videoCodecId (e.g. kaltura:1926081:0_c03e1b5g |
308 | # -f mp4-56) | |
309 | vcodec = 'none' if 'videoCodecId' not in f and f.get( | |
310 | 'frameRate') == 0 else f.get('videoCodecId') | |
01b06aed S |
311 | formats.append({ |
312 | 'format_id': '%(fileExt)s-%(bitrate)s' % f, | |
d80a39ce S |
313 | 'ext': f.get('fileExt'), |
314 | 'tbr': int_or_none(f['bitrate']), | |
315 | 'fps': int_or_none(f.get('frameRate')), | |
01b06aed S |
316 | 'filesize_approx': int_or_none(f.get('size'), invscale=1024), |
317 | 'container': f.get('containerFormat'), | |
1d16035b | 318 | 'vcodec': vcodec, |
d80a39ce S |
319 | 'height': int_or_none(f.get('height')), |
320 | 'width': int_or_none(f.get('width')), | |
01b06aed S |
321 | 'url': video_url, |
322 | }) | |
81953d1a RA |
323 | if '/playManifest/' in data_url: |
324 | m3u8_url = sign_url(data_url.replace( | |
325 | 'format/url', 'format/applehttp')) | |
326 | formats.extend(self._extract_m3u8_formats( | |
327 | m3u8_url, entry_id, 'mp4', 'm3u8_native', | |
328 | m3u8_id='hls', fatal=False)) | |
bb4b8c57 | 329 | |
0d97ef43 NJ |
330 | self._sort_formats(formats) |
331 | ||
1094074c RA |
332 | subtitles = {} |
333 | if captions: | |
334 | for caption in captions.get('objects', []): | |
1094074c | 335 | # Continue if caption is not ready |
65547082 | 336 | if caption.get('status') != 2: |
1094074c | 337 | continue |
92d4cfa3 S |
338 | if not caption.get('id'): |
339 | continue | |
340 | caption_format = int_or_none(caption.get('format')) | |
1094074c RA |
341 | subtitles.setdefault(caption.get('languageCode') or caption.get('language'), []).append({ |
342 | 'url': '%s/api_v3/service/caption_captionasset/action/serve/captionAssetId/%s' % (self._SERVICE_URL, caption['id']), | |
92d4cfa3 | 343 | 'ext': caption.get('fileExt') or self._CAPTION_TYPES.get(caption_format) or 'ttml', |
1094074c RA |
344 | }) |
345 | ||
0d97ef43 | 346 | return { |
ee3ec091 | 347 | 'id': entry_id, |
0d97ef43 NJ |
348 | 'title': info['name'], |
349 | 'formats': formats, | |
1094074c | 350 | 'subtitles': subtitles, |
bdceea7a | 351 | 'description': clean_html(info.get('description')), |
0d97ef43 NJ |
352 | 'thumbnail': info.get('thumbnailUrl'), |
353 | 'duration': info.get('duration'), | |
354 | 'timestamp': info.get('createdAt'), | |
26e40542 | 355 | 'uploader_id': info.get('userId') if info.get('userId') != 'None' else None, |
0d97ef43 NJ |
356 | 'view_count': info.get('plays'), |
357 | } |