]>
Commit | Line | Data |
---|---|---|
21fbf0f9 | 1 | import functools |
34541395 | 2 | import itertools |
df2a5633 | 3 | import math |
21fbf0f9 | 4 | import operator |
125cfd78 | 5 | import re |
6 | ||
7 | from .common import InfoExtractor | |
278d061a | 8 | from .openload import PhantomJSwrapper |
3d2623a8 | 9 | from ..networking import Request |
10 | from ..networking.exceptions import HTTPError | |
1cc79574 | 11 | from ..utils import ( |
ac668111 | 12 | NO_DEFAULT, |
13 | ExtractorError, | |
d0fb4bd1 | 14 | clean_html, |
b8526c78 | 15 | determine_ext, |
e0ddbd02 | 16 | format_field, |
ed8648a3 | 17 | int_or_none, |
cd85a1bb | 18 | merge_dicts, |
8f9a477e | 19 | orderedSet, |
e1e35d1a | 20 | remove_quotes, |
4f2a58c9 | 21 | remove_start, |
0320ddc1 | 22 | str_to_int, |
2181983a | 23 | update_url_query, |
4938c8d5 | 24 | url_or_none, |
ac668111 | 25 | urlencode_postdata, |
125cfd78 | 26 | ) |
125cfd78 | 27 | |
9933b574 | 28 | |
71a1f617 | 29 | class PornHubBaseIE(InfoExtractor): |
2181983a | 30 | _NETRC_MACHINE = 'pornhub' |
4c968755 | 31 | _PORNHUB_HOST_RE = r'(?:(?P<host>pornhub(?:premium)?\.(?:com|net|org))|pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd\.onion)' |
2181983a | 32 | |
71a1f617 S |
33 | def _download_webpage_handle(self, *args, **kwargs): |
34 | def dl(*args, **kwargs): | |
35 | return super(PornHubBaseIE, self)._download_webpage_handle(*args, **kwargs) | |
36 | ||
29f7c58a | 37 | ret = dl(*args, **kwargs) |
38 | ||
39 | if not ret: | |
40 | return ret | |
41 | ||
42 | webpage, urlh = ret | |
71a1f617 S |
43 | |
44 | if any(re.search(p, webpage) for p in ( | |
45 | r'<body\b[^>]+\bonload=["\']go\(\)', | |
46 | r'document\.cookie\s*=\s*["\']RNKEY=', | |
47 | r'document\.location\.reload\(true\)')): | |
48 | url_or_request = args[0] | |
3d2623a8 | 49 | url = (url_or_request.url |
50 | if isinstance(url_or_request, Request) | |
71a1f617 S |
51 | else url_or_request) |
52 | phantom = PhantomJSwrapper(self, required_version='2.0') | |
53 | phantom.get(url, html=webpage) | |
54 | webpage, urlh = dl(*args, **kwargs) | |
55 | ||
56 | return webpage, urlh | |
57 | ||
2181983a | 58 | def _real_initialize(self): |
59 | self._logged_in = False | |
60 | ||
62beefa8 N |
61 | def _set_age_cookies(self, host): |
62 | self._set_cookie(host, 'age_verified', '1') | |
63 | self._set_cookie(host, 'accessAgeDisclaimerPH', '1') | |
1d3d579c | 64 | self._set_cookie(host, 'accessAgeDisclaimerUK', '1') |
62beefa8 N |
65 | self._set_cookie(host, 'accessPH', '1') |
66 | ||
2181983a | 67 | def _login(self, host): |
68 | if self._logged_in: | |
69 | return | |
70 | ||
71 | site = host.split('.')[0] | |
72 | ||
73 | # Both sites pornhub and pornhubpremium have separate accounts | |
74 | # so there should be an option to provide credentials for both. | |
75 | # At the same time some videos are available under the same video id | |
76 | # on both sites so that we have to identify them as the same video. | |
77 | # For that purpose we have to keep both in the same extractor | |
78 | # but under different netrc machines. | |
79 | username, password = self._get_login_info(netrc_machine=site) | |
80 | if username is None: | |
81 | return | |
82 | ||
add96eb9 | 83 | login_url = 'https://www.{}/{}login'.format(host, 'premium/' if 'premium' in host else '') |
2181983a | 84 | login_page = self._download_webpage( |
add96eb9 | 85 | login_url, None, f'Downloading {site} login page') |
2181983a | 86 | |
87 | def is_logged(webpage): | |
88 | return any(re.search(p, webpage) for p in ( | |
de954c1b | 89 | r'id="profileMenuDropdown"', |
90 | r'class="ph-icon-logout"')) | |
2181983a | 91 | |
92 | if is_logged(login_page): | |
93 | self._logged_in = True | |
94 | return | |
95 | ||
96 | login_form = self._hidden_inputs(login_page) | |
97 | ||
98 | login_form.update({ | |
518c1afc | 99 | 'email': username, |
2181983a | 100 | 'password': password, |
101 | }) | |
102 | ||
103 | response = self._download_json( | |
add96eb9 | 104 | f'https://www.{host}/front/authenticate', None, |
105 | f'Logging in to {site}', | |
2181983a | 106 | data=urlencode_postdata(login_form), |
107 | headers={ | |
108 | 'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8', | |
109 | 'Referer': login_url, | |
110 | 'X-Requested-With': 'XMLHttpRequest', | |
111 | }) | |
112 | ||
113 | if response.get('success') == '1': | |
114 | self._logged_in = True | |
115 | return | |
116 | ||
117 | message = response.get('message') | |
118 | if message is not None: | |
119 | raise ExtractorError( | |
add96eb9 | 120 | f'Unable to login: {message}', expected=True) |
2181983a | 121 | |
122 | raise ExtractorError('Unable to log in') | |
123 | ||
71a1f617 S |
124 | |
125 | class PornHubIE(PornHubBaseIE): | |
bc4b2d75 | 126 | IE_DESC = 'PornHub and Thumbzilla' |
add96eb9 | 127 | _VALID_URL = rf'''(?x) |
bc4b2d75 S |
128 | https?:// |
129 | (?: | |
ed807c18 | 130 | (?:[^/]+\.)? |
add96eb9 | 131 | {PornHubBaseIE._PORNHUB_HOST_RE} |
ed807c18 | 132 | /(?:(?:view_video\.php|video/show)\?viewkey=|embed/)| |
bc4b2d75 S |
133 | (?:www\.)?thumbzilla\.com/video/ |
134 | ) | |
b52c9ef1 | 135 | (?P<id>[\da-z]+) |
add96eb9 | 136 | ''' |
bfd973ec | 137 | _EMBED_REGEX = [r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?pornhub(?:premium)?\.(?:com|net|org)/embed/[\da-z]+)'] |
360075e2 | 138 | _TESTS = [{ |
9933b574 | 139 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=648719015', |
cd85a1bb | 140 | 'md5': 'a6391306d050e4547f62b3f485dd9ba9', |
9933b574 | 141 | 'info_dict': { |
249efaf4 PH |
142 | 'id': '648719015', |
143 | 'ext': 'mp4', | |
611c1dd9 | 144 | 'title': 'Seductive Indian beauty strips down and fingers her pink pussy', |
ed8648a3 | 145 | 'uploader': 'Babes', |
3661ebf2 | 146 | 'upload_date': '20130628', |
cd85a1bb | 147 | 'timestamp': 1372447216, |
ed8648a3 S |
148 | 'duration': 361, |
149 | 'view_count': int, | |
150 | 'like_count': int, | |
151 | 'dislike_count': int, | |
152 | 'comment_count': int, | |
153 | 'age_limit': 18, | |
6bb05b32 YCH |
154 | 'tags': list, |
155 | 'categories': list, | |
d0fb4bd1 | 156 | 'cast': list, |
6c376029 S |
157 | }, |
158 | }, { | |
159 | # non-ASCII title | |
160 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=1331683002', | |
161 | 'info_dict': { | |
162 | 'id': '1331683002', | |
163 | 'ext': 'mp4', | |
164 | 'title': '重庆婷婷女王足交', | |
3661ebf2 | 165 | 'upload_date': '20150213', |
cd85a1bb | 166 | 'timestamp': 1423804862, |
6c376029 S |
167 | 'duration': 1753, |
168 | 'view_count': int, | |
169 | 'like_count': int, | |
170 | 'dislike_count': int, | |
171 | 'comment_count': int, | |
172 | 'age_limit': 18, | |
6bb05b32 YCH |
173 | 'tags': list, |
174 | 'categories': list, | |
6c376029 S |
175 | }, |
176 | 'params': { | |
177 | 'skip_download': True, | |
178 | }, | |
10db0d2f | 179 | 'skip': 'Video has been flagged for verification in accordance with our trust and safety policy', |
4938c8d5 GF |
180 | }, { |
181 | # subtitles | |
182 | 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5af5fef7c2aa7', | |
183 | 'info_dict': { | |
184 | 'id': 'ph5af5fef7c2aa7', | |
185 | 'ext': 'mp4', | |
186 | 'title': 'BFFS - Cute Teen Girls Share Cock On the Floor', | |
187 | 'uploader': 'BFFs', | |
188 | 'duration': 622, | |
189 | 'view_count': int, | |
190 | 'like_count': int, | |
191 | 'dislike_count': int, | |
192 | 'comment_count': int, | |
193 | 'age_limit': 18, | |
194 | 'tags': list, | |
195 | 'categories': list, | |
196 | 'subtitles': { | |
197 | 'en': [{ | |
add96eb9 | 198 | 'ext': 'srt', |
199 | }], | |
4938c8d5 GF |
200 | }, |
201 | }, | |
202 | 'params': { | |
203 | 'skip_download': True, | |
204 | }, | |
cd85a1bb | 205 | 'skip': 'This video has been disabled', |
4f2a58c9 L |
206 | }, { |
207 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph601dc30bae19a', | |
208 | 'info_dict': { | |
209 | 'id': 'ph601dc30bae19a', | |
210 | 'uploader': 'Projekt Melody', | |
211 | 'uploader_id': 'projekt-melody', | |
212 | 'upload_date': '20210205', | |
213 | 'title': '"Welcome to My Pussy Mansion" - CB Stream (02/03/21)', | |
214 | 'thumbnail': r're:https?://.+', | |
215 | }, | |
360075e2 S |
216 | }, { |
217 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph557bbb6676d2d', | |
218 | 'only_matching': True, | |
272e4db5 | 219 | }, { |
eaaaaec0 | 220 | # removed at the request of cam4.com |
272e4db5 S |
221 | 'url': 'http://fr.pornhub.com/view_video.php?viewkey=ph55ca2f9760862', |
222 | 'only_matching': True, | |
eaaaaec0 S |
223 | }, { |
224 | # removed at the request of the copyright owner | |
225 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=788152859', | |
226 | 'only_matching': True, | |
227 | }, { | |
228 | # removed by uploader | |
229 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph572716d15a111', | |
230 | 'only_matching': True, | |
195f0845 S |
231 | }, { |
232 | # private video | |
233 | 'url': 'http://www.pornhub.com/view_video.php?viewkey=ph56fd731fce6b7', | |
234 | 'only_matching': True, | |
bc4b2d75 S |
235 | }, { |
236 | 'url': 'https://www.thumbzilla.com/video/ph56c6114abd99a/horny-girlfriend-sex', | |
237 | 'only_matching': True, | |
a99cc4ca S |
238 | }, { |
239 | 'url': 'http://www.pornhub.com/video/show?viewkey=648719015', | |
240 | 'only_matching': True, | |
f97c0991 S |
241 | }, { |
242 | 'url': 'https://www.pornhub.net/view_video.php?viewkey=203640933', | |
243 | 'only_matching': True, | |
29f7c58a | 244 | }, { |
245 | 'url': 'https://www.pornhub.org/view_video.php?viewkey=203640933', | |
246 | 'only_matching': True, | |
fa9b8c66 TW |
247 | }, { |
248 | 'url': 'https://www.pornhubpremium.com/view_video.php?viewkey=ph5e4acdae54a82', | |
249 | 'only_matching': True, | |
2181983a | 250 | }, { |
251 | # Some videos are available with the same id on both premium | |
252 | # and non-premium sites (e.g. this and the following test) | |
253 | 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5f75b0f4b18e3', | |
254 | 'only_matching': True, | |
255 | }, { | |
256 | 'url': 'https://www.pornhubpremium.com/view_video.php?viewkey=ph5f75b0f4b18e3', | |
257 | 'only_matching': True, | |
ed807c18 | 258 | }, { |
259 | # geo restricted | |
260 | 'url': 'https://www.pornhub.com/view_video.php?viewkey=ph5a9813bfa7156', | |
261 | 'only_matching': True, | |
262 | }, { | |
4c968755 | 263 | 'url': 'http://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/view_video.php?viewkey=ph5a9813bfa7156', |
ed807c18 | 264 | 'only_matching': True, |
360075e2 | 265 | }] |
125cfd78 | 266 | |
0320ddc1 | 267 | def _extract_count(self, pattern, webpage, name): |
add96eb9 | 268 | return str_to_int(self._search_regex(pattern, webpage, f'{name} count', default=None)) |
0320ddc1 | 269 | |
125cfd78 | 270 | def _real_extract(self, url): |
5ad28e7f | 271 | mobj = self._match_valid_url(url) |
3430ff9b S |
272 | host = mobj.group('host') or 'pornhub.com' |
273 | video_id = mobj.group('id') | |
7399ca1f | 274 | |
2181983a | 275 | self._login(host) |
62beefa8 | 276 | self._set_age_cookies(host) |
125cfd78 | 277 | |
9a372f14 | 278 | def dl_webpage(platform): |
3430ff9b | 279 | self._set_cookie(host, 'platform', platform) |
9a372f14 | 280 | return self._download_webpage( |
add96eb9 | 281 | f'https://www.{host}/view_video.php?viewkey={video_id}', |
282 | video_id, f'Downloading {platform} webpage') | |
9a372f14 S |
283 | |
284 | webpage = dl_webpage('pc') | |
125cfd78 | 285 | |
50789175 | 286 | error_msg = self._html_search_regex( |
10db0d2f | 287 | (r'(?s)<div[^>]+class=(["\'])(?:(?!\1).)*\b(?:removed|userMessageSection)\b(?:(?!\1).)*\1[^>]*>(?P<error>.+?)</div>', |
288 | r'(?s)<section[^>]+class=["\']noVideo["\'][^>]*>(?P<error>.+?)</section>'), | |
3cb3b600 | 289 | webpage, 'error message', default=None, group='error') |
50789175 PH |
290 | if error_msg: |
291 | error_msg = re.sub(r'\s+', ' ', error_msg) | |
292 | raise ExtractorError( | |
add96eb9 | 293 | f'PornHub said: {error_msg}', |
50789175 PH |
294 | expected=True, video_id=video_id) |
295 | ||
ed807c18 | 296 | if any(re.search(p, webpage) for p in ( |
297 | r'class=["\']geoBlocked["\']', | |
298 | r'>\s*This content is unavailable in your country')): | |
299 | self.raise_geo_restricted() | |
300 | ||
6c376029 S |
301 | # video_title from flashvars contains whitespace instead of non-ASCII (see |
302 | # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying | |
303 | # on that anymore. | |
79367a98 | 304 | title = self._html_search_meta( |
46cc54ca S |
305 | 'twitter:title', webpage, default=None) or self._html_search_regex( |
306 | (r'(?s)<h1[^>]+class=["\']title["\'][^>]*>(?P<title>.+?)</h1>', | |
307 | r'<div[^>]+data-video-title=(["\'])(?P<title>(?:(?!\1).)+)\1', | |
308 | r'shareTitle["\']\s*[=:]\s*(["\'])(?P<title>(?:(?!\1).)+)\1'), | |
6c376029 S |
309 | webpage, 'title', group='title') |
310 | ||
79367a98 S |
311 | video_urls = [] |
312 | video_urls_set = set() | |
4938c8d5 | 313 | subtitles = {} |
79367a98 | 314 | |
ed8648a3 S |
315 | flashvars = self._parse_json( |
316 | self._search_regex( | |
03442072 | 317 | r'var\s+flashvars_\d+\s*=\s*({.+?});', webpage, 'flashvars', default='{}'), |
ed8648a3 S |
318 | video_id) |
319 | if flashvars: | |
4938c8d5 GF |
320 | subtitle_url = url_or_none(flashvars.get('closedCaptionsFile')) |
321 | if subtitle_url: | |
322 | subtitles.setdefault('en', []).append({ | |
323 | 'url': subtitle_url, | |
324 | 'ext': 'srt', | |
325 | }) | |
ed8648a3 S |
326 | thumbnail = flashvars.get('image_url') |
327 | duration = int_or_none(flashvars.get('video_duration')) | |
79367a98 S |
328 | media_definitions = flashvars.get('mediaDefinitions') |
329 | if isinstance(media_definitions, list): | |
330 | for definition in media_definitions: | |
331 | if not isinstance(definition, dict): | |
332 | continue | |
333 | video_url = definition.get('videoUrl') | |
add96eb9 | 334 | if not video_url or not isinstance(video_url, str): |
79367a98 S |
335 | continue |
336 | if video_url in video_urls_set: | |
337 | continue | |
338 | video_urls_set.add(video_url) | |
339 | video_urls.append( | |
340 | (video_url, int_or_none(definition.get('quality')))) | |
ed8648a3 | 341 | else: |
79367a98 S |
342 | thumbnail, duration = [None] * 2 |
343 | ||
0164cd5d | 344 | def extract_js_vars(webpage, pattern, default=NO_DEFAULT): |
79367a98 | 345 | assignments = self._search_regex( |
0164cd5d | 346 | pattern, webpage, 'encoded url', default=default) |
f4134726 S |
347 | if not assignments: |
348 | return {} | |
349 | ||
350 | assignments = assignments.split(';') | |
79367a98 S |
351 | |
352 | js_vars = {} | |
353 | ||
354 | def parse_js_value(inp): | |
355 | inp = re.sub(r'/\*(?:(?!\*/).)*?\*/', '', inp) | |
356 | if '+' in inp: | |
357 | inps = inp.split('+') | |
358 | return functools.reduce( | |
359 | operator.concat, map(parse_js_value, inps)) | |
360 | inp = inp.strip() | |
361 | if inp in js_vars: | |
362 | return js_vars[inp] | |
363 | return remove_quotes(inp) | |
364 | ||
365 | for assn in assignments: | |
366 | assn = assn.strip() | |
367 | if not assn: | |
368 | continue | |
369 | assn = re.sub(r'var\s+', '', assn) | |
370 | vname, value = assn.split('=', 1) | |
371 | js_vars[vname] = parse_js_value(value) | |
f4134726 | 372 | return js_vars |
79367a98 | 373 | |
f4134726 S |
374 | def add_video_url(video_url): |
375 | v_url = url_or_none(video_url) | |
376 | if not v_url: | |
377 | return | |
378 | if v_url in video_urls_set: | |
379 | return | |
380 | video_urls.append((v_url, None)) | |
381 | video_urls_set.add(v_url) | |
382 | ||
29f7c58a | 383 | def parse_quality_items(quality_items): |
384 | q_items = self._parse_json(quality_items, video_id, fatal=False) | |
385 | if not isinstance(q_items, list): | |
386 | return | |
387 | for item in q_items: | |
388 | if isinstance(item, dict): | |
389 | add_video_url(item.get('url')) | |
390 | ||
f4134726 | 391 | if not video_urls: |
29f7c58a | 392 | FORMAT_PREFIXES = ('media', 'quality', 'qualityItems') |
f4134726 | 393 | js_vars = extract_js_vars( |
add96eb9 | 394 | webpage, r'(var\s+(?:{})_.+)'.format('|'.join(FORMAT_PREFIXES)), |
0164cd5d | 395 | default=None) |
f4134726 S |
396 | if js_vars: |
397 | for key, format_url in js_vars.items(): | |
29f7c58a | 398 | if key.startswith(FORMAT_PREFIXES[-1]): |
399 | parse_quality_items(format_url) | |
400 | elif any(key.startswith(p) for p in FORMAT_PREFIXES[:2]): | |
f4134726 | 401 | add_video_url(format_url) |
0164cd5d S |
402 | if not video_urls and re.search( |
403 | r'<[^>]+\bid=["\']lockedPlayer', webpage): | |
404 | raise ExtractorError( | |
add96eb9 | 405 | f'Video {video_id} is locked', expected=True) |
f4134726 S |
406 | |
407 | if not video_urls: | |
408 | js_vars = extract_js_vars( | |
409 | dl_webpage('tv'), r'(var.+?mediastring.+?)</script>') | |
410 | add_video_url(js_vars['mediastring']) | |
79367a98 S |
411 | |
412 | for mobj in re.finditer( | |
413 | r'<a[^>]+\bclass=["\']downloadBtn\b[^>]+\bhref=(["\'])(?P<url>(?:(?!\1).)+)\1', | |
414 | webpage): | |
415 | video_url = mobj.group('url') | |
416 | if video_url not in video_urls_set: | |
417 | video_urls.append((video_url, None)) | |
418 | video_urls_set.add(video_url) | |
419 | ||
3661ebf2 | 420 | upload_date = None |
79367a98 | 421 | formats = [] |
10db0d2f | 422 | |
423 | def add_format(format_url, height=None): | |
f7ad7160 | 424 | ext = determine_ext(format_url) |
425 | if ext == 'mpd': | |
426 | formats.extend(self._extract_mpd_formats( | |
427 | format_url, video_id, mpd_id='dash', fatal=False)) | |
428 | return | |
429 | if ext == 'm3u8': | |
430 | formats.extend(self._extract_m3u8_formats( | |
431 | format_url, video_id, 'mp4', entry_protocol='m3u8_native', | |
432 | m3u8_id='hls', fatal=False)) | |
433 | return | |
ed807c18 | 434 | if not height: |
435 | height = int_or_none(self._search_regex( | |
436 | r'(?P<height>\d+)[pP]?_\d+[kK]', format_url, 'height', | |
437 | default=None)) | |
10db0d2f | 438 | formats.append({ |
439 | 'url': format_url, | |
a70635b8 | 440 | 'format_id': format_field(height, None, '%dp'), |
10db0d2f | 441 | 'height': height, |
10db0d2f | 442 | }) |
443 | ||
79367a98 | 444 | for video_url, height in video_urls: |
3661ebf2 S |
445 | if not upload_date: |
446 | upload_date = self._search_regex( | |
447 | r'/(\d{6}/\d{2})/', video_url, 'upload data', default=None) | |
448 | if upload_date: | |
449 | upload_date = upload_date.replace('/', '') | |
10db0d2f | 450 | if '/video/get_media' in video_url: |
451 | medias = self._download_json(video_url, video_id, fatal=False) | |
452 | if isinstance(medias, list): | |
453 | for media in medias: | |
454 | if not isinstance(media, dict): | |
455 | continue | |
456 | video_url = url_or_none(media.get('videoUrl')) | |
457 | if not video_url: | |
458 | continue | |
459 | height = int_or_none(media.get('quality')) | |
460 | add_format(video_url, height) | |
461 | continue | |
462 | add_format(video_url) | |
ed807c18 | 463 | |
4f2a58c9 L |
464 | model_profile = self._search_json( |
465 | r'var\s+MODEL_PROFILE\s*=', webpage, 'model profile', video_id, fatal=False) | |
0320ddc1 | 466 | video_uploader = self._html_search_regex( |
2d4fe594 | 467 | r'(?s)From: .+?<(?:a\b[^>]+\bhref=["\']/(?:(?:user|channel)s|model|pornstar)/|span\b[^>]+\bclass=["\']username)[^>]+>(.+?)<', |
4f2a58c9 | 468 | webpage, 'uploader', default=None) or model_profile.get('username') |
125cfd78 | 469 | |
29f7c58a | 470 | def extract_vote_count(kind, name): |
471 | return self._extract_count( | |
add96eb9 | 472 | (rf'<span[^>]+\bclass="votes{kind}"[^>]*>([\d,\.]+)</span>', |
473 | rf'<span[^>]+\bclass=["\']votes{kind}["\'][^>]*\bdata-rating=["\'](\d+)'), | |
29f7c58a | 474 | webpage, name) |
475 | ||
7700207e | 476 | view_count = self._extract_count( |
540b9f51 | 477 | r'<span class="count">([\d,\.]+)</span> [Vv]iews', webpage, 'view') |
29f7c58a | 478 | like_count = extract_vote_count('Up', 'like') |
479 | dislike_count = extract_vote_count('Down', 'dislike') | |
0320ddc1 | 480 | comment_count = self._extract_count( |
7700207e | 481 | r'All Comments\s*<span>\(([\d,.]+)\)', webpage, 'comment') |
0320ddc1 | 482 | |
5dda1ede | 483 | def extract_list(meta_key): |
d2d970d0 | 484 | div = self._search_regex( |
add96eb9 | 485 | rf'(?s)<div[^>]+\bclass=["\'].*?\b{meta_key}Wrapper[^>]*>(.+?)</div>', |
486 | webpage, meta_key, default=None) | |
d2d970d0 | 487 | if div: |
d0fb4bd1 | 488 | return [clean_html(x).strip() for x in re.findall(r'(?s)<a[^>]+\bhref=[^>]+>.+?</a>', div)] |
6bb05b32 | 489 | |
cd85a1bb S |
490 | info = self._search_json_ld(webpage, video_id, default={}) |
491 | # description provided in JSON-LD is irrelevant | |
492 | info['description'] = None | |
493 | ||
494 | return merge_dicts({ | |
125cfd78 | 495 | 'id': video_id, |
496 | 'uploader': video_uploader, | |
4f2a58c9 | 497 | 'uploader_id': remove_start(model_profile.get('modelProfileLink'), '/model/'), |
3661ebf2 | 498 | 'upload_date': upload_date, |
6c376029 | 499 | 'title': title, |
125cfd78 | 500 | 'thumbnail': thumbnail, |
ed8648a3 | 501 | 'duration': duration, |
0320ddc1 S |
502 | 'view_count': view_count, |
503 | 'like_count': like_count, | |
504 | 'dislike_count': dislike_count, | |
505 | 'comment_count': comment_count, | |
79367a98 | 506 | 'formats': formats, |
750e9833 | 507 | 'age_limit': 18, |
5dda1ede S |
508 | 'tags': extract_list('tags'), |
509 | 'categories': extract_list('categories'), | |
d0fb4bd1 | 510 | 'cast': extract_list('pornstars'), |
4938c8d5 | 511 | 'subtitles': subtitles, |
cd85a1bb | 512 | }, info) |
e66e1a00 S |
513 | |
514 | ||
71a1f617 | 515 | class PornHubPlaylistBaseIE(PornHubBaseIE): |
2181983a | 516 | def _extract_page(self, url): |
517 | return int_or_none(self._search_regex( | |
518 | r'\bpage=(\d+)', url, 'page', default=None)) | |
519 | ||
3430ff9b | 520 | def _extract_entries(self, webpage, host): |
475bcb22 S |
521 | # Only process container div with main playlist content skipping |
522 | # drop-down menu that uses similar pattern for videos (see | |
067aa17e | 523 | # https://github.com/ytdl-org/youtube-dl/issues/11594). |
475bcb22 S |
524 | container = self._search_regex( |
525 | r'(?s)(<div[^>]+class=["\']container.+)', webpage, | |
526 | 'container', default=webpage) | |
527 | ||
40e146aa | 528 | return [ |
3a23bae9 | 529 | self.url_result( |
add96eb9 | 530 | f'http://www.{host}/{video_url}', |
3a23bae9 S |
531 | PornHubIE.ie_key(), video_title=title) |
532 | for video_url, title in orderedSet(re.findall( | |
533 | r'href="/?(view_video\.php\?.*\bviewkey=[\da-z]+[^"]*)"[^>]*\s+title="([^"]+)"', | |
475bcb22 | 534 | container)) |
40e146aa | 535 | ] |
e66e1a00 | 536 | |
40e146aa | 537 | |
21b08463 | 538 | class PornHubUserIE(PornHubPlaylistBaseIE): |
add96eb9 | 539 | _VALID_URL = rf'(?P<url>https?://(?:[^/]+\.)?{PornHubBaseIE._PORNHUB_HOST_RE}/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/?#&]+))(?:[?#&]|/(?!videos)|$)' |
40e146aa | 540 | _TESTS = [{ |
21b08463 S |
541 | 'url': 'https://www.pornhub.com/model/zoe_ph', |
542 | 'playlist_mincount': 118, | |
543 | }, { | |
544 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious', | |
40e146aa | 545 | 'info_dict': { |
21b08463 | 546 | 'id': 'liz-vicious', |
40e146aa | 547 | }, |
21b08463 S |
548 | 'playlist_mincount': 118, |
549 | }, { | |
550 | 'url': 'https://www.pornhub.com/users/russianveet69', | |
9634de17 | 551 | 'only_matching': True, |
21b08463 S |
552 | }, { |
553 | 'url': 'https://www.pornhub.com/channels/povd', | |
9634de17 S |
554 | 'only_matching': True, |
555 | }, { | |
556 | 'url': 'https://www.pornhub.com/model/zoe_ph?abc=1', | |
557 | 'only_matching': True, | |
2181983a | 558 | }, { |
559 | # Unavailable via /videos page, but available with direct pagination | |
560 | # on pornstar page (see [1]), requires premium | |
561 | # 1. https://github.com/ytdl-org/youtube-dl/issues/27853 | |
562 | 'url': 'https://www.pornhubpremium.com/pornstar/sienna-west', | |
563 | 'only_matching': True, | |
564 | }, { | |
565 | # Same as before, multi page | |
566 | 'url': 'https://www.pornhubpremium.com/pornstar/lily-labeau', | |
567 | 'only_matching': True, | |
ed807c18 | 568 | }, { |
4c968755 | 569 | 'url': 'https://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/model/zoe_ph', |
ed807c18 | 570 | 'only_matching': True, |
21b08463 S |
571 | }] |
572 | ||
21b08463 | 573 | def _real_extract(self, url): |
5ad28e7f | 574 | mobj = self._match_valid_url(url) |
21b08463 | 575 | user_id = mobj.group('id') |
add96eb9 | 576 | videos_url = '{}/videos'.format(mobj.group('url')) |
62beefa8 | 577 | self._set_age_cookies(mobj.group('host')) |
2181983a | 578 | page = self._extract_page(url) |
579 | if page: | |
580 | videos_url = update_url_query(videos_url, {'page': page}) | |
21b08463 | 581 | return self.url_result( |
2181983a | 582 | videos_url, ie=PornHubPagedVideoListIE.ie_key(), video_id=user_id) |
21b08463 S |
583 | |
584 | ||
585 | class PornHubPagedPlaylistBaseIE(PornHubPlaylistBaseIE): | |
4bf568d3 S |
586 | @staticmethod |
587 | def _has_more(webpage): | |
588 | return re.search( | |
589 | r'''(?x) | |
590 | <li[^>]+\bclass=["\']page_next| | |
591 | <link[^>]+\brel=["\']next| | |
592 | <button[^>]+\bid=["\']moreDataBtn | |
593 | ''', webpage) is not None | |
594 | ||
2181983a | 595 | def _entries(self, url, host, item_id): |
596 | page = self._extract_page(url) | |
21b08463 | 597 | |
2181983a | 598 | VIDEOS = '/videos' |
599 | ||
600 | def download_page(base_url, num, fallback=False): | |
add96eb9 | 601 | note = 'Downloading page {}{}'.format(num, ' (switch to fallback)' if fallback else '') |
2181983a | 602 | return self._download_webpage( |
603 | base_url, item_id, note, query={'page': num}) | |
1f7a563a | 604 | |
2181983a | 605 | def is_404(e): |
3d2623a8 | 606 | return isinstance(e.cause, HTTPError) and e.cause.status == 404 |
2181983a | 607 | |
608 | base_url = url | |
609 | has_page = page is not None | |
610 | first_page = page if has_page else 1 | |
611 | for page_num in (first_page, ) if has_page else itertools.count(first_page): | |
21b08463 | 612 | try: |
2181983a | 613 | try: |
614 | webpage = download_page(base_url, page_num) | |
615 | except ExtractorError as e: | |
616 | # Some sources may not be available via /videos page, | |
617 | # trying to fallback to main page pagination (see [1]) | |
618 | # 1. https://github.com/ytdl-org/youtube-dl/issues/27853 | |
619 | if is_404(e) and page_num == first_page and VIDEOS in base_url: | |
620 | base_url = base_url.replace(VIDEOS, '') | |
621 | webpage = download_page(base_url, page_num, fallback=True) | |
622 | else: | |
623 | raise | |
21b08463 | 624 | except ExtractorError as e: |
2181983a | 625 | if is_404(e) and page_num != first_page: |
21b08463 S |
626 | break |
627 | raise | |
628 | page_entries = self._extract_entries(webpage, host) | |
629 | if not page_entries: | |
630 | break | |
2181983a | 631 | for e in page_entries: |
632 | yield e | |
21b08463 S |
633 | if not self._has_more(webpage): |
634 | break | |
635 | ||
2181983a | 636 | def _real_extract(self, url): |
5ad28e7f | 637 | mobj = self._match_valid_url(url) |
2181983a | 638 | host = mobj.group('host') |
639 | item_id = mobj.group('id') | |
640 | ||
641 | self._login(host) | |
62beefa8 | 642 | self._set_age_cookies(host) |
2181983a | 643 | |
644 | return self.playlist_result(self._entries(url, host, item_id), item_id) | |
21b08463 S |
645 | |
646 | ||
9634de17 | 647 | class PornHubPagedVideoListIE(PornHubPagedPlaylistBaseIE): |
add96eb9 | 648 | _VALID_URL = rf'https?://(?:[^/]+\.)?{PornHubBaseIE._PORNHUB_HOST_RE}/(?!playlist/)(?P<id>(?:[^/]+/)*[^/?#&]+)' |
21b08463 | 649 | _TESTS = [{ |
1f7a563a | 650 | 'url': 'https://www.pornhub.com/model/zoe_ph/videos', |
21b08463 | 651 | 'only_matching': True, |
34541395 S |
652 | }, { |
653 | 'url': 'http://www.pornhub.com/users/rushandlia/videos', | |
654 | 'only_matching': True, | |
21b08463 S |
655 | }, { |
656 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos', | |
657 | 'info_dict': { | |
9634de17 | 658 | 'id': 'pornstar/jenny-blighe/videos', |
21b08463 S |
659 | }, |
660 | 'playlist_mincount': 149, | |
1f7a563a S |
661 | }, { |
662 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos?page=3', | |
663 | 'info_dict': { | |
9634de17 | 664 | 'id': 'pornstar/jenny-blighe/videos', |
1f7a563a S |
665 | }, |
666 | 'playlist_mincount': 40, | |
f66df20c PV |
667 | }, { |
668 | # default sorting as Top Rated Videos | |
669 | 'url': 'https://www.pornhub.com/channels/povd/videos', | |
670 | 'info_dict': { | |
9634de17 | 671 | 'id': 'channels/povd/videos', |
f66df20c PV |
672 | }, |
673 | 'playlist_mincount': 293, | |
674 | }, { | |
675 | # Top Rated Videos | |
676 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=ra', | |
677 | 'only_matching': True, | |
678 | }, { | |
679 | # Most Recent Videos | |
680 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=da', | |
681 | 'only_matching': True, | |
682 | }, { | |
683 | # Most Viewed Videos | |
684 | 'url': 'https://www.pornhub.com/channels/povd/videos?o=vi', | |
685 | 'only_matching': True, | |
92ded33a S |
686 | }, { |
687 | 'url': 'http://www.pornhub.com/users/zoe_ph/videos/public', | |
688 | 'only_matching': True, | |
21b08463 S |
689 | }, { |
690 | # Most Viewed Videos | |
691 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=mv', | |
692 | 'only_matching': True, | |
693 | }, { | |
694 | # Top Rated Videos | |
695 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=tr', | |
696 | 'only_matching': True, | |
697 | }, { | |
698 | # Longest Videos | |
699 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=lg', | |
700 | 'only_matching': True, | |
701 | }, { | |
702 | # Newest Videos | |
703 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos?o=cm', | |
704 | 'only_matching': True, | |
21b08463 S |
705 | }, { |
706 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/paid', | |
707 | 'only_matching': True, | |
708 | }, { | |
709 | 'url': 'https://www.pornhub.com/pornstar/liz-vicious/videos/fanonly', | |
710 | 'only_matching': True, | |
9634de17 S |
711 | }, { |
712 | 'url': 'https://www.pornhub.com/video', | |
713 | 'only_matching': True, | |
714 | }, { | |
715 | 'url': 'https://www.pornhub.com/video?page=3', | |
716 | 'only_matching': True, | |
717 | }, { | |
718 | 'url': 'https://www.pornhub.com/video/search?search=123', | |
719 | 'only_matching': True, | |
720 | }, { | |
721 | 'url': 'https://www.pornhub.com/categories/teen', | |
722 | 'only_matching': True, | |
723 | }, { | |
724 | 'url': 'https://www.pornhub.com/categories/teen?page=3', | |
725 | 'only_matching': True, | |
726 | }, { | |
727 | 'url': 'https://www.pornhub.com/hd', | |
728 | 'only_matching': True, | |
729 | }, { | |
730 | 'url': 'https://www.pornhub.com/hd?page=3', | |
731 | 'only_matching': True, | |
732 | }, { | |
733 | 'url': 'https://www.pornhub.com/described-video', | |
734 | 'only_matching': True, | |
735 | }, { | |
736 | 'url': 'https://www.pornhub.com/described-video?page=2', | |
737 | 'only_matching': True, | |
738 | }, { | |
739 | 'url': 'https://www.pornhub.com/video/incategories/60fps-1/hd-porn', | |
740 | 'only_matching': True, | |
ed807c18 | 741 | }, { |
4c968755 | 742 | 'url': 'https://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/model/zoe_ph/videos', |
ed807c18 | 743 | 'only_matching': True, |
40e146aa S |
744 | }] |
745 | ||
21b08463 S |
746 | @classmethod |
747 | def suitable(cls, url): | |
748 | return (False | |
9634de17 | 749 | if PornHubIE.suitable(url) or PornHubUserIE.suitable(url) or PornHubUserVideosUploadIE.suitable(url) |
add96eb9 | 750 | else super().suitable(url)) |
21b08463 | 751 | |
34541395 | 752 | |
21b08463 | 753 | class PornHubUserVideosUploadIE(PornHubPagedPlaylistBaseIE): |
add96eb9 | 754 | _VALID_URL = rf'(?P<url>https?://(?:[^/]+\.)?{PornHubBaseIE._PORNHUB_HOST_RE}/(?:(?:user|channel)s|model|pornstar)/(?P<id>[^/]+)/videos/upload)' |
21b08463 S |
755 | _TESTS = [{ |
756 | 'url': 'https://www.pornhub.com/pornstar/jenny-blighe/videos/upload', | |
757 | 'info_dict': { | |
758 | 'id': 'jenny-blighe', | |
759 | }, | |
760 | 'playlist_mincount': 129, | |
761 | }, { | |
762 | 'url': 'https://www.pornhub.com/model/zoe_ph/videos/upload', | |
763 | 'only_matching': True, | |
ed807c18 | 764 | }, { |
4c968755 | 765 | 'url': 'http://pornhubvybmsymdol4iibwgwtkpwmeyd6luq2gxajgjzfjvotyt5zhyd.onion/pornstar/jenny-blighe/videos/upload', |
ed807c18 | 766 | 'only_matching': True, |
21b08463 | 767 | }] |
df2a5633 | 768 | |
769 | ||
770 | class PornHubPlaylistIE(PornHubPlaylistBaseIE): | |
add96eb9 | 771 | _VALID_URL = rf'(?P<url>https?://(?:[^/]+\.)?{PornHubBaseIE._PORNHUB_HOST_RE}/playlist/(?P<id>[^/?#&]+))' |
df2a5633 | 772 | _TESTS = [{ |
773 | 'url': 'https://www.pornhub.com/playlist/44121572', | |
774 | 'info_dict': { | |
775 | 'id': '44121572', | |
776 | }, | |
777 | 'playlist_count': 77, | |
778 | }, { | |
779 | 'url': 'https://www.pornhub.com/playlist/4667351', | |
780 | 'only_matching': True, | |
781 | }, { | |
782 | 'url': 'https://de.pornhub.com/playlist/4667351', | |
783 | 'only_matching': True, | |
784 | }, { | |
785 | 'url': 'https://de.pornhub.com/playlist/4667351?page=2', | |
786 | 'only_matching': True, | |
787 | }] | |
788 | ||
789 | def _entries(self, url, host, item_id): | |
790 | webpage = self._download_webpage(url, item_id, 'Downloading page 1') | |
791 | playlist_id = self._search_regex(r'var\s+playlistId\s*=\s*"([^"]+)"', webpage, 'playlist_id') | |
792 | video_count = int_or_none( | |
793 | self._search_regex(r'var\s+itemsCount\s*=\s*([0-9]+)\s*\|\|', webpage, 'video_count')) | |
794 | token = self._search_regex(r'var\s+token\s*=\s*"([^"]+)"', webpage, 'token') | |
795 | page_count = math.ceil((video_count - 36) / 40.) + 1 | |
796 | page_entries = self._extract_entries(webpage, host) | |
797 | ||
798 | def download_page(page_num): | |
add96eb9 | 799 | note = f'Downloading page {page_num}' |
800 | page_url = f'https://www.{host}/playlist/viewChunked' | |
df2a5633 | 801 | return self._download_webpage(page_url, item_id, note, query={ |
802 | 'id': playlist_id, | |
803 | 'page': page_num, | |
804 | 'token': token, | |
805 | }) | |
806 | ||
807 | for page_num in range(1, page_count + 1): | |
808 | if page_num > 1: | |
809 | webpage = download_page(page_num) | |
810 | page_entries = self._extract_entries(webpage, host) | |
811 | if not page_entries: | |
812 | break | |
add96eb9 | 813 | yield from page_entries |
df2a5633 | 814 | |
815 | def _real_extract(self, url): | |
5ad28e7f | 816 | mobj = self._match_valid_url(url) |
df2a5633 | 817 | host = mobj.group('host') |
818 | item_id = mobj.group('id') | |
819 | ||
820 | self._login(host) | |
62beefa8 | 821 | self._set_age_cookies(host) |
df2a5633 | 822 | |
823 | return self.playlist_result(self._entries(mobj.group('url'), host, item_id), item_id) |