youtube_dl/extractor/instagram.py

   1 from __future__ import unicode_literals
   2
   3 import itertools
   4 import hashlib
   5 import json
   6 import re
   7
   8 from .common import InfoExtractor
   9 from ..compat import (
  10     compat_str,
  11     compat_HTTPError,
  12 )
  13 from ..utils import (
  14     ExtractorError,
  15     get_element_by_attribute,
  16     int_or_none,
  17     lowercase_escape,
  18     std_headers,
  19     try_get,
  20     url_or_none,
  21 )
  22
  23
  24 class InstagramIE(InfoExtractor):
  25     _VALID_URL = r'(?P<url>https?://(?:www\.)?instagram\.com/(?:p|tv|reel)/(?P<id>[^/?#&]+))'
  26     _TESTS = [{
  27         'url': 'https://instagram.com/p/aye83DjauH/?foo=bar#abc',
  28         'md5': '0d2da106a9d2631273e192b372806516',
  29         'info_dict': {
  30             'id': 'aye83DjauH',
  31             'ext': 'mp4',
  32             'title': 'Video by naomipq',
  33             'description': 'md5:1f17f0ab29bd6fe2bfad705f58de3cb8',
  34             'thumbnail': r're:^https?://.*\.jpg',
  35             'timestamp': 1371748545,
  36             'upload_date': '20130620',
  37             'uploader_id': 'naomipq',
  38             'uploader': 'Naomi Leonor Phan-Quang',
  39             'like_count': int,
  40             'comment_count': int,
  41             'comments': list,
  42         },
  43     }, {
  44         # missing description
  45         'url': 'https://www.instagram.com/p/BA-pQFBG8HZ/?taken-by=britneyspears',
  46         'info_dict': {
  47             'id': 'BA-pQFBG8HZ',
  48             'ext': 'mp4',
  49             'title': 'Video by britneyspears',
  50             'thumbnail': r're:^https?://.*\.jpg',
  51             'timestamp': 1453760977,
  52             'upload_date': '20160125',
  53             'uploader_id': 'britneyspears',
  54             'uploader': 'Britney Spears',
  55             'like_count': int,
  56             'comment_count': int,
  57             'comments': list,
  58         },
  59         'params': {
  60             'skip_download': True,
  61         },
  62     }, {
  63         # multi video post
  64         'url': 'https://www.instagram.com/p/BQ0eAlwhDrw/',
  65         'playlist': [{
  66             'info_dict': {
  67                 'id': 'BQ0dSaohpPW',
  68                 'ext': 'mp4',
  69                 'title': 'Video 1',
  70             },
  71         }, {
  72             'info_dict': {
  73                 'id': 'BQ0dTpOhuHT',
  74                 'ext': 'mp4',
  75                 'title': 'Video 2',
  76             },
  77         }, {
  78             'info_dict': {
  79                 'id': 'BQ0dT7RBFeF',
  80                 'ext': 'mp4',
  81                 'title': 'Video 3',
  82             },
  83         }],
  84         'info_dict': {
  85             'id': 'BQ0eAlwhDrw',
  86             'title': 'Post by instagram',
  87             'description': 'md5:0f9203fc6a2ce4d228da5754bcf54957',
  88         },
  89     }, {
  90         'url': 'https://instagram.com/p/-Cmh1cukG2/',
  91         'only_matching': True,
  92     }, {
  93         'url': 'http://instagram.com/p/9o6LshA7zy/embed/',
  94         'only_matching': True,
  95     }, {
  96         'url': 'https://www.instagram.com/tv/aye83DjauH/',
  97         'only_matching': True,
  98     }, {
  99         'url': 'https://www.instagram.com/reel/CDUMkliABpa/',
 100         'only_matching': True,
 101     }]
 102
 103     @staticmethod
 104     def _extract_embed_url(webpage):
 105         mobj = re.search(
 106             r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?instagram\.com/p/[^/]+/embed.*?)\1',
 107             webpage)
 108         if mobj:
 109             return mobj.group('url')
 110
 111         blockquote_el = get_element_by_attribute(
 112             'class', 'instagram-media', webpage)
 113         if blockquote_el is None:
 114             return
 115
 116         mobj = re.search(
 117             r'<a[^>]+href=([\'"])(?P<link>[^\'"]+)\1', blockquote_el)
 118         if mobj:
 119             return mobj.group('link')
 120
 121     def _real_extract(self, url):
 122         mobj = re.match(self._VALID_URL, url)
 123         video_id = mobj.group('id')
 124         url = mobj.group('url')
 125
 126         webpage = self._download_webpage(url, video_id)
 127
 128         (media, video_url, description, thumbnail, timestamp, uploader,
 129          uploader_id, like_count, comment_count, comments, height,
 130          width) = [None] * 12
 131
 132         shared_data = self._parse_json(
 133             self._search_regex(
 134                 r'window\._sharedData\s*=\s*({.+?});',
 135                 webpage, 'shared data', default='{}'),
 136             video_id, fatal=False)
 137         if shared_data:
 138             media = try_get(
 139                 shared_data,
 140                 (lambda x: x['entry_data']['PostPage'][0]['graphql']['shortcode_media'],
 141                  lambda x: x['entry_data']['PostPage'][0]['media']),
 142                 dict)
 143         # _sharedData.entry_data.PostPage is empty when authenticated (see
 144         # https://github.com/ytdl-org/youtube-dl/pull/22880)
 145         if not media:
 146             additional_data = self._parse_json(
 147                 self._search_regex(
 148                     r'window\.__additionalDataLoaded\s*\(\s*[^,]+,\s*({.+?})\s*\)\s*;',
 149                     webpage, 'additional data', default='{}'),
 150                 video_id, fatal=False)
 151             if additional_data:
 152                 media = try_get(
 153                     additional_data, lambda x: x['graphql']['shortcode_media'],
 154                     dict)
 155         if media:
 156             video_url = media.get('video_url')
 157             height = int_or_none(media.get('dimensions', {}).get('height'))
 158             width = int_or_none(media.get('dimensions', {}).get('width'))
 159             description = try_get(
 160                 media, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 161                 compat_str) or media.get('caption')
 162             thumbnail = media.get('display_src') or media.get('display_url')
 163             timestamp = int_or_none(media.get('taken_at_timestamp') or media.get('date'))
 164             uploader = media.get('owner', {}).get('full_name')
 165             uploader_id = media.get('owner', {}).get('username')
 166
 167             def get_count(keys, kind):
 168                 if not isinstance(keys, (list, tuple)):
 169                     keys = [keys]
 170                 for key in keys:
 171                     count = int_or_none(try_get(
 172                         media, (lambda x: x['edge_media_%s' % key]['count'],
 173                                 lambda x: x['%ss' % kind]['count'])))
 174                     if count is not None:
 175                         return count
 176             like_count = get_count('preview_like', 'like')
 177             comment_count = get_count(
 178                 ('preview_comment', 'to_comment', 'to_parent_comment'), 'comment')
 179
 180             comments = [{
 181                 'author': comment.get('user', {}).get('username'),
 182                 'author_id': comment.get('user', {}).get('id'),
 183                 'id': comment.get('id'),
 184                 'text': comment.get('text'),
 185                 'timestamp': int_or_none(comment.get('created_at')),
 186             } for comment in media.get(
 187                 'comments', {}).get('nodes', []) if comment.get('text')]
 188             if not video_url:
 189                 edges = try_get(
 190                     media, lambda x: x['edge_sidecar_to_children']['edges'],
 191                     list) or []
 192                 if edges:
 193                     entries = []
 194                     for edge_num, edge in enumerate(edges, start=1):
 195                         node = try_get(edge, lambda x: x['node'], dict)
 196                         if not node:
 197                             continue
 198                         node_video_url = url_or_none(node.get('video_url'))
 199                         if not node_video_url:
 200                             continue
 201                         entries.append({
 202                             'id': node.get('shortcode') or node['id'],
 203                             'title': 'Video %d' % edge_num,
 204                             'url': node_video_url,
 205                             'thumbnail': node.get('display_url'),
 206                             'width': int_or_none(try_get(node, lambda x: x['dimensions']['width'])),
 207                             'height': int_or_none(try_get(node, lambda x: x['dimensions']['height'])),
 208                             'view_count': int_or_none(node.get('video_view_count')),
 209                         })
 210                     return self.playlist_result(
 211                         entries, video_id,
 212                         'Post by %s' % uploader_id if uploader_id else None,
 213                         description)
 214
 215         if not video_url:
 216             video_url = self._og_search_video_url(webpage, secure=False)
 217
 218         formats = [{
 219             'url': video_url,
 220             'width': width,
 221             'height': height,
 222         }]
 223
 224         if not uploader_id:
 225             uploader_id = self._search_regex(
 226                 r'"owner"\s*:\s*{\s*"username"\s*:\s*"(.+?)"',
 227                 webpage, 'uploader id', fatal=False)
 228
 229         if not description:
 230             description = self._search_regex(
 231                 r'"caption"\s*:\s*"(.+?)"', webpage, 'description', default=None)
 232             if description is not None:
 233                 description = lowercase_escape(description)
 234
 235         if not thumbnail:
 236             thumbnail = self._og_search_thumbnail(webpage)
 237
 238         return {
 239             'id': video_id,
 240             'formats': formats,
 241             'ext': 'mp4',
 242             'title': 'Video by %s' % uploader_id,
 243             'description': description,
 244             'thumbnail': thumbnail,
 245             'timestamp': timestamp,
 246             'uploader_id': uploader_id,
 247             'uploader': uploader,
 248             'like_count': like_count,
 249             'comment_count': comment_count,
 250             'comments': comments,
 251         }
 252
 253
 254 class InstagramPlaylistIE(InfoExtractor):
 255     # A superclass for handling any kind of query based on GraphQL which
 256     # results in a playlist.
 257
 258     _gis_tmpl = None  # used to cache GIS request type
 259
 260     def _parse_graphql(self, webpage, item_id):
 261         # Reads a webpage and returns its GraphQL data.
 262         return self._parse_json(
 263             self._search_regex(
 264                 r'sharedData\s*=\s*({.+?})\s*;\s*[<\n]', webpage, 'data'),
 265             item_id)
 266
 267     def _extract_graphql(self, data, url):
 268         # Parses GraphQL queries containing videos and generates a playlist.
 269         def get_count(suffix):
 270             return int_or_none(try_get(
 271                 node, lambda x: x['edge_media_' + suffix]['count']))
 272
 273         uploader_id = self._match_id(url)
 274         csrf_token = data['config']['csrf_token']
 275         rhx_gis = data.get('rhx_gis') or '3c7ca9dcefcf966d11dacf1f151335e8'
 276
 277         cursor = ''
 278         for page_num in itertools.count(1):
 279             variables = {
 280                 'first': 12,
 281                 'after': cursor,
 282             }
 283             variables.update(self._query_vars_for(data))
 284             variables = json.dumps(variables)
 285
 286             if self._gis_tmpl:
 287                 gis_tmpls = [self._gis_tmpl]
 288             else:
 289                 gis_tmpls = [
 290                     '%s' % rhx_gis,
 291                     '',
 292                     '%s:%s' % (rhx_gis, csrf_token),
 293                     '%s:%s:%s' % (rhx_gis, csrf_token, std_headers['User-Agent']),
 294                 ]
 295
 296             # try all of the ways to generate a GIS query, and not only use the
 297             # first one that works, but cache it for future requests
 298             for gis_tmpl in gis_tmpls:
 299                 try:
 300                     json_data = self._download_json(
 301                         'https://www.instagram.com/graphql/query/', uploader_id,
 302                         'Downloading JSON page %d' % page_num, headers={
 303                             'X-Requested-With': 'XMLHttpRequest',
 304                             'X-Instagram-GIS': hashlib.md5(
 305                                 ('%s:%s' % (gis_tmpl, variables)).encode('utf-8')).hexdigest(),
 306                         }, query={
 307                             'query_hash': self._QUERY_HASH,
 308                             'variables': variables,
 309                         })
 310                     media = self._parse_timeline_from(json_data)
 311                     self._gis_tmpl = gis_tmpl
 312                     break
 313                 except ExtractorError as e:
 314                     # if it's an error caused by a bad query, and there are
 315                     # more GIS templates to try, ignore it and keep trying
 316                     if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 317                         if gis_tmpl != gis_tmpls[-1]:
 318                             continue
 319                     raise
 320
 321             edges = media.get('edges')
 322             if not edges or not isinstance(edges, list):
 323                 break
 324
 325             for edge in edges:
 326                 node = edge.get('node')
 327                 if not node or not isinstance(node, dict):
 328                     continue
 329                 if node.get('__typename') != 'GraphVideo' and node.get('is_video') is not True:
 330                     continue
 331                 video_id = node.get('shortcode')
 332                 if not video_id:
 333                     continue
 334
 335                 info = self.url_result(
 336                     'https://instagram.com/p/%s/' % video_id,
 337                     ie=InstagramIE.ie_key(), video_id=video_id)
 338
 339                 description = try_get(
 340                     node, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 341                     compat_str)
 342                 thumbnail = node.get('thumbnail_src') or node.get('display_src')
 343                 timestamp = int_or_none(node.get('taken_at_timestamp'))
 344
 345                 comment_count = get_count('to_comment')
 346                 like_count = get_count('preview_like')
 347                 view_count = int_or_none(node.get('video_view_count'))
 348
 349                 info.update({
 350                     'description': description,
 351                     'thumbnail': thumbnail,
 352                     'timestamp': timestamp,
 353                     'comment_count': comment_count,
 354                     'like_count': like_count,
 355                     'view_count': view_count,
 356                 })
 357
 358                 yield info
 359
 360             page_info = media.get('page_info')
 361             if not page_info or not isinstance(page_info, dict):
 362                 break
 363
 364             has_next_page = page_info.get('has_next_page')
 365             if not has_next_page:
 366                 break
 367
 368             cursor = page_info.get('end_cursor')
 369             if not cursor or not isinstance(cursor, compat_str):
 370                 break
 371
 372     def _real_extract(self, url):
 373         user_or_tag = self._match_id(url)
 374         webpage = self._download_webpage(url, user_or_tag)
 375         data = self._parse_graphql(webpage, user_or_tag)
 376
 377         self._set_cookie('instagram.com', 'ig_pr', '1')
 378
 379         return self.playlist_result(
 380             self._extract_graphql(data, url), user_or_tag, user_or_tag)
 381
 382
 383 class InstagramUserIE(InstagramPlaylistIE):
 384     _VALID_URL = r'https?://(?:www\.)?instagram\.com/(?P<id>[^/]{2,})/?(?:$|[?#])'
 385     IE_DESC = 'Instagram user profile'
 386     IE_NAME = 'instagram:user'
 387     _TEST = {
 388         'url': 'https://instagram.com/porsche',
 389         'info_dict': {
 390             'id': 'porsche',
 391             'title': 'porsche',
 392         },
 393         'playlist_count': 5,
 394         'params': {
 395             'extract_flat': True,
 396             'skip_download': True,
 397             'playlistend': 5,
 398         }
 399     }
 400
 401     _QUERY_HASH = '42323d64886122307be10013ad2dcc44',
 402
 403     @staticmethod
 404     def _parse_timeline_from(data):
 405         # extracts the media timeline data from a GraphQL result
 406         return data['data']['user']['edge_owner_to_timeline_media']
 407
 408     @staticmethod
 409     def _query_vars_for(data):
 410         # returns a dictionary of variables to add to the timeline query based
 411         # on the GraphQL of the original page
 412         return {
 413             'id': data['entry_data']['ProfilePage'][0]['graphql']['user']['id']
 414         }
 415
 416
 417 class InstagramTagIE(InstagramPlaylistIE):
 418     _VALID_URL = r'https?://(?:www\.)?instagram\.com/explore/tags/(?P<id>[^/]+)'
 419     IE_DESC = 'Instagram hashtag search'
 420     IE_NAME = 'instagram:tag'
 421     _TEST = {
 422         'url': 'https://instagram.com/explore/tags/lolcats',
 423         'info_dict': {
 424             'id': 'lolcats',
 425             'title': 'lolcats',
 426         },
 427         'playlist_count': 50,
 428         'params': {
 429             'extract_flat': True,
 430             'skip_download': True,
 431             'playlistend': 50,
 432         }
 433     }
 434
 435     _QUERY_HASH = 'f92f56d47dc7a55b606908374b43a314',
 436
 437     @staticmethod
 438     def _parse_timeline_from(data):
 439         # extracts the media timeline data from a GraphQL result
 440         return data['data']['hashtag']['edge_hashtag_to_media']
 441
 442     @staticmethod
 443     def _query_vars_for(data):
 444         # returns a dictionary of variables to add to the timeline query based
 445         # on the GraphQL of the original page
 446         return {
 447             'tag_name':
 448                 data['entry_data']['TagPage'][0]['graphql']['hashtag']['name']
 449         }