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