youtube_dlc/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         (video_url, description, thumbnail, timestamp, uploader,
 126          uploader_id, like_count, comment_count, comments, height,
 127          width) = [None] * 11
 128
 129         shared_data = try_get(webpage,
 130                               (lambda x: self._parse_json(
 131                                   self._search_regex(
 132                                       r'window\.__additionalDataLoaded\(\'/(?:p|tv)/(?:[^/?#&]+)/\',({.+?})\);',
 133                                       x, 'additional data', default='{}'),
 134                                   video_id, fatal=False),
 135                                lambda x: self._parse_json(
 136                                   self._search_regex(
 137                                       r'window\._sharedData\s*=\s*({.+?});',
 138                                       x, 'shared data', default='{}'),
 139                                   video_id, fatal=False)['entry_data']['PostPage'][0]),
 140                               None)
 141         if shared_data:
 142             media = try_get(
 143                 shared_data,
 144                 (lambda x: x['graphql']['shortcode_media'],
 145                  lambda x: x['media']),
 146                 dict)
 147             if media:
 148                 video_url = media.get('video_url')
 149                 height = int_or_none(media.get('dimensions', {}).get('height'))
 150                 width = int_or_none(media.get('dimensions', {}).get('width'))
 151                 description = try_get(
 152                     media, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 153                     compat_str) or media.get('caption')
 154                 thumbnail = media.get('display_src') or media.get('thumbnail_src')
 155                 timestamp = int_or_none(media.get('taken_at_timestamp') or media.get('date'))
 156                 uploader = media.get('owner', {}).get('full_name')
 157                 uploader_id = media.get('owner', {}).get('username')
 158
 159                 def get_count(key, kind):
 160                     return int_or_none(try_get(
 161                         media, (lambda x: x['edge_media_%s' % key]['count'],
 162                                 lambda x: x['%ss' % kind]['count'])))
 163                 like_count = get_count('preview_like', 'like')
 164                 comment_count = get_count('to_comment', 'comment')
 165
 166                 comments = [{
 167                     'author': comment.get('user', {}).get('username'),
 168                     'author_id': comment.get('user', {}).get('id'),
 169                     'id': comment.get('id'),
 170                     'text': comment.get('text'),
 171                     'timestamp': int_or_none(comment.get('created_at')),
 172                 } for comment in media.get(
 173                     'comments', {}).get('nodes', []) if comment.get('text')]
 174                 if not video_url:
 175                     edges = try_get(
 176                         media, lambda x: x['edge_sidecar_to_children']['edges'],
 177                         list) or []
 178                     if edges:
 179                         entries = []
 180                         for edge_num, edge in enumerate(edges, start=1):
 181                             node = try_get(edge, lambda x: x['node'], dict)
 182                             if not node:
 183                                 continue
 184                             node_video_url = url_or_none(node.get('video_url'))
 185                             if not node_video_url:
 186                                 continue
 187                             entries.append({
 188                                 'id': node.get('shortcode') or node['id'],
 189                                 'title': 'Video %d' % edge_num,
 190                                 'url': node_video_url,
 191                                 'thumbnail': node.get('display_url'),
 192                                 'width': int_or_none(try_get(node, lambda x: x['dimensions']['width'])),
 193                                 'height': int_or_none(try_get(node, lambda x: x['dimensions']['height'])),
 194                                 'view_count': int_or_none(node.get('video_view_count')),
 195                             })
 196                         return self.playlist_result(
 197                             entries, video_id,
 198                             'Post by %s' % uploader_id if uploader_id else None,
 199                             description)
 200
 201         if not video_url:
 202             video_url = self._og_search_video_url(webpage, secure=False)
 203
 204         formats = [{
 205             'url': video_url,
 206             'width': width,
 207             'height': height,
 208         }]
 209
 210         if not uploader_id:
 211             uploader_id = self._search_regex(
 212                 r'"owner"\s*:\s*{\s*"username"\s*:\s*"(.+?)"',
 213                 webpage, 'uploader id', fatal=False)
 214
 215         if not description:
 216             description = self._search_regex(
 217                 r'"caption"\s*:\s*"(.+?)"', webpage, 'description', default=None)
 218             if description is not None:
 219                 description = lowercase_escape(description)
 220
 221         if not thumbnail:
 222             thumbnail = self._og_search_thumbnail(webpage)
 223
 224         return {
 225             'id': video_id,
 226             'formats': formats,
 227             'ext': 'mp4',
 228             'title': 'Video by %s' % uploader_id,
 229             'description': description,
 230             'thumbnail': thumbnail,
 231             'timestamp': timestamp,
 232             'uploader_id': uploader_id,
 233             'uploader': uploader,
 234             'like_count': like_count,
 235             'comment_count': comment_count,
 236             'comments': comments,
 237         }
 238
 239
 240 class InstagramPlaylistIE(InfoExtractor):
 241     # A superclass for handling any kind of query based on GraphQL which
 242     # results in a playlist.
 243
 244     _gis_tmpl = None  # used to cache GIS request type
 245
 246     def _parse_graphql(self, webpage, item_id):
 247         # Reads a webpage and returns its GraphQL data.
 248         return self._parse_json(
 249             self._search_regex(
 250                 r'sharedData\s*=\s*({.+?})\s*;\s*[<\n]', webpage, 'data'),
 251             item_id)
 252
 253     def _extract_graphql(self, data, url):
 254         # Parses GraphQL queries containing videos and generates a playlist.
 255         def get_count(suffix):
 256             return int_or_none(try_get(
 257                 node, lambda x: x['edge_media_' + suffix]['count']))
 258
 259         uploader_id = self._match_id(url)
 260         csrf_token = data['config']['csrf_token']
 261         rhx_gis = data.get('rhx_gis') or '3c7ca9dcefcf966d11dacf1f151335e8'
 262
 263         cursor = ''
 264         for page_num in itertools.count(1):
 265             variables = {
 266                 'first': 12,
 267                 'after': cursor,
 268             }
 269             variables.update(self._query_vars_for(data))
 270             variables = json.dumps(variables)
 271
 272             if self._gis_tmpl:
 273                 gis_tmpls = [self._gis_tmpl]
 274             else:
 275                 gis_tmpls = [
 276                     '%s' % rhx_gis,
 277                     '',
 278                     '%s:%s' % (rhx_gis, csrf_token),
 279                     '%s:%s:%s' % (rhx_gis, csrf_token, std_headers['User-Agent']),
 280                 ]
 281
 282             # try all of the ways to generate a GIS query, and not only use the
 283             # first one that works, but cache it for future requests
 284             for gis_tmpl in gis_tmpls:
 285                 try:
 286                     json_data = self._download_json(
 287                         'https://www.instagram.com/graphql/query/', uploader_id,
 288                         'Downloading JSON page %d' % page_num, headers={
 289                             'X-Requested-With': 'XMLHttpRequest',
 290                             'X-Instagram-GIS': hashlib.md5(
 291                                 ('%s:%s' % (gis_tmpl, variables)).encode('utf-8')).hexdigest(),
 292                         }, query={
 293                             'query_hash': self._QUERY_HASH,
 294                             'variables': variables,
 295                         })
 296                     media = self._parse_timeline_from(json_data)
 297                     self._gis_tmpl = gis_tmpl
 298                     break
 299                 except ExtractorError as e:
 300                     # if it's an error caused by a bad query, and there are
 301                     # more GIS templates to try, ignore it and keep trying
 302                     if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 303                         if gis_tmpl != gis_tmpls[-1]:
 304                             continue
 305                     raise
 306
 307             edges = media.get('edges')
 308             if not edges or not isinstance(edges, list):
 309                 break
 310
 311             for edge in edges:
 312                 node = edge.get('node')
 313                 if not node or not isinstance(node, dict):
 314                     continue
 315                 if node.get('__typename') != 'GraphVideo' and node.get('is_video') is not True:
 316                     continue
 317                 video_id = node.get('shortcode')
 318                 if not video_id:
 319                     continue
 320
 321                 info = self.url_result(
 322                     'https://instagram.com/p/%s/' % video_id,
 323                     ie=InstagramIE.ie_key(), video_id=video_id)
 324
 325                 description = try_get(
 326                     node, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 327                     compat_str)
 328                 thumbnail = node.get('thumbnail_src') or node.get('display_src')
 329                 timestamp = int_or_none(node.get('taken_at_timestamp'))
 330
 331                 comment_count = get_count('to_comment')
 332                 like_count = get_count('preview_like')
 333                 view_count = int_or_none(node.get('video_view_count'))
 334
 335                 info.update({
 336                     'description': description,
 337                     'thumbnail': thumbnail,
 338                     'timestamp': timestamp,
 339                     'comment_count': comment_count,
 340                     'like_count': like_count,
 341                     'view_count': view_count,
 342                 })
 343
 344                 yield info
 345
 346             page_info = media.get('page_info')
 347             if not page_info or not isinstance(page_info, dict):
 348                 break
 349
 350             has_next_page = page_info.get('has_next_page')
 351             if not has_next_page:
 352                 break
 353
 354             cursor = page_info.get('end_cursor')
 355             if not cursor or not isinstance(cursor, compat_str):
 356                 break
 357
 358     def _real_extract(self, url):
 359         user_or_tag = self._match_id(url)
 360         webpage = self._download_webpage(url, user_or_tag)
 361         data = self._parse_graphql(webpage, user_or_tag)
 362
 363         self._set_cookie('instagram.com', 'ig_pr', '1')
 364
 365         return self.playlist_result(
 366             self._extract_graphql(data, url), user_or_tag, user_or_tag)
 367
 368
 369 class InstagramUserIE(InstagramPlaylistIE):
 370     _VALID_URL = r'https?://(?:www\.)?instagram\.com/(?P<id>[^/]{2,})/?(?:$|[?#])'
 371     IE_DESC = 'Instagram user profile'
 372     IE_NAME = 'instagram:user'
 373     _TEST = {
 374         'url': 'https://instagram.com/porsche',
 375         'info_dict': {
 376             'id': 'porsche',
 377             'title': 'porsche',
 378         },
 379         'playlist_count': 5,
 380         'params': {
 381             'extract_flat': True,
 382             'skip_download': True,
 383             'playlistend': 5,
 384         }
 385     }
 386
 387     _QUERY_HASH = '42323d64886122307be10013ad2dcc44',
 388
 389     @staticmethod
 390     def _parse_timeline_from(data):
 391         # extracts the media timeline data from a GraphQL result
 392         return data['data']['user']['edge_owner_to_timeline_media']
 393
 394     @staticmethod
 395     def _query_vars_for(data):
 396         # returns a dictionary of variables to add to the timeline query based
 397         # on the GraphQL of the original page
 398         return {
 399             'id': data['entry_data']['ProfilePage'][0]['graphql']['user']['id']
 400         }
 401
 402
 403 class InstagramTagIE(InstagramPlaylistIE):
 404     _VALID_URL = r'https?://(?:www\.)?instagram\.com/explore/tags/(?P<id>[^/]+)'
 405     IE_DESC = 'Instagram hashtag search'
 406     IE_NAME = 'instagram:tag'
 407     _TEST = {
 408         'url': 'https://instagram.com/explore/tags/lolcats',
 409         'info_dict': {
 410             'id': 'lolcats',
 411             'title': 'lolcats',
 412         },
 413         'playlist_count': 50,
 414         'params': {
 415             'extract_flat': True,
 416             'skip_download': True,
 417             'playlistend': 50,
 418         }
 419     }
 420
 421     _QUERY_HASH = 'f92f56d47dc7a55b606908374b43a314',
 422
 423     @staticmethod
 424     def _parse_timeline_from(data):
 425         # extracts the media timeline data from a GraphQL result
 426         return data['data']['hashtag']['edge_hashtag_to_media']
 427
 428     @staticmethod
 429     def _query_vars_for(data):
 430         # returns a dictionary of variables to add to the timeline query based
 431         # on the GraphQL of the original page
 432         return {
 433             'tag_name':
 434                 data['entry_data']['TagPage'][0]['graphql']['hashtag']['name']
 435         }