Raphaël G. Git Repositories - youtubedl/blob - 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/(?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
  97     @staticmethod
  98     def _extract_embed_url(webpage):
  99         mobj = re.search(
 100             r'<iframe[^>]+src=(["\'])(?P<url>(?:https?:)?//(?:www\.)?instagram\.com/p/[^/]+/embed.*?)\1',
 101             webpage)
 102         if mobj:
 103             return mobj.group('url')
 104
 105         blockquote_el = get_element_by_attribute(
 106             'class', 'instagram-media', webpage)
 107         if blockquote_el is None:
 108             return
 109
 110         mobj = re.search(
 111             r'<a[^>]+href=([\'"])(?P<link>[^\'"]+)\1', blockquote_el)
 112         if mobj:
 113             return mobj.group('link')
 114
 115     def _real_extract(self, url):
 116         mobj = re.match(self._VALID_URL, url)
 117         video_id = mobj.group('id')
 118         url = mobj.group('url')
 119
 120         webpage = self._download_webpage(url, video_id)
 121
 122         (video_url, description, thumbnail, timestamp, uploader,
 123          uploader_id, like_count, comment_count, comments, height,
 124          width) = [None] * 11
 125
 126         shared_data = self._parse_json(
 127             self._search_regex(
 128                 r'window\._sharedData\s*=\s*({.+?});',
 129                 webpage, 'shared data', default='{}'),
 130             video_id, fatal=False)
 131         if shared_data:
 132             media = try_get(
 133                 shared_data,
 134                 (lambda x: x['entry_data']['PostPage'][0]['graphql']['shortcode_media'],
 135                  lambda x: x['entry_data']['PostPage'][0]['media']),
 136                 dict)
 137             if media:
 138                 video_url = media.get('video_url')
 139                 height = int_or_none(media.get('dimensions', {}).get('height'))
 140                 width = int_or_none(media.get('dimensions', {}).get('width'))
 141                 description = try_get(
 142                     media, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 143                     compat_str) or media.get('caption')
 144                 thumbnail = media.get('display_src')
 145                 timestamp = int_or_none(media.get('taken_at_timestamp') or media.get('date'))
 146                 uploader = media.get('owner', {}).get('full_name')
 147                 uploader_id = media.get('owner', {}).get('username')
 148
 149                 def get_count(key, kind):
 150                     return int_or_none(try_get(
 151                         media, (lambda x: x['edge_media_%s' % key]['count'],
 152                                 lambda x: x['%ss' % kind]['count'])))
 153                 like_count = get_count('preview_like', 'like')
 154                 comment_count = get_count('to_comment', 'comment')
 155
 156                 comments = [{
 157                     'author': comment.get('user', {}).get('username'),
 158                     'author_id': comment.get('user', {}).get('id'),
 159                     'id': comment.get('id'),
 160                     'text': comment.get('text'),
 161                     'timestamp': int_or_none(comment.get('created_at')),
 162                 } for comment in media.get(
 163                     'comments', {}).get('nodes', []) if comment.get('text')]
 164                 if not video_url:
 165                     edges = try_get(
 166                         media, lambda x: x['edge_sidecar_to_children']['edges'],
 167                         list) or []
 168                     if edges:
 169                         entries = []
 170                         for edge_num, edge in enumerate(edges, start=1):
 171                             node = try_get(edge, lambda x: x['node'], dict)
 172                             if not node:
 173                                 continue
 174                             node_video_url = url_or_none(node.get('video_url'))
 175                             if not node_video_url:
 176                                 continue
 177                             entries.append({
 178                                 'id': node.get('shortcode') or node['id'],
 179                                 'title': 'Video %d' % edge_num,
 180                                 'url': node_video_url,
 181                                 'thumbnail': node.get('display_url'),
 182                                 'width': int_or_none(try_get(node, lambda x: x['dimensions']['width'])),
 183                                 'height': int_or_none(try_get(node, lambda x: x['dimensions']['height'])),
 184                                 'view_count': int_or_none(node.get('video_view_count')),
 185                             })
 186                         return self.playlist_result(
 187                             entries, video_id,
 188                             'Post by %s' % uploader_id if uploader_id else None,
 189                             description)
 190
 191         if not video_url:
 192             video_url = self._og_search_video_url(webpage, secure=False)
 193
 194         formats = [{
 195             'url': video_url,
 196             'width': width,
 197             'height': height,
 198         }]
 199
 200         if not uploader_id:
 201             uploader_id = self._search_regex(
 202                 r'"owner"\s*:\s*{\s*"username"\s*:\s*"(.+?)"',
 203                 webpage, 'uploader id', fatal=False)
 204
 205         if not description:
 206             description = self._search_regex(
 207                 r'"caption"\s*:\s*"(.+?)"', webpage, 'description', default=None)
 208             if description is not None:
 209                 description = lowercase_escape(description)
 210
 211         if not thumbnail:
 212             thumbnail = self._og_search_thumbnail(webpage)
 213
 214         return {
 215             'id': video_id,
 216             'formats': formats,
 217             'ext': 'mp4',
 218             'title': 'Video by %s' % uploader_id,
 219             'description': description,
 220             'thumbnail': thumbnail,
 221             'timestamp': timestamp,
 222             'uploader_id': uploader_id,
 223             'uploader': uploader,
 224             'like_count': like_count,
 225             'comment_count': comment_count,
 226             'comments': comments,
 227         }
 228
 229
 230 class InstagramPlaylistIE(InfoExtractor):
 231     # A superclass for handling any kind of query based on GraphQL which
 232     # results in a playlist.
 233
 234     _gis_tmpl = None  # used to cache GIS request type
 235
 236     def _parse_graphql(self, webpage, item_id):
 237         # Reads a webpage and returns its GraphQL data.
 238         return self._parse_json(
 239             self._search_regex(
 240                 r'sharedData\s*=\s*({.+?})\s*;\s*[<\n]', webpage, 'data'),
 241             item_id)
 242
 243     def _extract_graphql(self, data, url):
 244         # Parses GraphQL queries containing videos and generates a playlist.
 245         def get_count(suffix):
 246             return int_or_none(try_get(
 247                 node, lambda x: x['edge_media_' + suffix]['count']))
 248
 249         uploader_id = self._match_id(url)
 250         csrf_token = data['config']['csrf_token']
 251         rhx_gis = data.get('rhx_gis') or '3c7ca9dcefcf966d11dacf1f151335e8'
 252
 253         cursor = ''
 254         for page_num in itertools.count(1):
 255             variables = {
 256                 'first': 12,
 257                 'after': cursor,
 258             }
 259             variables.update(self._query_vars_for(data))
 260             variables = json.dumps(variables)
 261
 262             if self._gis_tmpl:
 263                 gis_tmpls = [self._gis_tmpl]
 264             else:
 265                 gis_tmpls = [
 266                     '%s' % rhx_gis,
 267                     '',
 268                     '%s:%s' % (rhx_gis, csrf_token),
 269                     '%s:%s:%s' % (rhx_gis, csrf_token, std_headers['User-Agent']),
 270                 ]
 271
 272             # try all of the ways to generate a GIS query, and not only use the
 273             # first one that works, but cache it for future requests
 274             for gis_tmpl in gis_tmpls:
 275                 try:
 276                     json_data = self._download_json(
 277                         'https://www.instagram.com/graphql/query/', uploader_id,
 278                         'Downloading JSON page %d' % page_num, headers={
 279                             'X-Requested-With': 'XMLHttpRequest',
 280                             'X-Instagram-GIS': hashlib.md5(
 281                                 ('%s:%s' % (gis_tmpl, variables)).encode('utf-8')).hexdigest(),
 282                         }, query={
 283                             'query_hash': self._QUERY_HASH,
 284                             'variables': variables,
 285                         })
 286                     media = self._parse_timeline_from(json_data)
 287                     self._gis_tmpl = gis_tmpl
 288                     break
 289                 except ExtractorError as e:
 290                     # if it's an error caused by a bad query, and there are
 291                     # more GIS templates to try, ignore it and keep trying
 292                     if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
 293                         if gis_tmpl != gis_tmpls[-1]:
 294                             continue
 295                     raise
 296
 297             edges = media.get('edges')
 298             if not edges or not isinstance(edges, list):
 299                 break
 300
 301             for edge in edges:
 302                 node = edge.get('node')
 303                 if not node or not isinstance(node, dict):
 304                     continue
 305                 if node.get('__typename') != 'GraphVideo' and node.get('is_video') is not True:
 306                     continue
 307                 video_id = node.get('shortcode')
 308                 if not video_id:
 309                     continue
 310
 311                 info = self.url_result(
 312                     'https://instagram.com/p/%s/' % video_id,
 313                     ie=InstagramIE.ie_key(), video_id=video_id)
 314
 315                 description = try_get(
 316                     node, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'],
 317                     compat_str)
 318                 thumbnail = node.get('thumbnail_src') or node.get('display_src')
 319                 timestamp = int_or_none(node.get('taken_at_timestamp'))
 320
 321                 comment_count = get_count('to_comment')
 322                 like_count = get_count('preview_like')
 323                 view_count = int_or_none(node.get('video_view_count'))
 324
 325                 info.update({
 326                     'description': description,
 327                     'thumbnail': thumbnail,
 328                     'timestamp': timestamp,
 329                     'comment_count': comment_count,
 330                     'like_count': like_count,
 331                     'view_count': view_count,
 332                 })
 333
 334                 yield info
 335
 336             page_info = media.get('page_info')
 337             if not page_info or not isinstance(page_info, dict):
 338                 break
 339
 340             has_next_page = page_info.get('has_next_page')
 341             if not has_next_page:
 342                 break
 343
 344             cursor = page_info.get('end_cursor')
 345             if not cursor or not isinstance(cursor, compat_str):
 346                 break
 347
 348     def _real_extract(self, url):
 349         user_or_tag = self._match_id(url)
 350         webpage = self._download_webpage(url, user_or_tag)
 351         data = self._parse_graphql(webpage, user_or_tag)
 352
 353         self._set_cookie('instagram.com', 'ig_pr', '1')
 354
 355         return self.playlist_result(
 356             self._extract_graphql(data, url), user_or_tag, user_or_tag)
 357
 358
 359 class InstagramUserIE(InstagramPlaylistIE):
 360     _VALID_URL = r'https?://(?:www\.)?instagram\.com/(?P<id>[^/]{2,})/?(?:$|[?#])'
 361     IE_DESC = 'Instagram user profile'
 362     IE_NAME = 'instagram:user'
 363     _TEST = {
 364         'url': 'https://instagram.com/porsche',
 365         'info_dict': {
 366             'id': 'porsche',
 367             'title': 'porsche',
 368         },
 369         'playlist_count': 5,
 370         'params': {
 371             'extract_flat': True,
 372             'skip_download': True,
 373             'playlistend': 5,
 374         }
 375     }
 376
 377     _QUERY_HASH = '42323d64886122307be10013ad2dcc44',
 378
 379     @staticmethod
 380     def _parse_timeline_from(data):
 381         # extracts the media timeline data from a GraphQL result
 382         return data['data']['user']['edge_owner_to_timeline_media']
 383
 384     @staticmethod
 385     def _query_vars_for(data):
 386         # returns a dictionary of variables to add to the timeline query based
 387         # on the GraphQL of the original page
 388         return {
 389             'id': data['entry_data']['ProfilePage'][0]['graphql']['user']['id']
 390         }
 391
 392
 393 class InstagramTagIE(InstagramPlaylistIE):
 394     _VALID_URL = r'https?://(?:www\.)?instagram\.com/explore/tags/(?P<id>[^/]+)'
 395     IE_DESC = 'Instagram hashtag search'
 396     IE_NAME = 'instagram:tag'
 397     _TEST = {
 398         'url': 'https://instagram.com/explore/tags/lolcats',
 399         'info_dict': {
 400             'id': 'lolcats',
 401             'title': 'lolcats',
 402         },
 403         'playlist_count': 50,
 404         'params': {
 405             'extract_flat': True,
 406             'skip_download': True,
 407             'playlistend': 50,
 408         }
 409     }
 410
 411     _QUERY_HASH = 'f92f56d47dc7a55b606908374b43a314',
 412
 413     @staticmethod
 414     def _parse_timeline_from(data):
 415         # extracts the media timeline data from a GraphQL result
 416         return data['data']['hashtag']['edge_hashtag_to_media']
 417
 418     @staticmethod
 419     def _query_vars_for(data):
 420         # returns a dictionary of variables to add to the timeline query based
 421         # on the GraphQL of the original page
 422         return {
 423             'tag_name':
 424                 data['entry_data']['TagPage'][0]['graphql']['hashtag']['name']
 425         }