X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/b4a0c9f9de9d715538a1718922d6ab01a40f7ce3..e96056e11025aad125c364bd96b2a2df071dd2ae:/youtube_dl/extractor/instagram.py diff --git a/youtube_dl/extractor/instagram.py b/youtube_dl/extractor/instagram.py index a77f619..b061850 100644 --- a/youtube_dl/extractor/instagram.py +++ b/youtube_dl/extractor/instagram.py @@ -1,20 +1,28 @@ from __future__ import unicode_literals import itertools +import hashlib +import json import re from .common import InfoExtractor -from ..compat import compat_str +from ..compat import ( + compat_str, + compat_HTTPError, +) from ..utils import ( + ExtractorError, get_element_by_attribute, int_or_none, lowercase_escape, + std_headers, try_get, + url_or_none, ) class InstagramIE(InfoExtractor): - _VALID_URL = r'(?Phttps?://(?:www\.)?instagram\.com/p/(?P[^/?#&]+))' + _VALID_URL = r'(?Phttps?://(?:www\.)?instagram\.com/(?:p|tv)/(?P[^/?#&]+))' _TESTS = [{ 'url': 'https://instagram.com/p/aye83DjauH/?foo=bar#abc', 'md5': '0d2da106a9d2631273e192b372806516', @@ -84,6 +92,9 @@ class InstagramIE(InfoExtractor): }, { 'url': 'http://instagram.com/p/9o6LshA7zy/embed/', 'only_matching': True, + }, { + 'url': 'https://www.instagram.com/tv/aye83DjauH/', + 'only_matching': True, }] @staticmethod @@ -163,7 +174,7 @@ class InstagramIE(InfoExtractor): node = try_get(edge, lambda x: x['node'], dict) if not node: continue - node_video_url = try_get(node, lambda x: x['video_url'], compat_str) + node_video_url = url_or_none(node.get('video_url')) if not node_video_url: continue entries.append({ @@ -219,55 +230,84 @@ class InstagramIE(InfoExtractor): } -class InstagramUserIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?instagram\.com/(?P[^/]{2,})/?(?:$|[?#])' - IE_DESC = 'Instagram user profile' - IE_NAME = 'instagram:user' - _TEST = { - 'url': 'https://instagram.com/porsche', - 'info_dict': { - 'id': 'porsche', - 'title': 'porsche', - }, - 'playlist_count': 5, - 'params': { - 'extract_flat': True, - 'skip_download': True, - 'playlistend': 5, - } - } +class InstagramPlaylistIE(InfoExtractor): + # A superclass for handling any kind of query based on GraphQL which + # results in a playlist. - def _entries(self, uploader_id): - query = { - '__a': 1, - } + _gis_tmpl = None # used to cache GIS request type - def get_count(kind): + def _parse_graphql(self, webpage, item_id): + # Reads a webpage and returns its GraphQL data. + return self._parse_json( + self._search_regex( + r'sharedData\s*=\s*({.+?})\s*;\s*[<\n]', webpage, 'data'), + item_id) + + def _extract_graphql(self, data, url): + # Parses GraphQL queries containing videos and generates a playlist. + def get_count(suffix): return int_or_none(try_get( - node, lambda x: x['%ss' % kind]['count'])) + node, lambda x: x['edge_media_' + suffix]['count'])) - for page_num in itertools.count(1): - page = self._download_json( - 'https://instagram.com/%s/' % uploader_id, uploader_id, - note='Downloading page %d' % page_num, - fatal=False, query=query) - if not page: - break + uploader_id = self._match_id(url) + csrf_token = data['config']['csrf_token'] + rhx_gis = data.get('rhx_gis') or '3c7ca9dcefcf966d11dacf1f151335e8' - nodes = try_get(page, lambda x: x['user']['media']['nodes'], list) - if not nodes: + cursor = '' + for page_num in itertools.count(1): + variables = { + 'first': 12, + 'after': cursor, + } + variables.update(self._query_vars_for(data)) + variables = json.dumps(variables) + + if self._gis_tmpl: + gis_tmpls = [self._gis_tmpl] + else: + gis_tmpls = [ + '%s' % rhx_gis, + '', + '%s:%s' % (rhx_gis, csrf_token), + '%s:%s:%s' % (rhx_gis, csrf_token, std_headers['User-Agent']), + ] + + # try all of the ways to generate a GIS query, and not only use the + # first one that works, but cache it for future requests + for gis_tmpl in gis_tmpls: + try: + json_data = self._download_json( + 'https://www.instagram.com/graphql/query/', uploader_id, + 'Downloading JSON page %d' % page_num, headers={ + 'X-Requested-With': 'XMLHttpRequest', + 'X-Instagram-GIS': hashlib.md5( + ('%s:%s' % (gis_tmpl, variables)).encode('utf-8')).hexdigest(), + }, query={ + 'query_hash': self._QUERY_HASH, + 'variables': variables, + }) + media = self._parse_timeline_from(json_data) + self._gis_tmpl = gis_tmpl + break + except ExtractorError as e: + # if it's an error caused by a bad query, and there are + # more GIS templates to try, ignore it and keep trying + if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403: + if gis_tmpl != gis_tmpls[-1]: + continue + raise + + edges = media.get('edges') + if not edges or not isinstance(edges, list): break - max_id = None - - for node in nodes: - node_id = node.get('id') - if node_id: - max_id = node_id - + for edge in edges: + node = edge.get('node') + if not node or not isinstance(node, dict): + continue if node.get('__typename') != 'GraphVideo' and node.get('is_video') is not True: continue - video_id = node.get('code') + video_id = node.get('shortcode') if not video_id: continue @@ -276,14 +316,14 @@ class InstagramUserIE(InfoExtractor): ie=InstagramIE.ie_key(), video_id=video_id) description = try_get( - node, [lambda x: x['caption'], lambda x: x['text']['id']], + node, lambda x: x['edge_media_to_caption']['edges'][0]['node']['text'], compat_str) thumbnail = node.get('thumbnail_src') or node.get('display_src') - timestamp = int_or_none(node.get('date')) + timestamp = int_or_none(node.get('taken_at_timestamp')) - comment_count = get_count('comment') - like_count = get_count('like') - view_count = int_or_none(node.get('video_views')) + comment_count = get_count('to_comment') + like_count = get_count('preview_like') + view_count = int_or_none(node.get('video_view_count')) info.update({ 'description': description, @@ -296,12 +336,93 @@ class InstagramUserIE(InfoExtractor): yield info - if not max_id: + page_info = media.get('page_info') + if not page_info or not isinstance(page_info, dict): break - query['max_id'] = max_id + has_next_page = page_info.get('has_next_page') + if not has_next_page: + break + + cursor = page_info.get('end_cursor') + if not cursor or not isinstance(cursor, compat_str): + break def _real_extract(self, url): - uploader_id = self._match_id(url) + user_or_tag = self._match_id(url) + webpage = self._download_webpage(url, user_or_tag) + data = self._parse_graphql(webpage, user_or_tag) + + self._set_cookie('instagram.com', 'ig_pr', '1') + return self.playlist_result( - self._entries(uploader_id), uploader_id, uploader_id) + self._extract_graphql(data, url), user_or_tag, user_or_tag) + + +class InstagramUserIE(InstagramPlaylistIE): + _VALID_URL = r'https?://(?:www\.)?instagram\.com/(?P[^/]{2,})/?(?:$|[?#])' + IE_DESC = 'Instagram user profile' + IE_NAME = 'instagram:user' + _TEST = { + 'url': 'https://instagram.com/porsche', + 'info_dict': { + 'id': 'porsche', + 'title': 'porsche', + }, + 'playlist_count': 5, + 'params': { + 'extract_flat': True, + 'skip_download': True, + 'playlistend': 5, + } + } + + _QUERY_HASH = '42323d64886122307be10013ad2dcc44', + + @staticmethod + def _parse_timeline_from(data): + # extracts the media timeline data from a GraphQL result + return data['data']['user']['edge_owner_to_timeline_media'] + + @staticmethod + def _query_vars_for(data): + # returns a dictionary of variables to add to the timeline query based + # on the GraphQL of the original page + return { + 'id': data['entry_data']['ProfilePage'][0]['graphql']['user']['id'] + } + + +class InstagramTagIE(InstagramPlaylistIE): + _VALID_URL = r'https?://(?:www\.)?instagram\.com/explore/tags/(?P[^/]+)' + IE_DESC = 'Instagram hashtag search' + IE_NAME = 'instagram:tag' + _TEST = { + 'url': 'https://instagram.com/explore/tags/lolcats', + 'info_dict': { + 'id': 'lolcats', + 'title': 'lolcats', + }, + 'playlist_count': 50, + 'params': { + 'extract_flat': True, + 'skip_download': True, + 'playlistend': 50, + } + } + + _QUERY_HASH = 'f92f56d47dc7a55b606908374b43a314', + + @staticmethod + def _parse_timeline_from(data): + # extracts the media timeline data from a GraphQL result + return data['data']['hashtag']['edge_hashtag_to_media'] + + @staticmethod + def _query_vars_for(data): + # returns a dictionary of variables to add to the timeline query based + # on the GraphQL of the original page + return { + 'tag_name': + data['entry_data']['TagPage'][0]['graphql']['hashtag']['name'] + }