]>
Commit | Line | Data |
---|---|---|
a78e3a57 | 1 | from __future__ import division, unicode_literals |
2 | ||
a78e3a57 | 3 | import json |
c60ee3a2 | 4 | import time |
a78e3a57 | 5 | |
6 | from .fragment import FragmentFD | |
82e3f6eb | 7 | from ..compat import compat_urllib_error |
273762c8 | 8 | from ..utils import ( |
9 | try_get, | |
c60ee3a2 | 10 | dict_get, |
11 | int_or_none, | |
273762c8 | 12 | RegexNotFoundError, |
13 | ) | |
82e3f6eb | 14 | from ..extractor.youtube import YoutubeBaseInfoExtractor as YT_BaseIE |
a78e3a57 | 15 | |
16 | ||
c60ee3a2 | 17 | class YoutubeLiveChatFD(FragmentFD): |
18 | """ Downloads YouTube live chats fragment by fragment """ | |
a78e3a57 | 19 | |
c60ee3a2 | 20 | FD_NAME = 'youtube_live_chat' |
a78e3a57 | 21 | |
22 | def real_download(self, filename, info_dict): | |
23 | video_id = info_dict['video_id'] | |
24 | self.to_screen('[%s] Downloading live chat' % self.FD_NAME) | |
25 | ||
82e3f6eb | 26 | fragment_retries = self.params.get('fragment_retries', 0) |
a78e3a57 | 27 | test = self.params.get('test', False) |
28 | ||
29 | ctx = { | |
30 | 'filename': filename, | |
31 | 'live': True, | |
32 | 'total_frags': None, | |
33 | } | |
34 | ||
273762c8 | 35 | ie = YT_BaseIE(self.ydl) |
a78e3a57 | 36 | |
c60ee3a2 | 37 | start_time = int(time.time() * 1000) |
38 | ||
273762c8 | 39 | def dl_fragment(url, data=None, headers=None): |
40 | http_headers = info_dict.get('http_headers', {}) | |
41 | if headers: | |
42 | http_headers = http_headers.copy() | |
43 | http_headers.update(headers) | |
44 | return self._download_fragment(ctx, url, info_dict, http_headers, data) | |
a78e3a57 | 45 | |
c60ee3a2 | 46 | def parse_actions_replay(live_chat_continuation): |
47 | offset = continuation_id = None | |
48 | processed_fragment = bytearray() | |
49 | for action in live_chat_continuation.get('actions', []): | |
50 | if 'replayChatItemAction' in action: | |
51 | replay_chat_item_action = action['replayChatItemAction'] | |
52 | offset = int(replay_chat_item_action['videoOffsetTimeMsec']) | |
53 | processed_fragment.extend( | |
54 | json.dumps(action, ensure_ascii=False).encode('utf-8') + b'\n') | |
55 | if offset is not None: | |
56 | continuation_id = try_get( | |
57 | live_chat_continuation, | |
58 | lambda x: x['continuations'][0]['liveChatReplayContinuationData']['continuation']) | |
59 | self._append_fragment(ctx, processed_fragment) | |
60 | return continuation_id, offset | |
61 | ||
d534c452 | 62 | def try_refresh_replay_beginning(live_chat_continuation): |
63 | # choose the second option that contains the unfiltered live chat replay | |
64 | refresh_continuation_id = try_get( | |
65 | live_chat_continuation, | |
66 | lambda x: x['header']['liveChatHeaderRenderer']['viewSelector']['sortFilterSubMenuRenderer']['subMenuItems'][1]['continuation']['reloadContinuationData']['continuation'], str) | |
67 | if refresh_continuation_id: | |
68 | # no data yet but required to call _append_fragment | |
69 | self._append_fragment(ctx, b'') | |
70 | return refresh_continuation_id, 0 | |
71 | return parse_actions_replay(live_chat_continuation) | |
72 | ||
c60ee3a2 | 73 | live_offset = 0 |
74 | ||
75 | def parse_actions_live(live_chat_continuation): | |
76 | nonlocal live_offset | |
77 | continuation_id = None | |
78 | processed_fragment = bytearray() | |
79 | for action in live_chat_continuation.get('actions', []): | |
80 | timestamp = self.parse_live_timestamp(action) | |
81 | if timestamp is not None: | |
82 | live_offset = timestamp - start_time | |
83 | # compatibility with replay format | |
84 | pseudo_action = { | |
85 | 'replayChatItemAction': {'actions': [action]}, | |
86 | 'videoOffsetTimeMsec': str(live_offset), | |
87 | 'isLive': True, | |
88 | } | |
89 | processed_fragment.extend( | |
90 | json.dumps(pseudo_action, ensure_ascii=False).encode('utf-8') + b'\n') | |
91 | continuation_data_getters = [ | |
92 | lambda x: x['continuations'][0]['invalidationContinuationData'], | |
93 | lambda x: x['continuations'][0]['timedContinuationData'], | |
94 | ] | |
95 | continuation_data = try_get(live_chat_continuation, continuation_data_getters, dict) | |
96 | if continuation_data: | |
97 | continuation_id = continuation_data.get('continuation') | |
98 | timeout_ms = int_or_none(continuation_data.get('timeoutMs')) | |
99 | if timeout_ms is not None: | |
100 | time.sleep(timeout_ms / 1000) | |
101 | self._append_fragment(ctx, processed_fragment) | |
102 | return continuation_id, live_offset | |
103 | ||
d534c452 | 104 | def download_and_parse_fragment(url, frag_index, request_data=None, headers=None): |
82e3f6eb | 105 | count = 0 |
106 | while count <= fragment_retries: | |
107 | try: | |
c60ee3a2 | 108 | success, raw_fragment = dl_fragment(url, request_data, headers) |
82e3f6eb | 109 | if not success: |
110 | return False, None, None | |
d534c452 | 111 | try: |
112 | data = ie._extract_yt_initial_data(video_id, raw_fragment.decode('utf-8', 'replace')) | |
113 | except RegexNotFoundError: | |
114 | data = None | |
115 | if not data: | |
116 | data = json.loads(raw_fragment) | |
82e3f6eb | 117 | live_chat_continuation = try_get( |
118 | data, | |
119 | lambda x: x['continuationContents']['liveChatContinuation'], dict) or {} | |
d534c452 | 120 | if info_dict['protocol'] == 'youtube_live_chat_replay': |
121 | if frag_index == 1: | |
122 | continuation_id, offset = try_refresh_replay_beginning(live_chat_continuation) | |
123 | else: | |
124 | continuation_id, offset = parse_actions_replay(live_chat_continuation) | |
125 | elif info_dict['protocol'] == 'youtube_live_chat': | |
126 | continuation_id, offset = parse_actions_live(live_chat_continuation) | |
82e3f6eb | 127 | return True, continuation_id, offset |
128 | except compat_urllib_error.HTTPError as err: | |
129 | count += 1 | |
130 | if count <= fragment_retries: | |
131 | self.report_retry_fragment(err, frag_index, count, fragment_retries) | |
132 | if count > fragment_retries: | |
133 | self.report_error('giving up after %s fragment retries' % fragment_retries) | |
134 | return False, None, None | |
135 | ||
a78e3a57 | 136 | self._prepare_and_start_frag_download(ctx) |
137 | ||
83b20a97 | 138 | success, raw_fragment = dl_fragment(info_dict['url']) |
a78e3a57 | 139 | if not success: |
140 | return False | |
273762c8 | 141 | try: |
142 | data = ie._extract_yt_initial_data(video_id, raw_fragment.decode('utf-8', 'replace')) | |
143 | except RegexNotFoundError: | |
144 | return False | |
82e3f6eb | 145 | continuation_id = try_get( |
146 | data, | |
147 | lambda x: x['contents']['twoColumnWatchNextResults']['conversationBar']['liveChatRenderer']['continuations'][0]['reloadContinuationData']['continuation']) | |
a78e3a57 | 148 | # no data yet but required to call _append_fragment |
149 | self._append_fragment(ctx, b'') | |
150 | ||
273762c8 | 151 | ytcfg = ie._extract_ytcfg(video_id, raw_fragment.decode('utf-8', 'replace')) |
152 | ||
153 | if not ytcfg: | |
154 | return False | |
155 | api_key = try_get(ytcfg, lambda x: x['INNERTUBE_API_KEY']) | |
156 | innertube_context = try_get(ytcfg, lambda x: x['INNERTUBE_CONTEXT']) | |
157 | if not api_key or not innertube_context: | |
158 | return False | |
c60ee3a2 | 159 | visitor_data = try_get(innertube_context, lambda x: x['client']['visitorData'], str) |
160 | if info_dict['protocol'] == 'youtube_live_chat_replay': | |
161 | url = 'https://www.youtube.com/youtubei/v1/live_chat/get_live_chat_replay?key=' + api_key | |
d534c452 | 162 | chat_page_url = 'https://www.youtube.com/live_chat_replay?continuation=' + continuation_id |
c60ee3a2 | 163 | elif info_dict['protocol'] == 'youtube_live_chat': |
164 | url = 'https://www.youtube.com/youtubei/v1/live_chat/get_live_chat?key=' + api_key | |
d534c452 | 165 | chat_page_url = 'https://www.youtube.com/live_chat?continuation=' + continuation_id |
273762c8 | 166 | |
82e3f6eb | 167 | frag_index = offset = 0 |
a78e3a57 | 168 | while continuation_id is not None: |
82e3f6eb | 169 | frag_index += 1 |
273762c8 | 170 | request_data = { |
171 | 'context': innertube_context, | |
172 | 'continuation': continuation_id, | |
173 | } | |
174 | if frag_index > 1: | |
175 | request_data['currentPlayerState'] = {'playerOffsetMs': str(max(offset - 5000, 0))} | |
d534c452 | 176 | headers = ie._generate_api_headers(ytcfg, visitor_data=visitor_data) |
177 | headers.update({'content-type': 'application/json'}) | |
178 | fragment_request_data = json.dumps(request_data, ensure_ascii=False).encode('utf-8') + b'\n' | |
179 | success, continuation_id, offset = download_and_parse_fragment( | |
180 | url, frag_index, fragment_request_data, headers) | |
181 | else: | |
182 | success, continuation_id, offset = download_and_parse_fragment(chat_page_url, frag_index) | |
82e3f6eb | 183 | if not success: |
184 | return False | |
185 | if test: | |
a78e3a57 | 186 | break |
187 | ||
188 | self._finish_frag_download(ctx) | |
a78e3a57 | 189 | return True |
c60ee3a2 | 190 | |
191 | @staticmethod | |
192 | def parse_live_timestamp(action): | |
193 | action_content = dict_get( | |
194 | action, | |
195 | ['addChatItemAction', 'addLiveChatTickerItemAction', 'addBannerToLiveChatCommand']) | |
196 | if not isinstance(action_content, dict): | |
197 | return None | |
198 | item = dict_get(action_content, ['item', 'bannerRenderer']) | |
199 | if not isinstance(item, dict): | |
200 | return None | |
201 | renderer = dict_get(item, [ | |
202 | # text | |
203 | 'liveChatTextMessageRenderer', 'liveChatPaidMessageRenderer', | |
204 | 'liveChatMembershipItemRenderer', 'liveChatPaidStickerRenderer', | |
205 | # ticker | |
206 | 'liveChatTickerPaidMessageItemRenderer', | |
207 | 'liveChatTickerSponsorItemRenderer', | |
208 | # banner | |
209 | 'liveChatBannerRenderer', | |
210 | ]) | |
211 | if not isinstance(renderer, dict): | |
212 | return None | |
213 | parent_item_getters = [ | |
214 | lambda x: x['showItemEndpoint']['showLiveChatItemEndpoint']['renderer'], | |
215 | lambda x: x['contents'], | |
216 | ] | |
217 | parent_item = try_get(renderer, parent_item_getters, dict) | |
218 | if parent_item: | |
219 | renderer = dict_get(parent_item, [ | |
220 | 'liveChatTextMessageRenderer', 'liveChatPaidMessageRenderer', | |
221 | 'liveChatMembershipItemRenderer', 'liveChatPaidStickerRenderer', | |
222 | ]) | |
223 | if not isinstance(renderer, dict): | |
224 | return None | |
225 | return int_or_none(renderer.get('timestampUsec'), 1000) |