#!/usr/bin/python3
# ythdd Invidious Translation Layer
# -----
# Translates requests sent through Invidious API at /api/invidious/
# to use internal extractors.
from flask import Response, request, redirect, url_for
from markupsafe import escape
from time import strftime, gmtime, time
from ythdd_globals import safeTraverse
from hashlib import md5
import json, datetime
import dateparser
import html
import invidious_formats
import math
import ythdd_globals
import ythdd_api_v1
import ythdd_extractor
import ythdd_struct_builder
import ythdd_struct_parser
# TODOs:
# ----------
# DONE:
# [✓] /api/v1/stats (stats())
# [✓] /streams/dQw4w9WgXcQ (does nothing)
# [✓] /vi/:videoIdXXXX/maxresdefault.jpg
# [✓] /api/v1/search?q=... (videos and playlists), pagination
# [✓] /api/v1/search/suggestions?q=...&pq=...
# [✓] /api/v1/channel/:ucid
# [✓] /api/v1/channel/:ucid/videos, shorts, playlists, streams
# [✓] /api/v1/comments/:videoid?continuation=...
# [✓] /api/v1/videos/:videoIdXXXX
# [✓] /api/v1/playlists/:plid
# [✓] /api/v1/channel/{videos, shorts, playlists, streams, latest?}/:ucid (rewrite)
# [✓] /api/v1/:videoIdXXXX/maxres.jpg redirects to best quality thumbnail
# [✓] /api/v1/storyboards/:videoIdXXXX
# ----------
# PLANNED:
# [X] /api/v1/videos/:videoIdXXXX does not depend on yt-dlp and offloads stream retrieval elsewhere (making initial response fast)
# [X] /api/v1/manifest/:videoIdXXXX (above is prerequisite)
# [X] rewrite the awful lookup logic
# [X] /api/v1/search?q=... complex filtering options (https://gitea.invidious.io/iv-org/invidious/src/branch/master/src/invidious/search/filters.cr)
# ----------
# IDEAS:
# [*] /api/v1/popular returns last requested videos by the IP (serving as multi-device history?)
# [*] /api/v1/trending returns recently archived videos
# [*] produce continuations instead of extracting them
# ----------
# NOT PLANNED/MAYBE IN THE FUTURE:
# [ ] /api/v1/auth/subscriptions (stub? db?)
# [ ] /api/v1/auth/feed?page=1 (stub? db?)
# [ ] /api/v1/auth/playlists (stub? db?)
DEFAULT_AVATAR = "https://yt3.ggpht.com/a/default-user=s176-c-k-c0x00ffffff-no-rj"
DEFAULT_VIDEO = "https://i.ytimg.com/img/no_thumbnail.jpg" # todo: replace this with a custom, local asset
def incrementBadRequests():
ythdd_globals.apiFailedRequests += 1
def greeting():
return 200, 'hello from Invidious TL!\nstats endpoint at /api/invidious/stats'
def send(status, response):
return Response(json.dumps(response), mimetype='application/json', status=status)
def notImplemented(data):
return send(501, {'error': f"not recognised/implemented in Invidious TL: {'/'.join(data)}"})
def stats():
data_to_send = {
"version": "2.0",
"software":
{
"name": "invidious",
"version": f"invidious TL, ythdd ({ythdd_globals.version})",
"branch": "https://gitea.7o7.cx/sherl/ythdd",
"tl_msg": "/api/invidious/api/v1/"
}
}
return send(200, data_to_send)
def videoIdSanityCheck(videoId: str):
if len(videoId) != 11:
incrementBadRequests()
return send(400, f'error: bad request. wrong videoId: {videoId} is {len(videoId)} characters long, but should be 11.')
# elif...?
def auth(data):
# can be either subscriptions, feed or playlists
match data[1]:
# NOT YET IMPLEMENTED
# TODO: make it use the internal db
case "subscriptions" | "feed" | "playlists":
return send(200, [])
case _:
incrementBadRequests()
return send(404, [])
def streams():
return send(200, '')
def epochToDate(epoch):
return strftime('%Y-%m-%dT%H:%M:%SZ', gmtime(epoch))
def dateToEpoch(date: str):
return datetime.datetime.fromisoformat(date).timestamp()
def trending():
return send(200, [{}])
def popular():
return send(200, [{}])
def getError(wdata: dict):
unknown_error = {"status": "Unknown error", "reason": "This is a generic ythdd error."}
error = ""
try:
playabilityStatus = safeTraverse(wdata, ['ec1', 'playabilityStatus'], default=unknown_error)
# check for error
if safeTraverse(playabilityStatus, ["status"], default="error") == "OK":
return None # no error? return None
error += f"({playabilityStatus['status']}) {playabilityStatus['reason']}"
except:
error += f"Generic error"
try:
errorScreen = safeTraverse(wdata, ['ec1', 'playabilityStatus', 'errorScreen', 'playerErrorMessageRenderer', 'subreason', 'runs'], default=[])
error += " - "
for x in errorScreen:
error += f"{x['text']} "
except:
pass
return error
def videos(data):
# an attempt on a faithful rewrite of
# https://github.com/iv-org/invidious/blob/master/src/invidious/videos/parser.cr
response = {}
# hls_url = safeTraverse(idata, ['stage1', 'streamingData', 'hlsManifestUrl'], default="")
# adaptive_formats = safeTraverse(idata, ['stage1', 'streamingData', 'adaptiveFormats'], default=[])
# if not hls_url or not adaptive_formats:
# print(f"serious error: couldn't get hls_url or adaptive_formats!\n"
# f"dumping idata:\n"
# f"{idata}")
# return send(500, {'error': getError(idata)})
time_start = time()
if len(data) < 4 or len(data[3]) != 11:
return send(400, {"error": "No valid video id."})
if ythdd_globals.config['general']['cache'] and data[3] in ythdd_globals.video_cache:
if ythdd_globals.video_cache[data[3]]['cacheTime'] + 3 * 60 * 60 > time():
response = ythdd_globals.video_cache[data[3]]
response['fromCache'] = True
return send(200, response)
else:
del ythdd_globals.video_cache[data[3]]
wdata = ythdd_extractor.WEBextractSinglePage(data[3])
age_restricted = False
error = getError(wdata)
if error is not None:
if error.startswith("(LOGIN_REQUIRED)") and "inappropriate for some users" in error:
# check if user provided age-gated cookies
if ythdd_globals.config["extractor"]["age_restricted_cookies_path"]:
ythdd_globals.print_debug(f"videos({data[3]}): using agegated cookies to bypass restriction")
ydata = ythdd_extractor.extract(data[3], use_cookies="agegated")
wdata = ythdd_extractor.WEBextractSinglePage(data[3], use_cookies="agegated")
age_restricted = True
else:
# return error if no age-gated cookies are provided
return send(500, {"status": "error", "error": error})
else:
# return error if it doesn't mention age restriction
return send(500, {"status": "error", "error": error})
else:
ydata = ythdd_extractor.extract(data[3])
#return send(200, {'ydata': ydata, 'wdata': wdata})
#return send(200, {'idata': idata, 'wdata': wdata})
main_results = wdata['ec2']['contents']['twoColumnWatchNextResults']
primary_results = safeTraverse(main_results, ['results', 'results', 'contents'])
# video_primary_renderer = safeTraverse(primary_results, [..., 'videoPrimaryInfoRenderer'])
video_secondary_renderer = safeTraverse(primary_results, [..., 'videoSecondaryInfoRenderer'])
video_details = safeTraverse(wdata, ['ec1', 'videoDetails'])
microformat = safeTraverse(wdata, ['ec1', 'microformat', 'playerMicroformatRenderer'], default={})
video_id = safeTraverse(video_details, ['videoId'], default=f"[{data[3]}] (errors occurred, check logs)")
title = safeTraverse(video_details, ['title'], default=video_id)
views = int(safeTraverse(video_details, ['viewCount'], default=0))
length = int(safeTraverse(video_details, ['lengthSeconds'], default=1))
published = dateToEpoch(safeTraverse(microformat, ['publishDate'], default="2000-01-01T00:00:00Z")) # ISO format to Unix timestamp
published_date = epochToDate(published)
premiere_timestamp = safeTraverse(microformat, ['liveBroadcastDetails', 'startTimestamp'], default=None) # let's ignore the nitty gritty for the time being
premiere_timestamp = premiere_timestamp if premiere_timestamp else safeTraverse(microformat, ['playabilityStatus', 'liveStreamability', 'liveStreamabilityRenderer', 'offlineSlate', 'liveStreamOfflineSlateRenderer', 'scheduledStartTime'], default=None)
live_now = safeTraverse(microformat, ['liveBroadcastDetails', 'isLiveNow'], default=False)
post_live_dvr = safeTraverse(video_details, ['isPostLiveDvr'], default=False)
allowed_regions = safeTraverse(microformat, ['availableCountries'], default=[])
allow_ratings = safeTraverse(video_details, ['allowRatings'], default=True)
family_friendly = safeTraverse(microformat, ['isFamilySafe'], default=True)
is_listed = safeTraverse(video_details, ['isCrawlable'], default=True)
is_upcoming = safeTraverse(video_details, ['isUpcoming'], default=False)
keywords = safeTraverse(video_details, ['keywords'], default=[])
related_raw = safeTraverse(wdata, ['ec2', 'contents', 'twoColumnWatchNextResults', 'secondaryResults', 'secondaryResults', 'results'], default=[]) # can possibly change in the future
related = []
for entry in related_raw[:-1]:
related_entry = {}
match safeTraverse(list(entry.keys()), [0], default=""):
case "compactVideoRenderer":
# legacy renderer, a/b tested and later phased out in summer 2025
continue
case "lockupViewModel":
y = safeTraverse(entry, ['lockupViewModel'])
if not isinstance(y, dict):
continue
if safeTraverse(y, ["contentType"], default="LOCKUP_CONTENT_TYPE_VIDEO") != "LOCKUP_CONTENT_TYPE_VIDEO":
# neither mixes nor playlists are currently supported by the invidious api
continue
# note: this model is similar, but not identical to the one in ythdd_struct_parser. perhaps they can be both handled in the struct parser some time.
lmvm = safeTraverse(y, ['metadata', 'lockupMetadataViewModel'], default=[])
related_entry['videoId'] = safeTraverse(y, ['contentId'])
related_entry['title'] = safeTraverse(lmvm, ['title', 'content'])
related_entry['videoThumbnails'] = ythdd_struct_builder.genThumbs(related_entry['videoId']) #safeTraverse(y, ['thumbnail', 'thumbnails'])
related_entry['author'] = safeTraverse(lmvm, ['metadata', 'contentMetadataViewModel', 'metadataRows', 0, 'metadataParts', 0, 'text', 'content'])
related_entry['authorId'] = safeTraverse(lmvm, ['image', 'decoratedAvatarViewModel', 'rendererContext', 'commandContext', 'onTap', 'innertubeCommand', 'browseEndpoint', 'browseId'], default="UNKNOWNCHANNELID")
related_entry['authorUrl'] = '/channel/' + related_entry['authorId']
related_entry['authorVerified'] = False if safeTraverse(lmvm, ['metadata', 'contentMetadataViewModel', 'metadataRows', 0, 'metadataParts', 0, 'text', 'attachmentRuns']) is None else True # seens to do the job
author_avatar_url = safeTraverse(lmvm, ['image', 'decoratedAvatarViewModel', 'avatar', 'avatarViewModel', 'image', 'sources', 0, 'url'], default=DEFAULT_AVATAR)
related_entry['authorThumbnails'] = ythdd_extractor.generateChannelAvatarsFromUrl(author_avatar_url)
related_entry['lengthSeconds'] = ythdd_struct_parser.parseLengthFromTimeBadge(safeTraverse(y, ['contentImage', 'thumbnailViewModel', 'overlays', 0, 'thumbnailOverlayBadgeViewModel', 'thumbnailBadges', 0, 'thumbnailBadgeViewModel', 'text'], default="0:0"))
related_entry['viewCountText'] = safeTraverse(lmvm, ['metadata', 'contentMetadataViewModel', 'metadataRows', 1, 'metadataParts', 0, 'text', 'content'], default="0").split(" ")[0]
related_entry['viewCount'] = ythdd_struct_parser.parseViewsFromViewText(related_entry['viewCountText'])
case _:
# unsupported model: print info into stdout
print("received an entry of unknown type during parsing of related videos:")
print(entry)
print("")
continue
related.append(related_entry)
likes = safeTraverse(ydata, ['like_count'], default=0)
description = safeTraverse(microformat, ['description', 'simpleText'], default="\n(ythdd: failed to retrieve description, perhaps it's empty?)")
short_description = safeTraverse(wdata, ['ec1', 'videoDetails', 'shortDescription'], default="(ythdd: failed to retrieve short description, perhaps it's empty?)")
description_html = html.escape(description).replace("\r\n", "
").replace("\n", "
") # still TODO: https://github.com/iv-org/invidious/blob/master/src/invidious/videos/parser.cr#L329
genre = safeTraverse(microformat, ['category'])
# TODO: genre blah blah blah...
author = safeTraverse(video_details, ['author'], default="Unknown Author")
ucid = safeTraverse(video_details, ['channelId'], default="UNKNOWNCHANNELID")
subs = ydata['channel_follower_count']
author_thumbnail = safeTraverse(video_secondary_renderer, ['owner', 'videoOwnerRenderer', 'thumbnail', 'thumbnails', 0, 'url'])
author_verified = ythdd_extractor.isVerified(safeTraverse(video_secondary_renderer, ['owner', 'videoOwnerRenderer', 'badges', 0], default=[]))
if author_thumbnail is None:
# there might be multiple authors (on a collaborative video)
# if so, then try to extract first channel's (uploader's) avatar
livm = safeTraverse(video_secondary_renderer, ["owner", "videoOwnerRenderer", "attributedTitle", "commandRuns", 0, "onTap", "innertubeCommand", "showDialogCommand", "panelLoadingStrategy", "inlineContent", "dialogViewModel", "customContent", "listViewModel", "listItems"], default=[])
author_thumbnail = safeTraverse(livm, [0, "listItemViewModel", "leadingAccessory", "avatarViewModel", "image", "sources", 0, "url"], default=DEFAULT_AVATAR)
author_verified = author_verified or safeTraverse(livm, [0, "listItemViewModel", "title", "attachmentRuns", 0, "element", "type", "imageType", "image", "sources", 0, "clientResource", "imageName"]) in ("AUDIO_BADGE", "CHECK_CIRCLE_FILLED")
author_thumbnail = ythdd_extractor.generateChannelAvatarsFromUrl(author_thumbnail)
wdata_streams = safeTraverse(wdata, ["ec1", "streamingData"], default=[])
adaptive_formats = []
format_streams = []
# adaptive_formats, format_streams = rebuildFormats(adaptive_formats)
if not live_now:
# adaptive_formats, format_streams = rebuildFormatsFromYtdlpApi(ydata)
initial_astreams_y = {} # itag is the key
initial_fstreams_y = {} # same here
initial_astreams_w = {}
initial_fstreams_w = {}
for video_stream in ydata["formats"]:
if video_stream["format_note"] in ("storyboard"):
# ignore non-audio/video formats (e.g. storyboards)
continue
if video_stream["format_id"] == "18": # todo: do this dynamically
initial_fstreams_y[int(video_stream["format_id"])] = video_stream
elif video_stream["format_id"].isdigit():
# filter out DRC audio
initial_astreams_y[int(video_stream["format_id"])] = video_stream
else:
continue
# format streams
for video_stream in wdata_streams["formats"]:
initial_fstreams_w[video_stream["itag"]] = video_stream
# adaptive streams
for video_stream in wdata_streams["adaptiveFormats"]:
initial_astreams_w[video_stream["itag"]] = video_stream
for itag in initial_astreams_y:
if itag in initial_astreams_w:
adaptive_formats.append(ythdd_struct_parser.parseAdaptiveStreams(initial_astreams_w[itag], initial_astreams_y[itag]))
for itag in initial_fstreams_y:
if itag in initial_fstreams_w:
format_streams.append( ythdd_struct_parser.parseFormatStreams( initial_fstreams_w[itag], initial_fstreams_y[itag]))
hls_url = None
else:
adaptive_formats, format_streams = [{"url": f"http://a/?expire={int(time_start + 5.9 * 60 * 60)}", "itag": "18", "type": "", "clen": "0", "lmt": "", "projectionType": "RECTANGULAR"}], [] # freetube/clipious shenanigans, see: https://github.com/FreeTubeApp/FreeTube/pull/5997 and https://github.com/lamarios/clipious/blob/b9e7885/lib/videos/models/adaptive_format.g.dart
hls_url = safeTraverse(ydata, ["url"], default="ythdd: unable to retrieve stream url")
if age_restricted:
if not adaptive_formats:
adaptive_formats = [{"url": f"http://a/?expire={int(time_start + 5.9 * 60 * 60)}", "itag": "18", "type": "", "clen": "0", "lmt": "", "projectionType": "RECTANGULAR"}] # same as above
description += " \n(ythdd: this video is age-restricted and thus only available in 360p - itag 18)"
description_html += "
(ythdd: this video is age-restricted and thus only available in 360p - itag 18)"
if live_now:
video_type = "livestream"
premiere_timestamp = published # ??? that works i guess
elif premiere_timestamp:
video_type = "scheduled"
published = dateToEpoch(premiere_timestamp) if premiere_timestamp else int(time())
else:
video_type = "video"
premium = False
if "YouTube Red" in keywords:
premium = True
# TODO: detect paywalled patron-only videos
# because we fetched the video's wdata, we might as
# well save it inside of general cache so that
# requests for the video's comments don't have to
# spawn an additional request for initial ctoken
ensure_comment_continuation(video_id, wdata)
storyboards = []
storyboards_extracted = ensure_storyboards(video_id, wdata, length=length)
if storyboards_extracted:
storyboards = ythdd_struct_builder.genStoryboards(video_id)
time_end = time()
response = {
"type": video_type,
"title": title,
"videoId": video_id,
"videoThumbnails": ythdd_struct_builder.genThumbs(video_id),
"storyboards": storyboards,
"description": description, # due to change (include ythdd metadata)
"descriptionHtml": description_html,
"published": published,
"publishedText": published_date,
"keywords": keywords,
"viewCount": views,
"viewCountText": str(views), # not implemented
"likeCount": likes,
"dislikeCount": 0,
"paid": False, # not implemented
"premium": premium,
"isFamilyFriendly": family_friendly,
"allowedRegions": allowed_regions,
"genre": genre,
"genreUrl": "/genreUrl/not/implemented/", # not implemented
"author": author,
"authorId": ucid,
"authorUrl": "/channel/" + ucid,
"authorVerified": author_verified,
"authorThumbnails": author_thumbnail,
"subCountText": str(subs),
"lengthSeconds": length,
"allowRatings": allow_ratings,
"rating": 0,
"isListed": is_listed,
"liveNow": live_now,
"isPostLiveDvr": post_live_dvr,
"isUpcoming": is_upcoming,
"dashUrl": ythdd_globals.config['general']['public_facing_url'] + "api/invidious/api/v1/manifest/" + video_id, # not implemented
"premiereTimestamp": premiere_timestamp,
"hlsUrl": hls_url, # broken after a change in iOS player, only usable for livestreams
"adaptiveFormats": adaptive_formats, # same as hlsUrl
"formatStreams": format_streams,
"captions": [], # not implemented
# "captions": [
# {
# "label": String,
# "language_code": String,
# "url": String
# }
# ],
# "musicTracks": [
# {
# "song": String,
# "artist": String,
# "album": String,
# "license": String
# }
# ],
"recommendedVideos": related,
"took": time_end - time_start
}
if ythdd_globals.config['general']['debug']:
response["ydata"] = ydata
response["wdata"] = wdata
if ythdd_globals.config['general']['cache']:
ythdd_globals.video_cache[data[3]] = response
ythdd_globals.video_cache[data[3]]['cacheTime'] = time()
# for debugging:
#return send(200, ythdd_extractor.WEBextractSinglePage(data[3]))
#return send(200, ythdd_extractor.IOSextract(data[3]))
#return send(200, {'idata': idata, 'wdata': wdata})
# if youtube returns not the videoId we aksed
# then it means that the instance is ratelimited
status_code = 200 if data[3] == response['videoId'] else 403
return send(status_code, response)
def search(data, req):
search_query = req.args.get('q')
# ignore paginated requests as we do nothing with the continuation token
page = req.args.get('page')
if page is not None and page != '1':
try:
page = int(page)
except:
return send(400, {"error": "Wrong page."})
else:
page = None # when page is "1"
if (data[-2].lower() != "search" or data[-1].lower() != "") and data[-1].lower() != "search":
previous_query = req.args.get('pq')
suggestions = ythdd_extractor.WEBgetSearchSuggestions(search_query, previous_query)
return send(200, suggestions)
results = ythdd_extractor.WEBextractSearchResults(search_query, page)
results_list = []
for entry in results:
parsed_entry = ythdd_struct_parser.parseRenderers(entry)
if parsed_entry is not None:
results_list.append(parsed_entry)
return send(200, results_list)
def get_channel_tab(requested_tab, ucid, req, only_json: bool = False):
# check for page/cont
ctoken = req.args.get('continuation')
# perhaps continuation tokens should be checked here (whether they are inside of general_cache)
# this way, malicious requests containing bogus ctokens can't be sent to potentially ban/ratelimit the instance (?)
# if ctoken is not None and ctoken not in ythdd_globals.general_cache...
# unique req fingerprint allows for this exact query to be cached in memory.
# md5 sum serves as a "unique" deterministic value which can be checked for cache hit/miss
unique_request_fingerprint = md5(f"{ucid}_{requested_tab}_{ctoken}".encode('utf-8')).hexdigest()
# if we haven't discovered parameters required for browsing a specific tab,
# then load them now
if ucid not in ythdd_globals.general_cache["continuations"]["channels"]:
channels(["", "", "", ucid], req, True)
# check if request has been cached within the last hour
if ythdd_globals.config['general']['cache'] and unique_request_fingerprint in ythdd_globals.general_cache["channels"]:
if ythdd_globals.general_cache["channels"][unique_request_fingerprint]['cacheTime'] + 1 * 60 * 60 > time():
response = ythdd_globals.general_cache["channels"][unique_request_fingerprint]
if only_json:
return response
else:
return send(200, response)
else:
del ythdd_globals.general_cache["channels"][unique_request_fingerprint]
# load relevant data from global (general) cache
param = safeTraverse(ythdd_globals.general_cache["continuations"]["channels"][ucid], ["tabs", requested_tab, "param"], default=None)
name = safeTraverse(ythdd_globals.general_cache["continuations"]["channels"][ucid], ["name"], default="")
avatar = safeTraverse(ythdd_globals.general_cache["continuations"]["channels"][ucid], ["avatar"], default=DEFAULT_AVATAR)
verified = safeTraverse(ythdd_globals.general_cache["continuations"]["channels"][ucid], ["verified"], default=False)
# if provided, ctoken will be used for browsing as well
wdata = ythdd_extractor.browseChannel(ucid, params=param, ctoken=ctoken)
# sanity check (whether we got what we requested)
received_tab = safeTraverse(wdata, ["responseContext", "serviceTrackingParams", 0, "params", 0, "value"])
if received_tab != f"channel.{requested_tab}":
# if that's not the case, either something changed in the innertube API,
# or content that was asked for isn't available
print(f"INFO: couldn't verify server returned channel data we asked for. "
f"Requested channel.{requested_tab}, got {received_tab}. Most likely we sent a request to Innertube which got rejected.")
# load requested tab
result = {}
if ctoken is None:
tabs = safeTraverse(wdata, ["contents", "twoColumnBrowseResultsRenderer", "tabs"], default=[])
for tab in tabs:
tab_name = safeTraverse(tab, ["tabRenderer", "title"], default="").lower()
# rewrite livestream tab for backwards compatibility with invidious (and clients like freetube)
if tab_name == "live":
tab_name = "streams"
if tab_name and tab_name == requested_tab:
result = safeTraverse(tab, ["tabRenderer", "content"], default=[])
break
items = []
inner_contents = []
new_continuation = ""
response = {}
match requested_tab:
case "videos" | "shorts" | "streams":
# videos/shorts/livestreams have actually the same response schema,
# only the renderers differ - but they are taken care of in ythdd_struct_parser.parseRenderers()
if ctoken is None:
inner_contents = safeTraverse(result, ["richGridRenderer", "contents"], default=[])
else:
inner_contents = safeTraverse(wdata, ["onResponseReceivedActions", 0, "appendContinuationItemsAction", "continuationItems"], default=[])
for entry in inner_contents:
# videos from videos tab have no owner info (?) or it's in another place. if it is somewhere, this expression can be made simpler by traversing something else in struct parser.
item = safeTraverse(entry, ["richItemRenderer", "content"])
if item is not None:
items.append(ythdd_struct_parser.parseRenderers(item, {"author_name": name, "author_ucid": ucid, "avatar": avatar}))
new_continuation = safeTraverse(inner_contents, [-1, "continuationItemRenderer", "continuationEndpoint", "continuationCommand", "token"], default="")
response = {
"videos": items
}
if new_continuation:
response["continuation"] = new_continuation
# cache response
if ythdd_globals.config['general']['cache']:
ythdd_globals.general_cache["channels"][unique_request_fingerprint] = response
ythdd_globals.general_cache["channels"][unique_request_fingerprint]['cacheTime'] = time()
# todo: save continuation(?)
# or... is there a usecase for saving it?
case "playlists":
# todo: replace this with an on-demand generated ctoken?
if ctoken is None:
inner_contents = safeTraverse(result, ["sectionListRenderer", "contents", 0, "itemSectionRenderer", "contents", 0, "gridRenderer", "items"], default=[])
else:
inner_contents = safeTraverse(wdata, ["onResponseReceivedActions", 0, "appendContinuationItemsAction", "continuationItems"], default=[])
for entry in inner_contents:
item = ythdd_struct_parser.parseRenderers(entry, {"author_name": name, "author_ucid": ucid, "avatar": avatar})
items.append(item)
new_continuation = safeTraverse(inner_contents, [-1, "continuationItemRenderer", "continuationEndpoint", "continuationCommand", "token"], default="")
response = {
"playlists": items
}
if new_continuation:
response["continuation"] = new_continuation
# cache response
if ythdd_globals.config['general']['cache']:
ythdd_globals.general_cache["channels"][unique_request_fingerprint] = response
ythdd_globals.general_cache["channels"][unique_request_fingerprint]['cacheTime'] = time()
case _:
# for all other renderers, which aren't currently supported
response = {
# "wdata": wdata
}
if ythdd_globals.config["general"]["debug"]:
response["wdata"] = wdata
if only_json:
return response
return send(200, response)
def get_comments(data, req, only_json: bool = False):
# get comment continuation
ctoken = req.args.get('continuation')
# perform some basic video id validation
if len(data) < 4 or len(data) >= 4 and len(data[3]) != 11:
return send(400, {"error": "Bad request: invalid videoId."})
video_id = data[3]
# if ctoken isn't provided, get it from the general cache
if ctoken is None or ctoken == '':
# but first ensure it's there
ensure_comment_continuation(video_id)
ctoken = ythdd_globals.general_cache["continuations"]["comments"][video_id][0]
# get joined video comment models
wdata, new_continuation = ythdd_extractor.WEBgetVideoComments(ctoken)
comments = []
for comment in wdata:
# parse the comment
parsed_comment = ythdd_struct_parser.customCommentRendererParser(comment)
if parsed_comment is not None:
comments.append(parsed_comment)
response = {
"videoId": video_id,
"comments": comments,
"continuation": new_continuation
}
if only_json:
return response
return send(200, response)
def ensure_comment_continuation(video_id: str, wdata = None):
# save continutation token for comments in global comment cache
if not video_id in ythdd_globals.general_cache["continuations"]["comments"]:
ythdd_globals.general_cache["continuations"]["comments"][video_id] = []
if wdata is None:
# perhaps saving related videos to cache might be not a bad idea?
wdata = ythdd_extractor.WEBextractSinglePage(video_id)
# search for "top comments" continuation token
# todo: replace this with on-demand continuation creation
comment_continuation = safeTraverse(wdata, ["ec2", "engagementPanels", 0, "engagementPanelSectionListRenderer", "header", "engagementPanelTitleHeaderRenderer", "menu", "sortFilterSubMenuRenderer", "subMenuItems", 0, "serviceEndpoint", "continuationCommand", "token"], default=None)
if comment_continuation is not None:
ythdd_globals.general_cache["continuations"]["comments"][video_id].append(comment_continuation)
else:
print(f"error: couldn't extract comment continuation token from video page ({video_id}). this video likely has comments disabled.")
ythdd_globals.general_cache["continuations"]["comments"][video_id].append("")
def ensure_storyboards(video_id: str, wdata = None, length = 60):
# Returns True on successful extraction, False when it failed.
# Storyboards don't expire. They can be cached indefinitely.
if not video_id in ythdd_globals.general_cache["storyboards"]:
ythdd_globals.general_cache["storyboards"][video_id] = None
if wdata is None:
wdata = ythdd_extractor.WEBextractSinglePage(video_id)
# get storyboard template string
storyboards = None
storyboard_template = safeTraverse(wdata, ["ec1", "storyboards", "playerStoryboardSpecRenderer", "spec"], default=None)
# silly sanity check, todo: do a regex one instead?
if isinstance(storyboard_template, str):
# sample storyboard template url structure, indented for readability
# https://i.ytimg.com/sb/:videoId/storyboard3_L$L/$N.jpg?sqp=b64encodedprotobuf
# | 48 # 27 # 100 # 10 # 10 # 0 # default # rs$datadatadatadatadatadatadatadatada
# | 80 # 45 # 55 # 10 # 10 # 1000 # M$M # rs$datadatadatadatadatadatadatadatada
# | 160 # 90 # 55 # 5 # 5 # 1000 # M$M # rs$datadatadatadatadatadatadatadatada
# | 320 # 180 # 55 # 3 # 3 # 1000 # M$M # rs$datadatadatadatadatadatadatadatada
# ^ width, height, thumb_count, columns, rows, interval, $N, sigh parameter. $L is just the index of a given storyboard, say, 0 for $N=default
# try to extract data from the storyboard template
try:
base_url, *formats = storyboard_template.split("|")
extracted_formats = []
for index, fmt in enumerate(formats):
fmt = fmt.split("#")
width = int(fmt[0])
height = int(fmt[1])
count = int(fmt[2])
columns = int(fmt[3])
rows = int(fmt[4])
interval = int(fmt[5])
name = fmt[6]
sigh = fmt[7]
thumbs_per_image = columns * rows
images_count = math.ceil(count / thumbs_per_image)
interval = interval if interval != 0 else int((length / count) * 1000) # calculated only for $N=default as it's the only one that has interval=0
extracted_formats.append({
"index": index,
"width": width,
"height": height,
"thumb_count": count,
"columns": columns,
"rows": rows,
"interval": interval,
"name": name,
"sigh": sigh,
"images_count": images_count
})
storyboards = {
"template_url": ythdd_globals.translateLinks(base_url, remove_params=False), # NOT removing params is crucial, otherwise sqp will be dropped!
"formats": extracted_formats
}
ythdd_globals.general_cache["storyboards"][video_id] = storyboards
return True
except:
print("error(ensure_storyboards): storyboard template url layout changed. please update ythdd for latest storyboard extraction fixes.")
return False
else:
print(f"error(ensure_storyboards: couldn't extract storyboards from video page ({video_id}). this video won't have storyboards.")
return False
def channels(data, req, only_json: bool = False):
# prevent potential out of bound read
if len(data) < 4:
return send(400, {"error": "No channel specified."})
if len(data) > 4 and len(data[4]) == 24 and data[4].startswith("UC"):
# reversed order (/api/v1/api/invidious/channels/videos/UC...)
data[3], data[4] = data[4], data[3]
# silly sanity check
if len(data[3]) != 24 or not data[3].startswith("UC"):
# typical order (/api/v1/api/invidious/channels/UC.../videos)
return send(404, {"error": "This channel does not exist."})
if len(data) > 4:
match data[4]:
case "videos" | "shorts" | "playlists" | "podcasts" | "streams":
return get_channel_tab( data[4], data[3], req)
case "live":
return get_channel_tab("streams", data[3], req)
case "latest":
return get_channel_tab( "videos", data[3], req)
case _:
return send(400, {"error": f"Bad request, unrecognized/unsupported tab \"{data[4]}\"."})
wdata = ythdd_extractor.browseChannel(data[3])
channel_meta = safeTraverse(wdata, ["metadata", "channelMetadataRenderer"])
banners = safeTraverse(wdata, ["header", "pageHeaderRenderer", "content", "pageHeaderViewModel", "banner", "imageBannerViewModel", "image", "sources"], default=[])
avatar = safeTraverse(wdata, ["header", "pageHeaderRenderer", "content", "pageHeaderViewModel", "image", "decoratedAvatarViewModel", "avatar", "avatarViewModel", "image", "sources", 0, "url"], default=DEFAULT_AVATAR)
subscribers = ythdd_struct_parser.parseViewsFromViewText(safeTraverse(wdata, ["header", "pageHeaderRenderer", "content", "pageHeaderViewModel", "metadata", "contentMetadataViewModel", "metadataRows", 1, "metadataParts", 0, "text", "content"], default="0"))
verified = False # to be replaced later with ythdd_extractor.isVerified(...)
author_name = safeTraverse(channel_meta, ["title"], default="Unknown Channel")
author_ucid = safeTraverse(channel_meta, ["externalId"], default=data[3]) # prevent recursion with fallback to provided ucid
ythdd_globals.general_cache["continuations"]["channels"][author_ucid] = {
"avatar": avatar,
"name": author_name,
"tabs": {},
"verified": verified
}
tabs = safeTraverse(wdata, ["contents", "twoColumnBrowseResultsRenderer", "tabs"], default=[])
tab_names = []
for tab in tabs:
# collect tab names
tab_name = safeTraverse(tab, ["tabRenderer", "title"], default="").lower()
if tab_name:
# same as in get_channel_tab
if tab_name == "live":
tab_name = "streams"
tab_names.append(tab_name)
# and their params (used to retrieve data about them)
ythdd_globals.general_cache["continuations"]["channels"][author_ucid]["tabs"][tab_name] = dict()
ythdd_globals.general_cache["continuations"]["channels"][author_ucid]["tabs"][tab_name]["param"] = safeTraverse(tab, ["tabRenderer", "endpoint", "browseEndpoint", "params"], default=None)
latest_videos = get_channel_tab("videos", data[3], req, only_json=True)
for banner in banners:
banner["url"] = ythdd_globals.translateLinks(banner["url"])
avatars = ythdd_extractor.generateChannelAvatarsFromUrl(avatar)
response = {
"author": author_name,
"authorId": author_ucid,
"authorUrl": "https://www.youtube.com/channel/" + author_ucid,
"authorBanners": banners,
"authorThumbnails": avatars,
"subCount": subscribers,
"totalViews": 0,
"joined": 0,
"autoGenerated": False, # todo: actually check this
"ageGated": False,
"isFamilyFriendly": safeTraverse(channel_meta, ["isFamilySafe"], default=False),
"description": safeTraverse(channel_meta, ["description"], default="ythdd: no channel description"),
"descriptionHtml": html.escape(safeTraverse(channel_meta, ["description"], default="ythdd: no channel description (html)")).replace("\r\n", "
").replace("\n", "
"),
"allowedRegions": safeTraverse(channel_meta, ["availableCountryCodes"], default=[]),
"tabs": tab_names,
"tags": [safeTraverse(channel_meta, ["keywords"], default="")],
"authorVerified": verified,
"latestVideos": latest_videos["videos"], # using struct parser
"relatedChannels": []
}
if ythdd_globals.config["general"]["debug"]:
response["wdata"] = wdata
# todo: cache response
if only_json:
return response
return send(200, response)
def playlists(data, req, only_json: bool = False):
# read playlist id and sanity check
if len(data) < 4:
return send(400, {"error": "No playlist specified."})
# todo: make clipious stop spamming requests for paginated response
page = req.args.get('page')
# if page is not None and page != '1':
# return send(404, {"error": "Paginated queries are not supported."})
plid = data[3]
# no info about what length can playlists be
# settle for a basic sanity check instead
# if len(plid) not in (18, 34, 36):
if not (plid.startswith("PL") or plid.startswith("OL")):
response = {"error": "Only standard playlists are currently supported (no mixes, video-based playlists, etc.)"}
if only_json:
return response
return send(400, response)
# check if request has been cached within the last hour
if ythdd_globals.config['general']['cache'] and plid in ythdd_globals.general_cache['playlists']:
if ythdd_globals.general_cache['playlists'][plid]['cacheTime'] + 1 * 60 * 60 > time():
response = ythdd_globals.general_cache['playlists'][plid].copy()
if page is not None and page != '1':
response['videos'] = []
if only_json:
return response
else:
return send(200, response)
else:
del ythdd_globals.general_cache['playlists'][plid]
# browse the playlist iteratively, first fetch is without any continuation
all_unparsed_videos = []
meta, new_continuation, videos = ythdd_extractor.WEBextractPlaylist(plid=plid)
if isinstance(videos, list):
all_unparsed_videos = videos.copy()
while new_continuation != None:
# fetch subsequent playlist videos
_, new_continuation, videos = ythdd_extractor.WEBextractPlaylist(ctoken=new_continuation)
if videos is not None:
all_unparsed_videos.extend(videos)
# process videos
parsed_videos = []
for video in all_unparsed_videos:
parsed_video = ythdd_struct_parser.parseRenderers(video)
if parsed_video is not None:
parsed_videos.append(parsed_video)
# process metadata
primary_sidebar = safeTraverse(meta, ["sidebar", "playlistSidebarRenderer", "items", 0, "playlistSidebarPrimaryInfoRenderer"], default={})
secondary_sidebar = safeTraverse(meta, ["sidebar", "playlistSidebarRenderer", "items", 1, "playlistSidebarSecondaryInfoRenderer"], default={})
# apparently fields can be stored inside of simpleText one time, only to be stored inside of runs another time
title = ythdd_struct_parser.extractTextFromSimpleOrRuns(safeTraverse(primary_sidebar, ["title"]), default="Unknown playlist title")
playlist_thumb = ythdd_globals.translateLinks(safeTraverse(primary_sidebar, ["thumbnailRenderer", "playlistVideoThumbnailRenderer", "thumbnail", "thumbnails", -1, "url"], default=DEFAULT_VIDEO))
author = safeTraverse(secondary_sidebar, ["videoOwner", "videoOwnerRenderer", "title", "runs", 0, "text"], default="Unknown channel")
author_ucid = safeTraverse(secondary_sidebar, ["videoOwner", "videoOwnerRenderer", "title", "runs", 0, "navigationEndpoint", "browseEndpoint", "browseId"], default="UNKNOWNCHANNELID")
author_avatars = ythdd_extractor.generateChannelAvatarsFromUrl(safeTraverse(secondary_sidebar, ["videoOwner", "videoOwnerRenderer", "thumbnail", "thumbnails", 0, "url"], default=DEFAULT_AVATAR))
description = safeTraverse(meta, ["microformat", "microformatDataRenderer", "description"], default="(ythdd: failed to retrieve description, perhaps it's empty?)")
if author_ucid == "UNKNOWNCHANNELID" and author == "Unknown channel":
# most likely a collaborative playlist
author = safeTraverse(secondary_sidebar, ["videoOwner", "videoOwnerRenderer", "title", "simpleText"], default="by Unknown user and others").removeprefix("by ")
author_orig = author[:author.rfind(" and ")]
description += f"\n(ythdd: This is a collaborative playlist by \"{author_orig}\" {author.removeprefix(author_orig + ' ')}. You can't view the authors' channels.)"
author = author_orig
description_html = html.escape(description).replace("\r\n", "
").replace("\n", "
")
video_count = ythdd_struct_parser.parseViewsFromViewText(ythdd_struct_parser.extractTextFromSimpleOrRuns(safeTraverse(primary_sidebar, ["stats", 0]), default="No videos"))
view_count = ythdd_struct_parser.parseViewsFromViewText(ythdd_struct_parser.extractTextFromSimpleOrRuns(safeTraverse(primary_sidebar, ["stats", 1]), default="No views"))
updated = ythdd_struct_parser.extractTextFromSimpleOrRuns(safeTraverse(primary_sidebar, ["stats", 2]), default="2000-01-01").removeprefix("Last updated on ").removeprefix("Updated ")
updated = int(dateparser.parse(updated).timestamp())
is_unlisted = safeTraverse(primary_sidebar, ["badges", 0, "metadataBadgeRenderer", "icon", "iconType"], default="PRIVACY_LISTED") == "PRIVACY_UNLISTED" # this needs further research https://gitea.invidious.io/iv-org/invidious/src/commit/325e013e0d9e5670fa0df7635ff30a0ee029e05e/src/invidious/playlists.cr#L133
response = {
"type": "playlist",
"title": title,
"playlistId": plid,
"playlistThumbnail": playlist_thumb,
"author": author,
"authorId": author_ucid,
"authorUrl": "/channel/" + author_ucid,
"subtitle": None, # todo?
"authorThumbnails": author_avatars,
"description": description,
"descriptionHtml": description_html,
"videoCount": video_count,
"viewCount": view_count,
"updated": updated,
"isListed": not is_unlisted,
"videos": parsed_videos
}
# todo: cache videos and metadata separately, so that paginated queries can be supported as well
if ythdd_globals.config['general']['cache']:
ythdd_globals.general_cache['playlists'][plid] = response.copy()
ythdd_globals.general_cache['playlists'][plid]['cacheTime'] = time()
if page is not None or page == '1':
response['videos'] = []
if only_json:
return response
return send(200, response)
def storyboards(data, req):
height = req.args.get("height")
width = req.args.get("width")
video_id = data[3]
try:
height = int(height)
except:
pass
try:
width = int(width)
except:
pass
resp = ythdd_struct_builder.genWebvttStoryboard(video_id, width, height)
return Response(resp, mimetype="text/vtt", status=200)
def lookup(data, req):
# possibly TODO: rewrite this mess
if len(data) > 2:
if (data[0], data[1]) == ("api", "v1"):
match data[2]:
case 'stats' | '': # /api/invidious/api/v1/stats and /api/invidious/api/v1/
return stats()
case 'trending':
return trending()
case 'popular':
return popular()
case 'videos':
return videos(data)
case 'auth':
return auth(data)
case 'search':
return search(data, req)
case 'channels':
return channels(data, req)
case 'comments':
return get_comments(data, req)
case 'playlists':
return playlists(data, req)
case 'storyboards':
return storyboards(data, req)
case _:
incrementBadRequests()
return notImplemented(data)
elif data[0] in ('ggpht', 'vi'):
# for some reason the Materialous client
# and FreeTube keep making requests to these
if data[1] in ('ggpht', 'vi'):
return redirect('/' + "/".join(data[1:]))
return redirect('/' + "/".join(data[0:]))
else:
incrementBadRequests()
return notImplemented(data)
elif len(data) == 2:
if (data[0], data[1]) == ("api", "v1"): # /api/invidious/api/v1
return stats()
elif data[0] == "streams":
return streams()
elif data[0] in ('ggpht', 'vi'):
return redirect('/' + "/".join(data[0:]))
else:
incrementBadRequests()
return notImplemented(data)
elif len(data) == 1:
if data[0] == "videoplayback":
return redirect(url_for('videoplaybackProxy', **req.args))
return stats() # /api/invidious/something