Support new YouTube HTML output

This commit is contained in:
Mattéo Delabre 2021-01-04 20:55:57 +01:00
parent 1490ea6ad7
commit 13e49d9a59
Signed by: matteo
GPG Key ID: AE3FBD02DC583ABB
2 changed files with 40 additions and 57 deletions

View File

@ -7,17 +7,19 @@ import { retryable, exclusive } from './retry.mjs';
import { TemporaryError } from './retry.mjs';
const log = debug('youtube-maze:api');
const PLAYER_RESP_REGEX = /window\["ytInitialPlayerResponse"\] = (\{.*?\});/;
const PLAYER_CONFIG_REGEX = /ytplayer\.config = (\{.*?\});/;
const METADATA_REGEXES = [
/window\["ytInitialPlayerResponse"\] = (\{.*?\});/,
/var ytInitialPlayerResponse = (\{.*?\});/,
];
/**
* Fetch the `ytplayer.config` object for a YouTube video.
* Fetch metadata about a YouTube video.
*
* @async
* @param String videoId Identifier of the video to fetch.
* @return Object The player configuration object.
* @return Object The video metadata object.
*/
const bareGetPlayerConfig = async (videoId) =>
const bareGetMetadata = async (videoId) =>
{
const url = util.format(WATCH_BASE, encodeURIComponent(videoId));
log(`Fetching ${videoId} (${url})`);
@ -33,36 +35,24 @@ const bareGetPlayerConfig = async (videoId) =>
}
const body = await res.text();
const searchResults = METADATA_REGEXES.map(regex => body.match(regex));
const searchResult = searchResults.find(item => item !== null);
// Look for the initial player response object to check whether the
// video was found
const responseSearch = body.match(PLAYER_RESP_REGEX);
if (responseSearch === null)
if (searchResult === null)
{
throw new TemporaryError(`Invalid YouTube response for video ${videoId}`);
throw new TemporaryError(`Incomplete YouTube response for video ${videoId}`);
}
const response = JSON.parse(responseSearch[1]);
const metadata = JSON.parse(searchResult[1]);
if (response.playabilityStatus.status !== 'OK')
if (metadata.playabilityStatus.status !== 'OK')
{
throw new Error(`Video ${videoId} is not available; \
status: ${response.playabilityStatus.status}; \
reason: "${response.playabilityStatus.reason}"`);
status: ${metadata.playabilityStatus.status}; \
reason: "${metadata.playabilityStatus.reason}"`);
}
// Look for the definition of ytplayer.config and unserialize it
const configSearch = body.match(PLAYER_CONFIG_REGEX);
if (configSearch === null)
{
throw new TemporaryError(`Unable to find player for video ${videoId}`);
}
const config = JSON.parse(configSearch[1]);
config.args.player_response = JSON.parse(config.args.player_response);
const actualId = config.args.player_response.videoDetails.videoId;
const actualId = metadata.videoDetails.videoId;
if (videoId !== actualId)
{
@ -70,40 +60,36 @@ reason: "${response.playabilityStatus.reason}"`);
}
log(`Done fetching ${videoId}`);
return config;
return metadata;
};
export const getPlayerConfig = exclusive(retryable(bareGetPlayerConfig));
export const getMetadata = exclusive(retryable(bareGetMetadata));
/**
* Get metadata about a YouTube video.
* Get video ID and title.
*
* @param config The `ytplayer.config` object corresponding to the source
* YouTube video, as obtained from `getPlayerConfig`.
* @return Object containing the video metadata.
* @param Object metadata The video metadata object from `getMetadata`.
* @return Object Containing the video ID and title.
*/
export const getVideoMeta = config => ({
videoId: config.args.player_response.videoDetails.videoId,
title: config.args.player_response.videoDetails.title,
export const getVideoInfo = metadata => ({
videoId: metadata.videoDetails.videoId,
title: metadata.videoDetails.title,
});
/**
* Find videos linked from the endscreen of a YouTube video.
*
* @param config The `ytplayer.config` object corresponding to the source
* YouTube video, as obtained from `getPlayerConfig`.
* @param Object metadata The video metadata object from `getMetadata`.
* @return List of identifiers of linked videos.
*/
export const getEndScreenVideos = config =>
export const getEndScreenVideos = metadata =>
{
const response = config.args.player_response;
if (!('endscreen' in response))
if (!('endscreen' in metadata))
{
return [];
}
return response.endscreen.endscreenRenderer.elements
return metadata.endscreen.endscreenRenderer.elements
.map(elt => elt.endscreenElementRenderer)
.filter(rdr => 'watchEndpoint' in rdr.endpoint)
.map(rdr => rdr.endpoint.watchEndpoint.videoId);
@ -112,20 +98,17 @@ export const getEndScreenVideos = config =>
/**
* Find videos linked from as cards from a YouTube video.
*
* @param config The `ytplayer.config` object corresponding to the source
* YouTube video, as obtained from `getPlayerConfig`.
* @param Object metadata The video metadata object from `getMetadata`.
* @return List of identifiers of linked videos.
*/
export const getCardVideos = config =>
export const getCardVideos = metadata =>
{
const response = config.args.player_response;
if (!('cards' in response))
if (!('cards' in metadata))
{
return [];
}
return response.cards.cardCollectionRenderer.cards
return metadata.cards.cardCollectionRenderer.cards
.map(card => card.cardRenderer.content)
.filter(content => 'videoInfoCardContentRenderer' in content)
.map(content => content.videoInfoCardContentRenderer)

View File

@ -31,19 +31,19 @@ export const exploreVideos = async (videoId, onUpdate) =>
if (!(currentId in videosNodes))
{
const config = await api.getPlayerConfig(currentId);
const meta = api.getVideoMeta(config);
const metadata = await api.getMetadata(currentId);
const info = api.getVideoInfo(metadata);
videosNodes[currentId] = meta;
videosNodes[currentId] = info;
nextVideos[currentId] = new Set();
// Add links between this video and the linked ones
api.getEndScreenVideos(config)
.forEach(nextId => nextVideos[meta.videoId].add(nextId));
api.getCardVideos(config)
.forEach(nextId => nextVideos[meta.videoId].add(nextId));
api.getEndScreenVideos(metadata)
.forEach(nextId => nextVideos[info.videoId].add(nextId));
api.getCardVideos(metadata)
.forEach(nextId => nextVideos[info.videoId].add(nextId));
for (let nextId of nextVideos[meta.videoId])
for (let nextId of nextVideos[info.videoId])
{
queue.push(nextId);
}