Skip to main content

Add New Fetchers

APM uses a modularized approach to organize media fetchers, similar to muisicfree. While there are potentials to make APM support extra fetchers via user-defined plugins, I don't have the time or interest to maintain any plain string evaled into commonjs functions and am currently sticking with hardcoded fetchers. However if you would like to add your own fetchers in your compiled APM, read along!

APM's Fetcher Structure​

All of APM's fetchers are in the src/utils/mediafetch directory. Each site's fetcher is a single ts file, and exports a dict with fields: regexSearchMatch, regexFetch, regexResolveURLMatch, resolveURL, refreshSong. The fetcher itself is loaded at searchBiliURLs in src/utils/BiliSearch.ts (where search via the top search bar happens, using regexSearchMatch, regexFetch), and fetchPlayUrlPromise in src/utils/Data.js (where resolving the song's streamable URL happens, using regexResolveURLMatch, resolveURL).

In searchBiliURLs, user input is regexed against regexSearchMatch. If there is a match, regexFetch is called with the extracted match to return songs.

In fetchPlayUrlPromise, song id is regexed against regexResolveURLMatch. If there is a match, resolveURL is called with the song object to return the resolved streamable URL.

refreshSong is a function work in progress.

regexSearchMatch​

Based on needs, this regexp may include extracted portions, such as extracting a single field using /(BV[^/?]+)/, extracting two fields using /space.bilibili\.com\/(\d+)\/channel\/collectiondetail\?sid=(\d+)/, or not extracting anything at all, merely matching a string using /steria.vplayer.tk/.

It accepts up to four parameters:

reExtracted, the regexp extracted object;

progressEmitter, emits the current progress. input is a number from 0 - 100.

favList, a list of Song.bvids to prevent duplicates getting re-parsed by regexFetch.

useBiliTag, a special tag for bilibili fetches.

regexFetch​

regexFetch is simply a wrapper of two functions: songFetch and paginatedFetch. paginatedFetch takes the user input and uses API to return VideoInfo objects, and songFetch turns the VideoInfo objects into Song objects. The seemlying redundant VideoInfo object is specifically due to Bilibili having video inside of video disguising as a video list, and is redundant anywhere else. Using Bilibili channel (bilichannel.ts) as an example:

regexFetch parses user input and returns the channel's mid;

paginatedFetch uses the bilibili channel API and mid to loop through all pages of the user's bilibili channel, and return a list of VideoInfos, one for each BVID;

Because one BVID may have multiple sub videos, each VideoInfo is extracted by songFetch to return Songs.

paginatedFetch​

There are two common functions written for paginatedFetch in paginatedfetch.ts, fetchPaginatedAPI and fetchAwaitPaginatedAPI. fetchAwaitPaginatedAPI stops the first time a duplicate is found in favList, while fetchPaginatedAPI resolves all pages regardless. Both have the same props, where:

url, the API url. this must have a {pn} to be replaced by a page number, starting from 1.

getMediaCount, the total number of songs to be fetched from the API. this can be a hardcoded number to limit the total number of fetched songs, or more commonly a field from the response.

getPageSize, the total number of songs in a single page of API. this together with getMediaCount, determines how many pages to be fetched.

getItems, parses the response json to get each item from the response.

resolveBiliBVID = async () => [], parses the item from getItems, into VideoInfo.

progressEmitter = () => undefined, emits the current progress. input is a number from 0 - 100.

favList = [], a list of Song.bvids to prevent duplicates getting re-parsed by regexFetch.

limiter = biliApiLimiter, Bottleneck throttler for each page request. biliApiLimiter is a min 200ms, max concurrent 5 throttler.

params = undefined, : custom headers for fetch.

jsonify = res => res.json(), a custom command to parse response json.

getBVID = (val: any) => val.bvid, parses an item from getItems, to a bvid to be compared in favList for duplicates.

fetcher = bfetch, the fetch method to be used. typically is just fetch, or your wrapper of fetch that adds extra headers like UA and referrer, as in my bfetch.

songFetch​

For most sites this is simply a re-cast of VideoInfo into Song. see steriatk.ts's songFetch.

Note when casting to a Song object, cid must be always unique; bvid is used for de-duplication purposes. regexResolveURLMatch will match for cid. My way of managing this is use a special prefix for cid. eg. in steriatk.ts all cid has a steriatk- prefix, and regexResolveURLMatch is /^steriatk-/.

regexResolveURLMatch​

see above.

resolveURL​

resolves the streamable URL when regexResolveURLMatch is matched. For personal websites like steria.vplayer.tk its a static URL, and one can either put that URL in bvid or cid for simple parsing; for bigger streaming platforms like bilibili or youtube this needs to be a customized resolver that uses other APIs, similar to fetchVideoPlayUrlPromise in Data.js.