search

songlink / docs

[Deprecated, possibly still accessed] Documentation for Songlink products and services: API, embed, etc. Links to these READMEs might exist in the wild, so we should maintain and update them to new documentation.
83 stars 3 forks source link

YouTube links - consider scraping the website for the song link #6

Open chrirauch opened 2 years ago

chrirauch commented 2 years ago

Hi there

First of all, thanks for your great service. I'm using your public api for a personal signal bot. Basically we have a group in which we paste YouTube links to songs or music videos which I send to your api and return the song.link url.

I found out that a lot of those results where only links to YouTube and YouTube music are generated, can be circumvented by passing the song_url metadata value to the song.link api instead of the original url.

Thought I leave it here because you could probably use that internally to link together different urls of YouTube music videos

I don't know how you would get this with a regular api call, but the ytdl-core package returns it and I guess it shouldn't be too hard to figure out where they get this info if you are interested

I made a small demo on runkit

In the demo, 3 out of 4 urls work after passing the url from the metadata instead of the original url, the last one (Eric Clapton - cocaine) has no song_url property https://runkit.com/chrirauch/songurl

kweiberth commented 2 years ago

Thanks for providing this code snippet. Do you know how the song_url property is being calculated or parsed? It looks like it's probably coming from the ytdl-core package, and not the YouTube API. I wonder if it's being scraped from the web page or network requests from the youtube.com web pages.

chrirauch commented 2 years ago

Hi,

Unfortunately no. Didn't find this in the official api docs. I just stumbled upon this while looking for a way to get better results from youtube links. But i will have a look at the ytdl-core package when i find some time. It looks like it's not a huge package and hopefully it shouldn't be too much hassle to find out where they get it from.

Will report back when i found something. But it'll probably be on the weekend, as my schedule during the week is quite packed

chrirauch commented 2 years ago

It looks like they scrape it from the webpage. In the case of the union13 song https://www.youtube.com/watch?v=ILWBMmgmTj4 i found some references in the html source code to the working url (https://www.youtube.com/watch?v=q6pV6nceHR4). For example the song name in the more info section is linked to the url that ytdl-core returns as song_url

Bildschirmfoto 2022-02-28 um 11 59 50