search

Bionus / imgbrd-grabber

Very customizable imageboard/booru downloader with powerful filenaming features.
https://www.bionus.org/imgbrd-grabber/
Apache License 2.0
2.58k stars 220 forks source link

[Source request] - Cure Gecko #1803

Open RiabikovR opened 5 years ago

RiabikovR commented 5 years ago

Cure Gecko's Booru Search is a service that provides a metasearch of the most popular boorus. It is basically a database of direct links to the images hosted on boorus. I have been using it for a very long time and found it very convenient for casual browsing, I still often use it as an alternative to grabber. It allows querying more than two tags at a time from danbooru and also matches your query to the words that are contained in tags as opposed to full tags, which is less precise, but I personally find it way more convenient. It has an api, described here, that can provide results in html, json and rss formats. It also used to be able to return plaintext and html list of links, but as of now this part of an api appears to be broken.

Bionus commented 5 years ago

It indeed looks like a good web-based Grabber-like website. But I'm not sure of what it would add compared to simply using Grabber on the sources individually.

You do mention a few interesting advantages:

  1. It allows querying more than two tags at a time from danbooru
  2. [It] matches your query to the words that are contained in tags as opposed to full tags, which is less precise

Which raises a few questions:

  1. Does that mean that it keeps a local dump of the whole Danbooru "database"? Because if they did as Grabber does, that is to query Danbooru for every request, that would be impossible. This leads to a question of "Grabber as a web service", which allows for similar caching.
  2. Same question, since Danbooru's API does not allow it, but keeping a local dump would.

Adding the source would be quite quick, especially given it supports JSON and RSS. The question is whether we should add such "aggregating" sources, if the added value is small.

Does it have other advantages than the two you listed? For web-based, I also see that it has a very clean and "no-bullshit" interface, but that doesn't really matter for a Grabber source.

RiabikovR commented 5 years ago

I always thought that they just keep a database, scraping the post data from the source sites and then the images themselves are downloaded by the user's browser, but that does not appear to be the case now that I've looked into it. It looks like there is an image server.

As for advantages, there aren't really any other than what I've already mentioned. It feels really quick as far as both search and navigation go, but that probably has to do with the interface. If it really has it's own image server, that could help with sankaku's throttling, but it appears to only use it when serving the webinterface, returning the source site links instead when answering the api calls, so that would probably be a bottleneck of it's own.

RiabikovR commented 5 years ago

So, it looks like the author has a local image database, and the images that are on there get pulled from the database, with the ones that are not being pulled from the original source.