if you (or someone else) run your own bot, you'll need to:
- update the bot to the latest version, preferably via git
- run main.py once
- edit config.json so that "mention_handling" is 0
if i host your bot, and you'd like it to stop creating fake mentions, let me know!
it's now possible to prevent your bot from creating fake mentions.
preliminary testing using starwall-e
looks like it works! it's not generating any fake mentions anymore!
a new version of Mastodon.py has been released for the first time in months. mstdn-ebooks, OCRbot, etc. will be updated shortly to work with it. please don't try updating your Mastodon.py version yet.
info about the new release:
github issue: https://github.com/Lynnesbian/mstdn-ebooks/issues/15
a bug with OCRbot that prevented it from working with chinese text has now been fixed!
heads up: the free ebooks bots are being shut down in around 60 hours. this only applies to the bots i host, and not any hosted by other people.
if you'd like your ebooks bot to stay up, you have several options:
- you can pledge the ebooks tier on my patreon: https://patreon.com/lynnesbian this is US$2.50 per month.
- you can ask me for the bot credentials and files, and host it yourself, using this guide: https://cloud.lynnesbian.space/s/jozbRi69t4TpD95
- you can ask around and see if anyone would like to host your bot for you
thank you for understanding
okay i've decided that i'm turning off all of the free ebooks bots (that i host) in 72 hours. for the 5.5% of you who pay for yours, nothing will change, except they might reply a bit faster.
so what are your options? well:
- you can do nothing, and have them disappear. i'll be able to restore them if you decide to pay later.
- you can host it yourself. i'll give you the login credentials and the necessary files.
- you can find someone else to host if for you, either for less money or for free. i'm sure someone else is willing to do this for you.
here's a link to the tutorial on hosting the bot yourself: https://cloud.lynnesbian.space/s/jozbRi69t4TpD95
OCRbot now supports splitting captions over multiple replies. this means that you can tell it how many characters your instance has for the limit (e.g. 500) and i'll split posts up to ensure it doesn't hit the limit. this means that it's finally feasible to run OCRbot on instances that don't have ridiculously high character limits
one problem is that the error messages will be in english, and so will the instructions on how to use OCRbot. unfortunately i don't speak any other languages... i could rely on google translate i suppose, but that really seems like a bad idea.
OCRbot is far from perfect, and should not be relied on as the sole method of captioning images. It's much better and more reliable to caption images yourself, if at all possible. If it's too long to fit in the description field, consider posting it in a reply (or multiple replies).
Here is a list of caveats and other flaws to be aware of when using OCRbot: https://github.com/Lynnesbian/OCRbot/blob/master/README.md#caveats