ndrezn / wikipedia-pic-of-the-day Goto Github PK

Since Twitter is making API access paid, keep the bot running on other social media.

Occasionally, the POTD has multiple images, like on July 25 2022: https://en.wikipedia.org/wiki/Template:POTD/2022-07-25

Update the README.md with full documentation of the mechanics of the bot, dependencies, attribution, etc.

You might want to document on GitHub how the attribution works. Schwede66 3:20 pm, Today (UTC−4)

Thanks Schwede 🙂

Occasionally, videos come up, for example from August 5 2022: https://en.wikipedia.org/wiki/Template:POTD/2022-08-05

This one is kinda gross. But! It is a pic of the day -- so -- we must post it! The file type tends to be .webm.

Occassionally authors for photographs are annotated using templates, see: https://commons.wikimedia.org/wiki/File:Red-and-green_macaw_(Ara_chloropterus)_juvenile.JPG on https://en.wikipedia.org/wiki/Template:POTD/2022-09-07 and https://en.wikipedia.org/wiki/Template:POTD/2022-09-06. These templates should be parsed and converted to strings as part of the attribution reply. For now attribution in these cases only includes the link to the photograph.

I see an issue with posting to potd_test. This is a shame, because potd_test is a nice way to make sure this bot is ship-shape without posting to main, specifically a json.decoder.JSONDecoder error. I will investigate but I am suspicious that this is caused by my Twitter API permissions not yet propagating for the new account.

Right now, captions are generated by concatanating sentences until the 240 character limit is reached. This works fine, but occassionally the first sentence is >240 characters and there's no prose attached to the tweet. In this circumstance, the caption should be tokenized by word rather than character and split in the middle of the sentence, with the rest of the sentence added to the reply caption. See https://en.wikipedia.org/wiki/Template:POTD/2022-09-26 for an example.

As noted by andrybak over on Wikipedia:

for better compliance with licenses of the images (mostly variants of Creative Commons) I recommend that you include a link to the file page in the tweets. Either to Wikipedia https://en.wikipedia.org/wiki/File:Le_Voyage_dans_la_lune_(black_and_white,_1902).webm or Wikimedia Commons: https://commons.wikimedia.org/wiki/File:Le_Voyage_dans_la_lune_(black_and_white,_1902).webm. Quiet a few of POTD pictures are in the public domain (including today's Template:POTD/2022-09-01), but some are published under CC licenses that require attribution. For example, Template:POTD/2022-08-27, license of which was violated by https://twitter.com/wiki_potd/status/1563476127624007681. —⁠andrybak (talk)

I think this could be resolved by adding a second reply including the name of the author and a link to the image on WP or WM.

Video support should be updated to check video length. If >2:20, crop down to 2:30 and add a fade. If <2:20, leave as-is. Twitter has a max length of 2:20 for videos.

Right now, videos <2:20 will have the last frame extended for the length it takes to fill the 2:20, which is pretty sloppy...

Since Twitter is making API access paid, keep the bot running on other social media.