slaninas / tostr

Bot that forwards tweets to nostr

tostr (doesn't work due to mandatory login policy)

/ˈtəʊstə(r)/

Twitter to nostr. Bot that forwards tweets to nostr.

Reply to its message with !help and it will show you all available commands.

Powered by nostr-bot.

How to run using Docker

git clone https://github.com/slaninas/tostr/ && cd tostr
# Now add secret (hex private key) to config file, tune config if you wish to
./build_and_run.sh --clearnet|tor

Now the bot should be running and waiting for mentions. Just reply to its message to interact, see Commands. It relays only new tweets that were posted after you launched it.

Tor

In case --tor is used connections to both relay and Twitter should be going through tor. But if you need full anonymity please check yourself there are no leaks.

Known limitations/issues

• Heavy CPU load when starting the bot that already follows lot of users
• in update_user function, since value may not correspond to the previous until value (seems it breaks shortly after a new tweet is found), this may lead to tweets being forwarded twice or not at all
• twint is a Twitter scraper that currently works but who knows for how long
• Doesn't work for retweets by users you follow (twint has --retweets option but it's extremely slow 1.5 vs 30 s for some accounts)
• There are multiple processes spawned for each account check and relaying, twint also takes some time to process so it's slow, I tested it with 40 accounts and it took almost a minute to check if there were any new tweets for them. (under 4.5 s now, only twint process is spawned now)
• But it shows replies by people you follow (is that good or bad?)
• Tweets containing ' or " are not relayed

TODOs

• Check if tweets containing ' and " are forwarded
• After reconnection to relay, request events that were send when disconnected
• Check the relay's response after sending subscription request
• Better handling of handled event ids (add timestamp, remove old ones)
• Don't send set_metadata again after reconnect
• Cleanup
• Set timeout for connection
• Error handling
• If tweets fetch fails include the time inverval from failed attempt in the next check
• Proxy support
• Use existing websocket crate instead of spawning websocat process
• Find solution better than spawning nostril process
• Parallelization, async?
• Follow Twitters redirects and send original url to nostr
• Add proper logging
• Check timestamps that are used for twint's --since option, if the new tweets check takes too long some tweets may get ignored during the next check
• Read "hello" message from a config instead of using hardcoded one and post it when bot starts