482f72a4d1
execution: factor out chat completion process
2024-04-26 04:03:34 +02:00
d3567c3607
execution: handle undefined message in logUsedTokens
2024-04-26 04:02:09 +02:00
67a6e4d486
execution+configDefault: retrofit for tooll_calls api
2024-04-26 03:27:15 +02:00
370b7623b5
Dockerfile: chmod dist to 777 for config.ts compilation
2024-04-24 02:38:52 +02:00
b567e13f2a
functionManager: use json-schema-to-ts to derive arguments for OpenAIFunctions
2024-04-24 02:27:31 +02:00
d9bee2dcf2
check-quota: define integration types and context to any
2024-04-23 21:11:40 +02:00
3c10f4ed6f
Allow chatting with bot through direct messages
...
Added new command to make the bot see direct messages,
because by default these are not visible,
unless bot opens direct message with the user.
2024-04-23 21:10:52 +02:00
63cb52e7f4
Command: Add integration and context types
...
While discord api supports that, discord api types, which discord.js depends on does not.
2024-04-23 20:56:55 +02:00
0e5c8d22cc
functionManager: retrofit for tool calls api
2024-04-23 17:46:24 +02:00
91232e99a7
Update dependencies
2024-04-23 16:47:08 +02:00
dc01146ee8
ignore @everyone mentions
2023-12-25 16:09:35 +01:00
74359067d0
functionManager: don't return empty function array for openai
...
openai doesn't accept empty function parameter,
it was discovered after I removed the testing buildin function
2023-12-13 21:08:09 +01:00
24e85f535a
"Gracefully" exit when receiving sigint
2023-12-13 20:56:23 +01:00
02a5b0f7b9
Remove getTime model function
...
it lead to higher usage of tokens.
2023-12-13 17:42:59 +01:00
722862ded3
Add shell.nix
...
to development util on nixos system
2023-12-13 17:06:09 +01:00
c911e567c6
Update dependencies
2023-12-13 17:04:18 +01:00
6e1cb8c956
Merge branch 'refactor/scripts'
2023-10-12 12:01:52 +02:00
a186ba9e80
scripts/pushCommands: Refactor how commands are being read
...
this should reduce errors when reading commands
also this won't fail if reading one command fails
2023-10-12 11:55:53 +02:00
a05047ab7d
script/pushCommands: make sure the script exits
2023-09-30 14:47:10 +02:00
4729f7f563
/check-limit: rename to check-quota and change description
...
to match it to the new reality, limits are named quotas now
description of an option previously was longer than 100 characters
which discord api didn't like it
2023-09-30 14:46:12 +02:00
6a31473d22
scripts: move scripts to the project root directory
...
I believe, that these kind of scripts should reside there
instead of src directory
2023-09-30 14:46:12 +02:00
b7f6a5fe91
toOpenAIMessages: Add fallback to use user id on empty resolved name
...
Previously if someone named whose current mechanism
resolved to empty string ("Джо" for example)
then the openai would scream that the name
does not match their server side regex.
Meaning that the bot could not be used
in a given channel until that person
changes it's display name
or it will be forgotten for the bot.
2023-09-27 20:45:23 +02:00
ffa79e798e
Fix import grouping in execution.ts
...
I'm silly silly bad cutie
that doesn't check commits
when I push them
2023-09-27 20:30:08 +02:00
23ceca5cd3
Update OpenAI dependency
2023-09-27 17:14:17 +02:00
8ed2e758f8
Update dependencies
...
OpenAI package will be updated soon,
as it needs codebase migration
2023-09-27 15:52:30 +02:00
03a1c62cd5
Execution: Stringify response error on axios error
...
I'm sick of not understanding the 400 problem
2023-09-26 18:17:12 +02:00
6272c7f551
/check-limit: change description of recovery-for paramteter
...
to mach it better for the current default behaviour
2023-09-25 11:32:52 +02:00
49b074f98e
/check-limit: show when the user can use the bot again by default
...
fixes #12
2023-09-25 11:10:21 +02:00
80f4f18eab
quota/tokenCount: allow arbitrary multipliers for tokens
...
I didn't notice that OpenAI changed pricing again
2023-09-25 09:58:54 +02:00
552143e345
quota/tokenCount: Rename single variable name to something that has more sense
...
I forgot to rename it when I commited it last time ;-;
2023-09-25 09:30:52 +02:00
94992743e8
quota/tokenCount: fix "null from a database"
2023-09-21 21:51:03 +02:00
2629659ffc
Interactions: fix on failed reply to be ephemeral
...
fixes #17
2023-09-21 20:57:40 +02:00
96dd7bce95
Merge branch 'feature/quota'
2023-09-21 20:52:06 +02:00
18646b9dc6
config: fix imports not working correctly
...
this patch moves the rootDir of the typescript project up a directory
this moves all content in the dist directory inside the new src directory
I couldn't find other way
2023-09-21 20:51:38 +02:00
16fb74ec4b
quota/tokenCount: Implement the "request tokens as half" case
2023-09-21 20:42:06 +02:00
4abdaebf70
quota/tokenCount: I always forget about debug logs...
2023-09-21 20:38:50 +02:00
b2ee156028
/check-limit: fix wrong unit name in field name
2023-09-21 20:08:52 +02:00
6792c05959
Quota: add tokenCount QuotaEngine
2023-09-21 20:07:35 +02:00
e194c1d81a
messageCount: Add docs
2023-09-21 09:31:23 +02:00
339ef06ff9
Quota: Refactor how Quotas are being handled
...
also renamed limits to quota
I believe this new approach would allow me and bot hosters
to add, implement or change the quota behavior more easily.
Reimplemented the currently existing "Message count" limit
to use the new IQuota, refactoring a code *a little*.
2023-09-21 07:11:15 +02:00
46bb5c867d
/summon: fix command being canceled wrongly in execution
...
the problem was that undefined value was compared to number,
making the expression return true
2023-09-19 20:30:55 +02:00
74fe8e3e8b
Init: Remove debug logger
2023-09-18 13:33:41 +02:00
29318592b0
Execution: Don't execute the summon command if bot is writing already
...
fixes #14
2023-09-18 12:44:55 +02:00
b6eb476162
Fix typo again
2023-09-18 12:26:29 +02:00
1ae09945c0
Add Dockerfile
2023-09-18 11:26:37 +02:00
58a054d137
Prisma: fix typo [breaking change]
...
also: bump version because of breaking change
2023-09-18 11:24:35 +02:00
7ff4abc3c0
Configuration: recactor how it is written and handled
2023-09-18 11:22:10 +02:00
13e993b964
meta: change eslint styling and tsconfig project config
2023-09-18 10:41:54 +02:00
e6d6764e34
add user request limit to config, rename previous limit to read limit
2023-09-13 04:32:14 +02:00
0931fe103d
Wait some time before retrying an OpenAI request
...
OpenAI recommended prectice is to wait some time before retrying
the request that ended with http 5xx code
the bot was not doing that before, now it does
this might fix some issues when we were retrying the request too fast
2023-08-22 21:22:18 +02:00
