Server-side token counting and message trimming #19
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![sweep-ai[bot]](https://avatars.githubusercontent.com/in/307814?s=60&v=4){kind=small}
Sandbox Executions |
Apply Sweep Rules to your PR?
|
4 tasks
❌ Deploy Preview for endless-chat failed.
|
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
|
|
Hi there! 👋 Thanks for opening a PR. 🎉 To get the most out of Senior Dev, please sign up in our Web App, connect your GitHub account, and add/join your organization CNSeniorious000. After that, you will receive code reviews beginning on your next opened PR. 🚀 |
Pull Request ReportHey there! 👋 Here's a report on the changes made in the Changes
Suggestions to Improve Code
Bugs
Improvements
RatingI would rate the code a 7 out of 10 based on the following criteria:
Keep up the good work! If you have any questions or need further assistance, feel free to ask. 😄 Code DiffsDiff-1- import { ProxyAgent, fetch } from 'undici'
+ import { ProxyAgent, fetch } from 'undici'Diff-2- import { generatePayload, parseOpenAIStream } from '@/utils/openAI'
+ import { generatePayload, parseOpenAIStream } from '@/utils/openAI'Diff-3- import { verifySignature } from '@/utils/auth'
+ import { verifySignature } from '@/utils/auth'Diff-4- import type { APIRoute } from 'astro'
+ import type { APIRoute } from 'astro'Diff-5- const apiKey = import.meta.env.OPENAI_API_KEY
+ const apiKey = import.meta.env.OPENAI_API_KEYDiff-6- const httpsProxy = import.meta.env.HTTPS_PROXY
+ const httpsProxy = import.meta.env.HTTPS_PROXYDiff-7- const baseUrl = ((import.meta.env.OPENAI_API_BASE_URL) || 'https://api.openai.com').trim().replace(/\/$/, '')
+ const baseUrl = ((import.meta.env.OPENAI_API_BASE_URL) || 'https://api.openai.com').trim().replace(/\/$/, '')Diff-8- const sitePassword = import.meta.env.SITE_PASSWORD
+ const sitePassword = import.meta.env.SITE_PASSWORDDiff-9+ const ua = import.meta.env.UNDICI_UADiff-10+ if (!messages) {
+ return new Response(JSON.stringify({
+ error: {
+ message: 'No input text.',
+ },
+ }), { status: 400 })
+ }Diff-11+ if (sitePassword && sitePassword !== pass) {
+ return new Response(JSON.stringify({
+ error: {
+ message: 'Invalid password.',
+ },
+ }), { status: 401 })
+ }Diff-12+ if (import.meta.env.PROD && !await verifySignature({ t: time, m: messages?.[messages.length - 1]?.content || '' }, sign)) {
+ return new Response(JSON.stringify({
+ error: {
+ message: 'Invalid signature.',
+ },
+ }), { status: 401 })
+ }Diff-13+ const encoder = new Encoder('en');Diff-14+ const minMessages = 5;
+ const maxTokens = 1024;Diff-15+ function countTokens(message) {
+ return encoder.encode(message).length;
+ }Diff-16+ let totalTokens = messages.reduce((sum, msg) => sum + countTokens(msg.content), 0);
+ while (totalTokens > maxTokens && messages.length > minMessages) {
+ totalTokens -= countTokens(messages[0].content);
+ messages.shift();
+ }Diff-17+ const initOptions = generatePayload(request.headers.get('Authorization') ?? `Bearer ${apiKey}`, messages, model);Diff-18+ if (baseUrl) request.headers.forEach((val, key) => (FORWARD_HEADERS.includes(key) || key.startsWith('sec-') || key |
❌ Deploy Preview for chat-for-free failed.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR Feedback (click)
Description
This pull request includes changes to the
generate.tsfile in thesrc/pages/apidirectory. The changes involve adding server-side token counting and message trimming logic to the existing code.Summary
ProxyAgentandfetchfrom theundicilibrary.generatePayloadandparseOpenAIStreamfunctions from the@/utils/openAImodule.verifySignaturefunction from the@/utils/authmodule.APIRoutetype from theastromodule.apiKey,httpsProxy,baseUrl,sitePassword, anduausing values fromimport.meta.env.FORWARD_HEADERScontaining headers to be forwarded.postfunction that handles the API request.messagesfield and returned an error response if it is empty.encoderinstance from thetiktoken-jslibrary.minMessagesandmaxTokensvariables for message truncation.countTokensfunction to count the tokens in a message using theencoderinstance.maxTokenslimit.initOptionsvariable with the newmessagesarray.headersobject with the appropriate values.baseUrland forwarded specific headers if it exists.uaheader if it is defined.httpsProxyand updated theinitOptionsobject with a newdispatcherproperty if it exists.Fixes #17.
🎉 Latest improvements to Sweep:
ropelibrary to refactor Python! Check out Large Language Models are Bad at Refactoring Code. To have Sweep refactor your code, trysweep: Refactor <your_file>.py!💡 To get Sweep to edit this pull request, you can: