fix: max prompt length per model

package.json

@@ -1,6 +1,6 @@
 {
   "name": "code-review-gpt",
-  "version": "0.0.25",
+  "version": "0.0.26",
   "description": "Your AI code reviewer",
   "bin": {
     "code-review-gpt": "./dist/index.js"

src/review/constants.ts

@@ -54,7 +54,24 @@ Please summarise the review using 3 emojis.
 export const signOff =
   "#### Powered by [Code Review GPT](https://github.com/mattzcarey/code-review-gpt)";
 
-export const maxPromptLength = 30000; //max is 8k tokens which is about 40k characters
+export const modelInfo = [
+  {
+    model: "gpt4",
+    maxPromptLength: 21000, //8k tokens
+  },
+  {
+    model: "gpt-4-32k",
+    maxPromptLength: 90000, //32k tokens
+  },
+  {
+    model: "gpt-3.5-turbo",
+    maxPromptLength: 9000, //4k tokens
+  },
+  {
+    model: "gpt-3.5-turbo-16k",
+    maxPromptLength: 45000, //16k tokens
+  },
+]; // Response needs about 1k tokens ~= 3k characters
 
 export const supportedFiles = new Set([
   ".js",

src/review/index.ts

@@ -1,7 +1,8 @@
 import { commentOnPR } from "./ci/commentOnPR";
+import { askAI } from "./llm/askAI";
 import { constructPromptsArray } from "./prompt/constructPrompt";
 import { getFileNames } from "./prompt/getFileNames";
-import { askAI } from "./llm/askAI";
+import { getMaxPromptLength } from "./prompt/getMaxPromptLength";
 
 interface ReviewArgs {
   [x: string]: unknown;
@@ -14,8 +15,10 @@ export const review = async (yargs: ReviewArgs) => {
   const isCi = yargs.ci;
   const modelName = yargs.model as string;
 
+  const maxPromptLength = getMaxPromptLength(modelName);
+
   const fileNames = await getFileNames(isCi);
-  const prompts = await constructPromptsArray(fileNames);
+  const prompts = await constructPromptsArray(fileNames, maxPromptLength);
 
   const response = await askAI(prompts, modelName);
 

src/review/prompt/constructPrompt.ts

@@ -1,11 +1,11 @@
-import {
-  instructionPrompt,
-  filePromptTemplate,
-  maxPromptLength,
-} from "../constants";
 import { readFile } from "fs/promises";
+import { filePromptTemplate, instructionPrompt } from "../constants";
 
-const appendToLastPrompt = (prompts: string[], text: string): string[] => {
+const appendToLastPrompt = (
+  prompts: string[],
+  text: string,
+  maxPromptLength: number
+): string[] => {
   const newPrompts = [...prompts];
   const lastPromptIndex = newPrompts.length - 1;
 
@@ -18,8 +18,20 @@ const appendToLastPrompt = (prompts: string[], text: string): string[] => {
   return newPrompts;
 };
 
+const splitFileContentsIntoChunks = (
+  fileContents: string,
+  maxChunkSize: number
+): string[] => {
+  const chunks = [];
+  for (let i = 0; i < fileContents.length; i += maxChunkSize) {
+    chunks.push(fileContents.slice(i, i + maxChunkSize));
+  }
+  return chunks;
+};
+
 export const constructPromptsArray = async (
-  fileNames: string[]
+  fileNames: string[],
+  maxPromptLength: number
 ): Promise<string[]> => {
   console.info("Constructing prompt...");
 
@@ -28,18 +40,27 @@ export const constructPromptsArray = async (
   for (const fileName of fileNames) {
     try {
       const fileContents = await readFile(fileName, "utf8");
-      const filePrompt = filePromptTemplate
-        .replace("{fileName}", fileName)
-        .replace("{fileContents}", fileContents);
-
-      if (filePrompt.length > maxPromptLength) {
-        console.warn(
-          `The file ${fileName} is too large and may cause unexpected behavior such as cut off responses. Skipping this file.`
-        );
-        continue;
-      }
+      const fileChunks = splitFileContentsIntoChunks(
+        fileContents,
+        maxPromptLength
+      );
 
-      prompts = appendToLastPrompt(prompts, filePrompt);
+      for (let i = 0; i < fileChunks.length; i++) {
+        const fileChunk = fileChunks[i];
+        let chunkFileName = fileName;
+        if (i !== 0) {
+          console.warn(
+            `File ${fileName} is too large to fit in a single prompt. Splitting into ${fileChunks.length} chunks.`
+          );
+          chunkFileName += ` Continued part ${i + 1}`;
+        }
+
+        const filePrompt = filePromptTemplate
+          .replace("{fileName}", chunkFileName)
+          .replace("{fileContents}", fileChunk);
+
+        prompts = appendToLastPrompt(prompts, filePrompt, maxPromptLength);
+      }
     } catch (error) {
       console.error(`Failed to process file ${fileName}:`, error);
       throw error;

src/review/prompt/getMaxPromptLength.ts

@@ -0,0 +1,17 @@
+import { modelInfo } from "../constants";
+
+export const getMaxPromptLength = (modelName: string): number => {
+  const maxPromptLength = modelInfo.find(
+    (info) => info.model === modelName
+  )?.maxPromptLength;
+
+  if (!maxPromptLength) {
+    throw new Error(
+      `Model ${modelName} not found. Please choose one of ${modelInfo.map(
+        (info) => info.model
+      )} or make a PR to add a new model.`
+    );
+  }
+
+  return maxPromptLength;
+};