Review Agent

packages/agents/reviewAgent/.gitignore

@@ -0,0 +1,136 @@
+# Logs
+logs
+*.log
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+lerna-debug.log*
+.pnpm-debug.log*
+
+# Diagnostic reports (https://nodejs.org/api/report.html)
+report.[0-9]*.[0-9]*.[0-9]*.[0-9]*.json
+
+# Runtime data
+pids
+*.pid
+*.seed
+*.pid.lock
+
+# Directory for instrumented libs generated by jscoverage/JSCover
+lib-cov
+
+# Coverage directory used by tools like istanbul
+coverage
+*.lcov
+
+# nyc test coverage
+.nyc_output
+
+# Grunt intermediate storage (https://gruntjs.com/creating-plugins#storing-task-files)
+.grunt
+
+# Bower dependency directory (https://bower.io/)
+bower_components
+
+# node-waf configuration
+.lock-wscript
+
+# Compiled binary addons (https://nodejs.org/api/addons.html)
+build/Release
+
+# Dependency directories
+node_modules/
+jspm_packages/
+
+# Snowpack dependency directory (https://snowpack.dev/)
+web_modules/
+
+# TypeScript cache
+*.tsbuildinfo
+
+# Optional npm cache directory
+.npm
+
+# Optional eslint cache
+.eslintcache
+
+# Optional stylelint cache
+.stylelintcache
+
+# Microbundle cache
+.rpt2_cache/
+.rts2_cache_cjs/
+.rts2_cache_es/
+.rts2_cache_umd/
+
+# Optional REPL history
+.node_repl_history
+
+# Output of 'npm pack'
+*.tgz
+
+# Yarn Integrity file
+.yarn-integrity
+
+# dotenv environment variable files
+.env
+.env.development.local
+.env.test.local
+.env.production.local
+.env.local
+
+# parcel-bundler cache (https://parceljs.org/)
+.cache
+.parcel-cache
+
+# Next.js build output
+.next
+out
+
+# Nuxt.js build / generate output
+.nuxt
+dist
+
+# Gatsby files
+.cache/
+# Comment in the public line in if your project uses Gatsby and not Next.js
+# https://nextjs.org/blog/next-9-1#public-directory-support
+# public
+
+# vuepress build output
+.vuepress/dist
+
+# vuepress v2.x temp and cache directory
+.temp
+.cache
+
+# vitepress build output
+**/.vitepress/dist
+
+# vitepress cache directory
+**/.vitepress/cache
+
+# Docusaurus cache and generated files
+.docusaurus
+
+# Serverless directories
+.serverless/
+
+# FuseBox cache
+.fusebox/
+
+# DynamoDB Local files
+.dynamodb/
+
+# TernJS port file
+.tern-port
+
+# Stores VSCode versions used for testing VSCode extensions
+.vscode-test
+
+# yarn v2
+.yarn/cache
+.yarn/unplugged
+.yarn/build-state.yml
+.yarn/install-state.gz
+.pnp.*

packages/agents/reviewAgent/app.ts

@@ -0,0 +1,66 @@
+import dotenv from 'dotenv';
+import { App, Octokit } from "octokit";
+import { createNodeMiddleware } from "@octokit/webhooks";
+import fs from "fs";
+import http from "http";
+import { WebhookEventDefinition } from "@octokit/webhooks/types";
+import { generate_pr_reviews } from './nodes/generate_pr_reviews.js';
+import { github_push_pr_reviews } from './nodes/github_push_pr_reviews.js';
+import { github_pr_parser } from './nodes/github_pr_parser.js';
+
+dotenv.config();
+const appId = process.env.APP_ID as string;
+const webhookSecret = process.env.WEBHOOK_SECRET as string;
+const privateKeyPath = process.env.PRIVATE_KEY_PATH as string;
+
+const privateKey = fs.readFileSync(privateKeyPath, "utf8");
+
+const app = new App({
+    appId: appId,
+    privateKey: privateKey,
+    webhooks: {
+        secret: webhookSecret
+    },
+});
+
+const rules = [
+    "Do NOT provide general feedback, summaries, explanations of changes, or praises for making good additions.",
+    "Do NOT provide any advice that is not actionable or directly related to the changes.",
+    "Focus solely on offering specific, objective insights based on the given context and refrain from making broad comments about potential impacts on the system or question intentions behind the changes.",
+    "Keep comments concise and to the point. Every comment must highlight a specific issue and provide a clear and actionable solution to the developer.",
+    "If there are no issues found on a line range, do NOT respond with any comments. This includes comments such as \"No issues found\" or \"LGTM\"."
+]
+
+async function handlePullRequestOpened({
+    octokit,
+    payload,
+}: {
+    octokit: Octokit;
+    payload: WebhookEventDefinition<"pull-request-opened"> | WebhookEventDefinition<"pull-request-synchronize">;
+}) {
+    console.log(`Received a pull request event for #${payload.pull_request.number}`);
+
+    const prPayload = await github_pr_parser(octokit, payload);
+    const fileDiffReviews = await generate_pr_reviews(prPayload, rules);
+    await github_push_pr_reviews(app, prPayload, fileDiffReviews);
+}
+
+app.webhooks.on("pull_request.opened", handlePullRequestOpened);
+app.webhooks.on("pull_request.synchronize", handlePullRequestOpened);
+
+app.webhooks.onError((error) => {
+    console.error(error);
+});
+
+
+const port = 3050;
+const host = 'localhost';
+const path = "/api/webhook";
+const localWebhookUrl = `http://${host}:${port}${path}`;
+
+const middleware = createNodeMiddleware(app.webhooks, { path });
+
+http.createServer(middleware).listen(port, () => {
+    console.log(`Server is listening for events at: ${localWebhookUrl}`);
+    console.log('Press Ctrl + C to quit.')
+});

packages/agents/reviewAgent/nodes/fetch_file_content.ts

@@ -0,0 +1,49 @@
+import { sourcebot_context, sourcebot_pr_payload } from "../types.js";
+import { z } from "zod";
+
+// TODO: use original Sourcebot schemas instead of redefining here
+const fileSourceResponseSchema = z.object({
+    source: z.string(),
+    language: z.string(),
+});
+
+const base64Decode = (base64: string): string => {
+    const binString = atob(base64);
+    return Buffer.from(Uint8Array.from(binString, (m) => m.codePointAt(0)!).buffer).toString();
+}
+
+export const fetch_file_content = async (pr_payload: sourcebot_pr_payload, filename: string): Promise<sourcebot_context> => {
+    console.log("Executing fetch_file_content");
+
+    const fileSourceRequest = {
+        fileName: filename,
+        repository: pr_payload.hostDomain + "/" + pr_payload.owner + "/" + pr_payload.repo,
+    }
+    console.log(JSON.stringify(fileSourceRequest, null, 2));
+
+    const response = await fetch('http://localhost:3000/api/source', {
+        method: 'POST',
+        headers: {
+            'Content-Type': 'application/json',
+            'X-Org-Domain': '~'
+        },
+        body: JSON.stringify(fileSourceRequest)
+    });
+
+    if (!response.ok) {
+        throw new Error(`Failed to fetch file content: ${response.statusText}`);
+    }
+
+    const responseData = await response.json();
+    const fileSourceResponse = fileSourceResponseSchema.parse(responseData);
+    const fileContent = base64Decode(fileSourceResponse.source);
+
+    const fileContentContext: sourcebot_context = {
+        type: "file_content",
+        description: `The content of the file ${filename}`,
+        context: fileContent,
+    }
+
+    console.log("Completed fetch_file_content");
+    return fileContentContext;
+}

packages/agents/reviewAgent/nodes/generate_diff_review_prompt.ts

@@ -0,0 +1,44 @@
+import { sourcebot_diff, sourcebot_context, sourcebot_diff_review_schema } from "../types.js";
+import { zodToJsonSchema } from "zod-to-json-schema";
+
+export const generate_diff_review_prompt = async (diff: sourcebot_diff, context: sourcebot_context[], rules: string[]) => {
+    console.log("Executing generate_diff_review_prompt");
+
+    const prompt = `
+    You are an expert software engineer that excells at reviewing code changes. Given the input, additional context, and rules defined below, review the code changes and provide a detailed review. The review you provide
+    must conform to all of the rules defined below. The output format of your review must conform to the output format defined below.
+
+    # Input
+
+    The input is the old and new code snippets, which represent a single hunk from a git diff. The old code snippet is the code before the changes were made, and the new code snippet is the code after the changes were made. Each code snippet
+    is a sequence of lines each with a line number.
+
+    ## Old Code Snippet
+
+    \`\`\`
+    ${diff.oldSnippet}
+    \`\`\`
+
+    ## New Code Snippet
+
+    \`\`\`
+    ${diff.newSnippet}
+    \`\`\`
+
+    # Additional Context
+
+    ${context.map(c => `${c.type}: ${c.description}\n\n${c.context}`).join("\n\n----------------------\n\n")}
+
+    # Rules
+
+    - ${rules.join("\n- ")}
+
+    # Output Format (JSON Schema)
+    The output must be a valid JSON object that conforms to the following JSON schema. Do NOT respond with anything other than the JSON object. Do NOT respond with
+    the JSON object in a markdown code block.
+    ${JSON.stringify(zodToJsonSchema(sourcebot_diff_review_schema), null, 2)}
+    `;
+
+    console.log("Completed generate_diff_review_prompt");
+    return prompt;
+}

packages/agents/reviewAgent/nodes/generate_pr_reviews.ts

@@ -0,0 +1,46 @@
+import { sourcebot_pr_payload, sourcebot_diff_review, sourcebot_file_diff_review, sourcebot_context } from "../types.js";
+import { generate_diff_review_prompt } from "./generate_diff_review_prompt.js";
+import { invoke_diff_review_llm } from "./invoke_diff_review_llm.js";
+import { fetch_file_content } from "./fetch_file_content.js";
+
+export const generate_pr_reviews = async (pr_payload: sourcebot_pr_payload, rules: string[]): Promise<sourcebot_file_diff_review[]> => {
+    console.log("Executing generate_pr_reviews");
+
+    const file_diff_reviews: sourcebot_file_diff_review[] = [];
+    for (const file_diff of pr_payload.file_diffs) {
+        const reviews: sourcebot_diff_review[] = [];
+
+        for (const diff of file_diff.diffs) {
+            const fileContentContext = await fetch_file_content(pr_payload, file_diff.to);
+            const context: sourcebot_context[] = [
+                {
+                    type: "pr_title",
+                    description: "The title of the pull request",
+                    context: pr_payload.title,
+                },
+                {
+                    type: "pr_description",
+                    description: "The description of the pull request",
+                    context: pr_payload.description,
+                },
+                fileContentContext,
+            ];
+
+            const prompt = await generate_diff_review_prompt(diff, context, rules);
+            console.log(prompt);
+
+            const diffReview = await invoke_diff_review_llm(prompt, file_diff.to);
+            reviews.push(diffReview);
+        }
+
+        if (reviews.length > 0) {
+            file_diff_reviews.push({
+                filename: file_diff.to,
+                reviews: reviews,
+            });
+        }
+    }
+
+    console.log("Completed generate_pr_reviews");
+    return file_diff_reviews;
+}