fix(perf): Add indices to hot paths for repository querying (#526)

Author: brendan-kellam

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Changed
+- Improved repository query performance by adding db indices. [#526](https://github.com/sourcebot-dev/sourcebot/pull/526)
+
 ## [4.7.1] - 2025-09-19
 
 ### Fixed

packages/db/prisma/migrations/20250919221836_add_indexes_to_repo_tables/migration.sql

@@ -0,0 +1,5 @@
+-- CreateIndex
+CREATE INDEX "Repo_orgId_idx" ON "Repo"("orgId");
+
+-- CreateIndex
+CREATE INDEX "RepoToConnection_repoId_connectionId_idx" ON "RepoToConnection"("repoId", "connectionId");

packages/db/prisma/schema.prisma

@@ -71,6 +71,7 @@ model Repo {
   searchContexts SearchContext[]
 
   @@unique([external_id, external_codeHostUrl, orgId])
+  @@index([orgId])
 }
 
 model SearchContext {
@@ -119,6 +120,7 @@ model RepoToConnection {
   repoId Int
 
   @@id([connectionId, repoId])
+  @@index([repoId, connectionId])
 }
 
 model Invite {

packages/db/tools/scriptRunner.ts

@@ -4,6 +4,8 @@ import { migrateDuplicateConnections } from "./scripts/migrate-duplicate-connect
 import { injectAuditData } from "./scripts/inject-audit-data";
 import { confirmAction } from "./utils";
 import { createLogger } from "@sourcebot/logger";
+import { injectRepoData } from "./scripts/inject-repo-data";
+import { testRepoQueryPerf } from "./scripts/test-repo-query-perf";
 
 export interface Script {
     run: (prisma: PrismaClient) => Promise<void>;
@@ -12,6 +14,8 @@ export interface Script {
 export const scripts: Record<string, Script> = {
     "migrate-duplicate-connections": migrateDuplicateConnections,
     "inject-audit-data": injectAuditData,
+    "inject-repo-data": injectRepoData,
+    "test-repo-query-perf": testRepoQueryPerf,
 }
 
 const parser = new ArgumentParser();

packages/db/tools/scripts/inject-repo-data.ts

@@ -0,0 +1,64 @@
+import { Script } from "../scriptRunner";
+import { PrismaClient } from "../../dist";
+import { createLogger } from "@sourcebot/logger";
+
+const logger = createLogger('inject-repo-data');
+
+const NUM_REPOS = 100000;
+
+export const injectRepoData: Script = {
+    run: async (prisma: PrismaClient) => {
+        const orgId = 1;
+        
+        // Check if org exists
+        const org = await prisma.org.findUnique({
+            where: { id: orgId }
+        });
+
+        if (!org) {
+            await prisma.org.create({
+                data: {
+                    id: orgId,
+                    name: 'Test Org',
+                    domain: 'test-org.com'
+                }
+            });
+        }
+
+        const connection = await prisma.connection.create({
+            data: {
+                orgId,
+                name: 'test-connection',
+                connectionType: 'github',
+                config: {}
+            }
+        });
+
+
+        logger.info(`Creating ${NUM_REPOS} repos...`);
+
+        for (let i = 0; i < NUM_REPOS; i++) {
+            await prisma.repo.create({
+                data: {
+                    name: `test-repo-${i}`,
+                    isFork: false,
+                    isArchived: false,
+                    metadata: {},
+                    cloneUrl: `https://github.com/test-org/test-repo-${i}`,
+                    webUrl: `https://github.com/test-org/test-repo-${i}`,
+                    orgId,
+                    external_id: `test-repo-${i}`,
+                    external_codeHostType: 'github',
+                    external_codeHostUrl: 'https://github.com',
+                    connections: {
+                        create: {
+                            connectionId: connection.id,
+                        }
+                    }
+                }
+            });
+        }
+
+        logger.info(`Created ${NUM_REPOS} repos.`);
+    }
+}; 

packages/db/tools/scripts/test-repo-query-perf.ts

@@ -0,0 +1,28 @@
+import { Script } from "../scriptRunner";
+import { PrismaClient } from "../../dist";
+import { createLogger } from "@sourcebot/logger";
+
+const logger = createLogger('test-repo-query-perf');
+
+export const testRepoQueryPerf: Script = {
+    run: async (prisma: PrismaClient) => {
+
+
+        const start = Date.now();
+        const allRepos = await prisma.repo.findMany({
+            where: {
+                orgId: 1,
+            },
+            include: {
+                connections: {
+                    include: {
+                        connection: true,
+                    }
+                }
+            }
+        });
+
+        const durationMs = Date.now() - start;
+        logger.info(`Found ${allRepos.length} repos in ${durationMs}ms`);
+    }
+};