Ensure don't go over sqlite max like pattern length

core/autocomplete/cache.ts

@@ -1,14 +1,14 @@
 import { Mutex } from "async-mutex";
 import { open } from "sqlite";
 import sqlite3 from "sqlite3";
-import { DatabaseConnection } from "../indexing/refreshIndex.js";
+import { DatabaseConnection, truncateSqliteLikePattern } from "../indexing/refreshIndex.js";
 import { getTabAutocompleteCacheSqlitePath } from "../util/paths.js";
 
 export class AutocompleteLruCache {
   private static capacity = 1000;
   private mutex = new Mutex();
 
-  constructor(private db: DatabaseConnection) {}
+  constructor(private db: DatabaseConnection) { }
 
   static async get(): Promise<AutocompleteLruCache> {
     const db = await open({
@@ -37,7 +37,7 @@
     // Have to make sure we take the key with shortest length
     const result = await this.db.get(
       "SELECT key, value FROM cache WHERE ? LIKE key || '%' ORDER BY LENGTH(key) DESC LIMIT 1",
-      prefix,
+      truncateSqliteLikePattern(prefix),
     );
 
     // Validate that the cached compeltion is a valid completion for the prefix

core/indexing/CodeSnippetsIndex.ts

@@ -14,7 +14,7 @@ import {
   getParserForFile,
   getQueryForFile,
 } from "../util/treeSitter";
-import { DatabaseConnection, SqliteDb, tagToString } from "./refreshIndex";
+import { DatabaseConnection, SqliteDb, tagToString, truncateSqliteLikePattern } from "./refreshIndex";
 import {
   IndexResultType,
   MarkCompleteCallback,
@@ -28,7 +28,7 @@ export class CodeSnippetsCodebaseIndex implements CodebaseIndex {
   relativeExpectedTime: number = 1;
   artifactId = "codeSnippets";
 
-  constructor(private readonly ide: IDE) {}
+  constructor(private readonly ide: IDE) { }
 
   private static async _createTables(db: DatabaseConnection) {
     await db.exec(`CREATE TABLE IF NOT EXISTS code_snippets (
@@ -383,7 +383,7 @@ export class CodeSnippetsCodebaseIndex implements CodebaseIndex {
     const db = await SqliteDb.get();
     await CodeSnippetsCodebaseIndex._createTables(db);
 
-    const likePatterns = workspaceDirs.map((dir) => `${dir}%`);
+    const likePatterns = workspaceDirs.map((dir) => truncateSqliteLikePattern(`${dir}%`));
     const placeholders = likePatterns.map(() => "?").join(" OR path LIKE ");
 
     const query = `

core/indexing/refreshIndex.test.ts

@@ -0,0 +1,34 @@
+import { truncateToLastNBytes } from "./refreshIndex";
+
+describe("truncateToLastNBytes", () => {
+  it("should return full string if maxBytes greater than string byte length", () => {
+    const input = "Hello World";
+    const result = truncateToLastNBytes(input, 100);
+    expect(result).toBe("Hello World");
+  });
+
+  it("should truncate ASCII string correctly", () => {
+    const input = "Hello World";
+    const result = truncateToLastNBytes(input, 5);
+    expect(result).toBe("World");
+  });
+
+  it("should handle empty string", () => {
+    const input = "";
+    const result = truncateToLastNBytes(input, 5);
+    expect(result).toBe("");
+  });
+
+  it("should handle UTF-8 characters correctly", () => {
+    const input = "👋 Hello";
+    // 👋 is 4 bytes, space is 1 byte
+    const result = truncateToLastNBytes(input, 5);
+    expect(result).toBe("Hello");
+  });
+
+  it("should handle maxBytes of 0", () => {
+    const input = "Hello World";
+    const result = truncateToLastNBytes(input, 0);
+    expect(result).toBe("");
+  });
+});

core/indexing/refreshIndex.ts

@@ -469,3 +469,24 @@ export class GlobalCacheCodeBaseIndex implements CodebaseIndex {
     );
   }
 }
+
+const SQLITE_MAX_LIKE_PATTERN_LENGTH = 50000;
+
+export function truncateToLastNBytes(input: string, maxBytes: number): string {
+  let bytes = 0;
+  let startIndex = 0;
+
+  for (let i = input.length - 1; i >= 0; i--) {
+    bytes += new TextEncoder().encode(input[i]).length;
+    if (bytes > maxBytes) {
+      startIndex = i + 1;
+      break;
+    }
+  }
+
+  return input.substring(startIndex, input.length);
+}
+
+export function truncateSqliteLikePattern(input: string) {
+  return truncateToLastNBytes(input, SQLITE_MAX_LIKE_PATTERN_LENGTH);
+}