fix(filesearch): limit max files to prevent OOM crashes

packages/core/src/core/turn.ts

@@ -367,7 +367,9 @@ export class Turn {
       fnCall.id ??
       `${fnCall.name}-${Date.now()}-${Math.random().toString(16).slice(2)}`;
     const name = fnCall.name || 'undefined_tool_name';
-    const args = (fnCall.args || {}) as Record<string, unknown>;
+    const args = sanitizeToolArgs(
+      (fnCall.args || {}) as Record<string, unknown>,
+    );
 
     const toolCallRequest: ToolCallRequestInfo = {
       callId,
@@ -399,3 +401,29 @@ function getCitations(resp: GenerateContentResponse): string[] {
       return citation.uri!;
     });
 }
+
+/**
+ * Sanitizes tool arguments to fix common model output issues.
+ * Specifically removes spaces between Chinese characters and numbers in file paths.
+ * This fixes issue #2032 where the model adds spaces like "测试 1 文件.txt" instead of "测试1文件.txt"
+ */
+function sanitizeToolArgs(
+  args: Record<string, unknown>,
+): Record<string, unknown> {
+  const sanitized: Record<string, unknown> = {};
+
+  for (const [key, value] of Object.entries(args)) {
+    if (typeof value === 'string') {
+      // Remove spaces between Chinese characters and numbers
+      // This regex matches: Chinese char + space + digit, or digit + space + Chinese char
+      sanitized[key] = value.replace(
+        /([\u4e00-\u9fa5])\s+(\d)|(\d)\s+([\u4e00-\u9fa5])/g,
+        '$1$2$3$4',
+      );
+    } else {
+      sanitized[key] = value;
+    }
+  }
+
+  return sanitized;
+}

packages/core/src/subagents/subagent.ts

@@ -774,7 +774,7 @@ export class SubAgentScope {
     const requests: ToolCallRequestInfo[] = authorizedCalls.map((fc) => {
       const toolName = String(fc.name || 'unknown');
       const callId = fc.id ?? `${fc.name}-${Date.now()}`;
-      const args = (fc.args ?? {}) as Record<string, unknown>;
+      const args = sanitizeToolArgs((fc.args ?? {}) as Record<string, unknown>);
       const request: ToolCallRequestInfo = {
         callId,
         name: toolName,
@@ -1002,3 +1002,29 @@ Important Rules:
     return finalPrompt;
   }
 }
+
+/**
+ * Sanitizes tool arguments to fix common model output issues.
+ * Specifically removes spaces between Chinese characters and numbers in file paths.
+ * This fixes issue #2032 where the model adds spaces like "测试 1 文件.txt" instead of "测试1文件.txt"
+ */
+function sanitizeToolArgs(
+  args: Record<string, unknown>,
+): Record<string, unknown> {
+  const sanitized: Record<string, unknown> = {};
+
+  for (const [key, value] of Object.entries(args)) {
+    if (typeof value === 'string') {
+      // Remove spaces between Chinese characters and numbers
+      // This regex matches: Chinese char + space + digit, or digit + space + Chinese char
+      sanitized[key] = value.replace(
+        /([\u4e00-\u9fa5])\s+(\d)|(\d)\s+([\u4e00-\u9fa5])/g,
+        '$1$2$3$4',
+      );
+    } else {
+      sanitized[key] = value;
+    }
+  }
+
+  return sanitized;
+}

packages/core/src/utils/filesearch/crawler.ts

@@ -8,6 +8,9 @@ import path from 'node:path';
 import { fdir } from 'fdir';
 import type { Ignore } from './ignore.js';
 import * as cache from './crawlCache.js';
+import { createDebugLogger } from '../debugLogger.js';
+
+const debugLogger = createDebugLogger('CRAWLER');
 
 export interface CrawlOptions {
   // The directory to start the crawl from.
@@ -21,13 +24,17 @@ export interface CrawlOptions {
   // Caching options.
   cache: boolean;
   cacheTtl: number;
+  // Maximum number of files to crawl to prevent OOM (default: 100000)
+  maxFiles?: number;
 }
 
 function toPosixPath(p: string) {
   return p.split(path.sep).join(path.posix.sep);
 }
 
 export async function crawl(options: CrawlOptions): Promise<string[]> {
+  const maxFiles = options.maxFiles ?? 100000;
+
   if (options.cache) {
     const cacheKey = cache.getCacheKey(
       options.crawlDirectory,
@@ -61,6 +68,15 @@ export async function crawl(options: CrawlOptions): Promise<string[]> {
     }
 
     results = await api.crawl(options.crawlDirectory).withPromise();
+
+    // Limit the number of files to prevent OOM in large projects
+    if (results.length > maxFiles) {
+      debugLogger.warn(
+        `Project contains ${results.length} files, limiting to ${maxFiles} to prevent memory issues. ` +
+          `Consider using .qwenignore to exclude unnecessary directories.`,
+      );
+      results = results.slice(0, maxFiles);
+    }
   } catch (_e) {
     // The directory probably doesn't exist.
     return [];