@customize-agent/knowledge 4.0.24 → 4.0.26

package/dist/core/multi-project-manager.js

@@ -62,7 +62,7 @@ export class MultiProjectManager {
     }
     async search(projectRoot, query, options = {}) {
         const limit = options.limit ?? 10;
-        const scope = options.scope ?? 'all';
+        const scope = options.scope ?? 'project';
         const project = await this.getProject(projectRoot);
         await this.ensureFreshForSearch(projectRoot, project);
         if (scope === 'project')
@@ -83,7 +83,7 @@ export class MultiProjectManager {
         };
     }
     async semanticSearch(projectRoot, query, options = {}) {
-        const scope = options.scope ?? 'all';
+        const scope = options.scope ?? 'project';
         const project = await this.getProject(projectRoot);
         await this.ensureFreshForSearch(projectRoot, project);
         if (scope === 'project') {

package/dist/extraction/ocr-providers.d.ts

@@ -38,6 +38,7 @@ export declare class TesseractJsProvider implements OcrProvider {
     private _available;
     private worker;
     private workerPromise;
+    private workerLock;
     private warnings;
     get available(): boolean;
     recognize(input: {

package/dist/extraction/ocr-providers.js

@@ -11,7 +11,11 @@ import { fileURLToPath } from 'node:url';
 import { resolveAndImport, resolvePackage } from './module-resolver.js';
 // ─── 路径工具 ───────────────────────────────────────────────────
 const knowledgeDir = path.dirname(fileURLToPath(import.meta.url));
-const OCR_NATIVE_NOISE_PATTERNS = [/^Image too small to scale!!/u, /^Line cannot be recognized!!$/u];
+const OCR_NATIVE_NOISE_PATTERNS = [
+    /Image too small to scale/u,
+    /Line cannot be recognized/u,
+    /empty image/iu
+];
 const OCR_NOISE_SUPPRESSION_KEY = Symbol.for('customize-agent.ocr-noise-suppression');
 function ocrNoiseSuppressionState() {
     const globalState = globalThis;
@@ -30,8 +34,7 @@ function isNativeOcrNoise(chunk) {
     const text = Buffer.isBuffer(chunk) ? chunk.toString('utf8') : chunk instanceof Uint8Array ? Buffer.from(chunk).toString('utf8') : typeof chunk === 'string' ? chunk : '';
     if (!text)
         return false;
-    const lines = text.split(/\r?\n/u).map(line => line.trim()).filter(Boolean);
-    return lines.length > 0 && lines.every(line => OCR_NATIVE_NOISE_PATTERNS.some(pattern => pattern.test(line)));
+    return OCR_NATIVE_NOISE_PATTERNS.some(pattern => pattern.test(text));
 }
 async function suppressNativeOcrNoise(operation) {
     const state = ocrNoiseSuppressionState();
@@ -64,12 +67,48 @@ async function suppressNativeOcrNoise(operation) {
         }
     }
 }
-// ─── Tesseract.js Provider（主 OCR 引擎，跨平台） ──────────────
+// 临时屏蔽 Tesseract 底层 WASM 污染日志
+function withSilencedConsole(fn) {
+    const originalLog = console.log;
+    const originalWarn = console.warn;
+    const originalError = console.error;
+    const filter = (orig) => (...args) => {
+        const msg = String(args[0] || '');
+        if (msg.includes('Line cannot be recognized') || msg.includes('Image too small to scale')) {
+            return;
+        }
+        orig(...args);
+    };
+    console.log = filter(originalLog);
+    console.warn = filter(originalWarn);
+    console.error = filter(originalError);
+    try {
+        const result = fn();
+        if (result instanceof Promise) {
+            return result.finally(() => {
+                console.log = originalLog;
+                console.warn = originalWarn;
+                console.error = originalError;
+            });
+        }
+        console.log = originalLog;
+        console.warn = originalWarn;
+        console.error = originalError;
+        return result;
+    }
+    catch (error) {
+        console.log = originalLog;
+        console.warn = originalWarn;
+        console.error = originalError;
+        throw error;
+    }
+}
 export class TesseractJsProvider {
     id = 'tesseract.js';
     _available = null;
     worker = null;
     workerPromise = null;
+    workerLock = Promise.resolve();
     warnings = [];
     get available() {
         if (this._available !== null)
@@ -115,8 +154,19 @@ export class TesseractJsProvider {
             return { text: '', confidence: 0, regions: [], warnings: [`image too small for OCR: ${width}x${height}`] };
         }
         try {
-            const worker = await this.getWorker();
-            const result = await suppressNativeOcrNoise(() => worker.recognize(pngPath));
+            const worker = await withSilencedConsole(() => this.getWorker());
+            let unlock;
+            const nextLock = new Promise(resolve => { unlock = resolve; });
+            const currentLock = this.workerLock;
+            this.workerLock = currentLock.then(() => nextLock).catch(() => nextLock);
+            await currentLock;
+            let result;
+            try {
+                result = await withSilencedConsole(() => suppressNativeOcrNoise(() => worker.recognize(pngPath)));
+            }
+            finally {
+                unlock();
+            }
             const text = (result.data.text ?? '').trim();
             const lines = (result.data.lines ?? []);
             const regions = lines

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@customize-agent/knowledge",
-  "version": "4.0.24",
+  "version": "4.0.26",
   "description": "Local knowledge base infrastructure for customize-agent",
   "type": "module",
   "main": "./dist/index.js",