@customize-agent/knowledge 4.0.23 → 4.0.25

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/content-extractor.d.ts

@@ -48,6 +48,7 @@ export declare class ContentExtractor {
     private extractOfficeZip;
     private extractRasterImage;
     private tryPaddleOcrLayout;
+    private filterOcrNativeNoise;
     private formatBoundingBox;
     private validateRasterImage;
     private extractPdf;

package/dist/extraction/content-extractor.js

@@ -7,6 +7,7 @@ import { resolveAndImport, resolvePackage } from './module-resolver.js';
 import { createOcrProvider } from './ocr-providers.js';
 const CAD_INTERNAL_TOKEN_RE = /\b(?:TDbPipe|TDbPipeValve|TDbPipeFitting|TDbWellh|AcDb\w+|Dwg\w+|ObjectId|Handle|ByLayer|Continuous|Model|Layout\d*)\b/giu;
 const CAD_INTERNAL_LINE_RE = /^(?:TDb\w+|AcDb\w+|[A-F0-9]{8,}|\d+|Model|Layout\d*|ByLayer|Continuous)$/iu;
+const OCR_NATIVE_NOISE_PATTERNS = [/^Image too small to scale!!/u, /^Line cannot be recognized!!$/u];
 /** 文件内容提取器，支持文档、表格、图片、CAD 等多种文件格式的内容抽取 */
 export class ContentExtractor {
     async extract(file) {
@@ -977,18 +978,22 @@ export class ContentExtractor {
         if (!command)
             return undefined;
         const result = spawnSync(command, [filePath], { encoding: 'utf8', timeout: 0, maxBuffer: 50 * 1024 * 1024, shell: true });
-        if (result.status !== 0 || !result.stdout.trim())
+        const stdout = this.filterOcrNativeNoise(result.stdout || '');
+        if (result.status !== 0 || !stdout.trim())
             return undefined;
         try {
-            const parsed = JSON.parse(result.stdout);
+            const parsed = JSON.parse(stdout);
             const lines = parsed.map((region, index) => `区域 ${index + 1} [${region.type ?? 'text'}] ${this.formatBoundingBox(region.bbox)}: ${region.text ?? ''}`);
             return { text: ['OCR 版面分析区域:', ...lines].join('\n'), regionCount: lines.length };
         }
         catch {
-            const lines = result.stdout.split(/\r?\n/u).filter(Boolean);
+            const lines = stdout.split(/\r?\n/u).filter(Boolean);
             return { text: ['OCR 版面分析区域:', ...lines].join('\n'), regionCount: lines.length };
         }
     }
+    filterOcrNativeNoise(value) {
+        return value.split(/\r?\n/u).map(line => line.trim()).filter(line => line && !OCR_NATIVE_NOISE_PATTERNS.some(pattern => pattern.test(line))).join('\n');
+    }
     formatBoundingBox(value) {
         if (!value || typeof value !== 'object')
             return '';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@customize-agent/knowledge",
-  "version": "4.0.23",
+  "version": "4.0.25",
   "description": "Local knowledge base infrastructure for customize-agent",
   "type": "module",
   "main": "./dist/index.js",