dsh-knowledge-library 0.1.1 → 0.3.0

package/dist/standalone.js

@@ -5,13 +5,21 @@ import { join } from 'node:path';
 import { Library } from './core/library.js';
 import { ImportQueue } from './core/jobs.js';
 import { API_PATH, createHandler, serveAsset } from './http.js';
-import { fetchPublicPage } from './core/fetch.js';
+import { fetchPublicImage, fetchPublicPage } from './core/fetch.js';
+import { ImageCapture } from './core/image-capture.js';
+import { createPreviewSettings } from './preview-settings.js';
+import { KnowledgeQueue } from './core/knowledge.js';
+import { previewKnowledgeModel } from './knowledge-model.js';
 const root = resolve(process.env.KNOWLEDGE_LIBRARY_ROOT || join(homedir(), 'Documents', 'DSH-Knowledge-Library'));
 const library = new Library(root);
-const queue = new ImportQueue(library, (url, signal) => fetchPublicPage(url, signal, { allowProxyFakeIp: process.env.KNOWLEDGE_LIBRARY_ALLOW_FAKE_IP === '1' }));
+const model = previewKnowledgeModel();
+const knowledge = new KnowledgeQueue(library, model);
+const settings = createPreviewSettings(library, { allowProxyFakeIp: process.env.KNOWLEDGE_LIBRARY_ALLOW_FAKE_IP === '1' });
+const images = new ImageCapture(library, model, (url, signal, _policy, referer) => fetchPublicImage(url, signal, settings.fetchPolicy(), referer));
+const queue = new ImportQueue(library, (url, signal) => fetchPublicPage(url, signal, settings.fetchPolicy()), images);
 const port = Number(process.env.PORT || 4317);
 let origin = `http://127.0.0.1:${port}`;
-const api = createHandler(library, queue, () => origin);
+const api = createHandler(library, queue, () => origin, settings, knowledge);
 const server = createServer((req, res) => {
     if (req.url?.startsWith(API_PATH))
         void api(req, res);
@@ -30,7 +38,7 @@ async function shutdown() {
     stopping = true;
     server.close();
     server.closeAllConnections();
-    await queue.close();
+    await Promise.all([queue.close(), knowledge.close()]);
     library.close();
 }
 process.on('SIGINT', () => void shutdown());

package/dist/tools.d.ts

@@ -1,4 +1,5 @@
 import { type ToolDefinition } from '@deepseek-ai/dsh-tools';
 import type { Library } from './core/library.js';
 import type { ImportQueue } from './core/jobs.js';
-export declare function createTools(library: Library, queue: ImportQueue): ToolDefinition[];
+import type { KnowledgeQueue } from './core/knowledge.js';
+export declare function createTools(library: Library, queue: ImportQueue, knowledge?: KnowledgeQueue): ToolDefinition[];

package/dist/tools.js

@@ -12,7 +12,7 @@ function resultSummary(result) {
     const { content, segments, path, ...material } = result.material;
     return { ...material, duplicate: result.duplicate, segmentCount: segments.length, path };
 }
-export function createTools(library, queue) {
+export function createTools(library, queue, knowledge) {
     const tool = (name, description, parameters, execute, readOnly = false) => defineTool({
         name, description, parameters,
         output: { schema: { type: 'json' }, render: (_args, value) => [{ type: 'text', text: JSON.stringify(value) }] },
@@ -58,5 +58,14 @@ export function createTools(library, queue) {
             retryId: { type: 'string' },
         }, args => args.retryId ? queue.retry(args.retryId) : queue.list()),
         tool('library_overview', '查看资料库统计、分类、专题、标签和索引警告。', {}, () => library.stats(), true),
+        ...(knowledge ? [
+            tool('library_knowledge_generate', '选择1至10份原始资料。多篇默认smart：逐篇分析、智能分组，等待用户确认分类后按组生成独立草稿；单篇默认compose直接生成草稿。已知同主题也可显式compose。正式笔记均待用户确认，不自动发布。', {
+                sourceIds: { type: 'array', required: true, items: { type: 'string' } }, topic: { type: 'string', description: '围绕什么问题整理，可选' },
+                mode: { type: 'string', enum: ['smart', 'compose'], description: 'smart智能分组，compose同主题直接综合；省略时多篇smart、单篇compose' },
+            }, args => knowledge.add(args.sourceIds, args.topic, args.mode)),
+            tool('library_knowledge_status', '查看知识整理状态、失败原因和待确认草稿；提醒用户到资料库确认保存，不代替用户确认。', {
+                id: { type: 'string', description: '知识整理任务ID；省略查看总览' },
+            }, args => args.id ? knowledge.get(args.id) : knowledge.overview(), true),
+        ] : []),
     ];
 }

package/dist/types.d.ts

@@ -24,10 +24,14 @@ export interface Material {
     author?: string;
     createdAt: string;
     updatedAt: string;
+    deletedAt?: string;
+    deletionId?: string;
+    knowledgeJobId?: string;
+    knowledgeUpdateJobIds?: string[];
     contentHash: string;
     textChecksum?: string;
     segmentsChecksum?: string;
-    revision: 1;
+    revision: number;
     originalName?: string;
     originalFile?: string;
     citations: Citation[];
@@ -37,6 +41,61 @@ export interface MaterialDetail extends Material {
     content: string;
     segments: Segment[];
     path: string;
+    images?: PageImageView[];
+}
+export interface PageImageCandidate {
+    url: string;
+    alt: string;
+}
+export interface PageImageRecord extends PageImageCandidate {
+    id: string;
+    status: 'pending' | 'downloading' | 'recognizing' | 'completed' | 'failed' | 'skipped';
+    attempts: number;
+    error?: string;
+    fileName?: string;
+    checksum?: string;
+    mediaType?: 'image/png' | 'image/jpeg' | 'image/webp' | 'image/gif';
+    width?: number;
+    height?: number;
+    bytes?: number;
+    finalUrl?: string;
+    text?: string;
+    description?: string;
+    uncertainty?: string;
+}
+export interface PageImageManifest {
+    schemaVersion: 1;
+    model: string;
+    limit: number;
+    excluded: number;
+    records: PageImageRecord[];
+}
+export type PageImageView = Omit<PageImageRecord, 'text' | 'description' | 'fileName' | 'checksum'> & {
+    segmentId?: string;
+    originalAvailable: boolean;
+    model: string;
+};
+export interface ImageCaptureSettings {
+    enabled: boolean;
+    maxImages: number;
+    model?: string;
+    version: number;
+}
+export interface ImageCaptureStatus extends ImageCaptureSettings {
+    supported: boolean;
+    available: boolean;
+    currentModel?: string;
+    reason?: string;
+}
+export interface TrashSnapshot {
+    count: number;
+    pending: number;
+    token: string;
+}
+export interface EmptyTrashResult {
+    deleted: number;
+    pending: number;
+    errors: string[];
 }
 export interface CaptureFields {
     title?: string;
@@ -52,6 +111,7 @@ export interface SearchOptions {
     tag?: string;
     collection?: string;
     kind?: string;
+    trashed?: boolean;
     limit?: number;
     offset?: number;
 }
@@ -75,4 +135,142 @@ export interface ImportJob {
     attempts: number;
     materialId?: string;
     error?: string;
+    errorCode?: 'PAGE_REQUIRES_BROWSER';
+    progress?: string;
+    imageCapture?: {
+        model: string;
+        maxImages: number;
+    };
+    imageSummary?: {
+        total: number;
+        completed: number;
+        failed: number;
+        skipped: number;
+    };
+}
+export interface KnowledgeDraft {
+    title: string;
+    content: string;
+    tags: string[];
+    citations: Citation[];
+    category?: string;
+}
+export interface KnowledgeSource {
+    id: string;
+    title: string;
+    fingerprint: string;
+    category?: string;
+    tags?: string[];
+}
+export interface NoteIntegration {
+    recommendation: 'new' | 'append' | 'duplicate';
+    choice: 'new' | 'append' | 'duplicate';
+    reason: string;
+    overlaps: string[];
+    additions: string[];
+    conflicts: string[];
+    content: string;
+    search: {
+        scanned: number;
+        matched: number;
+        compared: number;
+        oversized: number;
+    };
+    target?: {
+        id: string;
+        title: string;
+        category: string;
+        revision: number;
+        fingerprint: string;
+        content: string;
+    };
+}
+export interface SourceAnalysis {
+    sourceId: string;
+    theme: string;
+    concepts: string[];
+    summary: string;
+    usage: string;
+    category: string;
+    tags: string[];
+    certainty: 'clear' | 'uncertain';
+    uncertainty: string;
+    citations: Citation[];
+}
+export interface KnowledgeGroup {
+    id: string;
+    title: string;
+    reason: string;
+    sourceIds: string[];
+    category: string;
+    tags: string[];
+}
+export interface GroupingPlan {
+    groups: KnowledgeGroup[];
+    unassigned: Array<{
+        sourceId: string;
+        reason: string;
+    }>;
+    applyClassification?: boolean;
+}
+export interface ClassificationStep {
+    sourceId: string;
+    before: {
+        category: string;
+        tags: string[];
+    };
+    after: {
+        category: string;
+        tags: string[];
+    };
+    done: boolean;
+}
+export interface ClassificationJournal {
+    apply: boolean;
+    version: number;
+    direction: 'apply' | 'rollback';
+    steps: ClassificationStep[];
+}
+export interface KnowledgeJob {
+    id: string;
+    sourceIds: string[];
+    sources: KnowledgeSource[];
+    topic: string;
+    status: 'queued' | 'running' | 'grouping' | 'classifying' | 'grouped' | 'draft' | 'completed' | 'failed' | 'cancelled' | 'discarded';
+    mode?: 'compose' | 'smart';
+    parentId?: string;
+    groupId?: string;
+    analyses?: SourceAnalysis[];
+    plan?: GroupingPlan;
+    planVersion?: number;
+    classification?: ClassificationJournal;
+    children?: Array<{
+        groupId: string;
+        jobId: string;
+    }>;
+    createdAt: string;
+    updatedAt: string;
+    progress: string;
+    model?: string;
+    modelCalls?: number;
+    incremental?: boolean;
+    integration?: NoteIntegration;
+    error?: string;
+    draft?: KnowledgeDraft;
+    draftVersion: number;
+    noteId?: string;
+}
+export interface KnowledgeState {
+    status: 'pending' | 'running' | 'grouping' | 'draft' | 'completed' | 'failed';
+    jobId?: string;
+    noteIds: string[];
+}
+export interface KnowledgeOverview {
+    supported: boolean;
+    smartSupported?: boolean;
+    available: boolean;
+    model?: string;
+    reason?: string;
+    jobs: KnowledgeJob[];
+    states: Record<string, KnowledgeState>;
 }