@dsh-libing/knowledge-base 0.1.3 → 0.1.4

package/lib/store.js

@@ -32,7 +32,8 @@ export class Store {
     }
     async trash(kbId, expectedName) {
         return this.serialized(async () => {
-            const registry = await readJson(join(this.root, 'registry.json'), registrySchema), kb = registry.kbs.find(k => k.kbId === id(kbId) && !k.deletedAt);
+            const registry = await readJson(join(this.root, 'registry.json'), registrySchema);
+            const kb = registry.kbs.find(k => k.kbId === id(kbId) && !k.deletedAt);
             if (!kb)
                 throw new Error('知识库不存在，请刷新');
             if (kb.name !== expectedName)
@@ -46,7 +47,8 @@ export class Store {
     }
     async restore(kbId) {
         return this.serialized(async () => {
-            const registry = await readJson(join(this.root, 'registry.json'), registrySchema), kb = registry.kbs.find(k => k.kbId === id(kbId) && k.deletedAt);
+            const registry = await readJson(join(this.root, 'registry.json'), registrySchema);
+            const kb = registry.kbs.find(k => k.kbId === id(kbId) && k.deletedAt);
             if (!kb)
                 throw new Error('知识库不在回收站，请刷新');
             if (kb.purge)
@@ -80,13 +82,15 @@ export class Store {
         if (confirmed !== true)
             throw new Error('必须明确确认彻底删除不可恢复');
         return this.serialized(async () => {
-            const registry = await readJson(join(this.root, 'registry.json'), registrySchema), kb = registry.kbs.find(k => k.kbId === id(kbId) && k.deletedAt);
+            const registry = await readJson(join(this.root, 'registry.json'), registrySchema);
+            const kb = registry.kbs.find(k => k.kbId === id(kbId) && k.deletedAt);
             if (!kb)
                 throw new Error('只能彻底删除回收站中的知识库');
             if (kb.name !== expectedName)
                 throw new Error('知识库名称已变化，请重新确认彻底删除');
             const targets = [...new Set([...(kb.purge?.jobIds || []), ...jobIds])];
-            const kbPath = await this.purgeTarget('kbs', kb.kbId), jobPaths = await Promise.all(targets.map(key => this.purgeTarget('jobs', key)));
+            const kbPath = await this.purgeTarget('kbs', kb.kbId);
+            const jobPaths = await Promise.all(targets.map(key => this.purgeTarget('jobs', key)));
             // Persist intent before removing files: a crash or partial failure must not allow restoration.
             kb.purge = { startedAt: kb.purge?.startedAt || now(), jobIds: targets };
             await atomicWrite(join(this.root, 'registry.json'), registry);
@@ -110,7 +114,8 @@ export class Store {
         });
     }
     async kb(selector) {
-        const kbs = await this.list(), matches = kbs.filter(k => k.kbId === selector || k.name === selector);
+        const kbs = await this.list();
+        const matches = kbs.filter(k => k.kbId === selector || k.name === selector);
         if (matches.length !== 1)
             throw new Error(matches.length ? '知识库名称重复，请使用 kbId' : '知识库不存在，请先创建或选择');
         await safePath(this.root, join('kbs', id(required(matches[0]).kbId)), true);
@@ -136,7 +141,8 @@ export class Store {
     kbRoot(kbId) { return join(this.root, 'kbs', id(kbId, 'kbId')); }
     async inspect(kbId, revision) { return this.inspectKnown(await this.kb(kbId), revision); }
     async inspectKnown(kb, revision) {
-        const kbPath = await safePath(this.root, join('kbs', kb.kbId)), current = await this.current(kb.kbId);
+        const kbPath = await safePath(this.root, join('kbs', kb.kbId));
+        const current = await this.current(kb.kbId);
         if (!revision && !current)
             return { location: { kbPath, revision: null, wikiPath: null, indexPath: null }, summary: { revision: null, documentCount: 0, chunkCount: 0, coverageIncomplete: false } };
         const snapshot = await this.revisionKnown(kb, revision || required(current).revision);
@@ -166,7 +172,8 @@ export class Store {
         filename = text(basename(filename), '文件名', 200);
         if (!/\.(?:docx?|pdf)$/i.test(filename) || data.length > this.config.maxInputBytes)
             throw new Error('只支持大小限制内的 .docx / .doc / .pdf 文件');
-        const receiptId = `u-${randomUUID()}`, file = join(this.root, 'uploads', receiptId + '.bin');
+        const receiptId = `u-${randomUUID()}`;
+        const file = join(this.root, 'uploads', receiptId + '.bin');
         await atomicWrite(file, data);
         await atomicWrite(join(this.root, 'uploads', receiptId + '.json'), { receiptId, filename, createdAt: now(), bytes: data.length, hash: hash(data) });
         return { receiptId, filename, bytes: data.length };
@@ -183,7 +190,8 @@ export class Store {
     async open(selector, requested) { const kb = await this.kb(selector); return { kb, snapshot: await this.revisionKnown(kb, requested) }; }
     async revision(kbId, requested) { return (await this.open(kbId, requested)).snapshot; }
     async revisionKnown(kb, requested) {
-        const current = requested ? null : await this.current(kb.kbId), revision = requested || current?.revision;
+        const current = requested ? null : await this.current(kb.kbId);
+        const revision = requested || current?.revision;
         if (!revision)
             throw new Error('知识库尚无已发布版本');
         id(revision, 'revision');
@@ -206,7 +214,8 @@ export class Store {
         return this.revisionCache.chunk(await safePath(snapshot.root, chunk.path), chunk.contentHash, fresh);
     }
     async asset(kbId, revision, docId, assetId) {
-        const snapshot = await this.revision(kbId, revision), doc = snapshot.manifest.documents.find(d => d.docId === docId);
+        const snapshot = await this.revision(kbId, revision);
+        const doc = snapshot.manifest.documents.find(d => d.docId === docId);
         const asset = doc?.assets.find(a => a.assetId === assetId);
         if (!asset)
             throw new Error('图片不在所选文档版本中');
@@ -267,7 +276,8 @@ export class Store {
         return docs;
     }
     async publish(kbId, revision, stage, manifest, operation) {
-        const kbRoot = this.kbRoot(kbId), target = join(kbRoot, 'revisions', id(revision));
+        const kbRoot = this.kbRoot(kbId);
+        const target = join(kbRoot, 'revisions', id(revision));
         await atomicWrite(join(stage, 'manifest.json'), manifest);
         await mkdir(join(kbRoot, 'revisions'), { recursive: true });
         try {
@@ -284,7 +294,9 @@ export class Store {
         // The pointer owns visibility. Replaying the same job after a crash is idempotent.
         await atomicWrite(join(kbRoot, 'current.json'), { revision, publishedAt: now(), buildKey: manifest.buildKey, coverageIncomplete: manifest.coverageIncomplete });
         await this.serialized(async () => {
-            const logFile = join(kbRoot, 'log.md'), old = await readFile(logFile, 'utf8'), line = `- ${operation.jobId} ${now()} ${operation.kind} ${revision} ← ${operation.docId}\n`;
+            const logFile = join(kbRoot, 'log.md');
+            const old = await readFile(logFile, 'utf8');
+            const line = `- ${operation.jobId} ${now()} ${operation.kind} ${revision} ← ${operation.docId}\n`;
             if (!old.includes(`- ${operation.jobId} `))
                 await atomicWrite(logFile, old + line);
         });
@@ -306,11 +318,13 @@ export class Store {
     }
     /** Cleanup runs with the sole owner before accepting uploads or starting jobs. Inputs are independent copies. */
     async cleanUploads(at = Date.now()) {
-        const root = await safePath(this.root, 'uploads'), cutoff = at - this.config.uploadRetentionDays * 86400000;
+        const root = await safePath(this.root, 'uploads');
+        const cutoff = at - this.config.uploadRetentionDays * 86400000;
         for (const entry of await readdir(root, { withFileTypes: true })) {
             if (!entry.isFile() || !/^u-[a-f0-9-]{36}\.json$/u.test(entry.name))
                 continue;
-            const file = await safePath(root, entry.name), receipt = await readJson(file, uploadSchema);
+            const file = await safePath(root, entry.name);
+            const receipt = await readJson(file, uploadSchema);
             if (entry.name !== receipt.receiptId + '.json')
                 throw new Error('上传收据身份不一致');
             const time = Date.parse(receipt.createdAt);
@@ -326,7 +340,8 @@ export class Store {
     }
     async jobDirectories() { return (await readdir(await safePath(this.root, 'jobs'), { withFileTypes: true })).filter(d => d.isDirectory()).map(d => join(this.root, 'jobs', d.name)); }
     async acquireOwner() {
-        const dir = join(this.root, '.owner.lock'), nonce = randomUUID();
+        const dir = join(this.root, '.owner.lock');
+        const nonce = randomUUID();
         try {
             await mkdir(dir);
         }

package/lib/tools.js

@@ -4,10 +4,12 @@ import { z } from 'zod';
 import { prepareImage } from "./figures/process.js";
 import { buildSchema, jobActionSchema, catalogQuerySchema, readQuerySchema, searchSchema, assetQuerySchema } from "./shared.js";
 import { checkAbort, message } from "./util.js";
+import { selectedKnowledge } from "./agent-access.js";
 const str = (description) => ({ type: 'string', description });
 const bool = (description) => ({ type: 'boolean', description });
 const int = (description) => ({ type: 'integer', description });
-const selector = str('已登记知识库 ID 或唯一库名'), revision = str('固定的知识库版本；首次检索可省略');
+const selector = str('已登记知识库 ID 或唯一库名');
+const revision = str('固定的知识库版本；首次检索可省略');
 const canonical = (value) => z.json().parse(JSON.parse(JSON.stringify(value)));
 function jobView(job) {
     const { jobId, kbId, docId, revision, state, error, outputPath, progress, routes, usage, report, failures } = job;
@@ -20,7 +22,10 @@ export function tools(runtime) {
     const define = (name, title, description, parameters, execute, render) => defineTool({
         name: `libing_${name}`, description, parameters, output: { schema: { type: 'json' }, render: render ?? ((_args, value) => [{ type: 'text', text: JSON.stringify(value) }]) },
         presentCall: args => ({ card: 'generic', title, kind: 'other', rawInput: args }),
-        execute: async (args, exec) => { await ready; checkAbort(exec.signal); return runtime.run(async (signal) => canonical(await execute(args, { ...exec, signal })), exec.signal); },
+        execute: async (args, exec) => { await ready; checkAbort(exec.signal); return runtime.run(async (signal) => { const policy = await selectedKnowledge(runtime, exec.agent, name, args); const scoped = { ...args, ...policy.kbId ? { kbId: policy.kbId } : {} }; const result = await execute(scoped, { ...exec, signal }); if (policy.allowed !== undefined && name === 'kb_catalog' && policy.kbId === undefined) {
+            const value = z.object({ knowledgeBases: z.array(z.object({ kbId: z.string() }).passthrough()), truncated: z.boolean() }).parse(result);
+            return canonical({ ...value, knowledgeBases: value.knowledgeBases.filter(kb => policy.allowed?.includes(kb.kbId)) });
+        } return canonical(result); }, exec.signal); },
     });
     return [
         define('kb_build', '构建知识库', '将指定 Word / PDF 构建为知识库；返回后台 jobId。仅明确允许时建库、跳过图片或接受缺口。', {
@@ -55,8 +60,11 @@ export function tools(runtime) {
         define('kb_read_image', '核对知识库原图', '读取所选文档版本原图及裁片。当前模型须声明图片输入能力；忽略的图片不可作为证据。', {
             kbId: { ...selector, required: true }, revision, docId: { ...str('文档ID'), required: true }, assetId: { ...str('图片ID'), required: true },
         }, async (raw, exec) => {
-            const args = assetQuerySchema.parse(raw), kb = await store.kb(args.kbId), asset = await store.asset(kb.kbId, args.revision, args.docId, args.assetId);
-            const figures = asset.doc.figures.filter(f => f.assetId === args.assetId && f.status !== 'ignored'), region = asset.asset.source && asset.doc.sourceFormat === 'pdf';
+            const args = assetQuerySchema.parse(raw);
+            const kb = await store.kb(args.kbId);
+            const asset = await store.asset(kb.kbId, args.revision, args.docId, args.assetId);
+            const figures = asset.doc.figures.filter(f => f.assetId === args.assetId && f.status !== 'ignored');
+            const region = asset.asset.source && asset.doc.sourceFormat === 'pdf';
             if (!figures.length && (!region || asset.doc.figures.some(f => f.assetId === args.assetId && f.status === 'ignored')))
                 throw new Error('此版本的图片已忽略，请在图片复核恢复导入');
             const route = exec.agent?.session.requestHeader()?.config ?? exec.agent?.options;

package/lib/util.js

@@ -48,7 +48,8 @@ export async function atomicWrite(file, value) {
 export async function safePath(root, rel = '.', existing = true) {
     if (typeof rel !== 'string' || rel.includes('\0') || isAbsolute(rel))
         throw new Error('无效相对路径');
-    const base = await realpath(root), target = resolve(base, rel);
+    const base = await realpath(root);
+    const target = resolve(base, rel);
     const inside = (value) => { const r = relative(base, value); return r !== '..' && !r.startsWith('..' + sep) && !isAbsolute(r); };
     if (!inside(target))
         throw new Error('路径越界');

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dsh-libing/knowledge-base",
-  "version": "0.1.3",
+  "version": "0.1.4",
   "description": "知识治理：Word/PDF 导入、图表复核、可恢复任务与固定版本证据检索",
   "type": "module",
   "icon": "icon.svg",