@kweaver-ai/kweaver-sdk 0.7.2 → 0.7.3

package/dist/templates/dataflow/unstructured/manifest.json

@@ -0,0 +1,11 @@
+{
+  "name": "unstructured",
+  "type": "dataflow",
+  "description": "非结构化文档处理流程",
+  "arguments": [
+    { "name": "title", "required": true, "description": "数据流标题", "type": "string" },
+    { "name": "content_dataset_id", "required": true, "description": "内容数据集 ID", "type": "string" },
+    { "name": "document_dataset_id", "required": true, "description": "文档数据集 ID", "type": "string" },
+    { "name": "element_dataset_id", "required": true, "description": "元素数据集 ID", "type": "string" }
+  ]
+}

package/dist/templates/dataflow/unstructured/template.json

@@ -0,0 +1,63 @@
+{
+  "title": "{{title}}",
+  "steps": [
+    {
+      "id": "0",
+      "title": "",
+      "operator": "@trigger/dataflow-doc"
+    },
+    {
+      "id": "1",
+      "title": "",
+      "operator": "@content/file_parse",
+      "parameters": {
+        "docid": "{{__0.id}}",
+        "model": "embedding",
+        "slice_vector": "slice_vector",
+        "source_type": "docid",
+        "version": "{{__0.rev}}"
+      }
+    },
+    {
+      "id": "1001",
+      "title": "写入向量",
+      "operator": "@dataset/write-docs",
+      "parameters": {
+        "dataset_id": "{{content_dataset_id}}",
+        "documents": "{{__1.chunks}}"
+      }
+    },
+    {
+      "id": "1002",
+      "title": "写入元素",
+      "operator": "@dataset/write-docs",
+      "parameters": {
+        "dataset_id": "{{element_dataset_id}}",
+        "documents": "{{__1.content_list}}"
+      }
+    },
+    {
+      "id": "1003",
+      "title": "写入文件元信息",
+      "operator": "@dataset/write-docs",
+      "parameters": {
+        "dataset_id": "{{document_dataset_id}}",
+        "documents": [
+          {
+            "document_id": "{{__0.id}}",
+            "doc_name": "{{__0.name}}"
+          }
+        ]
+      }
+    }
+  ],
+  "trigger_config": {
+    "operator": "@trigger/manual",
+    "dataSource": {
+      "operator": "",
+      "parameters": {
+        "accessorid": "00000000-0000-0000-0000-000000000000"
+      }
+    }
+  }
+}

package/dist/templates/dataset/document/manifest.json

@@ -0,0 +1,10 @@
+{
+  "name": "document",
+  "type": "dataset",
+  "description": "文档元信息数据集",
+  "arguments": [
+    { "name": "name", "required": true, "description": "数据集名称", "type": "string" },
+    { "name": "catalog_id", "required": false, "default": "adp_bkn_catalog", "description": "所属目录ID", "type": "string" },
+    { "name": "source_identifier", "required": false, "default": "", "description": "数据源标识符，为空时自动生成", "type": "string" }
+  ]
+}

package/dist/templates/dataset/document/template.json

@@ -0,0 +1,23 @@
+{
+  "catalog_id": "{{catalog_id}}",
+  "name": "{{name}}",
+  "category": "dataset",
+  "status": "active",
+  "description": "文档元信息数据集",
+  "source_identifier": "{{source_identifier}}",
+  "schema_definition": [
+    { "name": "id", "type": "keyword" },
+    { "name": "document_id", "type": "keyword" },
+    { "name": "doc_name", "type": "text", "features": [
+      { "name": "doc_name_keyword", "feature_type": "keyword", "ref_property": "doc_name" },
+      { "name": "doc_name_fulltext", "feature_type": "fulltext", "ref_property": "doc_name", "config": { "analyzer": "standard" } }
+    ]},
+    { "name": "doc_md5", "type": "keyword" },
+    { "name": "pages", "type": "integer" },
+    { "name": "file_type", "type": "keyword" },
+    { "name": "creator_id", "type": "keyword" },
+    { "name": "created_at", "type": "text" },
+    { "name": "updated_at", "type": "text" },
+    { "name": "@timestamp", "type": "long" }
+  ]
+}

package/dist/templates/dataset/document-content/manifest.json

@@ -0,0 +1,10 @@
+{
+  "name": "document-content",
+  "type": "dataset",
+  "description": "文档切片及向量数据集",
+  "arguments": [
+    { "name": "name", "required": true, "description": "数据集名称", "type": "string" },
+    { "name": "catalog_id", "required": false, "default": "adp_bkn_catalog", "description": "所属目录ID", "type": "string" },
+    { "name": "source_identifier", "required": false, "default": "", "description": "数据源标识符，为空时自动生成", "type": "string" }
+  ]
+}

package/dist/templates/dataset/document-content/template.json

@@ -0,0 +1,29 @@
+{
+  "catalog_id": "{{catalog_id}}",
+  "name": "{{name}}",
+  "category": "dataset",
+  "status": "active",
+  "description": "文档切片及向量数据集",
+  "source_identifier": "{{source_identifier}}",
+  "schema_definition": [
+    { "name": "id", "type": "keyword" },
+    { "name": "document_id", "type": "keyword" },
+    { "name": "slice_md5", "type": "keyword" },
+    { "name": "deduplication_id", "type": "keyword" },
+    { "name": "segment_id", "type": "integer" },
+    { "name": "slice_type", "type": "integer" },
+    { "name": "slice_content", "type": "text", "features": [
+      { "name": "slice_content_fulltext", "feature_type": "fulltext", "ref_property": "slice_content", "config": { "analyzer": "standard" } }
+    ]},
+    { "name": "text_vector", "type": "vector", "features": [
+      { "name": "text_vector", "feature_type": "vector", "ref_property": "text_vector", "config": { "dimension": 768, "method": { "name": "hnsw", "engine": "lucene", "parameters": { "ef_construction": 256 } } } }
+    ]},
+    { "name": "img_path", "type": "keyword" },
+    { "name": "image_vector", "type": "vector", "features": [
+      { "name": "image_vector", "feature_type": "vector", "ref_property": "image_vector", "config": { "dimension": 512, "method": { "name": "hnsw", "engine": "lucene", "parameters": { "ef_construction": 256 } } } }
+    ]},
+    { "name": "created_at", "type": "text" },
+    { "name": "updated_at", "type": "text" },
+    { "name": "@timestamp", "type": "long" }
+  ]
+}

package/dist/templates/dataset/document-element/manifest.json

@@ -0,0 +1,10 @@
+{
+  "name": "document-element",
+  "type": "dataset",
+  "description": "文档结构化元素数据集",
+  "arguments": [
+    { "name": "name", "required": true, "description": "数据集名称", "type": "string" },
+    { "name": "catalog_id", "required": false, "default": "adp_bkn_catalog", "description": "所属目录ID", "type": "string" },
+    { "name": "source_identifier", "required": false, "default": "", "description": "数据源标识符，为空时自动生成", "type": "string" }
+  ]
+}

package/dist/templates/dataset/document-element/template.json

@@ -0,0 +1,21 @@
+{
+  "catalog_id": "{{catalog_id}}",
+  "name": "{{name}}",
+  "category": "dataset",
+  "status": "active",
+  "description": "文档结构化元素数据集",
+  "source_identifier": "{{source_identifier}}",
+  "schema_definition": [
+    { "name": "id", "type": "keyword" },
+    { "name": "element_id", "type": "keyword" },
+    { "name": "document_id", "type": "keyword" },
+    { "name": "element_type", "type": "keyword" },
+    { "name": "parent_id", "type": "keyword" },
+    { "name": "level", "type": "integer" },
+    { "name": "content", "type": "text", "features": [
+      { "name": "content_fulltext", "feature_type": "fulltext", "ref_property": "content", "config": { "analyzer": "standard" } }
+    ]},
+    { "name": "metadata", "type": "object" },
+    { "name": "@timestamp", "type": "long" }
+  ]
+}

package/dist/utils/template-loader.d.ts

@@ -0,0 +1,40 @@
+export interface TemplateManifest {
+    name: string;
+    type: "dataset" | "bkn" | "dataflow";
+    description: string;
+    arguments: Array<{
+        name: string;
+        required: boolean;
+        description: string;
+        type: "string" | "integer" | "boolean" | "array";
+        default?: unknown;
+    }>;
+}
+export interface LoadedTemplate {
+    template: Record<string, unknown>;
+    manifest: TemplateManifest;
+    templatePath: string;
+}
+/**
+ * Generate a unique source identifier with prefix
+ */
+export declare function generateSourceIdentifier(prefix: string): string;
+/**
+ * Render template with arguments, applying defaults and validation
+ */
+export declare function renderTemplate(template: Record<string, unknown>, manifest: TemplateManifest, args: Record<string, unknown>): Record<string, unknown>;
+/**
+ * Load template from directory
+ */
+export declare function loadTemplate(templateName: string, templateType: "dataset" | "bkn" | "dataflow", templatesDir: string): Promise<LoadedTemplate | null>;
+/**
+ * List all available templates of a given type
+ */
+export declare function listTemplates(templateType: "dataset" | "bkn" | "dataflow", templatesDir: string): Promise<Array<{
+    name: string;
+    description: string;
+}>>;
+/**
+ * Get the templates directory path (relative to dist or src)
+ */
+export declare function getTemplatesDir(): string;

package/dist/utils/template-loader.js

@@ -0,0 +1,129 @@
+import { readFile, access } from "node:fs/promises";
+import { join } from "node:path";
+import { constants } from "node:fs";
+/**
+ * Generate a unique source identifier with prefix
+ */
+export function generateSourceIdentifier(prefix) {
+    const random = Math.random().toString(36).substring(2, 15);
+    const timestamp = Date.now().toString(36);
+    return `${prefix}_${timestamp}_${random}`;
+}
+/**
+ * Replace {{placeholder}} with actual values in a string
+ */
+function replacePlaceholders(str, values) {
+    return str.replace(/\{\{(\w+)\}\}/g, (_, key) => {
+        if (values[key] !== undefined) {
+            return String(values[key]);
+        }
+        return `{{${key}}}`;
+    });
+}
+/**
+ * Deep replace placeholders in an object
+ */
+function deepReplace(obj, values) {
+    if (typeof obj === "string") {
+        return replacePlaceholders(obj, values);
+    }
+    if (Array.isArray(obj)) {
+        return obj.map(item => deepReplace(item, values));
+    }
+    if (obj !== null && typeof obj === "object") {
+        const result = {};
+        for (const [key, value] of Object.entries(obj)) {
+            result[key] = deepReplace(value, values);
+        }
+        return result;
+    }
+    return obj;
+}
+/**
+ * Render template with arguments, applying defaults and validation
+ */
+export function renderTemplate(template, manifest, args) {
+    // Merge args with defaults
+    const merged = {};
+    const missing = [];
+    for (const arg of manifest.arguments) {
+        if (args[arg.name] !== undefined) {
+            merged[arg.name] = args[arg.name];
+        }
+        else if (arg.default !== undefined) {
+            merged[arg.name] = arg.default;
+        }
+        else if (arg.required) {
+            missing.push(arg.name);
+        }
+    }
+    if (missing.length > 0) {
+        throw new Error(`Missing required argument(s): ${missing.join(", ")}`);
+    }
+    // Deep replace placeholders
+    return deepReplace(template, merged);
+}
+/**
+ * Load template from directory
+ */
+export async function loadTemplate(templateName, templateType, templatesDir) {
+    const templateDir = join(templatesDir, templateType, templateName);
+    try {
+        await access(templateDir, constants.R_OK);
+    }
+    catch {
+        return null;
+    }
+    const templatePath = join(templateDir, "template.json");
+    const manifestPath = join(templateDir, "manifest.json");
+    try {
+        const [templateContent, manifestContent] = await Promise.all([
+            readFile(templatePath, "utf-8"),
+            readFile(manifestPath, "utf-8"),
+        ]);
+        return {
+            template: JSON.parse(templateContent),
+            manifest: JSON.parse(manifestContent),
+            templatePath: templateDir,
+        };
+    }
+    catch {
+        return null;
+    }
+}
+/**
+ * List all available templates of a given type
+ */
+export async function listTemplates(templateType, templatesDir) {
+    const { readdir } = await import("node:fs/promises");
+    const typeDir = join(templatesDir, templateType);
+    try {
+        const entries = await readdir(typeDir, { withFileTypes: true });
+        const templates = [];
+        for (const entry of entries) {
+            if (!entry.isDirectory())
+                continue;
+            const loaded = await loadTemplate(entry.name, templateType, templatesDir);
+            if (loaded) {
+                templates.push({
+                    name: loaded.manifest.name,
+                    description: loaded.manifest.description,
+                });
+            }
+        }
+        return templates;
+    }
+    catch {
+        return [];
+    }
+}
+/**
+ * Get the templates directory path (relative to dist or src)
+ */
+export function getTemplatesDir() {
+    // When running from dist, templates are copied to dist/templates
+    // When running from src (tsx), templates are in src/templates
+    const { url } = import.meta;
+    const baseDir = join(new URL(url).pathname, "..", "..", "templates");
+    return baseDir;
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kweaver-ai/kweaver-sdk",
-  "version": "0.7.2",
+  "version": "0.7.3",
   "description": "KWeaver TypeScript SDK — CLI tool and programmatic API for knowledge networks and Decision Agents.",
   "type": "module",
   "main": "./dist/index.js",