@magda/semantic-indexer-sdk 6.0.0-alpha.6 → 6.0.0-alpha.7

package/dist/index.d.ts

@@ -1,6 +1,6 @@
 import { DeleteByQuery_Request } from '@opensearch-project/opensearch/api/_core/deleteByQuery.js';
 import { DeleteByQuery_Response } from '@opensearch-project/opensearch/api/_core/deleteByQuery.js';
-import { ResponseBody } from '@opensearch-project/opensearch/api/_types/_core.search.js';
+import { Search_Response } from '@opensearch-project/opensearch/api/index.js';
 import { TransportRequestOptions } from '@opensearch-project/opensearch/lib/Transport.js';
 import { default as URI_2 } from 'urijs';
 import urijs from 'urijs';
@@ -205,14 +205,15 @@ declare class DeleteResult {
 }
 
 declare class EmbeddingApiClient extends BaseApiClient {
-    private taskSize;
+    taskSize: number;
     constructor(options: EmbeddingApiClientConfig);
-    private testConnection;
+    testConnection(): Promise<boolean>;
     get(text: string): Promise<number[]>;
     get(textList: string[]): Promise<number[][]>;
 }
 
 declare interface EmbeddingApiClientConfig extends BaseApiClientConfig {
+    taskSize?: number;
 }
 
 export declare type EmbeddingText = {
@@ -375,6 +376,7 @@ declare class OpensearchApiClient {
     private client;
     constructor(config: OpensearchConfig);
     static getOpensearchApiClient(openSearchConfig: OpensearchConfig): Promise<OpensearchApiClient>;
+    ping(): Promise<void>;
     createIndex(indexDefinition: {
         indexName: string;
         settings?: any;
@@ -384,9 +386,9 @@ declare class OpensearchApiClient {
     indexExists(indexName: string): Promise<boolean>;
     indexDocument(indexName: string, document: any): Promise<void>;
     bulkIndexDocument(indexName: string, documents: any[]): Promise<void>;
-    searchDocuments(indexName: string, query: string): Promise<ResponseBody>;
     deleteDocument(indexName: string, documentId: string): Promise<void>;
     deleteByQuery(params: DeleteByQuery_Request, options?: TransportRequestOptions): Promise<DeleteByQuery_Response>;
+    search(indexName: string, searchBody: Record<string, unknown>): Promise<Search_Response>;
 }
 
 declare interface OpensearchConfig {

package/dist/index.js

@@ -69051,9 +69051,9 @@ var require_Helpers = __commonJS({
   }
 });
 
-// ../../node_modules/@opensearch-project/opensearch/node_modules/secure-json-parse/index.js
+// ../../node_modules/secure-json-parse/index.js
 var require_secure_json_parse = __commonJS({
-  "../../node_modules/@opensearch-project/opensearch/node_modules/secure-json-parse/index.js"(exports, module) {
+  "../../node_modules/secure-json-parse/index.js"(exports, module) {
     "use strict";
     init_cjs_shim();
     var hasBuffer = typeof Buffer !== "undefined";
@@ -114030,7 +114030,7 @@ var getDefaultProjectName = () => {
 };
 
 // ../../magda-semantic-indexer-framework/node_modules/langsmith/dist/index.js
-var __version__ = "0.3.49";
+var __version__ = "0.3.58";
 
 // ../../magda-semantic-indexer-framework/node_modules/langsmith/dist/utils/env.js
 var globalEnv;
@@ -115505,10 +115505,10 @@ var Client = class _Client {
     if (patch) {
       const sampled = [];
       for (const run of runs) {
-        if (!this.filteredPostUuids.has(run.id)) {
+        if (!this.filteredPostUuids.has(run.trace_id)) {
           sampled.push(run);
-        } else {
-          this.filteredPostUuids.delete(run.id);
+        } else if (run.id === run.trace_id) {
+          this.filteredPostUuids.delete(run.trace_id);
         }
       }
       return sampled;
@@ -115880,8 +115880,8 @@ var Client = class _Client {
       ["patch", preparedUpdateParams]
     ]) {
       for (const originalPayload of payloads) {
-        const { inputs, outputs, events: events2, attachments, ...payload } = originalPayload;
-        const fields = { inputs, outputs, events: events2 };
+        const { inputs, outputs, events: events2, extra, error, serialized, attachments, ...payload } = originalPayload;
+        const fields = { inputs, outputs, events: events2, extra, error, serialized };
         const stringifiedPayload = serialize(payload, `Serializing for multipart ingestion of run with id: ${payload.id}`);
         accumulatedParts.push({
           name: `${method}.${payload.id}`,
@@ -118982,16 +118982,52 @@ function _getWriteReplicasFromEnv() {
     return [];
   try {
     const parsed = JSON.parse(envVar);
-    _checkEndpointEnvUnset(parsed);
-    return Object.entries(parsed).map(([url2, key]) => ({
-      apiUrl: url2.replace(/\/$/, ""),
-      apiKey: key
-    }));
+    if (Array.isArray(parsed)) {
+      const replicas = [];
+      for (const item of parsed) {
+        if (typeof item !== "object" || item === null) {
+          console.warn(`Invalid item type in LANGSMITH_RUNS_ENDPOINTS: expected object, got ${typeof item}`);
+          continue;
+        }
+        if (typeof item.api_url !== "string") {
+          console.warn(`Invalid api_url type in LANGSMITH_RUNS_ENDPOINTS: expected string, got ${typeof item.api_url}`);
+          continue;
+        }
+        if (typeof item.api_key !== "string") {
+          console.warn(`Invalid api_key type in LANGSMITH_RUNS_ENDPOINTS: expected string, got ${typeof item.api_key}`);
+          continue;
+        }
+        replicas.push({
+          apiUrl: item.api_url.replace(/\/$/, ""),
+          apiKey: item.api_key
+        });
+      }
+      return replicas;
+    } else if (typeof parsed === "object" && parsed !== null) {
+      _checkEndpointEnvUnset(parsed);
+      const replicas = [];
+      for (const [url2, key] of Object.entries(parsed)) {
+        const cleanUrl = url2.replace(/\/$/, "");
+        if (typeof key === "string") {
+          replicas.push({
+            apiUrl: cleanUrl,
+            apiKey: key
+          });
+        } else {
+          console.warn(`Invalid value type in LANGSMITH_RUNS_ENDPOINTS for URL ${url2}: expected string, got ${typeof key}`);
+          continue;
+        }
+      }
+      return replicas;
+    } else {
+      console.warn(`Invalid LANGSMITH_RUNS_ENDPOINTS \u2013 must be valid JSON array of objects with api_url and api_key properties, or object mapping url->apiKey, got ${typeof parsed}`);
+      return [];
+    }
   } catch (e2) {
     if (isConflictingEndpointsError(e2)) {
       throw e2;
     }
-    console.warn("Invalid LANGSMITH_RUNS_ENDPOINTS \u2013 must be valid JSON mapping of url->apiKey");
+    console.warn("Invalid LANGSMITH_RUNS_ENDPOINTS \u2013 must be valid JSON array of objects with api_url and api_key properties, or object mapping url->apiKey");
     return [];
   }
 }
@@ -120458,10 +120494,9 @@ var AIMessageChunk = class _AIMessageChunk extends BaseMessageChunk {
       };
     } else {
       const groupedToolCallChunk = fields.tool_call_chunks.reduce((acc, chunk) => {
-        if (!chunk.id)
-          return acc;
-        acc[chunk.id] = acc[chunk.id] ?? [];
-        acc[chunk.id].push(chunk);
+        const chunkId = chunk.id || `fallback-${chunk.index || 0}`;
+        acc[chunkId] = acc[chunkId] ?? [];
+        acc[chunkId].push(chunk);
         return acc;
       }, {});
       const toolCalls = [];
@@ -120471,6 +120506,7 @@ var AIMessageChunk = class _AIMessageChunk extends BaseMessageChunk {
         const name = chunks[0]?.name ?? "";
         const joinedArgs = chunks.map((c) => c.args || "").join("");
         const argsStr = joinedArgs.length ? joinedArgs : "{}";
+        const originalId = chunks[0]?.id || id;
         try {
           parsedArgs = parsePartialJson(argsStr);
           if (parsedArgs === null || typeof parsedArgs !== "object" || Array.isArray(parsedArgs)) {
@@ -120479,14 +120515,14 @@ var AIMessageChunk = class _AIMessageChunk extends BaseMessageChunk {
           toolCalls.push({
             name,
             args: parsedArgs,
-            id,
+            id: originalId,
             type: "tool_call"
           });
         } catch (e2) {
           invalidToolCalls.push({
             name,
             args: argsStr,
-            id,
+            id: originalId,
             error: "Malformed args.",
             type: "invalid_tool_call"
           });
@@ -134405,30 +134441,37 @@ var EmbeddingApiClient = class extends BaseApiClient {
   constructor(options) {
     options.baseApiUrl = options.baseApiUrl || "http://localhost:3000";
     super(options);
+    this.taskSize = options.taskSize || 10;
     this.testConnection();
   }
   async testConnection() {
     try {
       await this.get("test");
-      console.log(`Successfully connected to embedding API: ${this.getBaseApiUri().toString()}`);
+      return true;
     } catch (err) {
       throw new Error(`Failed to connect to embedding API: ${this.getBaseApiUri().toString()}, error: ${err}`);
     }
   }
   async get(input) {
-    if (Array.isArray(input)) {
-      const result = [];
-      for (let i2 = 0; i2 < input.length; i2 += this.taskSize) {
-        const chunk = input.slice(i2, i2 + this.taskSize);
-        const body = { input: chunk };
-        const response = await fetchRequest("post", this.getBaseApiUri().segmentCoded("v1").segmentCoded("embeddings").toString(), body, "application/json", false, this.addAuthHeader());
-        result.push(...response.data.map((d) => d.embedding));
+    const url2 = this.getBaseApiUri().segmentCoded("v1").segmentCoded("embeddings").toString();
+    try {
+      if (Array.isArray(input)) {
+        const result = [];
+        for (let i2 = 0; i2 < input.length; i2 += this.taskSize) {
+          const chunk = input.slice(i2, i2 + this.taskSize);
+          const res = await fetchRequest("post", url2, { input: chunk }, "application/json", false, this.addAuthHeader());
+          result.push(...res.data.map((d) => d.embedding));
+        }
+        return result;
+      } else {
+        const res = await fetchRequest("post", url2, { input }, "application/json", false, this.addAuthHeader());
+        return res.data[0].embedding;
       }
-      return result;
-    } else {
-      const body = { input };
-      const response = await fetchRequest("post", this.getBaseApiUri().segmentCoded("v1").segmentCoded("embeddings").toString(), body, "application/json", false, this.addAuthHeader());
-      return response.data[0].embedding;
+    } catch (e2) {
+      if (e2 instanceof ServerError) {
+        throw new Error(`Embedding API error: ${e2.statusCode}: ${e2.message}`);
+      }
+      throw new Error(`Embedding API request failed: ${e2 instanceof Error ? e2.message : String(e2)}`);
     }
   }
 };
@@ -134469,6 +134512,9 @@ var OpensearchApiClient = class _OpensearchApiClient {
     }
     return instance;
   }
+  async ping() {
+    await this.client.ping();
+  }
   async createIndex(indexDefinition) {
     await this.client.indices.create({
       index: indexDefinition.indexName,
@@ -134512,20 +134558,6 @@ var OpensearchApiClient = class _OpensearchApiClient {
       }
     });
   }
-  async searchDocuments(indexName, query) {
-    const searchResult = await this.client.search({
-      index: indexName,
-      body: {
-        query: {
-          match: {
-            text: query
-          }
-        }
-      }
-    });
-    console.log(`Search query: ${query}, result: ${JSON.stringify(searchResult.body)}`);
-    return searchResult.body;
-  }
   async deleteDocument(indexName, documentId) {
     await this.client.delete({
       index: indexName,
@@ -134540,6 +134572,12 @@ var OpensearchApiClient = class _OpensearchApiClient {
     });
     return response;
   }
+  async search(indexName, searchBody) {
+    return this.client.search({
+      index: indexName,
+      body: searchBody
+    });
+  }
 };
 
 // ../../magda-semantic-indexer-framework/dist/semanticIndexerOptions.js
@@ -140043,7 +140081,7 @@ async function semanticIndexer(userConfig) {
         async: true,
         dereference: false,
         includeEvents: false,
-        includeRecords: false,
+        includeRecords: true,
         onRecordFound,
         maxRetries: 3
       };

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@magda/semantic-indexer-sdk",
   "description": "MAGDA Semantic Indexer SDK",
-  "version": "6.0.0-alpha.6",
+  "version": "6.0.0-alpha.7",
   "type": "module",
   "exports": {
     ".": "./dist/index.js",
@@ -23,7 +23,7 @@
   "author": "",
   "license": "Apache-2.0",
   "devDependencies": {
-    "@magda/semantic-indexer-framework": "^6.0.0-alpha.6",
+    "@magda/semantic-indexer-framework": "^6.0.0-alpha.7",
     "@microsoft/api-extractor": "~7.39.0",
     "esbuild": "^0.19.10",
     "ts-loader": "^9.5.1",