@toxplanet/pegasus-sdk 1.2.0 → 1.2.2

package/config/environment.acc.js

@@ -5,11 +5,10 @@ module.exports = {
   sourceService: 'pegasus-sdk',
   secretArn: 'arn:aws:secretsmanager:us-east-1:605134466764:secret:rds!cluster-9b502dde-5e2a-49db-b2c5-9801141ee40b-gkHbLm',
   clusterArn: 'arn:aws:rds:us-east-1:605134466764:cluster:cr-chemicals-acc',
+  openSearchLambdaArn: 'arn:aws:lambda:us-east-1:605134466764:function:pegasus-os-bridge-acc',
   database: {
     name: 'chemicals'
   },
-  openSearchEndpoint: 'https://1pbu0yqr197lq07hfcjh.us-east-1.aoss.amazonaws.com',
-  openSearchIndex: 'chemicals',
   indexRoutes: {
     chemicals: ['chemicals*'],
     documents: ['documents*'],

package/config/environment.dev.js

@@ -5,11 +5,10 @@ module.exports = {
   sourceService: 'pegasus-sdk',
   secretArn: 'arn:aws:secretsmanager:us-east-1:292931567094:secret:rds!cluster-b851c3ce-58cc-41cd-aeae-05cc7f5e031a-ZYSjiI',
   clusterArn: 'arn:aws:rds:us-east-1:292931567094:cluster:cr-chemicals',
+  openSearchLambdaArn: 'arn:aws:lambda:us-east-1:292931567094:function:pegasus-os-bridge-dev',
   database: {
     name: 'chemicals'
   },
-  openSearchEndpoint: 'https://war8lk73nzswquk8dcz1.us-east-1.aoss.amazonaws.com',
-  openSearchIndex: 'chemicals',
   indexRoutes: {
     chemicals: ['chemicals*'],
     documents: ['documents*'],

package/config/environment.prod.js

@@ -4,12 +4,11 @@ module.exports = {
   awsAccountId: '147997144422',
   sourceService: 'pegasus-sdk',
   secretArn: 'arn:aws:secretsmanager:us-east-1:964963729446:secret:rds!cluster-bd301b0f-93b7-4dcb-a4fa-ebf753fd1c00-atsPOm',
-  clusterArn: 'arn:aws:rds:us-east-1:147997144422:cluster:cr-chemicals-prod',
+  clusterArn: 'arn:aws:rds:us-east-1:964963729446:cluster:cr-chemicals-prod',
+  openSearchLambdaArn: 'arn:aws:lambda:us-east-1:964963729446:function:pegasus-os-bridge-prod',
   database: {
     name: 'chemicals'
   },
-  openSearchEndpoint: 'https://aq6ftqi0hawm42795fci.us-east-1.aoss-fips.amazonaws.com',
-  openSearchIndex: 'chemicals',
   indexRoutes: {
     chemicals: ['chemicals*'],
     documents: ['documents*'],

package/config/environment.qa.js

@@ -5,11 +5,10 @@ module.exports = {
   sourceService: 'pegasus-sdk',
   secretArn: 'arn:aws:secretsmanager:us-east-1:147997144422:secret:rds!cluster-25483b3f-3758-43ed-9548-26c91de16c2d-oYjysU',
   clusterArn: 'arn:aws:rds:us-east-1:147997144422:cluster:cr-chemicals-qa',
+  openSearchLambdaArn: 'arn:aws:lambda:us-east-1:147997144422:function:pegasus-os-bridge-qa',
   database: {
     name: 'chemicals'
   },
-  openSearchEndpoint: 'https://odusb11s00j5hyy5r6.us-east-1.aoss.amazonaws.com',
-  openSearchIndex: 'chemicals',
   indexRoutes: {
     chemicals: ['chemicals*'],
     documents: ['documents*'],

package/config/index.js

@@ -17,12 +17,8 @@ function loadConfig(envOverride = null) {
     config.secretName = process.env.PEGASUS_SDK_DB_SECRET_ARN;
   }
 
-  if (process.env.PEGASUS_SDK_OPENSEARCH_ENDPOINT) {
-    config.openSearchEndpoint = process.env.PEGASUS_SDK_OPENSEARCH_ENDPOINT;
-  }
-
-  if (process.env.PEGASUS_SDK_OPENSEARCH_INDEX) {
-    config.openSearchIndex = process.env.PEGASUS_SDK_OPENSEARCH_INDEX;
+  if (process.env.PEGASUS_SDK_OPENSEARCH_LAMBDA_ARN) {
+    config.openSearchLambdaArn = process.env.PEGASUS_SDK_OPENSEARCH_LAMBDA_ARN;
   }
 
   if (process.env.PEGASUS_SDK_DATABASE_HOST) {

package/lib/chemicals.js

@@ -738,11 +738,8 @@ class ChemicalsService {
     }
 
     try {
-      const opensearchClient = this.connection.getOpenSearchClient();
-      const indexName = this.connection.getOpenSearchIndex();
-      
-      const response = await opensearchClient.search({
-        index: indexName,
+      const result = await this.connection.invokeOpenSearch({
+        operation: 'search',
         body: {
           size: limit,
           query: {
@@ -751,7 +748,11 @@ class ChemicalsService {
                 { term: { 'chemical_name.keyword': { value: searchTerm, boost: SEARCH_BOOST_EXACT_PRIMARY, case_insensitive: true } } },
                 { prefix: { 'chemical_name.keyword': { value: searchTerm, boost: SEARCH_BOOST_PREFIX_PRIMARY, case_insensitive: true } } },
                 { term: { 'synonyms.keyword': { value: searchTerm, boost: SEARCH_BOOST_EXACT_SECONDARY, case_insensitive: true } } },
-                { prefix: { 'synonyms.keyword': { value: searchTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } }
+                { prefix: { 'synonyms.keyword': { value: searchTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } },
+                { match: { 'chemical_name': { query: searchTerm, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+                { match: { 'synonyms': { query: searchTerm, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+                { match_phrase_prefix: { 'chemical_name': { query: searchTerm, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+                { match_phrase_prefix: { 'synonyms': { query: searchTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
               ],
               minimum_should_match: 1
             }
@@ -760,7 +761,7 @@ class ChemicalsService {
         }
       });
 
-      const hits = response.body?.hits?.hits || [];
+      const hits = result?.hits?.hits || [];
       const results = hits.map((hit) => ({
         id: hit._source.postgres_id,
         name: hit._source.chemical_name,
@@ -789,11 +790,8 @@ class ChemicalsService {
     }
 
     try {
-      const opensearchClient = this.connection.getOpenSearchClient();
-      const indexName = this.connection.getOpenSearchIndex();
-      
-      const response = await opensearchClient.search({
-        index: indexName,
+      const result = await this.connection.invokeOpenSearch({
+        operation: 'search',
         body: {
           size: limit,
           query: {
@@ -802,7 +800,11 @@ class ChemicalsService {
                 { term: { 'synonyms.keyword': { value: synonymTerm, boost: SEARCH_BOOST_EXACT_PRIMARY, case_insensitive: true } } },
                 { prefix: { 'synonyms.keyword': { value: synonymTerm, boost: SEARCH_BOOST_PREFIX_PRIMARY, case_insensitive: true } } },
                 { term: { 'chemical_name.keyword': { value: synonymTerm, boost: SEARCH_BOOST_EXACT_SECONDARY, case_insensitive: true } } },
-                { prefix: { 'chemical_name.keyword': { value: synonymTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } }
+                { prefix: { 'chemical_name.keyword': { value: synonymTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } },
+                { match: { 'synonyms': { query: synonymTerm, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+                { match: { 'chemical_name': { query: synonymTerm, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+                { match_phrase_prefix: { 'synonyms': { query: synonymTerm, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+                { match_phrase_prefix: { 'chemical_name': { query: synonymTerm, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
               ],
               minimum_should_match: 1
             }
@@ -811,7 +813,7 @@ class ChemicalsService {
         }
       });
 
-      const hits = response.body?.hits?.hits || [];
+      const hits = result?.hits?.hits || [];
       const results = hits.map((hit) => ({
         id: hit._source.postgres_id,
         name: hit._source.chemical_name,

package/lib/connection.js

@@ -1,7 +1,5 @@
-const { Client: OpenSearchClient } = require('@opensearch-project/opensearch');
 const { RDSDataClient, ExecuteStatementCommand } = require('@aws-sdk/client-rds-data');
-const { AwsSigv4Signer } = require('@opensearch-project/opensearch/aws');
-const { fromNodeProviderChain } = require('@aws-sdk/credential-providers');
+const { LambdaClient, InvokeCommand } = require('@aws-sdk/client-lambda');
 const { mapRecords } = require('./db');
 const { loadConfig } = require('../config');
 const { logInfo, logError } = require('@toxplanet/tphelper/logging');
@@ -15,12 +13,10 @@ class PegasusConnection {
     this.region = this.config.region;
     this.secretArn = this.config.secretArn;
     this.clusterArn = this.config.clusterArn;
-    this.openSearchEndpoint = this.config.openSearchEndpoint;
-    this.openSearchIndex = this.config.openSearchIndex;
     this.databaseName = this.config.database?.name;
     
     this.rdsDataClient = null;
-    this.osClient = null;
+    this.lambdaClient = null;
     this.isConnected = false;
   }
 
@@ -30,6 +26,7 @@ class PegasusConnection {
     }
 
     this.rdsDataClient = new RDSDataClient({ region: this.region });
+    this.lambdaClient = new LambdaClient({ region: this.region });
 
     logInfo('pegasus-sdk', 'RDS Data API client initialized');
 
@@ -47,20 +44,6 @@ class PegasusConnection {
       throw err;
     }
 
-    if (this.openSearchEndpoint) {
-      this.osClient = new OpenSearchClient({
-        ...AwsSigv4Signer({
-          region: this.region,
-          service: 'aoss',
-          getCredentials: () => {
-            const credentialsProvider = fromNodeProviderChain();
-            return credentialsProvider();
-          }
-        }),
-        node: this.openSearchEndpoint
-      });
-    }
-
     this.isConnected = true;
   }
 
@@ -101,20 +84,49 @@ class PegasusConnection {
     }
 
     this.rdsDataClient = null;
-    this.osClient = null;
+    this.lambdaClient = null;
     this.isConnected = false;
     logInfo('pegasus-sdk', 'RDS Data API client disconnected');
   }
 
-  getOpenSearchClient() {
-    if (!this.osClient) {
-      throw new Error('OpenSearch connection not established. Call connect() first or provide openSearchEndpoint.');
+  async invokeOpenSearch(event) {
+    if (!this.lambdaClient) {
+      throw new Error('Lambda client not initialized. Call connect() first.');
     }
-    return this.osClient;
-  }
 
-  getOpenSearchIndex() {
-    return this.openSearchIndex || 'chemicals';
+    const lambdaArn = process.env.PEGASUS_SDK_OPENSEARCH_LAMBDA_ARN
+      || this.config.openSearchLambdaArn
+      || null;
+
+    if (!lambdaArn) {
+      throw new Error('No OpenSearch Lambda ARN configured. Set PEGASUS_SDK_OPENSEARCH_LAMBDA_ARN or provide awsAccountId in config.');
+    }
+
+    try {
+      const response = await this.lambdaClient.send(new InvokeCommand({
+        FunctionName: lambdaArn,
+        Payload: JSON.stringify(event)
+      }));
+
+      const rawPayload = response.Payload;
+      const payload = (rawPayload instanceof Uint8Array || Buffer.isBuffer(rawPayload))
+        ? Buffer.from(rawPayload).toString('utf-8')
+        : rawPayload;
+      const parsed = JSON.parse(payload);
+      const body = JSON.parse(parsed.body);
+
+      if (!body.success) {
+        const error = new Error(body.error);
+        error.statusCode = body.statusCode;
+        error.responseBody = body.body;
+        throw error;
+      }
+
+      return body.result;
+    } catch (error) {
+      logError('pegasus-sdk', 'PegasusConnection', 'invokeOpenSearch', error);
+      throw error;
+    }
   }
 
   async testConnection() {
@@ -142,28 +154,25 @@ class PegasusConnection {
       };
 
       let osStatus = null;
-      if (this.osClient) {
-        try {
-          const indexName = this.getOpenSearchIndex();
-          const testSearch = await this.osClient.search({
-            index: indexName,
-            body: {
-              size: 1,
-              query: {
-                match: { chemical_name: 'benzene' }
-              }
+      try {
+        const testSearch = await this.invokeOpenSearch({
+          operation: 'search',
+          body: {
+            size: 1,
+            query: {
+              match: { chemical_name: 'benzene' }
             }
-          });
-          osStatus = {
-            connected: true,
-            resultsFound: testSearch.body.hits.total.value || 0
-          };
-        } catch (osError) {
-          osStatus = {
-            connected: false,
-            error: osError.message
-          };
-        }
+          }
+        });
+        osStatus = {
+          connected: true,
+          resultsFound: testSearch.hits.total.value || 0
+        };
+      } catch (osError) {
+        osStatus = {
+          connected: false,
+          error: osError.message
+        };
       }
 
       return {

package/lib/search.js

@@ -109,8 +109,6 @@ class SearchService {
     const synonymPrefix = options.synonymPrefix !== undefined ? options.synonymPrefix : 3;
 
     try {
-      const opensearchClient = this.connection.getOpenSearchClient();
-      
       // Get CAS number variations (if applicable)
       const queryVariations = getCasNumberVariations(query);
       
@@ -123,24 +121,28 @@ class SearchService {
       const shouldClauses = [];
       
       for (const queryVariation of queryVariations) {
-        // Exact matches (configurable priority)
         shouldClauses.push(
+          // keyword / array-of-keyword fields — exact term match
           { term: { 'cas_numbers': { value: queryVariation, boost: casExact } } },
-          { term: { 'chemical_name.keyword': { value: queryVariation, boost: nameExact, case_insensitive: true } } },
           { term: { 'identifier_values': { value: queryVariation, boost: identifierExact } } },
+          // keyword sub-fields — exact + prefix (present when index has .keyword multi-field)
+          { term: { 'chemical_name.keyword': { value: queryVariation, boost: nameExact, case_insensitive: true } } },
           { term: { 'synonyms.keyword': { value: queryVariation, boost: synonymExact, case_insensitive: true } } },
-          // Prefix matches (configurable priority)
-          { prefix: { 'cas_numbers': { value: queryVariation, boost: casPrefix } } },
           { prefix: { 'chemical_name.keyword': { value: queryVariation, boost: namePrefix, case_insensitive: true } } },
+          { prefix: { 'synonyms.keyword': { value: queryVariation, boost: synonymPrefix, case_insensitive: true } } },
+          // keyword field prefix matching for cas / identifiers
+          { prefix: { 'cas_numbers': { value: queryVariation, boost: casPrefix } } },
           { prefix: { 'identifier_values': { value: queryVariation, boost: identifierPrefix } } },
-          { prefix: { 'synonyms.keyword': { value: queryVariation, boost: synonymPrefix, case_insensitive: true } } }
+          // match queries work on plain text fields even without .keyword sub-fields
+          { match: { 'chemical_name': { query: queryVariation, boost: nameExact } } },
+          { match: { 'synonyms': { query: queryVariation, boost: synonymExact } } },
+          { match_phrase_prefix: { 'chemical_name': { query: queryVariation, boost: namePrefix } } },
+          { match_phrase_prefix: { 'synonyms': { query: queryVariation, boost: synonymPrefix } } }
         );
       }
       
-      const indexName = this.connection.getOpenSearchIndex();
-      
-      const response = await opensearchClient.search({
-        index: indexName,
+      const result = await this.connection.invokeOpenSearch({
+        operation: 'search',
         body: {
           size: limit,
           query: {
@@ -153,7 +155,7 @@ class SearchService {
         }
       });
 
-      const hits = response.body?.hits?.hits || [];
+      const hits = result?.hits?.hits || [];
       const results = hits.map((hit) => ({
         id: hit._source.postgres_id,
         name: hit._source.chemical_name,

package/package.json

@@ -1,13 +1,14 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.0",
+  "version": "1.2.2",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",
   "scripts": {
     "test": "vitest run",
     "test:watch": "vitest",
-    "test:ui": "vitest --ui"
+    "test:ui": "vitest --ui",
+    "test:e2e": "vitest run --config vitest.e2e.config.js"
   },
   "keywords": [
     "elasticsearch",
@@ -25,10 +26,9 @@
   "license": "MIT",
   "dependencies": {
     "@toxplanet/tphelper": "1.2.8",
-    "@opensearch-project/opensearch": "^2.5.0",
+    "@aws-sdk/client-lambda": "^3.490.0",
     "@aws-sdk/client-rds-data": "^3.490.0",
-    "@aws-sdk/client-sqs": "^3.490.0",
-    "@aws-sdk/credential-providers": "^3.490.0"
+    "@aws-sdk/client-sqs": "^3.490.0"
   },
   "engines": {
     "node": ">=18.0.0"