@toxplanet/pegasus-sdk 1.2.9 → 1.2.12

package/lib/chemicals.js

@@ -1,5 +1,6 @@
 const { logError, logInfo } = require('@toxplanet/tphelper/logging');
 const { SQSClient, SendMessageCommand } = require('@aws-sdk/client-sqs');
+const { getCasNumberVariations } = require('./search');
 
 const SEARCH_BOOST_EXACT_PRIMARY = 100;
 const SEARCH_BOOST_PREFIX_PRIMARY = 50;
@@ -23,6 +24,7 @@ function escapeLikePattern(value) {
 }
 
 function parsePostgresArray(str) {
+  if (Array.isArray(str)) return str.map((v) => String(v));
   if (!str || str === '{}') return [];
   const trimmed = str.slice(1, -1);
   if (!trimmed) return [];
@@ -214,7 +216,7 @@ class ChemicalsService {
         group.identifier_value.push(String(v));
       }
     }
-    const identifiers = this._expandIdentifierAliases(Array.from(identifierGroups.values()));
+    const identifiers = Array.from(identifierGroups.values());
 
     const metaList = Array.isArray(chemical.chemicalMeta) ? chemical.chemicalMeta : [];
     const meta = metaList.map((item) => {
@@ -223,22 +225,22 @@ class ChemicalsService {
       const valueArr = Array.isArray(rawValue)
         ? rawValue.map((v) => String(v))
         : (rawValue != null ? [String(rawValue)] : []);
-      const out = {
+      const unit = item && (item.unit || item.meta_value_unit);
+      if (unit && valueArr.length > 0) {
+        valueArr[valueArr.length - 1] = `${valueArr[valueArr.length - 1]} ${unit}`;
+      }
+      return {
         meta_key: String(key).toLowerCase(),
         meta_value_text: valueArr
       };
-      const unit = item && (item.unit || item.meta_value_unit);
-      if (unit) out.meta_value_unit = unit;
-      return out;
     });
 
     const synonyms = Array.isArray(chemical.chemicalSynonyms) ? chemical.chemicalSynonyms : [];
-    const names = [chemical.chemicalName, ...synonyms].filter(Boolean);
 
     return {
       chemical_set_identifier: chemical.sourceId || '',
       chemical_primary_name: chemical.chemicalName || '',
-      chemical_names: names,
+      chemical_names: synonyms,
       chemical_synonyms: synonyms,
       chemical_categories: chemical.chemicalCategories || [],
       chemical_identifiers: identifiers,
@@ -922,22 +924,32 @@ class ChemicalsService {
     }
 
     try {
+      const variations = getCasNumberVariations(searchTerm);
+      const shouldClauses = [];
+      for (const v of variations) {
+        shouldClauses.push(
+          { term: { 'chemical_name.keyword': { value: v, boost: SEARCH_BOOST_EXACT_PRIMARY, case_insensitive: true } } },
+          { prefix: { 'chemical_name.keyword': { value: v, boost: SEARCH_BOOST_PREFIX_PRIMARY, case_insensitive: true } } },
+          { term: { 'synonyms.keyword': { value: v, boost: SEARCH_BOOST_EXACT_SECONDARY, case_insensitive: true } } },
+          { prefix: { 'synonyms.keyword': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } },
+          { term: { 'cas_numbers': { value: v, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+          { prefix: { 'cas_numbers': { value: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+          { term: { 'identifier_values': { value: v, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+          { prefix: { 'identifier_values': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } },
+          { match: { 'chemical_name': { query: v, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+          { match: { 'synonyms': { query: v, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+          { match_phrase_prefix: { 'chemical_name': { query: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+          { match_phrase_prefix: { 'synonyms': { query: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
+        );
+      }
+
       const result = await this.connection.invokeOpenSearch({
         operation: 'search',
         body: {
           size: limit,
           query: {
             bool: {
-              should: [
-                { 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 } } },
-                { 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 } } }
-              ],
+              should: shouldClauses,
               minimum_should_match: 1
             }
           },
@@ -977,22 +989,32 @@ class ChemicalsService {
     }
 
     try {
+      const variations = getCasNumberVariations(synonymTerm);
+      const shouldClauses = [];
+      for (const v of variations) {
+        shouldClauses.push(
+          { term: { 'synonyms.keyword': { value: v, boost: SEARCH_BOOST_EXACT_PRIMARY, case_insensitive: true } } },
+          { prefix: { 'synonyms.keyword': { value: v, boost: SEARCH_BOOST_PREFIX_PRIMARY, case_insensitive: true } } },
+          { term: { 'chemical_name.keyword': { value: v, boost: SEARCH_BOOST_EXACT_SECONDARY, case_insensitive: true } } },
+          { prefix: { 'chemical_name.keyword': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY, case_insensitive: true } } },
+          { term: { 'cas_numbers': { value: v, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+          { prefix: { 'cas_numbers': { value: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+          { term: { 'identifier_values': { value: v, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+          { prefix: { 'identifier_values': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } },
+          { match: { 'synonyms': { query: v, boost: SEARCH_BOOST_EXACT_PRIMARY } } },
+          { match: { 'chemical_name': { query: v, boost: SEARCH_BOOST_EXACT_SECONDARY } } },
+          { match_phrase_prefix: { 'synonyms': { query: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+          { match_phrase_prefix: { 'chemical_name': { query: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
+        );
+      }
+
       const result = await this.connection.invokeOpenSearch({
         operation: 'search',
         body: {
           size: limit,
           query: {
             bool: {
-              should: [
-                { 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 } } },
-                { 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 } } }
-              ],
+              should: shouldClauses,
               minimum_should_match: 1
             }
           },
@@ -1283,7 +1305,7 @@ class ChemicalsService {
       search: async (params) => {
         let searchTerm = '';
         let limit = params.body?.size || 10;
-        
+
         const toLegacySource = (r) => this._toLegacyChemicalSource({
           name: r.name,
           cas: r.cas || [],
@@ -1291,11 +1313,21 @@ class ChemicalsService {
           id: r.id
         });
 
+        // Callers may pass an Elasticsearch query_string with reserved chars escaped
+        // (e.g. "71\-43\-2"). Strip the escapes plus a trailing wildcard so keyword
+        // term/prefix clauses against cas_numbers / identifier_values can match.
+        const normalizeQueryString = (raw) => {
+          if (!raw) return '';
+          let s = String(raw);
+          if (s.endsWith('*')) s = s.slice(0, -1);
+          return s.replace(/\\(.)/g, '$1');
+        };
+
         if (params.index === 'synonym_lookup_index') {
           const query = params.body?.query;
           searchTerm = query?.match?.chemical_name ||
                        query?.term?.chemical_name ||
-                       query?.query_string?.query || '';
+                       normalizeQueryString(query?.query_string?.query) || '';
           const searchResults = await this.searchBySynonym(searchTerm, limit);
 
           return {
@@ -1325,7 +1357,7 @@ class ChemicalsService {
           const query = params.body?.query;
           searchTerm = query?.match?.chemical_name ||
                        query?.term?.chemical_name ||
-                       query?.query_string?.query || '';
+                       normalizeQueryString(query?.query_string?.query) || '';
           const searchResults = await this.searchByName(searchTerm, limit);
 
           return {

package/lib/db/index.js

@@ -4,6 +4,16 @@ function getFieldValue(field) {
   if ('longValue' in field) return field.longValue;
   if ('doubleValue' in field) return field.doubleValue;
   if ('booleanValue' in field) return field.booleanValue;
+  if ('blobValue' in field) return field.blobValue;
+  if ('arrayValue' in field && field.arrayValue) {
+    const av = field.arrayValue;
+    if (Array.isArray(av.stringValues)) return av.stringValues;
+    if (Array.isArray(av.longValues)) return av.longValues;
+    if (Array.isArray(av.doubleValues)) return av.doubleValues;
+    if (Array.isArray(av.booleanValues)) return av.booleanValues;
+    if (Array.isArray(av.arrayValues)) return av.arrayValues;
+    return [];
+  }
   return null;
 }
 

package/lib/search.js

@@ -383,3 +383,4 @@ class SearchService {
 }
 
 module.exports = SearchService;
+module.exports.getCasNumberVariations = getCasNumberVariations;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.9",
+  "version": "1.2.12",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",