@toxplanet/pegasus-sdk 1.2.7 → 1.2.8

package/lib/chemicals.js

@@ -8,10 +8,6 @@ const SEARCH_BOOST_PREFIX_SECONDARY = 10;
 
 const ALLOWED_IDENTIFIER_TYPES = new Set(['CAS', 'SMILES', 'InChI', 'InChIKey', 'PubChem', 'CID', 'DTXSID', 'EINECS', 'EC']);
 
-const INCHIKEY_PATTERN = /^[A-Z0-9]{14}-[A-Z0-9]{8,10}-[A-Z0-9]$/;
-const EC_PATTERN = /^\d{3}-\d{3}-\d$/;
-const CID_PATTERN = /^CID/i;
-
 function escapeLikePattern(value) {
   return value.replace(/[%_\\]/g, '\\$&');
 }
@@ -65,18 +61,29 @@ function transformChemicalMeta(meta) {
 
 function transformChemicalIdentifiers(identifiers) {
   if (!identifiers || typeof identifiers !== 'object') return [];
+
   if (Array.isArray(identifiers)) {
-    // If it's already in new format, return as-is
-    if (identifiers.length > 0 && identifiers[0].type !== undefined) {
-      return identifiers;
-    }
-    // Transform from old format { identifier_key, identifier_value, ... } to new format { type, value }
+    if (identifiers.length === 0) return [];
+    if (identifiers[0].type !== undefined) return identifiers;
+    // Legacy { identifier_key, identifier_value } row form
     return identifiers.map(item => ({
       type: item.identifier_key || item.type,
       value: Array.isArray(item.identifier_value) ? item.identifier_value[0] : (item.value || item.identifier_value)
     }));
   }
-  return [];
+
+  // Legacy ES-doc form: { CAS: ["71-43-2"], CID: ["241"], InChIKey: "ABC..." }
+  // Flatten one entry per value, preserving multi-valued types.
+  const result = [];
+  for (const [type, value] of Object.entries(identifiers)) {
+    if (value == null) continue;
+    const values = Array.isArray(value) ? value : [value];
+    for (const v of values) {
+      if (v == null || v === '') continue;
+      result.push({ type, value: String(v) });
+    }
+  }
+  return result;
 }
 
 class ChemicalsService {
@@ -658,7 +665,7 @@ class ChemicalsService {
       }
 
       await this.connection.ensureConnected();
-      
+
       const sql = `SELECT * FROM chemicals WHERE chemical_identifiers->>'${identifierType}' = :value OR chemical_identifiers->'${identifierType}' ? :value`;
       const params = [{ name: 'value', value: { stringValue: identifierValue } }];
       const result = await this.connection.query(sql, params);
@@ -669,6 +676,36 @@ class ChemicalsService {
     }
   }
 
+  async findChemicalByIdentifier(identifierValue) {
+    if (!identifierValue) return null;
+    const result = await this.connection.invokeOpenSearch({
+      operation: 'search',
+      body: {
+        size: 1,
+        query: {
+          bool: {
+            should: [
+              { term: { 'cas_numbers.keyword': identifierValue } },
+              { term: { 'identifier_values.keyword': identifierValue } }
+            ],
+            minimum_should_match: 1
+          }
+        },
+        _source: ['postgres_id', 'chemical_name', 'cas_numbers', 'identifier_values', 'synonyms']
+      }
+    });
+
+    const hit = result?.hits?.hits?.[0]?._source;
+    if (!hit) return null;
+    return {
+      id: hit.postgres_id,
+      name: hit.chemical_name,
+      cas: hit.cas_numbers || [],
+      identifiers: hit.identifier_values || [],
+      synonyms: hit.synonyms || []
+    };
+  }
+
   async countByCollection(collectionName) {
     try {
       await this.connection.ensureConnected();
@@ -1065,29 +1102,24 @@ class ChemicalsService {
     
       get: async (params) => {
         const id = params.id;
-        let chemical = await this.getChemicalBySourceId(id);
-
-        if (!chemical) {
-          const casMatches = await this.getChemicalsByCAS(id);
-          if (casMatches.length > 0) chemical = casMatches[0];
-        }
+        let source = null;
 
-        if (!chemical) {
-          let identifierType = null;
-          if (CID_PATTERN.test(id)) {
-            identifierType = 'CID';
-          } else if (INCHIKEY_PATTERN.test(id)) {
-            identifierType = 'InChIKey';
-          } else if (EC_PATTERN.test(id)) {
-            identifierType = 'EC';
-          }
-          if (identifierType) {
-            const matches = await this.getChemicalsByIdentifier(identifierType, id);
-            if (matches.length > 0) chemical = matches[0];
+        const chemical = await this.getChemicalBySourceId(id);
+        if (chemical) {
+          source = this._chemicalRowToLegacySource(chemical);
+        } else {
+          const hit = await this.findChemicalByIdentifier(id);
+          if (hit) {
+            source = this._toLegacyChemicalSource({
+              name: hit.name,
+              cas: hit.cas,
+              identifiers: hit.identifiers,
+              id: hit.id
+            });
           }
         }
 
-        if (!chemical) {
+        if (!source) {
           return {
             body: {
               _index: params.index,
@@ -1104,7 +1136,7 @@ class ChemicalsService {
             _id: params.id,
             _version: 1,
             found: true,
-            _source: this._chemicalRowToLegacySource(chemical)
+            _source: source
           },
           statusCode: 200
         };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.7",
+  "version": "1.2.8",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",