@toxplanet/pegasus-sdk 1.2.6 → 1.2.7

package/lib/chemicals.js

@@ -6,7 +6,11 @@ const SEARCH_BOOST_PREFIX_PRIMARY = 50;
 const SEARCH_BOOST_EXACT_SECONDARY = 30;
 const SEARCH_BOOST_PREFIX_SECONDARY = 10;
 
-const ALLOWED_IDENTIFIER_TYPES = new Set(['CAS', 'SMILES', 'InChI', 'InChIKey', 'PubChem', 'DTXSID', 'EINECS', 'EC']);
+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, '\\$&');
@@ -94,6 +98,60 @@ class ChemicalsService {
     return d instanceof Date ? d.toISOString() : (d || new Date().toISOString());
   }
 
+  _toLegacyChemicalSource({ name = '', cas = [], identifiers = [], id = '', formula = null } = {}) {
+    return {
+      chemical_name: name,
+      chemical_name_sensitive: name,
+      chemical_name_sort: (name || '').toLowerCase(),
+      chemical_identifier: [...cas, ...identifiers],
+      chemical_set_identifier: cas[0] || id || '',
+      chemical_formula: formula
+    };
+  }
+
+  _chemicalRowToLegacySource(chemical) {
+    if (!chemical) return null;
+    const identifierObj = chemical.chemicalIdentifiers;
+    const cas = [];
+    const other = [];
+    let formula = null;
+
+    const visit = (type, value) => {
+      if (!type) return;
+      const values = Array.isArray(value) ? value : (value != null ? [value] : []);
+      const stringValues = values.map((v) => String(v)).filter((v) => v.length > 0);
+      if (type === 'CAS') {
+        cas.push(...stringValues);
+      } else if (type === 'formula' || type === 'Formula' || type === 'chemical_formula') {
+        if (formula == null && stringValues.length > 0) formula = stringValues[0];
+      } else {
+        other.push(...stringValues);
+      }
+    };
+
+    if (Array.isArray(identifierObj)) {
+      for (const item of identifierObj) {
+        if (item && typeof item === 'object') {
+          const type = item.type || item.identifier_key;
+          const value = item.value !== undefined ? item.value : item.identifier_value;
+          visit(type, value);
+        }
+      }
+    } else if (identifierObj && typeof identifierObj === 'object') {
+      for (const [type, value] of Object.entries(identifierObj)) {
+        visit(type, value);
+      }
+    }
+
+    return this._toLegacyChemicalSource({
+      name: chemical.chemicalName || '',
+      cas,
+      identifiers: other,
+      id: chemical.sourceId,
+      formula
+    });
+  }
+
   _mapChemicalRow(row) {
     if (!row) return null;
     return {
@@ -1006,9 +1064,30 @@ class ChemicalsService {
       },
     
       get: async (params) => {
-        const result = await this.getChemicalBySourceId(params.id);
+        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];
+        }
+
+        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];
+          }
+        }
 
-        if (!result) {
+        if (!chemical) {
           return {
             body: {
               _index: params.index,
@@ -1025,7 +1104,7 @@ class ChemicalsService {
             _id: params.id,
             _version: 1,
             found: true,
-            _source: result
+            _source: this._chemicalRowToLegacySource(chemical)
           },
           statusCode: 200
         };
@@ -1088,12 +1167,11 @@ class ChemicalsService {
         let searchTerm = '';
         let limit = params.body?.size || 10;
         
-        const toLegacySource = (r) => ({
-          chemical_name: r.name,
-          chemical_name_sensitive: r.name,
-          chemical_name_sort: (r.name || '').toLowerCase(),
-          chemical_identifier: [...(r.cas || []), ...(r.identifiers || [])],
-          chemical_set_identifier: (r.cas && r.cas[0]) || r.id || ''
+        const toLegacySource = (r) => this._toLegacyChemicalSource({
+          name: r.name,
+          cas: r.cas || [],
+          identifiers: r.identifiers || [],
+          id: r.id
         });
 
         if (params.index === 'synonym_lookup_index') {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.6",
+  "version": "1.2.7",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",