@toxplanet/pegasus-sdk 1.2.5 → 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 {
@@ -734,7 +792,7 @@ class ChemicalsService {
    */
   async searchByName(searchTerm, limit = 10) {
     if (!searchTerm) {
-      return { results: [] };
+      return { results: [], total: { value: 0, relation: 'eq' } };
     }
 
     try {
@@ -771,7 +829,10 @@ class ChemicalsService {
         score: hit._score
       }));
 
-      return { results };
+      return {
+        results,
+        total: result?.hits?.total ?? { value: results.length, relation: 'eq' }
+      };
     } catch (error) {
       logError('pegasus-sdk', 'ChemicalsService', 'searchByName', error);
       throw error;
@@ -786,7 +847,7 @@ class ChemicalsService {
    */
   async searchBySynonym(synonymTerm, limit = 10) {
     if (!synonymTerm) {
-      return { results: [] };
+      return { results: [], total: { value: 0, relation: 'eq' } };
     }
 
     try {
@@ -823,7 +884,10 @@ class ChemicalsService {
         score: hit._score
       }));
 
-      return { results };
+      return {
+        results,
+        total: result?.hits?.total ?? { value: results.length, relation: 'eq' }
+      };
     } catch (error) {
       logError('pegasus-sdk', 'ChemicalsService', 'searchBySynonym', error);
       throw error;
@@ -1000,9 +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 (!result) {
+        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 (!chemical) {
           return {
             body: {
               _index: params.index,
@@ -1019,7 +1104,7 @@ class ChemicalsService {
             _id: params.id,
             _version: 1,
             found: true,
-            _source: result
+            _source: this._chemicalRowToLegacySource(chemical)
           },
           statusCode: 200
         };
@@ -1082,6 +1167,13 @@ class ChemicalsService {
         let searchTerm = '';
         let limit = params.body?.size || 10;
         
+        const toLegacySource = (r) => this._toLegacyChemicalSource({
+          name: r.name,
+          cas: r.cas || [],
+          identifiers: r.identifiers || [],
+          id: r.id
+        });
+
         if (params.index === 'synonym_lookup_index') {
           const query = params.body?.query;
           searchTerm = query?.match?.chemical_name ||
@@ -1100,22 +1192,13 @@ class ChemicalsService {
                 failed: 0
               },
               hits: {
-                total: {
-                  value: searchResults.results.length,
-                  relation: 'eq'
-                },
+                total: searchResults.total ?? { value: searchResults.results.length, relation: 'eq' },
                 max_score: searchResults.results[0]?.score || 0,
                 hits: searchResults.results.map(result => ({
                   _index: params.index,
                   _id: result.id,
                   _score: result.score,
-                  _source: {
-                    postgres_id: result.id,
-                    chemical_name: result.name,
-                    cas_numbers: result.cas,
-                    identifier_values: result.identifiers,
-                    synonyms: result.synonyms
-                  }
+                  _source: toLegacySource(result)
                 }))
               }
             },
@@ -1139,22 +1222,13 @@ class ChemicalsService {
                 failed: 0
               },
               hits: {
-                total: {
-                  value: searchResults.results.length,
-                  relation: 'eq'
-                },
+                total: searchResults.total ?? { value: searchResults.results.length, relation: 'eq' },
                 max_score: searchResults.results[0]?.score || 0,
                 hits: searchResults.results.map(result => ({
                   _index: params.index,
                   _id: result.id,
                   _score: result.score,
-                  _source: {
-                    postgres_id: result.id,
-                    chemical_name: result.name,
-                    cas_numbers: result.cas,
-                    identifier_values: result.identifiers,
-                    synonyms: result.synonyms
-                  }
+                  _source: toLegacySource(result)
                 }))
               }
             },

package/lib/search.js

@@ -94,7 +94,7 @@ class SearchService {
    */
   async searchChemicals(query, options = {}) {
     if (!query) {
-      return { results: [] };
+      return { results: [], total: { value: 0, relation: 'eq' } };
     }
 
     // Extract options with defaults
@@ -165,7 +165,10 @@ class SearchService {
         score: hit._score
       }));
 
-      return { results };
+      return {
+        results,
+        total: result?.hits?.total ?? { value: results.length, relation: 'eq' }
+      };
     } catch (error) {
       logError('pegasus-sdk', 'SearchService', 'searchChemicals', error);
       throw error;
@@ -356,7 +359,7 @@ class SearchService {
                     chemical_set_identifier: (r.cas && r.cas[0]) || r.id || '',
                   }
                 })),
-                total: { value: pegasusResults.results.length, relation: 'eq' }
+                total: pegasusResults.total ?? { value: pegasusResults.results.length, relation: 'eq' }
               },
               timed_out: false,
               _shards: { total: 1, successful: 1, failed: 0 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.5",
+  "version": "1.2.7",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",