@toxplanet/pegasus-sdk 1.2.12 → 1.2.14

package/lib/chemicals.js

@@ -9,6 +9,25 @@ const SEARCH_BOOST_PREFIX_SECONDARY = 10;
 
 const ALLOWED_IDENTIFIER_TYPES = new Set(['CAS', 'SMILES', 'InChI', 'InChIKey', 'PubChem', 'CID', 'DTXSID', 'EINECS', 'EC']);
 
+// Heuristic: does the search term look like a chemical identifier (CAS, EC,
+// InChIKey, CID, etc.) rather than a chemical name? If so, restrict to keyword
+// matches so we don't get noise from tokenized text-field matches on names.
+function looksLikeIdentifier(term) {
+  if (!term) return false;
+  const t = String(term).trim();
+  if (!t) return false;
+  // CAS / EC: digits with hyphen or slash separators
+  if (/^\d{1,7}[-/]\d{1,3}[-/]\d{1,2}$/.test(t)) return true;   // CAS xxxxxxx-xx-x
+  if (/^\d{2,3}-\d{3}-\d$/.test(t)) return true;                // EC xxx-xxx-x
+  // InChIKey: 14-10-1 uppercase letters with hyphens
+  if (/^[A-Z]{14}-[A-Z]{10}-[A-Z]$/.test(t)) return true;
+  // PubChem CID
+  if (/^CID\d+$/i.test(t)) return true;
+  // DTXSID
+  if (/^DTXSID\d+$/i.test(t)) return true;
+  return false;
+}
+
 // Equivalence groups for CDI identifier_key normalization. When any member of
 // a group is present on a row, the others are backfilled with the same values
 // so consumers can look up by their preferred spelling.
@@ -776,6 +795,10 @@ class ChemicalsService {
         query: {
           bool: {
             should: [
+              // v2 mapping: cas_numbers / identifier_values are keyword type.
+              { term: { 'cas_numbers': identifierValue } },
+              { term: { 'identifier_values': identifierValue } },
+              // Back-compat with legacy mapping where these were dynamically text + .keyword.
               { term: { 'cas_numbers.keyword': identifierValue } },
               { term: { 'identifier_values.keyword': identifierValue } }
             ],
@@ -924,6 +947,7 @@ class ChemicalsService {
     }
 
     try {
+      const isIdentifier = looksLikeIdentifier(searchTerm);
       const variations = getCasNumberVariations(searchTerm);
       const shouldClauses = [];
       for (const v of variations) {
@@ -935,12 +959,22 @@ class ChemicalsService {
           { 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 } } }
+          { prefix: { 'identifier_values': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
         );
+        // Skip tokenized-text matches for identifier-shaped queries to avoid
+        // noise from the analyzer splitting "71-43-2" into 71/43/2 tokens.
+        if (!isIdentifier) {
+          shouldClauses.push(
+            // Whitespace-analyzer fields (preserve hyphenated compound names).
+            { 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 } } },
+            // Word-level (standard analyzer) sub-fields — let "ethane" still match "trichloroethane".
+            { match: { 'chemical_name.std': { query: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+            { match: { 'synonyms.std': { query: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
+          );
+        }
       }
 
       const result = await this.connection.invokeOpenSearch({
@@ -989,6 +1023,7 @@ class ChemicalsService {
     }
 
     try {
+      const isIdentifier = looksLikeIdentifier(synonymTerm);
       const variations = getCasNumberVariations(synonymTerm);
       const shouldClauses = [];
       for (const v of variations) {
@@ -1000,12 +1035,22 @@ class ChemicalsService {
           { 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 } } }
+          { prefix: { 'identifier_values': { value: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
         );
+        // Skip tokenized-text matches for identifier-shaped queries to avoid
+        // noise from the analyzer splitting "71-43-2" into 71/43/2 tokens.
+        if (!isIdentifier) {
+          shouldClauses.push(
+            // Whitespace-analyzer fields (preserve hyphenated compound names).
+            { 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 } } },
+            // Word-level (standard analyzer) sub-fields — let "ethane" still match "trichloroethane".
+            { match: { 'synonyms.std': { query: v, boost: SEARCH_BOOST_PREFIX_PRIMARY } } },
+            { match: { 'chemical_name.std': { query: v, boost: SEARCH_BOOST_PREFIX_SECONDARY } } }
+          );
+        }
       }
 
       const result = await this.connection.invokeOpenSearch({

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@toxplanet/pegasus-sdk",
-  "version": "1.2.12",
+  "version": "1.2.14",
   "description": "SDK for migrating chemical data to Pegasus PostgreSQL + OpenSearch architecture with Elasticsearch client compatibility",
   "main": "index.js",
   "type": "commonjs",