BugFix: Common Name Parsing (#246)

* Fix common names parsing bug * Add commonNames parser util function
bcgov · Jun 25, 2024 · f3bad73 · f3bad73
1 parent c8527c0
commit f3bad73
Show file tree

Hide file tree

Showing 4 changed files with 66 additions and 17 deletions.
diff --git a/api/src/services/itis-service.ts b/api/src/services/itis-service.ts
@@ -1,12 +1,12 @@
 import axios from 'axios';
-import { sortTaxonSearchResults } from '../utils/itis-sort';
+import { getItisTaxonCommonNames, sortTaxonSearchResults } from '../utils/itis-utils';
 import { getLogger } from '../utils/logger';
 import { TaxonSearchResult } from './taxonomy-service';
 
 const defaultLog = getLogger('services/itis-service');
 
 export type ItisSolrSearchResponse = {
-  commonNames: string[];
+  commonNames?: string[];
   kingdom: string;
   name: string;
   parentTSN: string;
@@ -83,8 +83,7 @@ export class ItisService {
    */
   _sanitizeItisData = (data: ItisSolrSearchResponse[]): TaxonSearchResult[] => {
     return data.map((item: ItisSolrSearchResponse) => {
-      const englishNames = item.commonNames?.filter((name) => name.split('$')[2] === 'English');
-      const commonNames = englishNames?.map((name) => name.split('$')[1]) ?? [];
+      const commonNames = getItisTaxonCommonNames(item.commonNames);
 
       return {
         tsn: Number(item.tsn),

diff --git a/api/src/services/taxonomy-service.ts b/api/src/services/taxonomy-service.ts
@@ -1,5 +1,6 @@
 import { IDBConnection } from '../database/db';
 import { TaxonomyRepository, TaxonRecord } from '../repositories/taxonomy-repository';
+import { getItisTaxonCommonNames } from '../utils/itis-utils';
 import { getLogger } from '../utils/logger';
 import { ItisService, ItisSolrSearchResponse } from './itis-service';
 
@@ -72,18 +73,7 @@ export class TaxonomyService {
    * @memberof TaxonomyService
    */
   async addItisTaxonRecord(itisSolrResponse: ItisSolrSearchResponse): Promise<TaxonRecord> {
-    const commonNames =
-      itisSolrResponse.commonNames
-        .filter((name) => name.split('$')[2] === 'English')
-        .map((name) => name.split('$')[1]) ?? [];
-    /* Sample itisResponse:
-     * commonNames: [
-     *   '$withered wooly milk-vetch$English$N$152846$2012-12-21 00:00:00$',
-     *   '$woolly locoweed$English$N$124501$2011-06-29 00:00:00$',
-     *   '$Davis Mountains locoweed$English$N$124502$2011-06-29 00:00:00$',
-     *   '$woolly milkvetch$English$N$72035$2012-12-21 00:00:00$'
-     * ]
-     */
+    const commonNames = getItisTaxonCommonNames(itisSolrResponse?.commonNames);
 
     return this.taxonRepository.addItisTaxonRecord(
       Number(itisSolrResponse.tsn),

diff --git a/api/src/utils/itis-sort.test.ts → api/src/utils/itis-utils.test.ts b/api/src/utils/itis-sort.test.ts → api/src/utils/itis-utils.test.ts
@@ -1,7 +1,7 @@
 import { expect } from 'chai';
 import { describe } from 'mocha';
 import { TaxonSearchResult } from '../services/taxonomy-service';
-import { sortTaxonSearchResults } from './itis-sort';
+import { getItisTaxonCommonNames, sortTaxonSearchResults } from './itis-utils';
 
 describe('itis-sort', () => {
   describe('sortTaxonSearchResults', () => {
@@ -78,3 +78,42 @@ describe('itis-sort', () => {
     });
   });
 });
+
+describe('getItisTaxonCommonNames', () => {
+  it('Returns an empty array if provided common names is undefined', () => {
+    const rawCommonNames = undefined;
+
+    const commonNames = getItisTaxonCommonNames(rawCommonNames);
+
+    expect(commonNames).to.eql([]);
+  });
+
+  it('Returns an empty array if the provided common names is empty', () => {
+    const rawCommonNames: string[] = [];
+
+    const commonNames = getItisTaxonCommonNames(rawCommonNames);
+
+    expect(commonNames).to.eql([]);
+  });
+
+  it('Returns an array of english common names', () => {
+    const rawCommonNames = [
+      '$withered wooly milk-vetch (German)$German$N$152846$2012-12-21 00:00:00$',
+      '$withered wooly milk-vetch$English$N$152846$2012-12-21 00:00:00$',
+      '$woolly locoweed$English$N$124501$2011-06-29 00:00:00$',
+      '$Davis Mountains locoweed (French)$French$N$124502$2011-06-29 00:00:00$',
+      '$Davis Mountains locoweed$English$N$124502$2011-06-29 00:00:00$',
+      '$woolly milkvetch$English$N$72035$2012-12-21 00:00:00$',
+      '$woolly milkvetch (French)$French$N$124502$2011-06-29 00:00:00$'
+    ];
+
+    const commonNames = getItisTaxonCommonNames(rawCommonNames);
+
+    expect(commonNames).to.eql([
+      'withered wooly milk-vetch',
+      'woolly locoweed',
+      'Davis Mountains locoweed',
+      'woolly milkvetch'
+    ]);
+  });
+});
diff --git a/api/src/utils/itis-sort.ts → api/src/utils/itis-utils.ts b/api/src/utils/itis-sort.ts → api/src/utils/itis-utils.ts
@@ -87,3 +87,24 @@ export const sortTaxonSearchResults = (
   // Sort the data by the score
   return taxonSearchResults.sort((a, b) => calculateScore(b) - calculateScore(a));
 };
+
+/**
+ * Parse the raw common names string from an ITIS taxon record into an array of english common names.
+ *
+ * @example
+ * const commonNames = [
+ *   '$withered wooly milk-vetch$English$N$152846$2012-12-21 00:00:00$',
+ *   '$woolly locoweed$English$N$124501$2011-06-29 00:00:00$',
+ *   '$Davis Mountains locoweed$English$N$124502$2011-06-29 00:00:00$',
+ *   '$woolly milkvetch$English$N$72035$2012-12-21 00:00:00$'
+ * ]
+ *
+ * const result = getItisTaxonCommonNames(commonNames)
+ * // result: ['withered wooly milk-vetch', 'woolly locoweed', 'Davis Mountains locoweed', 'woolly milkvetch']
+ *
+ * @param {string[]} [commonNames]
+ * @memberof TaxonomyService
+ */
+export const getItisTaxonCommonNames = (commonNames?: string[]): string[] => {
+  return commonNames?.filter((name) => name.split('$')[2] === 'English').map((name) => name.split('$')[1]) ?? [];
+};