fix(search): compare query and ticket words by the same singular key in the Jira floor

Rebuilding plural forms from a word's key missed singular words whose plural adds -es (release, database, license): their key drops the final e, so neither the word nor its plural was found and the floor dropped valid partial matches. Both sides now go through the same key, as an analyzer does for query and document, with a term containing punctuation matched as written.

Author: waleedlatif1

apps/sim/lib/sim-search/live/term-fallback.test.ts

@@ -250,6 +250,34 @@ describe('searchWithAnyTermFallback', () => {
       expect(result.documents.map((document) => document.id)).toEqual(['plurals'])
     })
 
+    it('counts a singular word whose plural adds -es, in either form', async () => {
+      const search = provider(
+        page(),
+        page(
+          { id: 'singular', title: 'Database release' },
+          { id: 'plural', title: 'Databases', content: 'Weekly releases' }
+        )
+      )
+      const result = await searchWithAnyTermFallback(
+        search,
+        { query: 'database release checklist owner', limit: 10, scopes: [] },
+        live,
+        halfTheTerms
+      )
+      expect(result.documents.map((document) => document.id).sort()).toEqual(['plural', 'singular'])
+    })
+
+    it('counts a term with punctuation inside it, such as ci/cd', async () => {
+      const search = provider(page(), page({ id: 'ci', title: 'CI/CD pipeline' }))
+      const result = await searchWithAnyTermFallback(
+        search,
+        { query: 'ci/cd pipeline failure owner', limit: 10, scopes: [] },
+        live,
+        halfTheTerms
+      )
+      expect(result.documents.map((document) => document.id)).toEqual(['ci'])
+    })
+
     it('returns the all-terms page unchanged when no partial match is close enough', async () => {
       const strict = page({ id: 'exact', title: 'Glean token creator role request' })
       const search = provider(strict, page({ id: 'loose', title: 'Role request' }))

apps/sim/lib/sim-search/live/text-match.ts

@@ -117,17 +117,15 @@ export function termCoverage(text: string, terms: readonly string[]): number {
 }
 
 /**
- * Whether the text holds at least half the query's distinct terms, a word and its regular plural
- * counting as one term as a stemmed search matches them.
+ * Whether the text holds at least half the query's distinct terms. Query and text words are
+ * compared by their singular key, so a word and its regular plural count as one term as a stemmed
+ * search matches them; a term with punctuation inside it is matched as written.
  */
 export function containsHalfTheTerms(text: string, query: string): boolean {
-  const terms = distinctTerms(query, singularKey).map(singularKey)
-  const present = terms.filter((term) => findTermMatches(text, wordForms(term)).length > 0)
+  const terms = distinctTerms(query, singularKey)
+  const words = new Set(text.match(/[\p{L}\p{N}_]+/gu)?.map(singularKey))
+  const present = terms.filter(
+    (term) => words.has(singularKey(term)) || findTermMatches(text, [term]).length > 0
+  )
   return present.length >= Math.ceil(terms.length / 2)
 }
-
-/** A singular word and its regular plural forms, longest first so a plural is matched whole. */
-function wordForms(singular: string): string[] {
-  const forms = [`${singular}es`, `${singular}s`, singular]
-  return singular.endsWith('y') ? [`${singular.slice(0, -1)}ies`, ...forms] : forms
-}