fix(inflekt): add compound *base word exceptions to singularize (codebases → codebase)

Author: pyramation

packages/inflekt/__tests__/inflection.test.ts

@@ -49,6 +49,19 @@ describe('singularize', () => {
     expect(singularize('SCHEMATA')).toBe('SCHEMA');
   });
 
+  it('should singularize compound *base words correctly', () => {
+    expect(singularize('codebases')).toBe('codebase');
+    expect(singularize('databases')).toBe('database');
+    expect(singularize('firebases')).toBe('firebase');
+    expect(singularize('knowledgebases')).toBe('knowledgebase');
+    expect(singularize('Codebases')).toBe('Codebase');
+    expect(singularize('Databases')).toBe('Database');
+  });
+
+  it('should still singularize "bases" to "basis"', () => {
+    expect(singularize('bases')).toBe('basis');
+  });
+
   it('should canonicalize malformed trailing triple-s words', () => {
     expect(singularize('classs')).toBe('class');
     expect(singularize('Classs')).toBe('Class');

packages/inflekt/src/pluralize.ts

@@ -25,6 +25,13 @@ const LATIN_SUFFIX_OVERRIDES: Array<[string, string]> = [
   ['data', 'datum'],
 ];
 
+/**
+ * Compound words ending in "base" that the inflection library incorrectly
+ * singularizes via its (b)a branch in the ses$ rule (e.g. codebases -> codebasis).
+ * We intercept these before delegating to inflection.singularize().
+ */
+const COMPOUND_BASE_REGEX = /(database|codebase|firebase|knowledgebase)s$/i;
+
 const TRAILING_TRIPLE_S_REGEX = /[sS]{3,}$/;
 const TRAILING_TRIPLE_S_BEFORE_ES_REGEX = /[sS]{3,}(?=e[sS]$)/;
 
@@ -96,6 +103,13 @@ export function singularize(word: string): string {
     }
   }
 
+  // Compound *base words: the inflection library's (b)a branch in the ses$
+  // rule incorrectly produces "codebasis" instead of "codebase".
+  const baseMatch = normalizedWord.match(COMPOUND_BASE_REGEX);
+  if (baseMatch) {
+    return normalizedWord.slice(0, -1);
+  }
+
   return normalizeMalformedDoubleS(inflection.singularize(normalizedWord));
 }