feat: cyrillic like query support (#1078)

src/mongoose/sanitizeFormattedValue.ts

@@ -97,7 +97,7 @@ export const sanitizeQueryValue = (schemaType: SchemaType, path: string, operato
     }
 
     if (operator === 'like' && typeof formattedValue === 'string') {
-      const $regex = wordBoundariesRegex(formattedValue)
+      const $regex = wordBoundariesRegex(formattedValue);
       formattedValue = { $regex };
     }
   }

src/utilities/wordBoundariesRegex.ts

@@ -1,7 +1,12 @@
 export default (input: string): RegExp => {
   const words = input.split(' ');
+
+  // Regex word boundaries that work for cyrillic characters - https://stackoverflow.com/a/47062016/1717697
+  const wordBoundaryBefore = '(?:(?<=[^\\p{L}\\p{N}])|^)';
+  const wordBoundaryAfter = '(?=[^\\p{L}\\p{N}]|$)';
+
   const regex = words.reduce((pattern, word, i) => {
-    return `${pattern}(?=.*\\b${word}.*\\b)${i + 1 === words.length ? '.+' : ''}`;
+    return `${pattern}(?=.*${wordBoundaryBefore}${word}.*${wordBoundaryAfter})${i + 1 === words.length ? '.+' : ''}`;
   }, '');
   return new RegExp(regex, 'i');
 };

test/collections-rest/int.spec.ts

@@ -373,6 +373,22 @@ describe('collections-rest', () => {
         expect(result.totalDocs).toEqual(1);
       });
 
+      it('like - cyrillic characters', async () => {
+        const post1 = await createPost({ title: 'Тест' });
+
+        const { status, result } = await client.find<Post>({
+          query: {
+            title: {
+              like: 'Тест',
+            },
+          },
+        });
+
+        expect(status).toEqual(200);
+        expect(result.docs).toEqual([post1]);
+        expect(result.totalDocs).toEqual(1);
+      });
+
       it('like - partial word match', async () => {
         const post = await createPost({ title: 'separate words should partially match' });