String helpers: performance and safety improvements

packages/helpers/src/strings.js

@@ -1,13 +1,9 @@
 const invisibleCharRegex = '[\\s\\p{C}\u034f\u17b4\u17b5\u2800\u115f\u1160\u3164\uffa0]';
-const trimStartRegex = new RegExp('^' + invisibleCharRegex, 'u');
+const trimStartRegex = new RegExp('^' + invisibleCharRegex + '+', 'u');
 const trimEndRegex = new RegExp(invisibleCharRegex + '$', 'u');
 
 export function trimStart(str) {
-  while (trimStartRegex.test(str)) {
-    str = str.replace(trimStartRegex, '');
-  }
-
-  return str;
+  return str.replace(trimStartRegex, '');
 }
 
 export function trimEnd(str) {
@@ -29,8 +25,15 @@ export function truncate(str, maxLength) {
 
   const ellipsis = ELLIPSIS.slice(0, maxLength);
 
-  const graphemes = Array.from(graphemeSegmenter.segment(str), (s) => s.segment);
-  if (graphemes.length < maxLength) return str;
+  const graphemes = [];
+
+  for (const s of graphemeSegmenter.segment(str)) {
+    if (graphemes.length > maxLength) {
+      return trimEnd(graphemes.slice(0, maxLength - ellipsis.length).join('')) + ellipsis;
+    }
 
-  return trimEnd(graphemes.slice(0, maxLength - ellipsis.length).join('')) + ellipsis;
+    graphemes.push(s.segment);
+  }
+
+  return str;
 }

packages/helpers/src/strings.test.js

@@ -3,30 +3,68 @@ import { trimEnd, trimStart, truncate } from './strings.js';
 describe('helpers', () => {
   describe('trimStart', () => {
     it('should remove invisible characters from the start of a string', () => {
-      const input = '\u034f\u034fHello';
-      const result = trimStart(input);
-      expect(result).toBe('Hello');
+      expect(trimStart(' A')).toBe('A');
+      expect(trimStart('\nB')).toBe('B');
+      expect(trimStart('\tC')).toBe('C');
+      expect(trimStart('\rD')).toBe('D');
+      expect(trimStart('\u034fE')).toBe('E');
+      expect(trimStart('\u17b4F')).toBe('F');
+      expect(trimStart('\u17b5G')).toBe('G');
+      expect(trimStart('\u2800F')).toBe('F');
+      expect(trimStart('\u115fH')).toBe('H');
+      expect(trimStart('\u1160I')).toBe('I');
+      expect(trimStart('\u3164J')).toBe('J');
+      expect(trimStart('\uffa0K')).toBe('K');
     });
 
     it('should return the same string if no invisible characters are at the start', () => {
       const input = 'Hello';
       const result = trimStart(input);
       expect(result).toBe('Hello');
     });
+
+    it('should not hang (ReDoS) on long string of invisible chars', () => {
+      const long = '\u034f'.repeat(1_000_000) + 'Hello';
+      const result = trimStart(long);
+      expect(result).toBe('Hello');
+    }, 200);
   });
 
   describe('trimEnd', () => {
     it('should remove invisible characters from the end of a string', () => {
-      const input = 'Hello\u034f\u034f';
-      const result = trimEnd(input);
-      expect(result).toBe('Hello');
+      expect(trimEnd('A ')).toBe('A');
+      expect(trimEnd('B\n')).toBe('B');
+      expect(trimEnd('C\t')).toBe('C');
+      expect(trimEnd('D\r')).toBe('D');
+      expect(trimEnd('E\u034f')).toBe('E');
+      expect(trimEnd('F\u17b4')).toBe('F');
+      expect(trimEnd('G\u17b5')).toBe('G');
+      expect(trimEnd('F\u2800')).toBe('F');
+      expect(trimEnd('H\u115f')).toBe('H');
+      expect(trimEnd('I\u1160')).toBe('I');
+      expect(trimEnd('J\u3164')).toBe('J');
+      expect(trimEnd('K\uffa0')).toBe('K');
     });
 
     it('should return the same string if no invisible characters are at the end', () => {
       const input = 'Hello';
       const result = trimEnd(input);
       expect(result).toBe('Hello');
     });
+
+    it('should not hang (ReDoS) on long string of invisible chars', () => {
+      const long1 = 'Hello' + '\u3164'.repeat(1_000_000);
+      const long2 = 'A'.repeat(1_000_000) + '\u17b5'.repeat(1_000_000);
+      const long3 = 'B' + '\u034f'.repeat(1_000_000) + 'C';
+      const long4 = '\u2800'.repeat(1_000_000) + 'D';
+      const long5 = 'E'.repeat(1_000_000) + '\u034f' + 'F';
+
+      expect.soft(trimEnd(long1)).toBe('Hello');
+      expect.soft(trimEnd(long2)).toBe('A'.repeat(1_000_000));
+      expect.soft(trimEnd(long3)).toBe(long3);
+      expect.soft(trimEnd(long4)).toBe(long4);
+      expect.soft(trimEnd(long5)).toBe(long5);
+    }, 1_000);
   });
 
   describe('truncate', () => {
@@ -116,5 +154,22 @@ describe('helpers', () => {
       expect(truncate(input, 19)).toBe('Truncate this 👩‍❤️‍💋‍👨...');
       expect(truncate(input, 20)).toBe('Truncate this 👩‍❤️‍💋‍👨 s...');
     });
+
+    it('should handle very long strings efficiently and correctly', () => {
+      const longString = '👩‍❤️‍💋‍👨a'.repeat(100_000); // > 1MB
+      const result = truncate(longString, 100);
+
+      expect.soft(result.endsWith('...')).toBeTruthy();
+      expect.soft(result.startsWith('👩‍❤️‍💋‍👨a'.repeat(48))).toBeTruthy();
+      expect.soft(result === '👩‍❤️‍💋‍👨a'.repeat(48) + '👩‍❤️‍💋‍👨' + '...').toBeTruthy();
+
+      // Truncate to a length less than ellipsis
+      const result2 = truncate(longString, 2);
+      expect.soft(result2).toBe('..');
+
+      // Truncate to a length greater than the string
+      const result3 = truncate(longString, 200_000);
+      expect(result3).toBe(longString);
+    }, 100);
   });
 });