Improve performance of data processing module

src/humanize-string.test.ts

@@ -1,49 +1,263 @@
-import { test, expect } from 'vitest';
+import { test, expect, describe } from 'vitest';
 import { humanizeString } from './humanize-string.js';
 
-test('Whitespaces', () => {
-  expect(humanizeString('Hello\u200b\xa0World!  ')).toStrictEqual({
-    text: 'Hello World!',
-    count: 4,
-  });
-  expect(
-    humanizeString('Hello\u200b\xa0World!  ', {
-      transformTrailingWhitespace: false,
-    }),
-  ).toStrictEqual({
-    text: 'Hello World!  ',
-    count: 2,
-  });
-  expect(
-    humanizeString('Hello\u200b\xa0World!  ', {
-      transformHidden: false,
-    }),
-  ).toStrictEqual({
-    text: 'Hello\u200b World!',
-    count: 3,
-  });
-  expect(
-    humanizeString('Hello\u200b\xa0World!  ', {
-      transformNbs: false,
-    }),
-  ).toStrictEqual({
-    text: 'Hello\xa0World!',
-    count: 3,
+describe('Existing functionality', () => {
+  test('Whitespaces', () => {
+    expect(humanizeString('Hello\u200b\xa0World!  ')).toStrictEqual({
+      text: 'Hello World!',
+      count: 4,
+    });
+    expect(
+      humanizeString('Hello\u200b\xa0World!  ', {
+        transformTrailingWhitespace: false,
+      }),
+    ).toStrictEqual({
+      text: 'Hello World!  ',
+      count: 2,
+    });
+    expect(
+      humanizeString('Hello\u200b\xa0World!  ', {
+        transformHidden: false,
+      }),
+    ).toStrictEqual({
+      text: 'Hello\u200b World!',
+      count: 3,
+    });
+    expect(
+      humanizeString('Hello\u200b\xa0World!  ', {
+        transformNbs: false,
+      }),
+    ).toStrictEqual({
+      text: 'Hello\xa0World!',
+      count: 3,
+    });
+  });
+
+  test('Dashes', () => {
+    expect(humanizeString('I — super — man – 💪')).toStrictEqual({
+      text: 'I - super - man - 💪',
+      count: 3,
+    });
+  });
+
+  test('Quotes', () => {
+    expect(
+      humanizeString('Angular "quote" «marks» looks„ like Christmas «« tree'),
+    ).toStrictEqual({
+      text: 'Angular "quote" "marks" looks" like Christmas "" tree',
+      count: 7,
+    });
+  });
+});
+
+describe('UTM Source Removal', () => {
+  test('Basic UTM source removal', () => {
+    const input = 'Check this out: https://de.wikipedia.org/wiki/BorgWarner?utm_source=chatgpt.com';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Check this out: https://de.wikipedia.org/wiki/BorgWarner');
+    expect(result.count).toBeGreaterThan(0);
+  });
+
+  test('UTM source with additional parameters', () => {
+    const input = 'Visit https://example.com?param1=value1&utm_source=chatgpt.com&param2=value2';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Visit https://example.com?param1=value1&param2=value2');
+    expect(result.count).toBeGreaterThan(0);
+  });
+
+  test('UTM source at the end of URL', () => {
+    const input = 'Check https://example.com?param1=value1&utm_source=chatgpt.com';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Check https://example.com?param1=value1');
+    expect(result.count).toBeGreaterThan(0);
+  });
+
+  test('UTM source as only parameter', () => {
+    const input = 'Visit https://example.com?utm_source=chatgpt.com';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Visit https://example.com');
+    expect(result.count).toBeGreaterThan(0);
+  });
+
+  test('Multiple URLs with UTM sources', () => {
+    const input = 'First: https://site1.com?utm_source=test and second: https://site2.com?utm_source=another';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('First: https://site1.com and second: https://site2.com');
+    expect(result.count).toBeGreaterThan(0);
+  });
+
+  test('URL without UTM source (no change)', () => {
+    const input = 'Visit https://example.com?param=value';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Visit https://example.com?param=value');
+    expect(result.count).toBe(0);
+  });
+
+  test('Text without URLs (no change)', () => {
+    const input = 'This is just plain text without any URLs';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('This is just plain text without any URLs');
+    expect(result.count).toBe(0);
+  });
+
+  test('UTM source removal disabled by default', () => {
+    const input = 'Check https://example.com?utm_source=chatgpt.com';
+    const result = humanizeString(input);
+
+    expect(result.text).toBe('Check https://example.com?utm_source=chatgpt.com');
+  });
+});
+
+describe('Trailing Dot Removal', () => {
+  test('Single trailing dot', () => {
+    const input = 'This is a sentence.';
+    const result = humanizeString(input, { removeTrailingDot: true });
+
+    expect(result.text).toBe('This is a sentence');
+    expect(result.count).toBe(1);
+  });
+
+  test('Multiple trailing dots', () => {
+    const input = 'This is a sentence...';
+    const result = humanizeString(input, { removeTrailingDot: true });
+
+    expect(result.text).toBe('This is a sentence');
+    expect(result.count).toBe(3);
+  });
+
+  test('No trailing dots (no change)', () => {
+    const input = 'This is a sentence';
+    const result = humanizeString(input, { removeTrailingDot: true });
+
+    expect(result.text).toBe('This is a sentence');
+    expect(result.count).toBe(0);
+  });
+
+  test('Dots in middle of text (preserved)', () => {
+    const input = 'This is a sentence. With more text.';
+    const result = humanizeString(input, { removeTrailingDot: true });
+
+    expect(result.text).toBe('This is a sentence. With more text');
+    expect(result.count).toBe(1);
+  });
+
+  test('Only dots', () => {
+    const input = '...';
+    const result = humanizeString(input, { removeTrailingDot: true });
+
+    expect(result.text).toBe('');
+    expect(result.count).toBe(3);
+  });
+
+  test('Trailing dot removal disabled by default', () => {
+    const input = 'This is a sentence.';
+    const result = humanizeString(input);
+
+    expect(result.text).toBe('This is a sentence.');
   });
 });
 
-test('Dashes', () => {
-  expect(humanizeString('I — super — man – 💪')).toStrictEqual({
-    text: 'I - super - man - 💪',
-    count: 3,
+describe('Combined Features', () => {
+  test('UTM source removal and trailing dot removal together', () => {
+    const input = 'Check out https://example.com?utm_source=chatgpt.com for more info.';
+    const result = humanizeString(input, { 
+      removeUtmSource: true, 
+      removeTrailingDot: true 
+    });
+
+    expect(result.text).toBe('Check out https://example.com for more info');
+    expect(result.count).toBeGreaterThan(1);
+  });
+
+  test('All transformations enabled', () => {
+    const input = 'Check — https://example.com?utm_source=test — "amazing" site...';
+    const result = humanizeString(input, {
+      transformDashes: true,
+      transformQuotes: true,
+      removeUtmSource: true,
+      removeTrailingDot: true
+    });
+
+    expect(result.text).toBe('Check - https://example.com - "amazing" site');
+    expect(result.count).toBeGreaterThan(0);
   });
 });
 
-test('Quotes', () => {
-  expect(
-    humanizeString('Angular “quote” «marks» looks„ like Christmas «« tree'),
-  ).toStrictEqual({
-    text: 'Angular "quote" "marks" looks" like Christmas "" tree',
-    count: 7,
+describe('Error Handling', () => {
+  test('Invalid input type throws error', () => {
+    expect(() => {
+      // @ts-expect-error - Testing invalid input
+      humanizeString(123);
+    }).toThrow(TypeError);
+  });
+
+  test('Invalid options type throws error', () => {
+    expect(() => {
+      // @ts-expect-error - Testing invalid options
+      humanizeString('test', 'invalid');
+    }).toThrow(TypeError);
+  });
+
+  test('Empty string handling', () => {
+    const result = humanizeString('');
+    expect(result).toStrictEqual({ count: 0, text: '' });
+  });
+
+  test('Null options handling', () => {
+    const result = humanizeString('test', undefined);
+    expect(result.text).toBe('test');
+  });
+
+  test('Partial options handling', () => {
+    const result = humanizeString('test...', { removeTrailingDot: true });
+    expect(result.text).toBe('test');
+    expect(result.count).toBe(3);
+  });
+
+  test('Invalid boolean options are ignored', () => {
+    const result = humanizeString('test...', { 
+      removeTrailingDot: true,
+      // @ts-expect-error - Testing invalid option value
+      invalidOption: 'not a boolean'
+    });
+    expect(result.text).toBe('test');
   });
 });
+
+describe('Performance and Edge Cases', () => {
+  test('Large text processing', () => {
+    const largeText = 'a'.repeat(10000) + '...';
+    const result = humanizeString(largeText, { removeTrailingDot: true });
+
+    expect(result.text).toBe('a'.repeat(10000));
+    expect(result.count).toBe(3);
+  });
+
+  test('Complex URLs with special characters', () => {
+    const input = 'Check https://example.com/path?param1=value1&utm_source=test%20source&param2=value2#anchor';
+    const result = humanizeString(input, { removeUtmSource: true });
+
+    expect(result.text).toBe('Check https://example.com/path?param1=value1&param2=value2#anchor');
+  });
+
+  test('Multiple transformations with count accuracy', () => {
+    const input = 'Text—with""special…chars https://test.com?utm_source=x.';
+    const result = humanizeString(input, {
+      transformDashes: true,
+      transformQuotes: true,
+      transformOther: true,
+      removeUtmSource: true,
+      removeTrailingDot: true
+    });
+
+    expect(result.text).toBe('Text-with""special...chars https://test.com');
+    expect(result.count).toBeGreaterThan(0);
+  });
+});