Assistant Regex avec Claude Code

Solving Regex Headaches with Claude Code

Regular expressions are powerful tools, but complex patterns are hard to read, write, and debug. With Claude Code, you can describe your requirements in natural language and get accurate regex patterns along with test code — all in one go.

If you’re not yet familiar with the basics, start with our Claude Code Getting Started Guide.

Basics: Generating Regex from Natural Language

In your project directory, enter a prompt like this:

> Create a TypeScript regex that matches US phone numbers
> (with and without dashes and parentheses). Include test code.

Example code Claude Code generates:

const phoneRegex = /^(\+1[-\s]?)?(\(\d{3}\)|\d{3})[-\s]?\d{3}[-\s]?\d{4}$/;

// Strict mobile pattern
const mobileRegex = /^(\+1[-\s]?)?\d{3}[-\s]?\d{3}[-\s]?\d{4}$/;

function validatePhone(input: string): boolean {
  return phoneRegex.test(input) || mobileRegex.test(input);
}

// Tests
const testCases = [
  { input: "(555) 123-4567", expected: true },
  { input: "555-123-4567", expected: true },
  { input: "+1 555 123 4567", expected: true },
  { input: "5551234567", expected: true },
  { input: "123-456", expected: false },
  { input: "abc-defg-hijk", expected: false },
];

testCases.forEach(({ input, expected }) => {
  const result = validatePhone(input);
  console.assert(
    result === expected,
    `Failed: ${input} => ${result}, expected ${expected}`
  );
});

console.log("All tests passed!");

Debugging Existing Regular Expressions

When a regex in your project isn’t working correctly, you can ask Claude Code directly.

> Investigate why the emailRegex in src/utils/validator.ts doesn't match [email protected] and fix it

Claude Code reads the file, identifies the issue in the regex, and fixes it. For example, it can instantly catch easy-to-miss bugs like an unescaped + character.

Practical Pattern: Building a URL Parser

Here’s an example of a URL parser useful for web scraping and log analysis.

interface ParsedUrl {
  protocol: string;
  host: string;
  port?: string;
  path: string;
  query?: Record<string, string>;
  fragment?: string;
}

function parseUrl(url: string): ParsedUrl | null {
  const regex =
    /^(https?):\/\/([^:/?#]+)(?::(\d+))?(\/[^?#]*)?(?:\?([^#]*))?(?:#(.*))?$/;
  const match = url.match(regex);

  if (!match) return null;

  const queryParams: Record<string, string> = {};
  if (match[5]) {
    match[5].split("&").forEach((pair) => {
      const [key, value] = pair.split("=");
      queryParams[decodeURIComponent(key)] = decodeURIComponent(value || "");
    });
  }

  return {
    protocol: match[1],
    host: match[2],
    port: match[3],
    path: match[4] || "/",
    query: match[5] ? queryParams : undefined,
    fragment: match[6],
  };
}

// Usage
const result = parseUrl("https://example.com:8080/api/users?page=1&limit=10#section");
console.log(result);
// {
//   protocol: "https",
//   host: "example.com",
//   port: "8080",
//   path: "/api/users",
//   query: { page: "1", limit: "10" },
//   fragment: "section"
// }

Improving Regex Readability

When you instruct Claude Code to “use named capture groups,” it generates more readable regular expressions.

// Date parser (using named capture groups)
const dateRegex =
  /^(?<year>\d{4})[-\/](?<month>0[1-9]|1[0-2])[-\/](?<day>0[1-9]|[12]\d|3[01])$/;

function parseDate(input: string) {
  const match = input.match(dateRegex);
  if (!match?.groups) return null;

  return {
    year: parseInt(match.groups.year),
    month: parseInt(match.groups.month),
    day: parseInt(match.groups.day),
  };
}

console.log(parseDate("2026-04-08"));
// { year: 2026, month: 4, day: 8 }

Prompt Tips

To get more accurate regex generation, include the following in your prompts. For more on effective prompting, see 5 Tips for Better Prompts.

1. Provide multiple examples of strings that should match
2. Explicitly state strings that should NOT match
3. Specify the use case (validation, extraction, replacement)
4. Mention language or framework constraints

> Create an email validation regex with the following conditions:
> - Should match: [email protected], [email protected]
> - Should NOT match: @example.com, user@, [email protected]
> - For use in TypeScript. Full RFC 5322 compliance is not required.

Summary

Using Claude Code as a regex assistant dramatically reduces the time needed to create complex patterns. By describing requirements in natural language and generating test cases simultaneously, you improve both the quality and maintainability of your regular expressions.

For more advanced usage, refer to the Claude Code official documentation.