Regex cheat sheet: all metacharacters and constructs

Regular expressions are a compact language for describing text patterns. Metacharacters, quantifiers, groups, and lookaround are easy to forget, even for experienced developers. This cheat sheet collects all the main constructs with brief examples. You can conveniently test patterns in the Regex Tester on reChecker. For in-depth theory, see the complete regex guide, and for practical examples, see the article with 20 useful patterns.

Metacharacters

Main table

Escaped characters

To use a metacharacter as a regular character, put \ before it:

\.  \*  \+  \?  \\  \(  \)  \[  \]  \{  \}  \|  \^  \$

Examples:

\d+\.\d+       — десятичное число: 3.14
\*\*\*         — три звёздочки
\[word\]       — [word] как текст

Character classes

Custom classes

Special characters inside []

• ^ at the beginning — negation: [^0-9]
• - between characters — range: [a-z]
• - at the beginning or end — literal: [-0-9] or [0-9-]
• ] — escape it: [\]\[]

Predefined classes

Note: in JavaScript, \w includes Unicode letters when the u flag is set. In other languages, it includes ASCII characters only.

Special sequences

Quantifiers

Basic quantifiers

Greediness

Example:

Текст: <b>жирный</b> и <b>ещё</b>

<b>.*</b>   → одно совпадение: <b>жирный</b> и <b>ещё</b>
<b>.*?</b>  → два совпадения: <b>жирный</b>, <b>ещё</b>

Anchors and boundaries

Examples:

^\d+$           — строка целиком из цифр
\bword\b         — слово «word» целиком
\Bcat\B          — «cat» внутри слова (concatenate)

Groups

Capturing groups

Examples:

(\d{2})\.(\d{2})\.(\d{4})     — дата, группы 1, 2, 3
(?:https?)://                   — протокол без захвата
(?<year>\d{4})-(?<month>\d{2}) — именованные группы

Backreferences

Examples:

(\w+)\s+\1              — повторяющееся слово
(['"]).*?\1              — строка в кавычках
<(\w+)>.*?</\1>          — парный HTML-тег

Lookahead and lookbehind

Lookahead

Examples:

\d+(?=\s*руб)     — число перед «руб»
\w+(?=\.)         — слово перед точкой
^(?!.*admin)      — строка без «admin»

Lookbehind

Examples:

(?<=\$)\d+        — число после $
(?<=class=")[^"]+ — значение атрибута class
(?<!\d)\d{3}      — 3 цифры не после цифры

Combining lookarounds

(?<=<b>).*?(?=</b>)   — содержимое между <b> и </b>
(?<!un)happy          — happy, но не unhappy

Flags (modifiers)

Usage

// JavaScript
'Hello'.match(/hello/i);           // ['Hello']
'1 2 3'.match(/\d/g);              // ['1', '2', '3']

# Python
re.search(r'hello', 'Hello', re.I)
re.findall(r'\d', '1 2 3')

// PHP
preg_match('/hello/i', 'Hello');
preg_match_all('/\d/', '1 2 3', $m);

Unicode character classes (extended)

In Unicode mode (u in JS, re.UNICODE in Python), these are available:

Negation: \P{L} — not a letter.

Examples:

\p{L}+              — последовательность букв (в т.ч. кириллица)
[\p{L}\p{N}\s]+     — слова с цифрами и пробелами
\p{Script=Cyrillic}+ — только кириллица

Special constructs

Conditional expressions (PCRE)

(?(condition)yes-pattern|no-pattern)

Example: match text in quotes or without quotes:

(?(?=['"])(['"]).*?\1|[\w\s]+)

Atomic groups (PCRE)

(?>pattern)

A group without backtracking — like a possessive quantifier applied to the whole pattern.

Comments (x flag)

\d{2}    # день
\.       # разделитель
\d{2}    # месяц

Operator precedence

From highest to lowest:

1. Quantifiers *, +, ?, {}
2. Concatenation (sequence)
3. Alternation |

Example: a|bc is interpreted as a or bc, not ab or c.

Escaping in different languages

JavaScript

const re = /\d+/;           // Литерал
const re2 = new RegExp('\\d+');  // Строка — двойное экранирование

Python

re.compile(r'\d+')   # raw string — один \
re.compile('\\d+')   # обычная строка — двойное \\

PHP

preg_match('/\d+/', $str);   # В строках PHP \d → d, нужен \\
preg_match('/\\d+/', $str);  # или один \ в паттерне

Performance

Recommendations for avoiding catastrophic backtracking:

1. Avoid nested quantifiers: (a+)+, (.*)*
2. Use atomic groups or possessive quantifiers
3. Narrow patterns: [^>]+ instead of .* inside tags
4. Specify boundaries explicitly: \d{4} instead of \d+ for a year

Quick reference: feature support

Modes in .NET

JavaScript has no built-in inline modifiers. In Python, use (?i) and equivalents.

Boundaries in different contexts

Word boundary \b

\b depends on the definition of “word.” In ASCII, this is [a-zA-Z0-9_]. In Unicode mode, it is an extended set. Be careful with Cyrillic: in older engines, \b may not work for non-ASCII characters.

Start/end boundaries

• ^ and $ — start and end of the entire string by default
• With the m flag — start and end of each line in multiline text
• \A and \Z — always the start and end of the text (independent of m)

Replacing with groups

Numbered groups

'2026-03-07'.replace(/(\d{4})-(\d{2})-(\d{2})/, '$3.$2.$1');
// '07.03.2026'

Named groups (ES2018)

'John Doe'.replace(/(?<first>\w+)\s+(?<last>\w+)/, '$<last>, $<first>');
// 'Doe, John'

Transforming during replacement

'hello world'.replace(/\b\w/g, m => m.toUpperCase());
// 'Hello World'

Common mistakes

Related materials

For an in-depth guide with theory and examples, see the complete regex guide. Ready-made patterns for common tasks are in 20 useful regular expressions. You can test a pattern in the Regex Tester.