Questions
Where can I test a pattern?
The regex tester highlights matches live, lists groups, previews a replacement and explains the pattern token by token.
Regex cheat sheet
The syntax shared by every mainstream engine, with the differences in one table at the end.
Fits on one A4 / Letter page in portrait; use “Save as PDF” in the print dialog to keep a copy.
Anchors and boundaries
| ^ $ | start / end of text (of line with m flag) |
| \A \z \Z | start / end of text regardless of flags (not in JS) |
| \b \B | word boundary / not |
Characters
| . | any char except newline (any with s flag) |
| \d \w \s | digit, word char, whitespace; \D \W \S negate |
| [abc] [a-z] [^abc] | set, range, negated set |
| [\]\\^-] | inside a set only ] \ ^ - need escaping |
| \t \n \r \xHH \uHHHH | tab, newline, CR, hex, Unicode |
| \p{L} \p{Lu} \p{N} \p{Script=Cyrillic} | Unicode properties (u flag in JS) |
| [[:alpha:]] [[:digit:]] | POSIX classes (grep, PCRE; not JS) |
| \. \* \( \\ | escaped metacharacter = literal |
Quantifiers
| * + ? | 0+, 1+, 0 or 1 |
| {n} {n,} {n,m} | exactly, at least, between |
| *? +? ?? {n,m}? | lazy: as few as possible |
| *+ ++ ?+ | possessive: no backtracking (PCRE, Java) |
Groups
| (abc) | capture, numbered from 1 |
| (?:abc) | no capture |
| (?<name>abc) | named (Python: (?P<name>…)) |
| \1 \k<name> | backreference (Python: (?P=name)) |
| a|b | alternation (lowest precedence) |
| (?>abc) | atomic group (PCRE, Java) |
| (?i)abc (?i:abc) | inline flag, scoped inline flag |
Lookaround (zero-width)
| (?=x) | followed by x |
| (?!x) | not followed by x |
| (?<=x) | preceded by x |
| (?<!x) | not preceded by x |
Flags
| g | all matches (JS, PHP via function) |
| i | ignore case |
| m | ^ $ match at line breaks |
| s | . matches newline (DOTALL) |
| x | ignore whitespace and # comments in the pattern (PCRE, Python; not JS) |
| u | Unicode (JS); required for \p{} |
| y | sticky (JS) |
Replacement
| $1 $<name> $& | JS, .NET: group, named, whole match |
| \1 \g<name> | Python re.sub |
| $1 \1 ${name} | PCRE / PHP preg_replace; Go uses $1 and ${name} |
| sed 's/(a)(b)/\2\1/' | sed -E (basic sed needs \( \)) |
Common patterns
| ^\s*$ | blank line |
| \s+$ | trailing whitespace |
| \b(?:\d{1,3}\.){3}\d{1,3}\b | IPv4 (loose) |
| [\w.+-]+@[\w-]+(?:\.[\w-]+)+ | email (practical) |
| https?://[^\s"'<>]+ | URL |
| \d{4}-\d{2}-\d{2} | ISO date |
| #(?:[0-9a-fA-F]{3}){1,2}\b | hex colour |
| "(?:[^"\\]|\\.)*" | double-quoted string with escapes |
| ^(?=.*\d)(?=.*[a-z])(?=.*[A-Z]).{8,}$ | password with digit, lower, upper, 8+ |
| (\w+)\s+\1 | repeated word |
Flavour differences
| JavaScript | no \A \z, no x flag, no possessive/atomic; lookbehind yes; named groups (?<n>) |
| PCRE (PHP, nginx, HAProxy) | everything; delimiters /…/ in PHP |
| Python re | (?P<n>), no possessive before 3.11, \Z = end |
| Go RE2, Rust regex | linear time; no lookaround, no backreferences |
| grep -E / sed -E | ERE: no \d (use [0-9]), POSIX classes, no lazy quantifiers |
| grep (basic) | \( \) \{ \} \+ \? escaped to be special |
| Java | possessive, atomic; strings need double backslashes |