{"id":1495,"date":"2023-05-21T22:00:00","date_gmt":"2023-05-21T20:00:00","guid":{"rendered":"https:\/\/demo.sidekick.be\/itenium\/regex-tutorial\/"},"modified":"2025-11-28T13:31:43","modified_gmt":"2025-11-28T12:31:43","slug":"regex-tutorial","status":"publish","type":"post","link":"https:\/\/demo.sidekick.be\/itenium\/blog\/productivity\/regex-tutorial\/","title":{"rendered":"Regular Expressions: Tutorial"},"content":{"rendered":"

Almost all of this stuff should work in most regex implementations.
Notable exceptions are: <<\/code> and ><\/code>, [a-Z]<\/code>, QE<\/code>. Your mileage may vary.<\/p>\n

Basic Syntax<\/h1>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Regex<\/th>\nMatches<\/th>\nRemarks<\/th>\n<\/tr>\n<\/thead>\n
Literals<\/strong><\/td>\n<\/td>\n<\/td>\n<\/tr>\n
abc<\/td>\nLiterals match themselves. Here: abc<\/code><\/td>\n<\/td>\n<\/tr>\n
<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
Metacharacters<\/strong><\/td>\n<\/td>\n<\/td>\n<\/tr>\n
^s<\/td>\ns<\/code> but only at the start of the input<\/td>\n<\/td>\n<\/tr>\n
e$<\/td>\ne<\/code> but only at the end.<\/td>\n^<\/code> and $<\/code> match a position.<\/td>\n<\/tr>\n
$<\/td>\nEscape character <\/code>: Match any $<\/code><\/td>\n<\/td>\n<\/tr>\n
.<\/td>\nMatch any one character<\/td>\nMatches rn<\/code> with dotall (s) flag<\/td>\n<\/tr>\n
<e<\/td>\nWord that starts with e<\/code><\/td>\nWord boundaries: [a-zA-Z0-9_]<\/code><\/td>\n<\/tr>\n
e><\/td>\nThat ends with e<\/code><\/td>\n<\/td>\n<\/tr>\n
b<\/td>\nWord boundary<\/td>\nBetween w<\/code> and W<\/code><\/td>\n<\/tr>\n
<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
Character Classes<\/strong><\/td>\n<\/td>\n<\/td>\n<\/tr>\n
[az]<\/td>\na<\/code> or z<\/code><\/td>\n<\/td>\n<\/tr>\n
[0-9]<\/td>\nA single digit<\/td>\nd<\/code><\/td>\n<\/tr>\n
[a-zA-Z]<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
[a-Z]<\/td>\nThis might<\/strong> also include []\\^_<\/code> and a literal `<\/td>\n<\/td>\n<\/tr>\n
[n$^-]<\/td>\nA newline, $<\/code>, ^<\/code> or a hyphen (-<\/code>)<\/td>\nA ^<\/code> at the start negates the class<\/td>\n<\/tr>\n
[^ab]<\/td>\nEverything but a<\/code> and \u2018backspace\u2019<\/td>\n[b]<\/code> (backspace) vs b<\/code> (word boundary)<\/td>\n<\/tr>\n
<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
Quantifiers<\/strong><\/td>\n<\/td>\n<\/td>\n<\/tr>\n
a?<\/td>\nZero or one<\/td>\n<\/td>\n<\/tr>\n
a*<\/td>\nZero or more<\/td>\n<\/td>\n<\/tr>\n
a+<\/td>\nOne or more<\/td>\n<\/td>\n<\/tr>\n
a{2,5}<\/td>\nTwo to five<\/td>\n<\/td>\n<\/tr>\n
a{4}<\/td>\nExactly four<\/td>\n<\/td>\n<\/tr>\n
a{5,}<\/td>\nFive or more<\/td>\na{,5}<\/code> for 0 to 5<\/td>\n<\/tr>\n
<\/td>\n<\/td>\n<\/td>\n<\/tr>\n
Grouping<\/strong><\/td>\n<\/td>\n<\/td>\n<\/tr>\n
(ab|yz)<\/td>\nLiteral ab<\/code> or yz<\/code>.<\/td>\nAlternation<\/td>\n<\/tr>\n
(?:ab)<\/td>\nab<\/code> but non-capturing group<\/td>\n<\/td>\n<\/tr>\n
(w+) 1<\/td>\n1<\/code> matches the same thing as (w+)<\/code><\/td>\nBackreferences (also 2<\/code> etc)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Shorthands<\/h2>\n\n\n\n\n\n\n\n\n\n\n\n\n
Shorthand<\/th>\nMeaning<\/th>\nRemarks<\/th>\n<\/tr>\n<\/thead>\n
d<\/td>\n[0-9]<\/td>\nD<\/code> -> [^0-9]<\/code><\/td>\n<\/tr>\n
w<\/td>\n[a-zA-Z0-9_]<\/td>\n<\/td>\n<\/tr>\n
s<\/td>\nWhitespace<\/td>\n<\/td>\n<\/tr>\n
t<\/td>\nTab<\/td>\n<\/td>\n<\/tr>\n
r<\/td>\nCarriage return<\/td>\n<\/td>\n<\/tr>\n
n<\/td>\nNewline<\/td>\n<\/td>\n<\/tr>\n
b<\/td>\nWord boundary<\/td>\nA<\/code> and Z<\/code>: Start\/End of string<\/td>\n<\/tr>\n
Q\u2026E<\/td>\nLiteral sequence<\/td>\nEx: Qlite[]ralE<\/code><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Shorthands can be inverted by capitalizing them: D<\/code> (not a digit)<\/p>\n

Also v<\/code> (vertical tab), f<\/code> (form feed).
Inside a character class b<\/code> matches backspace.<\/p>\n

Modifiers<\/h1>\n\n\n\n\n\n\n\n\n
Modifier<\/th>\nDescription<\/th>\n<\/tr>\n<\/thead>\n
g<\/code>(lobal)<\/td>\nJS: Match more than once<\/td>\n<\/tr>\n
i<\/code>(nsensitive)<\/td>\nCase insensitive matching<\/td>\n<\/tr>\n
m<\/code>(ultiline)<\/td>\n^$<\/code> match every line in the string (vs A<\/code> and Z<\/code>)<\/td>\n<\/tr>\n
s<\/code> (dotall)<\/td>\n.<\/code> matches r|n<\/code><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Replacement<\/h1>\n\n\n\n\n\n\n\n\n\n\n\n
Replacement<\/th>\nDescription<\/th>\n<\/tr>\n<\/thead>\n
$1<\/td>\nFirst captured group<\/td>\n<\/tr>\n
$2<\/td>\nSecond captured group<\/td>\n<\/tr>\n
$$<\/td>\nA literal $<\/code><\/td>\n<\/tr>\n
$&<\/td>\nEntire match<\/td>\n<\/tr>\n
$`<\/td>\nBefore matched string<\/td>\n<\/tr>\n
$\u2019<\/td>\nAfter matched string<\/td>\n<\/tr>\n
$+<\/td>\nLast matched string<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

Some implementations use <\/code> instead of $<\/code>.<\/p>\n

Use non-capturing groups (?:)<\/code> to keep your backreferences ($1<\/code>, $2<\/code>, \u2026) in check.
Or use named groups if supported in your regex implementation.<\/p>\n

Looking around<\/h1>\n

The last & thoughest feature each developer should definitely know: lookahead & lookbehind!<\/p>\n

\"Lookahead<\/p>\n

Example<\/strong>:<\/p>\n

A(?=B)   \u2013 Literal A followed by B\nA(?!B)   \u2013 Literal A not followed by B\n<\/code><\/pre>\n
Lookarounds do not match anything in themselves which make them very handy when you want to replace some text but only when it is (not) preceded\/succeeded by something else.<\/p>\n
Also in this series<\/h1>\n