Read This! | Back to Regular Expressions

VikiD’s RegExp Library

The library is no longer empty! Below is a working collection of regular expressions I actually use, with a note on each. For the theory, go back to Regular Expressions.

The core idea in plain words

A regular expression is a tiny pattern language for describing shapes of text: “three digits, a dash, four digits”, “a word at the start of a line followed by a colon”. A regex engine slides your pattern along the text and reports where it fits.

The catch is that there are several dialects. The ones you meet daily:

POSIX ERE — grep -E, sed -E, awk. No \d; use [0-9] or [[:digit:]].
PCRE / Perl-style — grep -P, Perl, PHP, and (close enough) Python’s re, JavaScript and Rust’s regex crate, which ripgrep (rg) uses. Has \d, \w, \s, non-greedy *?.
Emacs regexps — groups and alternation are backslashed: \(, \|. Or skip the backslash soup with the rx macro.

Unless noted, patterns below are PCRE-style and work in rg, Python and JavaScript.

An analogy: a stencil

A regex is a stencil you slide over a page. Literal characters are solid cut-outs that must match exactly; . is a hole that lets any character through; [aeiou] is a hole shaped to fit only vowels; + and * say “this hole can stretch”. Anchors (^, $, \b) pin the stencil to the edge of a line or a word.

The library

Whitespace and lines

Pattern	Matches
`^\s*$`	blank (or whitespace-only) line
`\s+$`	trailing whitespace
`^\s*#`	a comment line in shell/conf files
`^(?!\s*#).+`	a non-comment, non-empty line (PCRE lookahead)

Numbers

Pattern	Matches
`^-?\d+$`	an integer
`^-?\d+(\.\d+)?$`	an integer or decimal
`^0x[0-9A-Fa-f]+$`	a hex literal
`\b\d{1,3}(,\d{3})+\b`	thousands with commas: `1,234,567`

Dates and times

Pattern	Matches
`\b\d{4}-\d{2}-\d{2}\b`	ISO date shape `2026-10-22` (shape only; doesn’t reject month 13)
`\b\d{4}-(0[1-9]\|1[0-2])-(0[1-9]\|[12]\d\|3[01])\b`	ISO date with valid month/day ranges
`\b([01]\d\|2[0-3]):[0-5]\d\b`	24-hour time `21:30`

(The \| above is just a table escape; in real use it is a bare |.)

Network bits

Pattern	Matches
`\b(?:\d{1,3}\.){3}\d{1,3}\b`	IPv4 shape (allows 999)
`\b(?:(?:25[0-5]\|2[0-4]\d\|1?\d?\d)\.){3}(?:25[0-5]\|2[0-4]\d\|1?\d?\d)\b`	strict IPv4, 0–255
`https?://[^\s)>\]]+`	URLs in prose, stopping at a space or closing bracket
`^[^@\s]+@[^@\s]+\.[^@\s]+$`	“looks like an email” — good enough for a sanity check; don’t try to be RFC-perfect

Code and config

Pattern	Matches
`^\s*(def\|class)\s+(\w+)`	Python definitions; group 2 is the name
`^\(def(un\|var\|custom\|macro)\s+(\S+)`	Emacs Lisp top-level definitions
`\bTODO\b\|\bFIXME\b`	the things I promised myself
`^([A-Za-z_][A-Za-z0-9_])=(.)$`	`KEY=value` lines (env files, Void’s `/etc/rc.conf`)
`"(?:[^"\\]\|\\.)*"`	a double-quoted string with escapes

Logs

Pattern	Matches
`\b(ERROR\|FATAL\|panic)\b`	trouble
`^\S+ \S+ (\S+)\[(\d+)\]:`	syslog-style `host prog[pid]:`; groups are program and PID

A worked example: tidy a file, three ways

Task: in notes.org, find every ISO date and swap it from 2026-10-22 to 22/10/2026.

sed (POSIX ERE):

sed -E 's/\b([0-9]{4})-([0-9]{2})-([0-9]{2})\b/\3\/\2\/\1/g' notes.org

(GNU sed understands \b; add -i to edit in place once you’re happy.)

ripgrep, to preview first:

rg -n '\b(\d{4})-(\d{2})-(\d{2})\b' -r '$3/$2/$1' notes.org

-r only shows the replacement; it never touches the file.

Emacs: M-x replace-regexp, then

\([0-9]\{4\}\)-\([0-9]\{2\}\)-\([0-9]\{2\}\)  →  \3/\2/\1

or, readably, build the pattern with rx:

(rx (group (= 4 digit)) "-" (group (= 2 digit)) "-" (group (= 2 digit)))

Common misconceptions

".* is harmless." It is greedy: in <a><b>, <.*> matches the whole thing. Use <[^>]*> or the non-greedy <.*?>.
“Regex can parse HTML/JSON.” Not reliably — nesting defeats it. Use jq, pup or a real parser.
“All regex engines are the same.” \d does nothing in POSIX grep; | needs a backslash in Emacs. Know your dialect.
“A validation regex proves the data is valid.” It proves the shape. 2026-02-31 passes the date pattern above.
“Complex regex is clever.” It’s a liability. Comment it, or use rx or Python’s re.VERBOSE.

Connections

Regular Expressions — the theory and syntax.
awk and Bash — where many of these live.
VikiD’s Bash Cookbook — more one-liners.
Emacs — rx, occur, replace-regexp.
Linux Tools — grep, sed, ripgrep and friends.