Regex Match Numbers: Integers, Decimals and Negatives

The building blocks

\d+                    # one or more digits
-?\d+                  # optional minus
[+-]?\d+               # optional sign
\d+\.\d+               # decimal, digits required both sides
-?\d+(?:\.\d+)?        # optional decimal part
-?\d+(?:\.\d+)?(?:[eE][+-]?\d+)?   # scientific notation

Escape the dot. An unescaped . in \d+.\d+ matches any character, so 12x34 passes as a decimal.

Anchoring and boundaries

Unanchored, \d+ finds a number inside anything — abc123def matches. For validation, anchor with ^ and $; for extraction from text, use word boundaries \b\d+\b so abc123 is not treated as the number 123.

Note \b treats - as a boundary, so \b-?\d+\b will not capture the minus in -5. Use (?<![\w.])-?\d+ when the sign matters.

The Unicode surprise

In Python, Java and .NET, \d matches every Unicode decimal digit — including Arabic-Indic ٤٢ and Devanagari ४२. int() in Python will happily parse those, which may or may not be what you want.

Restrict to ASCII explicitly when the value feeds something that expects it: [0-9] instead of \d, or Python re.ASCII. In JavaScript, \d is always ASCII-only, which is one fewer thing to think about.

When not to use regex

Formatted numbers with thousands separators, currency symbols or locale-specific decimal commas get ugly fast — ^-?\d{1,3}(?:,\d{3})*(?:\.\d+)?$ handles one convention and fails on European formatting.

Regex is good at finding candidate numbers in text. Validating a range, checking precision or parsing a locale belongs in code: match loosely, then parse and check properly.