Regex for Extracting Numbers from a String — Integers, Decimals, and Negatives

Pulling numeric values out of free text — prices from product descriptions, quantities from chat messages, measurements from log lines — is a task every backend and scraping project runs into. This guide covers the regex for each numeric format you'll encounter.

Extracting Plain Integers

const text = 'Order #4521 has 3 items totaling 12 kg';
const integers = text.match(/\d+/g);
console.log(integers);
// ['4521', '3', '12']  — note these are strings, convert with Number() as needed

Extracting Decimals

const text = 'Price: 499.99, discount: 10.5%, rating 4.8 out of 5';
const decimals = text.match(/\d+\.?\d*/g);
console.log(decimals);
// ['499.99', '10.5', '4.8', '5']

Extracting Negative Numbers

A plain digit pattern ignores a leading minus sign — you need to explicitly allow it:

const text = 'Temperature dropped to -5.2°C, from a high of 21°C, change: -26.2';
const numbers = text.match(/-?\d+\.?\d*/g);
console.log(numbers);
// ['-5.2', '21', '-26.2']

Be careful: this pattern treats any digit sequence preceded directly by a hyphen as negative — including hyphens used as separators, like in order-42, which would incorrectly extract -42. Add a lookbehind or boundary check if your text contains hyphenated IDs.

// Safer: only treat '-' as a sign if preceded by whitespace, start of string, or another operator
const safeNegative = /(?<![a-zA-Z0-9])-?\d+\.?\d*/g;

'order-42 costs -5.50'.match(safeNegative);
// ['42', '-5.50']  — '-' in 'order-42' is correctly ignored as a sign

Extracting Numbers with Thousands Separators

const text = 'Revenue was ₹1,25,000 last month, up from ₹98,500';

// Indian numbering system uses irregular comma grouping (lakh/crore),
// so match any digit-comma sequence and clean up afterward
const raw = text.match(/[\d,]+/g);
console.log(raw);
// ['1,25,000', '98,500']

const asNumbers = raw.map((n) => Number(n.replace(/,/g, '')));
console.log(asNumbers);
// [125000, 98500]

Note the Indian numbering system groups digits as 2-2-3 (lakh/crore style) rather than the Western 3-3-3 pattern, so a fixed \d{1,3}(,\d{3})* pattern built for Western grouping will not match Indian-formatted numbers like 1,25,000. Matching any digit-comma run and stripping commas afterward, as above, sidesteps that issue.

Converting Matches to Actual Numbers

function extractNumbers(text) {
  const matches = text.match(/-?\d+\.?\d*/g) || [];
  return matches.map(Number);
}

extractNumbers('3 apples cost 45.50, discount -5');
// [3, 45.5, -5]

Common Pitfalls

• Forgetting matches are always strings — always wrap with Number() or parseFloat()
• Not handling the case where match() returns null when there are zero matches — always default to an empty array
• Trailing dots — a plain \d+\.?\d* can match a lone trailing period as part of a sentence (e.g. "the total is 5."); tighten to \d+(\.\d+)? to require digits after the dot
• Locale differences — some locales use a comma as the decimal separator instead of a dot; know your data source before choosing a pattern

Frequently Asked Questions