How to extract numbers from a string in Python?

python

regex

string-extraction

data-cleaning

byNikita Barsukov·Feb 17, 2025

Quickly extract numbers from a string in Python using re.findall() from the re module with a simple regex pattern \d+, capturing the sequence of any digits:

import re

numbers = re.findall(r'\d+', "Example string 123 with 456")
print(numbers)  # Outputs: ['123', '456']

For transforming the output to integers, we can utilize: list(map(int, numbers)).

More complex extraction: Negative numbers, floats

Positive integers are a piece of cake, but what if the game gets tougher - entering negative numbers and floats? Our regex pattern needs an upgrade: [-+]?[.]?[\d]+(?:,\d\d\d)*[\.]?\d*(?:[eE][-+]?\d+)?.

import re

complex_numbers = re.findall(r'[-+]?[.]?[\d]+(?:,\d\d\d)*[\.]?\d*(?:[eE][-+]?\d+)?', "Example -123 with +456.78 and 3.14e-10")
print(complex_numbers)  # Outputs:  ['-123', '+456.78', '3.14e-10']

To convert strings into numerical values, we can do:

numeric_values = [float(num) if '.' in num or 'e' in num.lower() else int(num) for num in complex_numbers]
print(numeric_values)  # Outputs: [-123, 456.78, 3.14e-10]

The isdigit approach: Extraction via list comprehension

When dealing with simpler situations looking for positive integers, list comprehension along with str.isdigit() method cases presents an alternative approach to regex:

extracted_numbers = [int(part) for part in "123 main st.".split() if part.isdigit()]
print(extracted_numbers)  # Outputs: [123]

Negotiating with formatted strings: Handling commas and separators

Ran into formatted numbers like "1,000" or "2.5M" in your journey? Worry not! With a few extra tricks, we can get past this obstacle:

import re

formatted_numbers = "The event gathered 1,000 enthusiasts and raised 2.5M dollars."

# Pre-process: Combat separators and format indicators
cleaned_string = formatted_numbers.replace(',', '').replace('M', '000000')

# Extract numbers
extracted_values = re.findall(r'\d+', cleaned_string)
print(extracted_values)  # Outputs: ['1000', '2500000']

Remember to adjust this strategy to tackle international number formats or specific application needs.

Robust extraction for Real-world applications: Dealing with Edge cases

Real-world data can be messy and unpredictable. To make our extraction process more robust, we should handle potential errors and unexpected inputs using try-except blocks:

raw_data = ["123", "4.5", "-inf", "NaN", "seven"]

def safe_convert(value):
    try:
        return float(value)
    except ValueError:
        # Oops! Float conversion failed. Return None.
        return None

converted_data = [safe_convert(item) for item in raw_data]