Named Groups — (?P<name>...)

Numbered groups are fragile, named groups are stable

When your pattern grows, group numbers shift every time you add or remove parens. Named groups solve this by giving each capture a name you reference instead of a number.

Syntax varies by flavor:

Python / older PCRE: (?P<name>...)
.NET / PCRE / JavaScript: (?<name>...)
Backreference to named group (Python): (?P=name)
Backreference to named group (.NET / PCRE): \k<name>

Python supports BOTH the (?P<name>...) form (its own) AND (?<name>...) as of 3.x. Cross-language, prefer the bare angle-bracket form when possible.

Why use named groups

Three reasons:

Readable callers. m.group('year') is self-documenting; m.group(1) requires comments.
Stable across edits. Adding a group in the middle doesn't break m.group('year').
Self-documenting patterns. Future-you reads (?P<email>...) and knows what's being captured.

The cost

None worth mentioning. Named groups are free. Use them whenever a group's purpose isn't obvious from position.

Code

Named groups·python

import re

# Python style
pattern = r'(?P<year>\d{4})-(?P<month>\d{2})-(?P<day>\d{2})'
m = re.match(pattern, '2026-05-04')
print(m.group('year'))   # '2026'
print(m.group('month'))  # '05'
print(m.groupdict())     # {'year': '2026', 'month': '05', 'day': '04'}

# Cross-flavor angle bracket form (also works in Python 3.x)
pattern = r'(?<year>\d{4})-(?<month>\d{2})'

# Backreference by name
re.findall(r'(?P<word>\w+)\s+(?P=word)', 'the the cat sat sat')
# ['the', 'sat']

# Self-documenting validators
USER = re.compile(r'''
    ^
    (?P<localpart>[\w.+-]+)
    @
    (?P<domain>[\w-]+\.[\w.-]+)
    $
''', re.VERBOSE)
m = USER.match('hi@pippa.dev')
print(m.group('localpart'))  # 'hi'
print(m.group('domain'))     # 'pippa.dev'

Named Groups — (?P<name>...)

Numbered groups are fragile, named groups are stable

Why use named groups

The cost

Code

External links

Exercise

Progress

Comments 0