Lazy (Non-Greedy) Matching — *? +? ??

Add ? to make a quantifier minimal-match

To flip greedy → lazy, append ? to the quantifier:

*? — 0 or more, but as few as possible
+? — 1 or more, but as few as possible
?? — 0 or 1, prefer 0
{n,m}? — between n and m, prefer n

Lazy quantifiers consume the MINIMUM and only extend if the rest of the pattern fails. Backwards from greedy, but same backtracking machinery.

The HTML disaster, fixed

Greedy .* ate both <p> blocks as one match. Lazy .*? matches each one separately — it stops at the first </p> instead of the last.

Performance note

Lazy is not free. The engine still backtracks; it just backtracks in the opposite direction (extending instead of shrinking). For some patterns lazy is faster, for others greedy is. The bigger win is correctness — lazy matches what you usually want.

The negated-class alternative

Often the cleanest fix is neither lazy nor greedy: it's a negated class. <p>([^<]*)</p> says "capture everything that's NOT a <" — no backtracking, no surprise. Track 8 will show this is the secret to ReDoS-safe patterns.

Code

Lazy fixes the greedy disaster·python

import re

html = '<p>first</p> <p>second</p>'

# Greedy — wrong
re.findall(r'<p>.*</p>', html)
# ['<p>first</p> <p>second</p>']

# Lazy — right
re.findall(r'<p>.*?</p>', html)
# ['<p>first</p>', '<p>second</p>']

# Better still — negated class (no backtracking)
re.findall(r'<p>[^<]*</p>', html)
# ['<p>first</p>', '<p>second</p>']

# Quoted strings — same lesson
re.findall(r'"[^"]*"', '"a" "b" "c"')
# ['"a"', '"b"', '"c"']  — clean and ReDoS-safe

Lazy (Non-Greedy) Matching — *? +? ??

Add ? to make a quantifier minimal-match

The HTML disaster, fixed

Performance note

The negated-class alternative

Code

External links

Exercise

Progress

Comments 0