TopolearnFlags fix the most common 'why doesn't it match' problems, and verbose mode keeps long patterns maintainable.
Flags modify how a pattern matches. Pass them as flags= and combine with |:
re.IGNORECASE(re.I) case-insensitive matching.re.MULTILINE(re.M) makes^and$match at the start and end of every line, not just the string.re.DOTALL(re.S) makes.match newlines too.re.VERBOSE(re.X) ignores whitespace in the pattern and allows#comments, so you can lay out complex patterns over several lines. Escape literal spaces as\or use[ ].re.ASCII(re.A) restricts\w,\d,\s,\bto ASCII.
Flags can also be written inline at the very start of the pattern: (?i), (?m), (?x); or scoped to part of it: (?i:abc)def.
re.compile(pattern, flags) returns a Pattern object with the same methods (search, findall, sub...). Module-level functions cache compiled patterns internally, so compiling is not mainly about speed. It helps readability: define named patterns once at module level, and the methods accept pos and endpos arguments for searching inside a region.
Never use plain-string flags in a positional slot: since 3.13, count, maxsplit and flags given positionally in sub, subn and split trigger a DeprecationWarning.
Example
import re
text = "Name: Ann\nname: Bob\nAGE: 31"
print(re.findall(r"^name: (\w+)", text, flags=re.I | re.M)) # ['Ann', 'Bob']
print(re.findall(r"^name: (\w+)", text)) # []
html = "<p>one\ntwo</p>"
print(re.search(r"<p>(.*)</p>", html)) # None
print(repr(re.search(r"<p>(.*)</p>", html, re.S).group(1))) # 'one\ntwo'
PHONE = re.compile(r"""
\(?(\d{3})\)? # area code, optional parentheses
[\s.-]? # separator
(\d{3})[\s.-]?(\d{4})
""", re.VERBOSE)
print(PHONE.findall("(555) 123-4567 and 555.987.6543"))
# [('555', '123', '4567'), ('555', '987', '6543')]
print(PHONE.search("x 555 123 4567", pos=2).groups()) # ('555', '123', '4567')Remember
re.I,re.M,re.S,re.X,re.Achange matching; combine with|- MULTILINE affects ^ and $; DOTALL affects .
re.compilegives reusable, documented Pattern objects; VERBOSE allows comments
Watch out for
- Forgetting that whitespace is ignored in VERBOSE mode, so a literal space in the pattern silently disappears.