Regular expressions
Metacharacters
.— any character except newline\w— word characters (letters, digits, underscore; may include CJK depending on engine/settings)\s— whitespace\b— word boundary*— 0 or more of the preceding atom+— 1 or more?— 0 or 1^— start of string (or negation inside[])
example: capture anchor text (non-greedy inner match) $reg='#<a[^>]*>([^<>]*)<\/a>#'; $str="<a href='ssss'>some</a>other<a href='ddd'>some</a>"; $res=preg_match_all($reg,$str,$m);
Character class
[]— match one character;?,*,.,{}usually need no extra escaping inside
Alternation
|— branches (can be multi-character); order matters
Capturing groups
(exp)— matchexpand capture into a numbered group(?<name>exp)— named capture; back-reference with\k<name>(?:exp)— match without capturing (no group number)
Group numbers are assigned left to right: unnamed groups first, then named ones.
(?=exp)— position beforeexp(lookahead)(?<=exp)— position afterexp(lookbehind)(?!exp)— position whereexpdoes not follow (negative lookahead)(?<!exp)— position whereexpdoes not precede (negative lookbehind)
Assertions match a position, not the text of
exp.
example: thousands separators (like `number_format`) $reg="#\\d{1,3}(?=(\\d{3})+(?:[.|$]))#"; $str="1234567891238.5646656567"; $res=preg_replace_callback($reg, function($matches){ return ($matches[0].','); },$str); example: negative lookahead — content inside tags that are **not** `<p>` $reg='#<(?!p)([^>\s]+)\s*(?:[^>]+)?>(.*?)</\1>#'; $str="<p>sdfd</p><div>dddiv</div>dom<img src='aaa'/><a href='ssss'>some</a>other"; $res=preg_match_all($reg,$str,$m);
(?#comment)— inline comment for readability
Back-references
\1,\2, … — repeat what an earlier captured group matched\k<name>or(?P=name)— named back-reference
example: quoted string where the opener and closer must match $reg="#(\"|').*?\\1#"; //$reg="#(?P<quote>\"|').*?(?P=quote)#"; $str="\"this is a 'string'\""; preg_match($reg,$str,$m);Lazy quantifiers
*?— repeat as few times as possible
example: without `?`, `.*` swallows up to the **last** `[/url]` $reg="#\\[url\\](.*?)\\[\\/url\\]#"; $str="[url]1.png[/url][url]2.png[/url][url]3.png[/url]"; $res=preg_replace($reg,"<img src='$1'/>",$str);Case sensitivity
- Ignore case globally:
#...#i - Local:
(?i)c— case-insensitive from the next character onward
- Ignore case globally:
Multiline
#...#m—^and$match line starts/ends;$normally means end of the whole string withoutm
Dotall (single-line)
#...#s—.matches newlines too (handy for multi-line blobs)
Lazy mode (PCRE)
#...#U— ungreedy quantifiers (similar effect to*?,+?, etc.)
Dollar and newline
#...#D— when$is used, it must not match before\n
UTF-8 mode
#...#u— treat pattern and subject as UTF-8
example: all-CJK string check $str="php百度"; $reg='#^[\x{4e00}-\x{9fa5}]+$#u'; if(preg_match($reg,$str,$m)){ echo 'all Chinese'; }else{ echo 'not all Chinese'; }
Tuning patterns
- Prefer character classes over alternation:
[a-c]beats(a|b|c)for speed - When alternates share a prefix, put the more specific branch first (e.g. 3- vs 4-digit postal codes)
- Greedy quantifiers backtrack:
\w*(\d+)oncopy2003ycan yield3because\w*eats the string first; use\w*?if you need the digits - Use
.only when the character set is unknown; use{n}when counts are known; prefer lazy quantifiers when greed hurts - Use
()sparingly; prefer(?:...)when you do not need captures - Anchor when you can:
^,$, and keep anchors outside groups when possible - Split huge patterns into smaller steps
- Reach for PHP builtins before regex when they fit
- Validation:
filter_var('admin@qq.com', FILTER_VALIDATE_EMAIL) - PHP source:
tokenizer_*instead of hand-rolled lexers
- Validation:
Questions? Leave a comment below or email: newbvirgil@gmail.com
Permalink: https://newbmiao.github.io/regex-notes.html
