regex and text patterns
A regular expression describes a text pattern with its own alphabet of character classes, quantifiers and anchors, and the same pattern can match differently depending on which flag is active, the detail that most often trips up anyone starting to write regex.
`invalid.email` fails to match because it lacks the `@` required between the username and the domain; the pattern requires at least 2 letters in the TLD with `{2,}`.
The `\d+(?:[.,]\d+)?` pattern with the `g` flag scans the entire text and extracts every number, with or without a decimal part separated by a comma or a period, ignoring the surrounding text.
The pattern requires `[1-5]` right after the second group and `[89abAB]` right after the third; without these 2 nibble restrictions, a string with the generic hyphenated shape would pass even without being a real UUID.
Yes. The idea is to let you build the expression, apply flags and inspect matches in real time, which is especially useful for debugging, learning and fine-tuning patterns.
`^` marks the start of the text (or of each line, with the `m` flag) and `$` marks the end; neither consumes a character, so an anchor alone never appears in the matched text, it only restricts where the rest of the pattern can start or end.
Without `g`, the test stops at the first match found in the text; with `g`, it keeps searching after each match and returns every occurrence, which is required to extract, for example, every number in a text instead of just the first.
A pattern with nested groups and overlapping quantifiers (like `(a+)+b` against text with no `b`) can trigger catastrophic backtracking, where the engine tries a number of combinations that grows exponentially with the text length; that is why a regex tester caps the number of matches returned, to keep a pathological pattern from freezing the tab.
Restrict the version nibble to `[1-5]` at the right spot (right after the second hyphen) and the variant nibble to `[89ab]` (right after the third), as RFC 4122 defines; without those 2 specific classes, the pattern accepts any correctly-shaped hexadecimal sequence, even one that is not a real UUID.
Everything runs locally in your browser. Nothing is sent to a server.