Regex Editor & Tester

Matches key-value style text in the form 'term : value', where the left side is one or more non-whitespace characters, followed by an optional space, a colon, another optional space, and a value of one or more non-whitespace characters (including the ▁ character commonly used in SentencePiece tokenization). With global and multiline flags, it's useful for extracting 'sentence: label' lines, tokenized NLP datasets, or colon-delimited pairs across multiple lines.