backreferences

Backreferences are a feature of regular expressions that allow a pattern to reuse text captured by a previous capturing group. In pattern matching, a backreference matches the exact substring previously captured by the corresponding group, enabling repetition of identical substrings without re-stating them. The most common form is a numeric backreference such as \1, \2, and so on, which refer to the first, second, etc., capturing groups created by parentheses in the pattern. Many engines also support named backreferences, using forms such as (?P=name), \k<name>, or similar syntax, with the exact form depending on the engine.

Backreferences are used to enforce constraints within a pattern, such as requiring two identical words in succession. For example, a pattern like (\b\w+)\s+\1 matches a word followed by whitespace and the same word again. Backreferences are also employed in replacement strings to insert or rearrange captured text. For instance, a substitution that swaps two adjacent characters can use a backreference to reinsert the captured groups in reverse order.

The exact syntax and capabilities of backreferences vary among regex engines. In many languages, the replacement string uses $1 or ${name} to refer to captured content, while the pattern may use \1 or (?P=name) for backreferences. Some engines support alternative forms such as \k<name>, and certain features like backreferences in lookaround contexts may be restricted or behave differently.

Performance considerations apply to backreferences, particularly when combined with optional or repeating parts, because they can trigger backtracking and slow down matching on certain inputs. Overall, backreferences are a powerful tool for repeating and reusing substrings, but their availability and syntax depend on the specific regex implementation.