Skip to main content

pcre_match

NAME​

pcre_match() - regular expression handler

SYNOPSIS​

mixed pcre_match(string|string *lines, string pattern, void|int|mapping flag, void|int|mapping pcre_flags);

DESCRIPTION​

analog with regexp efun for backwards compatibility reasons but utilizing
the PCRE2 library. UTF-8, Unicode properties, and JIT are on by default.

The optional `pcre_flags` are integer bits from `src/include/pcre_flags.h`
or a mapping of named PCRE2 options (see `pcre_config()`). For string
input, the 3rd argument is treated as `pcre_flags`; for array input, the
3rd argument remains the legacy flag (or a mapping of flags) and
`pcre_flags` is the 4th argument. Defaults to 0.
Behavior change: Unicode properties are on by default

Drivers before the PCRE2 upgrade compiled patterns with UTF matching only. Every pcre_* efun now also sets PCRE2_UCP, so the character-class escapes resolve against Unicode properties rather than ASCII:

  • \d matches any Nd digit (３), not just 0-9
  • \w matches any letter (café), not just [A-Za-z0-9_]
  • \s, \b and their negations widen to match

Mudlibs that use \d or \w to validate input (numeric IDs, object names, player names) will accept strings they used to reject. Restore the old behavior per call with the PCRE_NO_UCP flag bit or the no_ucp mapping option:

pcre_match(s, "^\\d+$", PCRE_NO_UCP);           // ASCII digits only
pcre_match(s, "^\\d+$", ([ "no_ucp": 1 ]));     // same, mapping form

SEE ALSO​

pcre_config(3), regexp(3)