HELP PCRE                      Perl-compatible regular expressions

    uses pcre;

Full modern regex syntax — \d \w \b, + ? {n,m}, capture groups,
alternation, (?i), lookaround — via a small PCRE2 shim.  One-time
setup:

    $ tools/build-poppcre.sh        (needs libpcre2-dev)

LIB * REGEXP is the zero-dependency sibling (the built-in Ved engine,
@-syntax); this library is for when you want the syntax the rest of
the world writes.  The two expose the same shape of API.

pcre_search(PAT, S, I) -> (START, NCHARS)
    First match at or after byte position I; both false if none.

pcre_matches(PAT, S) -> BOOL
pcre_first(PAT, S) -> SUBSTRING or false

pcre_groups(PAT, S) -> VECTOR or false
    Whole match plus each capture group as substrings (false for a
    group that did not participate):

        pcre_groups('(\\w+)@(\\w+)', 'bob@example') =>
        ** {bob@example bob example}

pcre_all(PAT, S) -> LIST
pcre_split(PAT, S) -> LIST
pcre_replace(PAT, S, NEW) -> S
    All matches / split around matches / every match replaced by the
    literal string NEW.

pcre_version() -> STRING

Remember Pop-11 string escapes: backslash-d is written '\\d' inside
a string literal.  Patterns are byte-oriented; put (*UTF) at the
front of a pattern for UTF-8 mode.  A malformed pattern mishaps with
PCRE2's error message and offset.

The most recently compiled pattern is cached in the shim, so loops
over one pattern compile it once.

See also: LIB * REGEXP, LIB * STRUTILS, tools/tests/test_pcre.p.

--- pop/help/pcre