NAME
    Hustle::Table - Cached general purpose dispatch and routing table

SYNOPSIS
      use Hustle::Table;

    Create a new table:

      my $table=Hustle::Table->new;

    Add entry as hash ref:

      $table->add( { matcher => qr/regex (match)/, value=> "a value"});

    Add entry as array ref (3 elements required):

       $table->add( [qr/another/, "another value", undef])

    Add entry as flat key value pairs:

            $table->add(matcher=>"jones", value=> sub {"one more"}, type=>"begin");

    Add entry as tuple

            $table->add(qr|magic matcher| => "to a value");

    Set the default entry:

            $table->set_default("default value");

    Prepare a dispatcher external cache:

            my %cache;
            my $dispatch = $table->prepare_dispatcher(cache=>\%cache);

    Call dispatcher to return the matching entry and any regex captures

      my ($entry, $captures)=$dispatch->("thing to match");

DESCRIPTION
    This module provides a class to construct a routing table and build a
    high performance dispatcher from it.

    A table can have any combination of regex, subroutine, exact string,
    begin string end string or numeric matching of entries. The order in
    which the entries are added defines their precedence. First in, first
    tested.

    In the case of no entries matching the input, a default/fallback entry
    always matches.

    Once all the entries have been added to the table, a dispatcher is
    prepared/created. The dispatcher is an anonymous subroutine, which tests
    its argument against the matcher in each entry in the table.

    It returns a list containing the first entry that matched, and if
    applicable, an anonymous array of any captures from regex matching.

    If more entries are required to be added to the table, the dispatcher
    must be prepared again.

    A cache (hash) is used to drastically improve table lookup performance.
    Entries are automatically added to the cache. A cache hit with a regex
    matcher will re execute the regexp to ensure the captures are returned
    as expected. Removal of cache entries is up to the user to implement on
    a application basis.

  API Change
    From v0.6.0: Regexp from non core Regexp engines are now usable as a
    matcher directly. In previous versions, these where not detected and
    processed as a string to be converted into a Perl core Regexp
    internally.

    In version v0.5.3 and earlier, the dispatcher would always return a two
    element list. The first being the match entry, and the second array ref
    of any captures from a regexp match. If the matcher type was 'begin',
    'end', 'exact', or 'numeric', the second element would always be an
    reference to an empty array.

    From v0.5.4 onwards to optimise performance of non regex matching, this
    is no longer the case. Only regex type matching will generate this
    second element. Other matching types will not.

    In other words when calling the dispatcher:

                    my ($entry, $captures)=$dispatcher->($input)

    The $captures variable above now will be "undef" instead of "[]", for
    non regex matching

CREATING A TABLE
    Calling the class constructor returns a new table. There are no required
    arguments:

            my $table=Hustle::Table->new;

    In this case, a default catch all entry (an undef value) is added
    automatically.

    If an argument is provided, it is the value used in the default/catch
    all entry:

            my $table=Hustle::Table->new($default);

ENTRIES
  Structure
    An entry is an anonymous array containing the following elements:

            [matcher, value, type, default]

    matcher
        "matcher" can be a regex, a subroutine, a string or a numeric value.

        When "matcher" is a regex, any captures are returned as the second
        item when calling the dispatcher

        When "matcher" is a subroutine, it is called with input to test and
        a reference to the "value" field in the entry as the two arguments.
        If it returns a true value it matches.

        When "matcher" is string or numeric value, the last field "type"
        specifies how to perform the match. See "type" below.

        If no "type" is specified or is "undef", the "matcher" is always
        treated as a regex

    value
        This is the data you want to retrieve from the table when the
        matches.

    type
        "type" is used to adjust how the matcher is interpreted. The
        possible values are:

                undef   =>      matcher treated as a regex or subroutine if possible
                                forces basic scalars to become a regexp

                "begin" =>      matcher string matches the begining of input string
                "end"   =>      matcher string matches the end of input string
                "exact" =>      matcher string matches string equality
                "numeric" =>    matcher number matches numeric equality

        If "matcher" is a precompiled regex (i.e. "qr{}"), or a subroutine
        (i.e. CODE reference), "type" is ignored.

        If "matcher" is a string or number, it is treated as a regex unless
        "type" is as above.

    default
        This is a flag indicating if the entry was the default entry. This
        can not be set

  Adding
    Entries are added in anonymous hash, anonymous array or flattened
    format, using the "add" method.

    Anonymous array entries must contain 3 elements, in the order of:

            $table->add([$matcher, $value, $type]);

    Anonymous hashes format only need to specify the matcher and value pairs

            $table->add({matcher=>$matcher, value=>$value, type=>$type});

    Single flattened format takes a list directly. It must contain 4
    elements

            $table->add(matcher=>$matcher, value=> $value);

    Single simple format takes two elements

            $table->add(qr{some matcher}=>$value);

    Or add multiple at once using mixed formats together

            $table->add(
                    [$matcher, $value, $type],
                    {matcher=> $matcher, value=>$value},
                    matcher=>$matcher, value=>$value
            );

    In any case,"matcher" and "value" are the only items which must be
    defined for subroutine and regex matchers. String matching will need the
    "type" also specified.

  Default Matcher
    Each list has a default matcher that will unconditionally match the
    input. This entry is specified by using "undef" as the matcher when
    adding an entry.

    To make it more explicit, the it can also be changed via the
    "set_default" method.

    The default "value" of the 'default' entry is undef

PREPARING A DISPATCHER
    Once all the entries are added to the table, the dispatcher can be
    constructed by calling "prepare_dispatcher":

            my $dispatcher=$table->prepare_dispatcher(%args);

    Arguments to this method include:

    cache
        The hash ref to use as the dispatchers cache. Specifying a hash
        allows external management. If no cache is specified an internal
        cache is used.

USING A DISPATCHER
    The dispatcher is simply a sub, which you call with the input to match
    against the table entries:

            my ($entry, $captures)=$dispatcher->("input");
            my $value=$entry->[1];

    The return from the dispatcher is a list of up to two elements.

    The first is the array reference to the table entry that matched (or the
    default entry if no match was found). The value associated with the
    table entry is located in position 1

    The second item, if present, is an anonymous array of any captures due
    to a matching regex.

    NOTE In version 0.5.3 and earlier: the second element was returned as a
    ref to an empty array even if the matcher was not a regex.

COMPARISON TO OTHER MODULES
    Solid performance compared to other Perl routing/dispatch modules.
    Faster in basic tests then other Perl modules:

    Smart::Dispatch Router::Simple Router::Boom

    If you need even more performance then checkout URI::Router

    TODO: make proper benchmark and comparison

AUTHOR
    Ruben Westerberg, <drclaw@mac.com>

COPYRIGHT AND LICENSE
    Copyright (C) 2022 by Ruben Westerberg

    Licensed under MIT

DISCLAIMER OF WARRANTIES
    THIS PACKAGE IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED
    WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
    MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.