package Unicode::CaseFold;

# ABSTRACT: Unicode case-folding for case-insensitive lookups.

BEGIN {
  our $VERSION = '1.01'; # VERSION
}
our $AUTHORITY = 'cpan:ARODLAND'; # AUTHORITY

use strict;
use warnings;

use 5.008001;

use Scalar::Util 1.11 ();
require Exporter;

our @ISA = qw(Exporter);
our @EXPORT_OK = qw(case_fold);
our @EXPORT = qw(fc);

our $SIMPLE_FOLDING;
our $XS = 0;

BEGIN {
  unless ($ENV{PERL_UNICODE_CASEFOLD_PP}) {
    local $@;
    eval {
      our $VERSION;
      require XSLoader;
      XSLoader::load(__PACKAGE__,
        exists $Unicode::CaseFold::{VERSION}
        ? ${ $Unicode::CaseFold::{VERSION} }
        : ()
      );
    };
    die $@ if $@ && $@ !~ /object version|loadable object/;
    $XS = 1 unless $@;
    $SIMPLE_FOLDING = 0 unless $@;
  }
  if (!$XS) {
    require Unicode::CaseFoldPP;
  }
}

sub fc {
  @_ = ($_) unless @_;
  goto &case_fold;
}

BEGIN {
  # Perl 5.10+ supports the (_) prototype which does the $_-defaulting for us,
  # and handles "lexical $_". Older perl doesn't, but we can fake it fairly
  # closely with a (;$) prototype. Older perl didn't have lexical $_ anyway.

  if ($^V ge v5.10.0) {
    Scalar::Util::set_prototype(\&fc, '_');
  } else {
    Scalar::Util::set_prototype(\&fc, ';$');
  }
}

1;

__END__

=pod

=encoding UTF-8

=head1 NAME

Unicode::CaseFold - Unicode case-folding for case-insensitive lookups.

=head1 VERSION

version 1.01

=head1 SYNOPSIS

    use Unicode::CaseFold;
    
    my $folded = fc $string;

=head2 What is Case-Folding?

In non-Unicode contexts, a common idiom to compare two strings
case-insensitively is C<lc($this) eq lc($that)>. Before comparing two strings
we I<normalize> them to an all-lowercase version. C<"Hello">, C<"HELLO">, and
C<"HeLlO"> all have the same lowercase form (C<"hello">), so it doesn't matter
which one we start with; they are all equal to one another after C<lc>.

In Unicode, things aren't so simple. A Unicode character might have mappings
for I<uppercase>, I<lowercase>, and I<titlecase>, and the lowercase mapping of
the uppercase mapping of a given character might not be the character that you
started with! For example C<< lc(uc("\N{LATIN SMALL LETTER SHARP S")) >> is
C<"ss">, not the eszett we started off with! Case-folding is a part of the
Unicode standard that allows any two strings that differ from one another only
by case to map to the same "case-folded" form, even when those strings include
characters with complex case-mappings.

=head2 Use for Case-insensitive Comparison

Simply write C<fc($this) eq fc($that)> instead of C<lc($this) eq lc($that)>.
You can also use C<index> on case-folded strings for substring search.

=head2 Use for String Lookups

Frequently we want to store data in a hash, or a database, or an external file
for later retrieval. Sometimes we want to be able to match the keys in this
data case-insensitively -- that is, we should be able to store some data under
the key "hello" and later retrieve it with the key "HELLO". Some databases
have complete support for collation, but in other databases the support is
missing or broken, and Perl hashes don't support it at all. By making
case-folding part of the process you use to normalize your keys before using
them to access a database or data structure, you get case-insensitive lookup.

    $roles{fc "Samuel L. Jackson"} = ["Gin Rummy", "Nick Fury", "Mace Windu"];
    
    $roles = $roles{fc "Samuel l. JACKSON"}; # Gets the data.

=head1 DESCRIPTION

This module provides Unicode case-folding for Perl. Case-folding is a tool
that allows a program to make case-insensitive string comparisons or do
case-insensitive lookups.

=head1 EXPORTS

=head2 fc($str)

Exported by default when you use the module. C<use Unicode::CaseFold ()> or
C<use Unicode::CaseFold qw(case_fold !fc)> if you don't want it to be
exported.

Returns the case-folded version of C<$str>. This function is prototyped to act
as much as possible like the built-ins C<lc> and C<uc>; it imposes a scalar
context on its argument, and if called with no argument it will return the
case-folded version of C<$_>.

=head2 case_fold($str)

Exported on request. Just like C<fc>, except that it has no prototype and
won't case-fold C<$_> if called without an argument.

=head1 VARIABLES

=head2 $Unicode::CaseFold::XS

Whether the XS extension is in use. The pure-perl implementation is 5-10 times
slower than the XS extension, and on versions of perl before 5.10.0 it will
use simple case-folding instead of full case-folding (see below).

=head2 $Unicode::CaseFold::SIMPLE_FOLDING

Is set to true if the perl version is prior to 5.10.0 and the XS extension is
not available. In this case, C<fc> will perform a simple case-folding instead
of a full case-folding. Although relatively few characters are affected,
strings case-folded using simple folding might not compare equal to the
corresponding strings case-folded with full folding, which may cause
compatibility issues.

Furthermore, when simple folding is in use, some strings
that would have case-folded to the same value when using full folding will
instead case-fold to different values. For example, C<fc("Wei\x{df}")> and
C<fc("Weiss")> both produce C<"weiss"> when full folding is in effect, but
the former produces C<"wei\x{df}"> when using simple folding.

If you want to check for this potentially dangerous situation, consult the
C<$Unicode::CaseFold::SIMPLE_FOLDING> variable.

=head1 COMPATIBILITY

=over 4

=item *

C<Unicode::CaseFold> requires Perl 5.8.1 or newer.

=item *

Different versions of perl include different versions of the Unicode database,
which is revised over time. If you are likely to be comparing strings that
have been folded using different versions of perl, you may need to consult the
changes for intervening Unicode standard versions to find out whether your
code will work correctly.

=item *

C<Unicode::CaseFold> uses "simple" rather than "full" case-folding when
operating in Pure-perl mode on perl versions previous to 5.10.0. For
compatibility implications, see L</$Unicode::CaseFold::SIMPLE_FOLDING>.

=back

=head1 SEE ALSO

=over 4

=item *

L<http://unicode.org/reports/tr21/tr21-5.html>: Unicode Standard Annex #21: Case Mappings

=back

=head1 AUTHOR

Andrew Rodland <arodland@cpan.org>

=head1 COPYRIGHT AND LICENSE

This software is copyright (c) 2017 by Andrew Rodland.

This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.

=cut