package FFI::C; use strict; use warnings; use 5.008001; use Carp (); use Ref::Util qw( is_ref is_plain_arrayref is_plain_hashref ); # ABSTRACT: C data types for FFI our $VERSION = '0.15'; # VERSION our %ffi; sub _ffi_get { my($filename) = @_; $ffi{$filename} ||= do { require FFI::Platypus; FFI::Platypus->new( api => 1 ); }; } sub ffi { my($class, $new) = @_; my(undef, $filename) = caller; if($new) { Carp::croak("Already have an FFI::Platypus instance for $filename") if defined $ffi{$filename}; return $ffi{$filename} = $new; } _ffi_get($filename); } our $def_class; sub _gen { shift; my($class, $filename) = caller; my($name, $members); my %extra = is_plain_hashref $_[-1] ? %{ pop() } : (); if(@_ == 2 && !is_ref $_[0] && is_plain_arrayref $_[1]) { ($name, $members) = @_; } elsif(@_ == 1 && is_plain_arrayref $_[0]) { $name = lcfirst [split /::/, $class]->[-1]; $name =~ s/([A-Z]+)/'_' . lc($1)/ge; $name .= "_t"; ($members) = @_; } else { my($method) = map { lc $_ } $def_class =~ /::([A-Za-z]+)Def$/; Carp::croak("usage: FFI::C->$method([\$name], \\\@members)"); } $def_class->new( _ffi_get($filename), %extra, name => $name, class => $class, members => $members, ); } sub struct { require FFI::C::StructDef; $def_class = 'FFI::C::StructDef'; goto &_gen; } sub union { require FFI::C::UnionDef; $def_class = 'FFI::C::UnionDef'; goto &_gen; } sub array { require FFI::C::ArrayDef; $def_class = 'FFI::C::ArrayDef'; goto &_gen; } sub enum { (undef) = shift; my $name = defined $_[0] && !is_ref $_[0] ? shift : undef; my @values = defined $_[0] && is_plain_arrayref $_[0] ? @{shift()} : (); my %config = defined $_[0] && is_plain_hashref $_[0] ? %{shift()} : (); my($class, $filename) = caller; unless(defined $name) { $name = lcfirst [split /::/, $class]->[-1]; $name =~ s/([A-Z]+)/'_' . lc($1)/ge; $name .= "_t"; } my $ffi = _ffi_get($filename), $config{package} ||= $class; my @maps; $config{maps} = \@maps; my $rev = $config{rev} ||= 'str'; $ffi->load_custom_type('::Enum', $name, \%config, @values); my($str_lookup, $int_lookup, $type) = @maps; require FFI::C::Def; $ffi->def('FFI::C::EnumDef', $name, FFI::C::EnumDef->new( str_lookup => $str_lookup, int_lookup => $int_lookup, type => $type, rev => $rev, ) ); } 1; __END__ =pod =encoding UTF-8 =head1 NAME FFI::C - C data types for FFI =head1 VERSION version 0.15 =head1 SYNOPSIS In C: #include typedef struct { uint8_t red; uint8_t green; uint8_t blue; } color_value_t; typedef struct { char name[22]; color_value_t value; } named_color_t; typedef named_color_t array_named_color_t[4]; typedef union { uint8_t u8; uint16_t u16; uint32_t u32; uint64_t u64; } anyint_t; In Perl: use FFI::C; package ColorValue { FFI::C->struct([ red => 'uint8', green => 'uint8', blue => 'uint8', ]); } package NamedColor { FFI::C->struct([ name => 'string(22)', value => 'color_value_t', ]); } package ArrayNamedColor { FFI::C->array(['named_color_t' => 4]); }; my $array = ArrayNamedColor->new([ { name => "red", value => { red => 255 } }, { name => "green", value => { green => 255 } }, { name => "blue", value => { blue => 255 } }, { name => "purple", value => { red => 255, blue => 255 } }, ]); # dim each color by 1/2 foreach my $color (@$array) { $color->value->red ( $color->value->red / 2 ); $color->value->green( $color->value->green / 2 ); $color->value->blue ( $color->value->blue / 2 ); } # print out the colors foreach my $color (@$array) { printf "%s [%02x %02x %02x]\n", $color->name, $color->value->red, $color->value->green, $color->value->blue; } package AnyInt { FFI::C->union([ u8 => 'uint8', u16 => 'uint16', u32 => 'uint32', u64 => 'uint64', ]); } my $int = AnyInt->new({ u8 => 42 }); print $int->u32; =head1 DESCRIPTION This distribution provides tools for building classes to interface for common C data types. Arrays, C, C and nested types based on those are supported. Core L also provides L for manipulating and passing structured data. Typically you want to use L instead, the main exception is when you need to pass structured data by value instead of by reference. To work with C APIs that work with C file pointers you can use L and L. For C APIs that expose the POSIX C structure use L. =head1 METHODS =head2 ffi FFI::C->ffi($ffi); my $ffi = FFI::C->ffi; Get or set the L instance used for the current Perl file (C<.pl> or C<.pm>). By default a new Platypus instance is created the on the first call to C, or when a new type is created with C, C or C below, so if you want to use your own Platypus instance make sure that you set it as soon as possible. The Platypus instance is file scoped because scoping on just one package doesn't make sense if you are defining multiple types in one file since each type must be in its own package. It also doesn't make sense to make the Platypus instance global, because different distributions would conflict. =head2 struct FFI::C->struct($name, \@members); FFI::C->struct(\@members); Generate a new L class with the given C<@members> into the calling package. (C<@members> should be a list of name/type pairs). You may optionally give a C<$name> which will be used for the L type name for the generated class. If you do not specify a C<$name>, a C style name will be generated from the last segment in the calling package name by converting to snake case and appending a C<_t> to the end. As an example, given: package MyLibrary::FooBar { FFI::C->struct([ a => 'uint8', b => 'float', ]); }; You can use C via the file scoped L instance using the type C. my $foobar = MyLibrary::FooBar->new({ a => 1, b => 3.14 }); $ffi->function( my_library_func => [ 'foo_bar_t' ] => 'void' )->call($foobar); =head2 union FFI::C->union($name, \@members); FFI::C->union(\@members); This works exactly like the C method above, except a L class is generated instead. =head2 array FFI::C->array($name, [$type, $count]); FFI::C->array($name, [$type]); FFI::C->array([$type, $count]); FFI::C->array([$type]); This is similar to C and C above, except L is generated. For an array you give it the member type and the element count. The element count is optional for variable length arrays, but keep in mind that when you create such an array you do need to provide a size. =head2 enum FFI::C->enum($name, \@values, \%config); FFI::C->enum(\@values, \%config); FFI::C->enum(\@values, \%config); FFI::C->enum(\@values); Defines an enum. The C<@values> and C<%config> are passed to L, except the constants are exported to the calling package by default. =head1 EXAMPLES =head2 unix time struct use FFI::Platypus 1.00; use FFI::C; my $ffi = FFI::Platypus->new( api => 1, lib => [undef], ); FFI::C->ffi($ffi); package Unix::TimeStruct { FFI::C->struct(tm => [ tm_sec => 'int', tm_min => 'int', tm_hour => 'int', tm_mday => 'int', tm_mon => 'int', tm_year => 'int', tm_wday => 'int', tm_yday => 'int', tm_isdst => 'int', tm_gmtoff => 'long', _tm_zone => 'opaque', ]); # For now 'string' is unsupported by FFI::C, but we # can cast the time zone from an opaque pointer to # string. sub tm_zone { my $self = shift; $ffi->cast('opaque', 'string', $self->_tm_zone); } # attach the C localtime function $ffi->attach( localtime => ['time_t*'] => 'tm', sub { my($inner, $class, $time) = @_; $time = time unless defined $time; $inner->(\$time); }); } # now we can actually use our My::UnixTime class my $time = Unix::TimeStruct->localtime; printf "time is %d:%d:%d %s\n", $time->tm_hour, $time->tm_min, $time->tm_sec, $time->tm_zone; =head1 CAVEATS L objects must be passed into C via L by pointers. So-called "pass-by-value" is not and will not be supported. For "pass-by-value" record types, you should instead use L. =head1 SEE ALSO =over 4 =item L =item L =item L =item L =item L =item L =item L =item L =item L =item L =item L =item L =back =head1 AUTHOR Graham Ollis =head1 COPYRIGHT AND LICENSE This software is copyright (c) 2020-2022 by Graham Ollis. 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