File Coverage

blib/lib/ShiftJIS/CP932/Correct.pm

Criterion	Covered	Total	%
statement	25	25	100.0
branch	4	4	100.0
condition			n/a
subroutine	7	7	100.0
pod	3	3	100.0
total	39	39	100.0

package ShiftJIS::CP932::Correct;

use strict;
use vars qw($VERSION $PACKAGE @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
use vars qw(%CorrCP932);
use Carp;
require Exporter;
@ISA = qw(Exporter);
@EXPORT      = qw(correct_cp932 is_corrected_cp932);
@EXPORT_OK   = qw(correct_cp932 is_corrected_cp932 is_cp932);
%EXPORT_TAGS = (all => \@EXPORT_OK);

$VERSION = '0.06';
$PACKAGE = 'ShiftJIS::CP932::Correct';

my $Err_too = "$PACKAGE: Too many arguments for %s";

my $Schar = '(?:[\x81-\x9F\xE0-\xFC][\x00-\xFF]|[\x00-\xFF])';

my $CP932 = '(?:[\x00-\x7f\xa1-\xdf]|'
    . '\x81[\x40-\x7e\x80-\xac\xb8-\xbf\xc8-\xce\xda-\xe8\xf0-\xf7\xfc]|'
    . '\x82[\x4f-\x58\x60-\x79\x81-\x9a\x9f-\xf1]|'
    . '\x83[\x40-\x7e\x80-\x96\x9f-\xb6\xbf-\xd6]|'
    . '\x84[\x40-\x60\x70-\x7e\x80-\x91\x9f-\xbe]|'
    . '\x88[\x9f-\xfc]|\x98[\x40-\x72\x9f-\xfc]|\xea[\x40-\x7e\x80-\xa4]|'
    . '[\x89-\x97\x99-\x9f\xe0-\xe9][\x40-\x7e\x80-\xfc]|'
    . '\x87[\x40-\x5d\x5f-\x75\x7e\x80-\x9c]|'
    . '\xed[\x40-\x7e\x80-\xfc]|\xee[\x40-\x7e\x80-\xec\xef-\xfc]|'
    . '[\xfa\xfb][\x40-\x7e\x80-\xfc]|\xfc[\x40-\x4b])';

my $CorrectCP932 = '(?:[\x00-\x7f\xa1-\xdf]|'
    . '\x81[\x40-\x7e\x80-\xac\xb8-\xbf\xc8-\xce\xda-\xe8\xf0-\xf7\xfc]|'
    . '\x82[\x4f-\x58\x60-\x79\x81-\x9a\x9f-\xf1]|'
    . '\x83[\x40-\x7e\x80-\x96\x9f-\xb6\xbf-\xd6]|'
    . '\x84[\x40-\x60\x70-\x7e\x80-\x91\x9f-\xbe]|'
    . '\x88[\x9f-\xfc]|\x98[\x40-\x72\x9f-\xfc]|\xea[\x40-\x7e\x80-\xa4]|'
    . '[\x89-\x97\x99-\x9f\xe0-\xe9][\x40-\x7e\x80-\xfc]|'
    . '\x87[\x40-\x5d\x5f-\x75\x7e\x80-\x8f\x93\x94\x98\x99]|'
    . '\xfa[\x40-\x49\x55-\x57\x5c-\x7e\x80-\xfc]|'
    . '\xfb[\x40-\x7e\x80-\xfc]|\xfc[\x40-\x4b])';

%CorrCP932 = (
"\x87\x90" => "\x81\xe0",
"\x87\x91" => "\x81\xdf",
"\x87\x92" => "\x81\xe7",
"\x87\x95" => "\x81\xe3",
"\x87\x96" => "\x81\xdb",
"\x87\x97" => "\x81\xda",
"\x87\x9a" => "\x81\xe6",
"\x87\x9b" => "\x81\xbf",
"\x87\x9c" => "\x81\xbe",
"\xed\x40" => "\xfa\x5c",
"\xed\x41" => "\xfa\x5d",
"\xed\x42" => "\xfa\x5e",
"\xed\x43" => "\xfa\x5f",
"\xed\x44" => "\xfa\x60",
"\xed\x45" => "\xfa\x61",
"\xed\x46" => "\xfa\x62",
"\xed\x47" => "\xfa\x63",
"\xed\x48" => "\xfa\x64",
"\xed\x49" => "\xfa\x65",
"\xed\x4a" => "\xfa\x66",
"\xed\x4b" => "\xfa\x67",
"\xed\x4c" => "\xfa\x68",
"\xed\x4d" => "\xfa\x69",
"\xed\x4e" => "\xfa\x6a",
"\xed\x4f" => "\xfa\x6b",
"\xed\x50" => "\xfa\x6c",
"\xed\x51" => "\xfa\x6d",
"\xed\x52" => "\xfa\x6e",
"\xed\x53" => "\xfa\x6f",
"\xed\x54" => "\xfa\x70",
"\xed\x55" => "\xfa\x71",
"\xed\x56" => "\xfa\x72",
"\xed\x57" => "\xfa\x73",
"\xed\x58" => "\xfa\x74",
"\xed\x59" => "\xfa\x75",
"\xed\x5a" => "\xfa\x76",
"\xed\x5b" => "\xfa\x77",
"\xed\x5c" => "\xfa\x78",
"\xed\x5d" => "\xfa\x79",
"\xed\x5e" => "\xfa\x7a",
"\xed\x5f" => "\xfa\x7b",
"\xed\x60" => "\xfa\x7c",
"\xed\x61" => "\xfa\x7d",
"\xed\x62" => "\xfa\x7e",
"\xed\x63" => "\xfa\x80",
"\xed\x64" => "\xfa\x81",
"\xed\x65" => "\xfa\x82",
"\xed\x66" => "\xfa\x83",
"\xed\x67" => "\xfa\x84",
"\xed\x68" => "\xfa\x85",
"\xed\x69" => "\xfa\x86",
"\xed\x6a" => "\xfa\x87",
"\xed\x6b" => "\xfa\x88",
"\xed\x6c" => "\xfa\x89",
"\xed\x6d" => "\xfa\x8a",
"\xed\x6e" => "\xfa\x8b",
"\xed\x6f" => "\xfa\x8c",
"\xed\x70" => "\xfa\x8d",
"\xed\x71" => "\xfa\x8e",
"\xed\x72" => "\xfa\x8f",
"\xed\x73" => "\xfa\x90",
"\xed\x74" => "\xfa\x91",
"\xed\x75" => "\xfa\x92",
"\xed\x76" => "\xfa\x93",
"\xed\x77" => "\xfa\x94",
"\xed\x78" => "\xfa\x95",
"\xed\x79" => "\xfa\x96",
"\xed\x7a" => "\xfa\x97",
"\xed\x7b" => "\xfa\x98",
"\xed\x7c" => "\xfa\x99",
"\xed\x7d" => "\xfa\x9a",
"\xed\x7e" => "\xfa\x9b",
"\xed\x80" => "\xfa\x9c",
"\xed\x81" => "\xfa\x9d",
"\xed\x82" => "\xfa\x9e",
"\xed\x83" => "\xfa\x9f",
"\xed\x84" => "\xfa\xa0",
"\xed\x85" => "\xfa\xa1",
"\xed\x86" => "\xfa\xa2",
"\xed\x87" => "\xfa\xa3",
"\xed\x88" => "\xfa\xa4",
"\xed\x89" => "\xfa\xa5",
"\xed\x8a" => "\xfa\xa6",
"\xed\x8b" => "\xfa\xa7",
"\xed\x8c" => "\xfa\xa8",
"\xed\x8d" => "\xfa\xa9",
"\xed\x8e" => "\xfa\xaa",
"\xed\x8f" => "\xfa\xab",
"\xed\x90" => "\xfa\xac",
"\xed\x91" => "\xfa\xad",
"\xed\x92" => "\xfa\xae",
"\xed\x93" => "\xfa\xaf",
"\xed\x94" => "\xfa\xb0",
"\xed\x95" => "\xfa\xb1",
"\xed\x96" => "\xfa\xb2",
"\xed\x97" => "\xfa\xb3",
"\xed\x98" => "\xfa\xb4",
"\xed\x99" => "\xfa\xb5",
"\xed\x9a" => "\xfa\xb6",
"\xed\x9b" => "\xfa\xb7",
"\xed\x9c" => "\xfa\xb8",
"\xed\x9d" => "\xfa\xb9",
"\xed\x9e" => "\xfa\xba",
"\xed\x9f" => "\xfa\xbb",
"\xed\xa0" => "\xfa\xbc",
"\xed\xa1" => "\xfa\xbd",
"\xed\xa2" => "\xfa\xbe",
"\xed\xa3" => "\xfa\xbf",
"\xed\xa4" => "\xfa\xc0",
"\xed\xa5" => "\xfa\xc1",
"\xed\xa6" => "\xfa\xc2",
"\xed\xa7" => "\xfa\xc3",
"\xed\xa8" => "\xfa\xc4",
"\xed\xa9" => "\xfa\xc5",
"\xed\xaa" => "\xfa\xc6",
"\xed\xab" => "\xfa\xc7",
"\xed\xac" => "\xfa\xc8",
"\xed\xad" => "\xfa\xc9",
"\xed\xae" => "\xfa\xca",
"\xed\xaf" => "\xfa\xcb",
"\xed\xb0" => "\xfa\xcc",
"\xed\xb1" => "\xfa\xcd",
"\xed\xb2" => "\xfa\xce",
"\xed\xb3" => "\xfa\xcf",
"\xed\xb4" => "\xfa\xd0",
"\xed\xb5" => "\xfa\xd1",
"\xed\xb6" => "\xfa\xd2",
"\xed\xb7" => "\xfa\xd3",
"\xed\xb8" => "\xfa\xd4",
"\xed\xb9" => "\xfa\xd5",
"\xed\xba" => "\xfa\xd6",
"\xed\xbb" => "\xfa\xd7",
"\xed\xbc" => "\xfa\xd8",
"\xed\xbd" => "\xfa\xd9",
"\xed\xbe" => "\xfa\xda",
"\xed\xbf" => "\xfa\xdb",
"\xed\xc0" => "\xfa\xdc",
"\xed\xc1" => "\xfa\xdd",
"\xed\xc2" => "\xfa\xde",
"\xed\xc3" => "\xfa\xdf",
"\xed\xc4" => "\xfa\xe0",
"\xed\xc5" => "\xfa\xe1",
"\xed\xc6" => "\xfa\xe2",
"\xed\xc7" => "\xfa\xe3",
"\xed\xc8" => "\xfa\xe4",
"\xed\xc9" => "\xfa\xe5",
"\xed\xca" => "\xfa\xe6",
"\xed\xcb" => "\xfa\xe7",
"\xed\xcc" => "\xfa\xe8",
"\xed\xcd" => "\xfa\xe9",
"\xed\xce" => "\xfa\xea",
"\xed\xcf" => "\xfa\xeb",
"\xed\xd0" => "\xfa\xec",
"\xed\xd1" => "\xfa\xed",
"\xed\xd2" => "\xfa\xee",
"\xed\xd3" => "\xfa\xef",
"\xed\xd4" => "\xfa\xf0",
"\xed\xd5" => "\xfa\xf1",
"\xed\xd6" => "\xfa\xf2",
"\xed\xd7" => "\xfa\xf3",
"\xed\xd8" => "\xfa\xf4",
"\xed\xd9" => "\xfa\xf5",
"\xed\xda" => "\xfa\xf6",
"\xed\xdb" => "\xfa\xf7",
"\xed\xdc" => "\xfa\xf8",
"\xed\xdd" => "\xfa\xf9",
"\xed\xde" => "\xfa\xfa",
"\xed\xdf" => "\xfa\xfb",
"\xed\xe0" => "\xfa\xfc",
"\xed\xe1" => "\xfb\x40",
"\xed\xe2" => "\xfb\x41",
"\xed\xe3" => "\xfb\x42",
"\xed\xe4" => "\xfb\x43",
"\xed\xe5" => "\xfb\x44",
"\xed\xe6" => "\xfb\x45",
"\xed\xe7" => "\xfb\x46",
"\xed\xe8" => "\xfb\x47",
"\xed\xe9" => "\xfb\x48",
"\xed\xea" => "\xfb\x49",
"\xed\xeb" => "\xfb\x4a",
"\xed\xec" => "\xfb\x4b",
"\xed\xed" => "\xfb\x4c",
"\xed\xee" => "\xfb\x4d",
"\xed\xef" => "\xfb\x4e",
"\xed\xf0" => "\xfb\x4f",
"\xed\xf1" => "\xfb\x50",
"\xed\xf2" => "\xfb\x51",
"\xed\xf3" => "\xfb\x52",
"\xed\xf4" => "\xfb\x53",
"\xed\xf5" => "\xfb\x54",
"\xed\xf6" => "\xfb\x55",
"\xed\xf7" => "\xfb\x56",
"\xed\xf8" => "\xfb\x57",
"\xed\xf9" => "\xfb\x58",
"\xed\xfa" => "\xfb\x59",
"\xed\xfb" => "\xfb\x5a",
"\xed\xfc" => "\xfb\x5b",
"\xee\x40" => "\xfb\x5c",
"\xee\x41" => "\xfb\x5d",
"\xee\x42" => "\xfb\x5e",
"\xee\x43" => "\xfb\x5f",
"\xee\x44" => "\xfb\x60",
"\xee\x45" => "\xfb\x61",
"\xee\x46" => "\xfb\x62",
"\xee\x47" => "\xfb\x63",
"\xee\x48" => "\xfb\x64",
"\xee\x49" => "\xfb\x65",
"\xee\x4a" => "\xfb\x66",
"\xee\x4b" => "\xfb\x67",
"\xee\x4c" => "\xfb\x68",
"\xee\x4d" => "\xfb\x69",
"\xee\x4e" => "\xfb\x6a",
"\xee\x4f" => "\xfb\x6b",
"\xee\x50" => "\xfb\x6c",
"\xee\x51" => "\xfb\x6d",
"\xee\x52" => "\xfb\x6e",
"\xee\x53" => "\xfb\x6f",
"\xee\x54" => "\xfb\x70",
"\xee\x55" => "\xfb\x71",
"\xee\x56" => "\xfb\x72",
"\xee\x57" => "\xfb\x73",
"\xee\x58" => "\xfb\x74",
"\xee\x59" => "\xfb\x75",
"\xee\x5a" => "\xfb\x76",
"\xee\x5b" => "\xfb\x77",
"\xee\x5c" => "\xfb\x78",
"\xee\x5d" => "\xfb\x79",
"\xee\x5e" => "\xfb\x7a",
"\xee\x5f" => "\xfb\x7b",
"\xee\x60" => "\xfb\x7c",
"\xee\x61" => "\xfb\x7d",
"\xee\x62" => "\xfb\x7e",
"\xee\x63" => "\xfb\x80",
"\xee\x64" => "\xfb\x81",
"\xee\x65" => "\xfb\x82",
"\xee\x66" => "\xfb\x83",
"\xee\x67" => "\xfb\x84",
"\xee\x68" => "\xfb\x85",
"\xee\x69" => "\xfb\x86",
"\xee\x6a" => "\xfb\x87",
"\xee\x6b" => "\xfb\x88",
"\xee\x6c" => "\xfb\x89",
"\xee\x6d" => "\xfb\x8a",
"\xee\x6e" => "\xfb\x8b",
"\xee\x6f" => "\xfb\x8c",
"\xee\x70" => "\xfb\x8d",
"\xee\x71" => "\xfb\x8e",
"\xee\x72" => "\xfb\x8f",
"\xee\x73" => "\xfb\x90",
"\xee\x74" => "\xfb\x91",
"\xee\x75" => "\xfb\x92",
"\xee\x76" => "\xfb\x93",
"\xee\x77" => "\xfb\x94",
"\xee\x78" => "\xfb\x95",
"\xee\x79" => "\xfb\x96",
"\xee\x7a" => "\xfb\x97",
"\xee\x7b" => "\xfb\x98",
"\xee\x7c" => "\xfb\x99",
"\xee\x7d" => "\xfb\x9a",
"\xee\x7e" => "\xfb\x9b",
"\xee\x80" => "\xfb\x9c",
"\xee\x81" => "\xfb\x9d",
"\xee\x82" => "\xfb\x9e",
"\xee\x83" => "\xfb\x9f",
"\xee\x84" => "\xfb\xa0",
"\xee\x85" => "\xfb\xa1",
"\xee\x86" => "\xfb\xa2",
"\xee\x87" => "\xfb\xa3",
"\xee\x88" => "\xfb\xa4",
"\xee\x89" => "\xfb\xa5",
"\xee\x8a" => "\xfb\xa6",
"\xee\x8b" => "\xfb\xa7",
"\xee\x8c" => "\xfb\xa8",
"\xee\x8d" => "\xfb\xa9",
"\xee\x8e" => "\xfb\xaa",
"\xee\x8f" => "\xfb\xab",
"\xee\x90" => "\xfb\xac",
"\xee\x91" => "\xfb\xad",
"\xee\x92" => "\xfb\xae",
"\xee\x93" => "\xfb\xaf",
"\xee\x94" => "\xfb\xb0",
"\xee\x95" => "\xfb\xb1",
"\xee\x96" => "\xfb\xb2",
"\xee\x97" => "\xfb\xb3",
"\xee\x98" => "\xfb\xb4",
"\xee\x99" => "\xfb\xb5",
"\xee\x9a" => "\xfb\xb6",
"\xee\x9b" => "\xfb\xb7",
"\xee\x9c" => "\xfb\xb8",
"\xee\x9d" => "\xfb\xb9",
"\xee\x9e" => "\xfb\xba",
"\xee\x9f" => "\xfb\xbb",
"\xee\xa0" => "\xfb\xbc",
"\xee\xa1" => "\xfb\xbd",
"\xee\xa2" => "\xfb\xbe",
"\xee\xa3" => "\xfb\xbf",
"\xee\xa4" => "\xfb\xc0",
"\xee\xa5" => "\xfb\xc1",
"\xee\xa6" => "\xfb\xc2",
"\xee\xa7" => "\xfb\xc3",
"\xee\xa8" => "\xfb\xc4",
"\xee\xa9" => "\xfb\xc5",
"\xee\xaa" => "\xfb\xc6",
"\xee\xab" => "\xfb\xc7",
"\xee\xac" => "\xfb\xc8",
"\xee\xad" => "\xfb\xc9",
"\xee\xae" => "\xfb\xca",
"\xee\xaf" => "\xfb\xcb",
"\xee\xb0" => "\xfb\xcc",
"\xee\xb1" => "\xfb\xcd",
"\xee\xb2" => "\xfb\xce",
"\xee\xb3" => "\xfb\xcf",
"\xee\xb4" => "\xfb\xd0",
"\xee\xb5" => "\xfb\xd1",
"\xee\xb6" => "\xfb\xd2",
"\xee\xb7" => "\xfb\xd3",
"\xee\xb8" => "\xfb\xd4",
"\xee\xb9" => "\xfb\xd5",
"\xee\xba" => "\xfb\xd6",
"\xee\xbb" => "\xfb\xd7",
"\xee\xbc" => "\xfb\xd8",
"\xee\xbd" => "\xfb\xd9",
"\xee\xbe" => "\xfb\xda",
"\xee\xbf" => "\xfb\xdb",
"\xee\xc0" => "\xfb\xdc",
"\xee\xc1" => "\xfb\xdd",
"\xee\xc2" => "\xfb\xde",
"\xee\xc3" => "\xfb\xdf",
"\xee\xc4" => "\xfb\xe0",
"\xee\xc5" => "\xfb\xe1",
"\xee\xc6" => "\xfb\xe2",
"\xee\xc7" => "\xfb\xe3",
"\xee\xc8" => "\xfb\xe4",
"\xee\xc9" => "\xfb\xe5",
"\xee\xca" => "\xfb\xe6",
"\xee\xcb" => "\xfb\xe7",
"\xee\xcc" => "\xfb\xe8",
"\xee\xcd" => "\xfb\xe9",
"\xee\xce" => "\xfb\xea",
"\xee\xcf" => "\xfb\xeb",
"\xee\xd0" => "\xfb\xec",
"\xee\xd1" => "\xfb\xed",
"\xee\xd2" => "\xfb\xee",
"\xee\xd3" => "\xfb\xef",
"\xee\xd4" => "\xfb\xf0",
"\xee\xd5" => "\xfb\xf1",
"\xee\xd6" => "\xfb\xf2",
"\xee\xd7" => "\xfb\xf3",
"\xee\xd8" => "\xfb\xf4",
"\xee\xd9" => "\xfb\xf5",
"\xee\xda" => "\xfb\xf6",
"\xee\xdb" => "\xfb\xf7",
"\xee\xdc" => "\xfb\xf8",
"\xee\xdd" => "\xfb\xf9",
"\xee\xde" => "\xfb\xfa",
"\xee\xdf" => "\xfb\xfb",
"\xee\xe0" => "\xfb\xfc",
"\xee\xe1" => "\xfc\x40",
"\xee\xe2" => "\xfc\x41",
"\xee\xe3" => "\xfc\x42",
"\xee\xe4" => "\xfc\x43",
"\xee\xe5" => "\xfc\x44",
"\xee\xe6" => "\xfc\x45",
"\xee\xe7" => "\xfc\x46",
"\xee\xe8" => "\xfc\x47",
"\xee\xe9" => "\xfc\x48",
"\xee\xea" => "\xfc\x49",
"\xee\xeb" => "\xfc\x4a",
"\xee\xec" => "\xfc\x4b",
"\xee\xef" => "\xfa\x40",
"\xee\xf0" => "\xfa\x41",
"\xee\xf1" => "\xfa\x42",
"\xee\xf2" => "\xfa\x43",
"\xee\xf3" => "\xfa\x44",
"\xee\xf4" => "\xfa\x45",
"\xee\xf5" => "\xfa\x46",
"\xee\xf6" => "\xfa\x47",
"\xee\xf7" => "\xfa\x48",
"\xee\xf8" => "\xfa\x49",
"\xee\xf9" => "\x81\xca",
"\xee\xfa" => "\xfa\x55",
"\xee\xfb" => "\xfa\x56",
"\xee\xfc" => "\xfa\x57",
"\xfa\x4a" => "\x87\x54",
"\xfa\x4b" => "\x87\x55",
"\xfa\x4c" => "\x87\x56",
"\xfa\x4d" => "\x87\x57",
"\xfa\x4e" => "\x87\x58",
"\xfa\x4f" => "\x87\x59",
"\xfa\x50" => "\x87\x5a",
"\xfa\x51" => "\x87\x5b",
"\xfa\x52" => "\x87\x5c",
"\xfa\x53" => "\x87\x5d",
"\xfa\x54" => "\x81\xca",
"\xfa\x58" => "\x87\x8a",
"\xfa\x59" => "\x87\x82",
"\xfa\x5a" => "\x87\x84",
"\xfa\x5b" => "\x81\xe6",
);

sub correct_cp932 ($) {
    my $result = "";
    my $str = shift;
    while ($str =~ /\G($Schar)/go) {
	my $c = $1;
	next unless $c =~ m|^$CP932$|o;
	$result .= defined $CorrCP932{ $c } ? $CorrCP932{ $c } : $c;
    }
    return $result;
}

sub is_corrected_cp932 ($) {
    my $str = shift;
    $str =~ s/\G$CorrectCP932//g;
    ! length $str;
}

sub is_cp932 ($) {
    my $str = shift;
    $str =~ s/\G$CP932//g;
    ! length $str;
}

1;
__END__