diff options
author | Rafael Garcia-Suarez <rgarciasuarez@gmail.com> | 2005-08-06 23:07:46 +0000 |
---|---|---|
committer | Rafael Garcia-Suarez <rgarciasuarez@gmail.com> | 2005-08-06 23:07:46 +0000 |
commit | 60f80d084f45e0a17d269ed04fc67c0168f5a001 (patch) | |
tree | 2f4f4f75a2f84e345b0bfaa9d6ca68b16538cecb /ext/Encode | |
parent | 4b6f7c8178ca97811f956dfcee6aa0d857a91238 (diff) | |
download | perl-60f80d084f45e0a17d269ed04fc67c0168f5a001.tar.gz |
Add files missed in change 25271
p4raw-id: //depot/perl@25273
Diffstat (limited to 'ext/Encode')
-rw-r--r-- | ext/Encode/lib/Encode/MIME/Header/ISO_2022_JP.pm | 127 | ||||
-rw-r--r-- | ext/Encode/t/mime_header_iso2022jp.t | 40 |
2 files changed, 167 insertions, 0 deletions
diff --git a/ext/Encode/lib/Encode/MIME/Header/ISO_2022_JP.pm b/ext/Encode/lib/Encode/MIME/Header/ISO_2022_JP.pm new file mode 100644 index 0000000000..5f637a3247 --- /dev/null +++ b/ext/Encode/lib/Encode/MIME/Header/ISO_2022_JP.pm @@ -0,0 +1,127 @@ +package Encode::MIME::Header::ISO_2022_JP; + +use strict; +use base qw(Encode::MIME::Header); + +$Encode::Encoding{'MIME-Header-ISO_2022_JP'} + = bless {encode => 'B', bpl => 76, Name => 'MIME-Header-ISO_2022_JP'} + => __PACKAGE__; + +use constant HEAD => '=?ISO-2022-JP?B?'; +use constant TAIL => '?='; + +use Encode::CJKConstants qw(%RE); + +our $VERSION = do { my @r = (q$Revision: 1.1 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r }; + + +# I owe the below codes totally to +# Jcode by Dan Kogai & http://www.din.or.jp/~ohzaki/perl.htm#JP_Base64 + +sub encode { + my $self = shift; + my $str = shift; + + utf8::encode($str) if( Encode::is_utf8($str) ); + Encode::from_to($str, 'utf8', 'euc-jp'); + + my($trailing_crlf) = ($str =~ /(\n|\r|\x0d\x0a)$/o); + + $str = _mime_unstructured_header($str, $self->{bpl}); + + not $trailing_crlf and $str =~ s/(\n|\r|\x0d\x0a)$//o; + + return $str; +} + + +sub _mime_unstructured_header { + my ($oldheader, $bpl) = @_; + my $crlf = $oldheader =~ /\n$/; + my($header, @words, @wordstmp, $i) = (''); + + $oldheader =~ s/\s+$//; + + @wordstmp = split /\s+/, $oldheader; + + for ($i = 0; $i < $#wordstmp; $i++){ + if( $wordstmp[$i] !~ /^[\x21-\x7E]+$/ and $wordstmp[$i + 1] !~ /^[\x21-\x7E]+$/){ + $wordstmp[$i + 1] = "$wordstmp[$i] $wordstmp[$i + 1]"; + } + else{ + push(@words, $wordstmp[$i]); + } + } + + push(@words, $wordstmp[-1]); + + for my $word (@words){ + if ($word =~ /^[\x21-\x7E]+$/) { + $header =~ /(?:.*\n)*(.*)/; + if (length($1) + length($word) > $bpl) { + $header .= "\n $word"; + } + else{ + $header .= $word; + } + } + else{ + $header = _add_encoded_word($word, $header, $bpl); + } + + $header =~ /(?:.*\n)*(.*)/; + + if(length($1) == $bpl){ + $header .= "\n "; + } + else { + $header .= ' '; + } + } + + $header =~ s/\n? $//mg; + + $crlf ? "$header\n" : $header; +} + + +sub _add_encoded_word { + my($str, $line, $bpl) = @_; + my $result = ''; + + while( length($str) ){ + my $target = $str; + $str = ''; + + if(length($line) + 22 + ($target =~ /^(?:$RE{EUC_0212}|$RE{EUC_C})/o) * 8 > $bpl){ + $line =~ s/[ \t\n\r]*$/\n/; + $result .= $line; + $line = ' '; + } + + while(1){ + my $iso_2022_jp = $target; + Encode::from_to($iso_2022_jp, 'euc-jp', 'iso-2022-jp'); + + my $encoded + = HEAD . MIME::Base64::encode_base64($iso_2022_jp, '') . TAIL; + + if(length($encoded) + length($line) > $bpl){ + $target =~ s/($RE{EUC_0212}|$RE{EUC_KANA}|$RE{EUC_C}|$RE{ASCII})$//o; + $str = $1 . $str; + } + else{ + $line .= $encoded; + last; + } + } + + } + + $result . $line; +} + + +1; +__END__ + diff --git a/ext/Encode/t/mime_header_iso2022jp.t b/ext/Encode/t/mime_header_iso2022jp.t new file mode 100644 index 0000000000..ffc77e07c5 --- /dev/null +++ b/ext/Encode/t/mime_header_iso2022jp.t @@ -0,0 +1,40 @@ + +use Test::More tests => 14; + +use strict; +use Encode; + +BEGIN{ + use_ok('Encode::MIME::Header::ISO_2022_JP'); +} + +require_ok('Encode::MIME::Header::ISO_2022_JP'); + +# below codes are from mime.t in Jcode + +my %mime = ( + "漢字、カタカナ、ひらがな" + => "=?ISO-2022-JP?B?GyRCNEE7eiEiJSslPyUrJUohIiRSJGkkLCRKGyhC?=", + "foo bar" + => "foo bar", + "漢字、カタカナ、ひらがなの混じったSubject Header." + => "=?ISO-2022-JP?B?GyRCNEE7eiEiJSslPyUrJUohIiRSJGkkLCRKJE46LiQ4JEMkPxsoQlN1?=\n =?ISO-2022-JP?B?YmplY3Q=?= Header.", +); + + +for my $k (keys %mime){ + $mime{"$k\n"} = $mime{$k} . "\n"; +} + + +for my $decoded (sort keys %mime){ + my $encoded = $mime{$decoded}; + + my $header = Encode::encode('MIME-Header-ISO_2022_JP', decode('euc-jp', $decoded)); + my $utf8 = Encode::decode('MIME-Header', $header); + + is(encode('euc-jp', $utf8), $decoded); + is($header, $encoded); +} + +__END__ |