package Encode::CN::HZ;
use strict;
-no warnings 'redefine'; # to quell the "use Encode" below
-use Encode::CN;
-use Encode qw|encode decode|;
-use base 'Encode::Encoding';
+use vars qw($VERSION);
+$VERSION = do { my @r = (q$Revision: 2.0 $ =~ /\d+/g); sprintf "%d."."%02d" x $#r, @r };
-# HZ is but escaped GB, so we implement it with the
-# GB2312(raw) encoding here. Cf. RFC 1842 & 1843.
+use Encode qw(:fallbacks);
-my $canon = 'hz';
-my $obj = bless {name => $canon}, __PACKAGE__;
-$obj->Define($canon);
+use base qw(Encode::Encoding);
+__PACKAGE__->Define('hz');
-sub decode
+# HZ is a combination of ASCII and escaped GB, so we implement it
+# with the GB2312(raw) encoding here. Cf. RFCs 1842 & 1843.
+
+# not ported for EBCDIC. Which should be used, "~" or "\x7E"?
+
+sub needs_lines { 1 }
+
+sub decode ($$;$)
{
my ($obj,$str,$chk) = @_;
- my $gb = Encode::find_encoding('gb2312');
-
- $str =~ s{~ # starting tilde
- (?:
- (~) # another tilde - escaped (set $1)
- | # or
- \n # \n - output nothing
- | # or
- \{ # opening brace of GB data
- ( # set $2 to any number of...
- (?:
- [^~] # non-tilde GB character
- | # or
- ~(?!\}) # tilde not followed by a closing brace
- )*
- )
- ~\} # closing brace of GB data
- | # XXX: invalid escape - maybe die on $chk?
- )
- }{
- (defined $1) ? '~' # two tildes make one tilde
- :
- (defined $2) ? $gb->decode($2, $chk) # decode the characters
- :
- '' # '' on ~\n and invalid escape
- }egx;
-
- return $str;
+
+ my $GB = Encode::find_encoding('gb2312-raw');
+ my $ret = '';
+ my $in_ascii = 1; # default mode is ASCII.
+
+ while (length $str) {
+ if ($in_ascii) { # ASCII mode
+ if ($str =~ s/^([\x00-\x7D\x7F]+)//) { # no '~' => ASCII
+ $ret .= $1;
+ # EBCDIC should need ascii2native, but not ported.
+ }
+ elsif ($str =~ s/^\x7E\x7E//) { # escaped tilde
+ $ret .= '~';
+ }
+ elsif ($str =~ s/^\x7E\cJ//) { # '\cJ' == LF in ASCII
+ 1; # no-op
+ }
+ elsif ($str =~ s/^\x7E\x7B//) { # '~{'
+ $in_ascii = 0; # to GB
+ }
+ else { # encounters an invalid escape, \x80 or greater
+ last;
+ }
+ }
+ else { # GB mode; the byte ranges are as in RFC 1843.
+ if ($str =~ s/^((?:[\x21-\x77][\x21-\x7E])+)//) {
+ $ret .= $GB->decode($1, $chk);
+ }
+ elsif ($str =~ s/^\x7E\x7D//) { # '~}'
+ $in_ascii = 1;
+ }
+ else { # invalid
+ last;
+ }
+ }
+ }
+ $_[1] = '' if $chk; # needs_lines guarantees no partial character
+ return $ret;
}
-sub encode
-{
- my ($obj,$str,$chk) = @_;
- my ($out, $in_gb);
- my $gb = Encode::find_encoding('gb2312');
+sub cat_decode {
+ my ($obj, undef, $src, $pos, $trm, $chk) = @_;
+ my ($rdst, $rsrc, $rpos) = \@_[1..3];
- $str =~ s/~/~~/g;
+ my $GB = Encode::find_encoding('gb2312-raw');
+ my $ret = '';
+ my $in_ascii = 1; # default mode is ASCII.
- # XXX: Since CHECK and partial decoding has not been implemented yet,
- # we'll use a very crude way to test for GB2312ness.
+ my $ini_pos = pos($$rsrc);
- for my $index (0 .. length($str) - 1) {
- no warnings 'utf8';
+ substr($src, 0, $pos) = '';
- my $char = substr($str, $index, 1);
- my $try = $gb->encode($char); # try encode this char
+ my $ini_len = bytes::length($src);
- if (defined($try)) { # is a GB character
- if ($in_gb) {
- $out .= $try; # in GB mode - just append it
+ # $trm is the first of the pair '~~', then 2nd tilde is to be removed.
+ # XXX: Is better C<$src =~ s/^\x7E// or die if ...>?
+ $src =~ s/^\x7E// if $trm eq "\x7E";
+
+ while (length $src) {
+ my $now;
+ if ($in_ascii) { # ASCII mode
+ if ($src =~ s/^([\x00-\x7D\x7F])//) { # no '~' => ASCII
+ $now = $1;
+ }
+ elsif ($src =~ s/^\x7E\x7E//) { # escaped tilde
+ $now = '~';
+ }
+ elsif ($src =~ s/^\x7E\cJ//) { # '\cJ' == LF in ASCII
+ next;
+ }
+ elsif ($src =~ s/^\x7E\x7B//) { # '~{'
+ $in_ascii = 0; # to GB
+ next;
+ }
+ else { # encounters an invalid escape, \x80 or greater
+ last;
+ }
+ }
+ else { # GB mode; the byte ranges are as in RFC 1843.
+ if ($src =~ s/^((?:[\x21-\x77][\x21-\x7F])+)//) {
+ $now = $GB->decode($1, $chk);
+ }
+ elsif ($src =~ s/^\x7E\x7D//) { # '~}'
+ $in_ascii = 1;
+ next;
+ }
+ else { # invalid
+ last;
}
- else {
- $out .= "~{$try"; # enter GB mode, then append it
- $in_gb = 1;
+ }
+
+ next if ! defined $now;
+
+ $ret .= $now;
+
+ if ($now eq $trm) {
+ $$rdst .= $ret;
+ $$rpos = $ini_pos + $pos + $ini_len - bytes::length($src);
+ pos($$rsrc) = $ini_pos;
+ return 1;
+ }
+ }
+
+ $$rdst .= $ret;
+ $$rpos = $ini_pos + $pos + $ini_len - bytes::length($src);
+ pos($$rsrc) = $ini_pos;
+ return ''; # terminator not found
+}
+
+
+sub encode($$;$)
+{
+ my ($obj,$str,$chk) = @_;
+
+ my $GB = Encode::find_encoding('gb2312-raw');
+ my $ret = '';
+ my $in_ascii = 1; # default mode is ASCII.
+
+ no warnings 'utf8'; # $str may be malformed UTF8 at the end of a chunk.
+
+ while (length $str) {
+ if ($str =~ s/^([[:ascii:]]+)//) {
+ my $tmp = $1;
+ $tmp =~ s/~/~~/g; # escapes tildes
+ if (! $in_ascii) {
+ $ret .= "\x7E\x7D"; # '~}'
+ $in_ascii = 1;
}
+ $ret .= pack 'a*', $tmp; # remove UTF8 flag.
}
- elsif ($in_gb) {
- $out .= "~}$char"; # leave GB mode, then append it
- $in_gb = 0;
+ elsif ($str =~ s/(.)//) {
+ my $tmp = $GB->encode($1, $chk);
+ last if !defined $tmp;
+ if (length $tmp == 2) { # maybe a valid GB char (XXX)
+ if ($in_ascii) {
+ $ret .= "\x7E\x7B"; # '~{'
+ $in_ascii = 0;
+ }
+ $ret .= $tmp;
+ }
+ elsif (length $tmp) { # maybe FALLBACK in ASCII (XXX)
+ if (!$in_ascii) {
+ $ret .= "\x7E\x7D"; # '~}'
+ $in_ascii = 1;
+ }
+ $ret .= $tmp;
+ }
}
- else {
- $out .= $char; # not in GB mode - just append it
+ else { # if $str is malformed UTF8 *and* if length $str != 0.
+ last;
}
}
+ $_[1] = $str if $chk;
- $out .= '~}' if $in_gb; # add closing brace as needed
+ # The state at the end of the chunk is discarded, even if in GB mode.
+ # That results in the combination of GB-OUT and GB-IN, i.e. "~}~{".
+ # Parhaps it is harmless, but further investigations may be required...
- return $out;
+ if (! $in_ascii) {
+ $ret .= "\x7E\x7D"; # '~}'
+ $in_ascii = 1;
+ }
+ return $ret;
}
1;
__END__
+
+=head1 NAME
+
+Encode::CN::HZ -- internally used by Encode::CN
+
+=cut
projects / p5sagit/p5-mst-13.2.git / blobdiff