2 BEGIN { @INC = '../../lib' };
14 return chr(($uv >> 6) | 0xC0).
15 chr(($uv & 0x3F) | 0x80);
17 return chr(($uv >> 12) | 0xE0).
18 chr((($uv >> 6) & 0x3F) | 0x80).
19 chr(($uv & 0x3F) | 0x80);
31 return chr($page).chr($ch);
37 return &encode_D if $page;
41 @ARGV = map(glob($_),@ARGV) if $^O eq 'MSWin32';
43 my $cname = shift(@ARGV);
44 chmod(0666,$cname) if -f $cname && !-w $cname;
45 open(C,">$cname") || die "Cannot open $cname:$!";
47 $dname =~ s/(\.[^\.]*)?$/.def/;
48 chmod(0666,$dname) if -f $cname && !-w $dname;
49 open(D,">$dname") || die "Cannot open $dname:$!";
51 $hname =~ s/(\.[^\.]*)?$/.h/;
52 chmod(0666,$hname) if -f $cname && !-w $hname;
53 open(H,">$hname") || die "Cannot open $hname:$!";
55 if ($cname =~ /(\w+)\.xs$/)
57 print C "#include <EXTERN.h>\n";
58 print C "#include <perl.h>\n";
59 print C "#include <XSUB.h>\n";
60 print C "#define U8 U8\n";
62 print C "#include \"encode.h\"\n";
69 if ($a =~ /^.*-(\d+)/)
72 if ($b =~ /^.*-(\d+)/)
81 foreach my $enc (sort cmp_name @ARGV)
83 my ($name) = $enc =~ /^.*?([\w-]+)(\.enc)$/;
86 compile(\*E,$name,\*C);
90 warn "Cannot open $enc for $name:$!";
94 foreach my $enc (sort cmp_name keys %encoding)
96 my $sym = "${enc}_encoding";
98 print C "encode_t $sym = \n";
99 print C " {",join(',',"\"$enc\"",@{$encoding{$enc}}),"};\n";
102 foreach my $enc (sort cmp_name keys %encoding)
104 my $sym = "${enc}_encoding";
106 print H "extern encode_t $sym;\n";
107 print D " Encode_Define(aTHX_ &$sym);\n";
110 if ($cname =~ /(\w+)\.xs$/)
113 print C "\nMODULE = Encode::$mod\tPACKAGE = Encode::$mod\n\n";
114 print C "BOOT:\n{\n";
115 print C "#include \"$dname\"\n";
126 my ($fh,$name,$ch) = @_;
131 while ($type = <$fh>)
133 last if $type !~ /^\s*#/;
136 return if $type eq 'E';
137 my ($def,$sym,$pages) = split(/\s+/,scalar(<$fh>));
138 warn "t=$type s=$sym d=$def p=$pages $name\n";
143 $rep = &{"encode_$type"}($v & 0xFF, ($v >> 8) & 0xffe);
149 my $page = hex($line);
151 for (my $i = 0; $i < 16; $i++)
154 for (my $j = 0; $j < 16; $j++)
157 my $ech = &{"encode_$type"}($ch,$page);
158 my $val = hex(substr($line,0,4,''));
159 if ($val || (!$ch && !$page))
161 my $uch = encode_U($val);
162 enter($e2u,$ech,$uch,$e2u);
163 enter($u2e,$uch,$ech,$u2e);
167 # No character at this position
168 # enter($e2u,$ech,undef,$e2u);
174 output($ch,$name.'_utf8',$e2u);
175 output($ch,'utf8_'.$name,$u2e);
176 $encoding{$name} = [$e2u->{Cname},$u2e->{Cname},
177 outstring($ch,$e2u->{Cname}.'_def',$rep),length($rep)];
182 my ($a,$s,$d,$t) = @_;
184 my $b = substr($s,0,1);
188 $e = [$b,$b,'',{},length($s),0];
193 enter($e->[3],substr($s,1),$d,$t);
199 $e->[5] = length($d);
206 my ($fh,$name,$s) = @_;
207 my $sym = $strings{$s};
210 foreach my $o (keys %strings)
212 my $i = index($o,$s);
216 $sym .= sprintf("+0x%02x",$i) if ($i);
220 $strings{$s} = $sym = $name;
221 printf $fh "static const U8 %s[%d] =\n",$name,length($s);
222 # Do in chunks of 16 chars to constrain line length
223 # Assumes ANSI C adjacent string litteral concatenation
226 my $c = substr($s,0,16,'');
227 print $fh '"',join('',map(sprintf('\x%02x',ord($_)),split(//,$c))),'"';
228 print $fh "\n" if length($s);
237 my ($fh,$name,$a) = @_;
240 my @keys = grep(ref($a->{$_}),sort keys %$a);
241 print $fh "\nextern encpage_t $name\[\];\n";
242 # print $fh "\nstatic encpage_t *$name;\n";
247 foreach my $b (@keys)
249 my ($s,$f,$out,$t,$end) = @{$a->{$b}};
251 ord($b) == ord($a->{$l}[1])+1 &&
252 $a->{$l}[3] == $a->{$b}[3] &&
253 $a->{$l}[4] == $a->{$b}[4] &&
254 $a->{$l}[5] == $a->{$b}[5]
255 # && length($a->{$l}[2]) < 16
258 my $i = ord($b)-ord($a->{$l}[0]);
260 $a->{$l}[2] .= $a->{$b}[2];
267 unless (exists $t->{Cname})
269 output($fh,sprintf("%s_%02x",$name,ord($s)),$t);
272 if (ord($keys[-1]) < 255)
274 my $t = chr(ord($keys[-1])+1);
275 $a->{$t} = [$t,chr(255),undef,$a,0,0];
281 next unless $a->{$b}[5];
282 my $s = ord($a->{$b}[0]);
283 my $e = ord($a->{$b}[1]);
284 outstring($fh,sprintf("%s__%02x_%02x",$name,$s,$e),$a->{$b}[2]);
288 print $fh "encpage_t $name\[",scalar(@ent),"] = {\n";
291 my ($s,$e,$out,$t,$end,$l) = @{$a->{$b}};
297 printf $fh outstring($fh,'',$out);
303 print $fh ",",$t->{Cname};
304 printf $fh ",0x%02x,0x%02x,$l,$end},\n",$sc,$ec;