2 BEGIN { @INC = '../../lib' };
7 # UTF-8 encocde long hand - only covers part of perl's range
15 return chr(($uv >> 6) | 0xC0).
16 chr(($uv & 0x3F) | 0x80);
18 return chr(($uv >> 12) | 0xE0).
19 chr((($uv >> 6) & 0x3F) | 0x80).
20 chr(($uv & 0x3F) | 0x80);
32 # encode double byte MS byte first
34 return chr($page).chr($ch);
39 # encode Multi-byte - single for 0..255 otherwise double
41 return &encode_D if $page;
45 # Win32 does not expand globs on command line
46 eval "\@ARGV = map(glob(\$_),\@ARGV)" if ($^O eq 'MSWin32');
48 my $cname = shift(@ARGV);
49 chmod(0666,$cname) if -f $cname && !-w $cname;
50 open(C,">$cname") || die "Cannot open $cname:$!";
52 $dname =~ s/(\.[^\.]*)?$/.def/;
53 chmod(0666,$dname) if -f $cname && !-w $dname;
54 open(D,">$dname") || die "Cannot open $dname:$!";
56 $hname =~ s/(\.[^\.]*)?$/.h/;
57 chmod(0666,$hname) if -f $cname && !-w $hname;
58 open(H,">$hname") || die "Cannot open $hname:$!";
60 foreach my $fh (\*C,\*D,\*H)
64 !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
65 This file was autogenerated by:
71 if ($cname =~ /(\w+)\.xs$/)
73 print C "#include <EXTERN.h>\n";
74 print C "#include <perl.h>\n";
75 print C "#include <XSUB.h>\n";
76 print C "#define U8 U8\n";
78 print C "#include \"encode.h\"\n";
85 if ($a =~ /^.*-(\d+)/)
88 if ($b =~ /^.*-(\d+)/)
97 foreach my $enc (sort cmp_name @ARGV)
99 my ($name) = $enc =~ /^.*?([\w-]+)(\.enc)$/;
102 compile_enc(\*E,lc($name),\*C);
106 warn "Cannot open $enc for $name:$!";
110 foreach my $enc (sort cmp_name keys %encoding)
112 my $sym = "${enc}_encoding";
114 print C "encode_t $sym = \n";
115 print C " {",join(',',"\"$enc\"",@{$encoding{$enc}}),"};\n\n";
118 foreach my $enc (sort cmp_name keys %encoding)
120 my $sym = "${enc}_encoding";
122 print H "extern encode_t $sym;\n";
123 print D " Encode_Define(aTHX_ &$sym);\n";
126 if ($cname =~ /(\w+)\.xs$/)
129 print C "\nMODULE = Encode::$mod\tPACKAGE = Encode::$mod\n\n";
130 print C "BOOT:\n{\n";
131 print C "#include \"$dname\"\n";
140 my ($fh,$name,$ch) = @_;
145 while ($type = <$fh>)
147 last if $type !~ /^\s*#/;
150 return if $type eq 'E';
151 my ($def,$sym,$pages) = split(/\s+/,scalar(<$fh>));
152 warn "$type encoded $name\n";
157 $rep = &{"encode_$type"}($v & 0xFF, ($v >> 8) & 0xffe);
163 my $page = hex($line);
165 for (my $i = 0; $i < 16; $i++)
168 for (my $j = 0; $j < 16; $j++)
171 my $ech = &{"encode_$type"}($ch,$page);
172 my $val = hex(substr($line,0,4,''));
173 if ($val || (!$ch && !$page))
175 my $uch = encode_U($val);
176 enter($e2u,$ech,$uch,$e2u);
177 enter($u2e,$uch,$ech,$u2e);
181 # No character at this position
182 # enter($e2u,$ech,undef,$e2u);
188 output($ch,$name.'_utf8',$e2u);
189 output($ch,'utf8_'.$name,$u2e);
190 $encoding{$name} = [$e2u->{Cname},$u2e->{Cname},
191 outstring($ch,$e2u->{Cname}.'_def',$rep),length($rep)];
196 my ($a,$s,$d,$t) = @_;
198 my $b = substr($s,0,1);
202 $e = [$b,$b,'',{},length($s),0];
207 enter($e->[3],substr($s,1),$d,$t);
213 $e->[5] = length($d);
219 my ($fh,$name,$s) = @_;
220 my $sym = $strings{$s};
223 foreach my $o (keys %strings)
225 my $i = index($o,$s);
229 $sym .= sprintf("+0x%02x",$i) if ($i);
233 $strings{$s} = $sym = $name;
234 printf $fh "\nstatic const U8 %s[%d] =\n",$name,length($s);
235 # Do in chunks of 16 chars to constrain line length
236 # Assumes ANSI C adjacent string litteral concatenation
239 my $c = substr($s,0,16,'');
240 print $fh '"',join('',map(sprintf('\x%02x',ord($_)),split(//,$c))),'"';
241 print $fh "\n" if length($s);
253 my @keys = grep(ref($a->{$_}),sort keys %$a);
256 foreach my $b (@keys)
258 my ($s,$f,$out,$t,$end) = @{$a->{$b}};
260 ord($b) == ord($a->{$l}[1])+1 &&
261 $a->{$l}[3] == $a->{$b}[3] &&
262 $a->{$l}[4] == $a->{$b}[4] &&
263 $a->{$l}[5] == $a->{$b}[5]
264 # && length($a->{$l}[2]) < 16
267 my $i = ord($b)-ord($a->{$l}[0]);
269 $a->{$l}[2] .= $a->{$b}[2];
276 if (exists $t->{Cname})
278 $t->{'Forward'} = 1 if $t != $a;
282 process(sprintf("%s_%02x",$name,ord($s)),$t);
285 if (ord($keys[-1]) < 255)
287 my $t = chr(ord($keys[-1])+1);
288 $a->{$t} = [$t,chr(255),undef,$a,0,0];
291 $a->{'Entries'} = \@ent;
297 my $name = $a->{'Cname'};
299 foreach my $b (@{$a->{'Entries'}})
301 next unless $a->{$b}[5];
302 my $s = ord($a->{$b}[0]);
303 my $e = ord($a->{$b}[1]);
304 outstring($fh,sprintf("%s__%02x_%02x",$name,$s,$e),$a->{$b}[2]);
308 print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"];\n";
311 foreach my $b (@{$a->{'Entries'}})
313 my ($s,$e,$out,$t,$end,$l) = @{$a->{$b}};
314 outtable($fh,$t) unless $t->{'Done'};
316 print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"] = {\n";
317 foreach my $b (@{$a->{'Entries'}})
319 my ($s,$e,$out,$t,$end,$l) = @{$a->{$b}};
325 printf $fh outstring($fh,'',$out);
331 print $fh ",",$t->{Cname};
332 printf $fh ",0x%02x,0x%02x,$l,$end},\n",$sc,$ec;
339 my ($fh,$name,$a) = @_;