2 BEGIN { @INC = '../../lib' };
9 # UTF-8 encode long hand - only covers part of perl's range
17 return chr(($uv >> 6) | 0xC0).
18 chr(($uv & 0x3F) | 0x80);
20 return chr(($uv >> 12) | 0xE0).
21 chr((($uv >> 6) & 0x3F) | 0x80).
22 chr(($uv & 0x3F) | 0x80);
34 # encode double byte MS byte first
36 return chr($page).chr($ch);
41 # encode Multi-byte - single for 0..255 otherwise double
43 return &encode_D if $page;
47 # Win32 does not expand globs on command line
48 eval "\@ARGV = map(glob(\$_),\@ARGV)" if ($^O eq 'MSWin32');
51 getopts('qo:f:n:',\%opt);
52 my $cname = (exists $opt{'o'}) ? $opt{'o'} : shift(@ARGV);
53 chmod(0666,$cname) if -f $cname && !-w $cname;
54 open(C,">$cname") || die "Cannot open $cname:$!";
58 $dname =~ s/(\.[^\.]*)?$/.def/;
60 my ($doC,$doEnc,$doUcm);
62 if ($cname =~ /\.(c|xs)$/)
65 chmod(0666,$dname) if -f $cname && !-w $dname;
66 open(D,">$dname") || die "Cannot open $dname:$!";
68 $hname =~ s/(\.[^\.]*)?$/.h/;
69 chmod(0666,$hname) if -f $cname && !-w $hname;
70 open(H,">$hname") || die "Cannot open $hname:$!";
72 foreach my $fh (\*C,\*D,\*H)
74 print $fh <<"END" unless $opt{'q'};
76 !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
77 This file was autogenerated by:
78 $^X $0 $cname @orig_ARGV
83 if ($cname =~ /(\w+)\.xs$/)
85 print C "#include <EXTERN.h>\n";
86 print C "#include <perl.h>\n";
87 print C "#include <XSUB.h>\n";
88 print C "#define U8 U8\n";
90 print C "#include \"encode.h\"\n";
92 elsif ($cname =~ /\.enc$/)
96 elsif ($cname =~ /\.ucm$/)
102 if (exists $opt{'f'})
104 # -F is followed by name of file containing list of filenames
105 my $flist = $opt{'f'};
106 open(FLIST,$flist) || die "Cannot open $flist:$!";
107 chomp(@encfiles = <FLIST>);
120 if ($a =~ /^.*-(\d+)/)
123 if ($b =~ /^.*-(\d+)/)
132 foreach my $enc (sort cmp_name @encfiles)
134 my ($name,$sfx) = $enc =~ /^.*?([\w-]+)\.(enc|ucm)$/;
135 $name = delete $opt{'n'} if exists $opt{'n'};
140 compile_enc(\*E,lc($name),\*C);
144 compile_ucm(\*E,lc($name),\*C);
149 warn "Cannot open $enc for $name:$!";
155 foreach my $enc (sort cmp_name keys %encoding)
157 my $sym = "${enc}_encoding";
159 print C "encode_t $sym = \n";
160 print C " {",join(',',"\"$enc\"",@{$encoding{$enc}}),"};\n\n";
163 foreach my $enc (sort cmp_name keys %encoding)
165 my $sym = "${enc}_encoding";
167 print H "extern encode_t $sym;\n";
168 print D " Encode_Define(aTHX_ &$sym);\n";
171 if ($cname =~ /(\w+)\.xs$/)
174 print C "\nMODULE = Encode::$mod\tPACKAGE = Encode::$mod\n\n";
175 print C "BOOT:\n{\n";
176 print C "#include \"$dname\"\n";
187 my ($fh,$name,$ch) = @_;
195 last if /^\s*CHARMAP\s*$/i;
196 if (/^\s*<(\w+)>\s+"?([^"]*)"?\s*$/i)
201 if (!defined($cs = $attr{'code_set_name'}))
203 warn "No <code_set_name> in $name\n";
213 if (exists $attr{'subchar'})
216 $attr{'subchar'} =~ /^\s*/cg;
217 push(@byte,$1) while $attr{'subchar'} =~ /\G\\x([0-9a-f]+)/icg;
218 $erep = join('',map(chr(hex($_)),@byte));
220 warn "Scanning $name ($cs)\n";
226 last if /^\s*END\s+CHARMAP\s*$/i;
230 $u = $1 if (/^<U([0-9a-f]+)>\s+/igc);
231 push(@byte,$1) while /\G\\x([0-9a-f]+)/igc;
232 $fb = $1 if /\G\s*(\|[0-3])/gc;
233 # warn "$_: $u @byte | $fb\n";
234 die "Bad line:$_" unless /\G\s*(#.*)?$/gc;
237 my $uch = encode_U(hex($u));
238 my $ech = join('',map(chr(hex($_)),@byte));
239 my $el = length($ech);
240 $max_el = $el if (!defined($max_el) || $el > $max_el);
241 $min_el = $el if (!defined($min_el) || $el < $min_el);
252 # $fb is fallback flag
253 # 0 - round trip safe
254 # 1 - fallback for unicode -> enc
255 # 2 - skip sub-char mapping
256 # 3 - fallback enc -> unicode
257 enter($u2e,$uch,$ech,$u2e,$fb+0) if ($fb =~ /[01]/);
258 enter($e2u,$ech,$uch,$e2u,$fb+0) if ($fb =~ /[03]/);
267 die "$nfb entries without fallback, $hfb entries with\n";
271 output($ch,$name.'_utf8',$e2u);
272 output($ch,'utf8_'.$name,$u2e);
273 $encoding{$name} = [$e2u->{Cname},$u2e->{Cname},
274 outstring($ch,$e2u->{Cname}.'_def',$erep),length($erep)];
278 output_enc($ch,$name,$e2u);
282 output_ucm($ch,$name,$u2e,$erep,$min_el,$max_el);
288 my ($fh,$name,$ch) = @_;
293 while ($type = <$fh>)
295 last if $type !~ /^\s*#/;
298 return if $type eq 'E';
299 my ($def,$sym,$pages) = split(/\s+/,scalar(<$fh>));
300 warn "$type encoded $name\n";
307 $rep = &{"encode_$type"}($v & 0xFF, ($v >> 8) & 0xffe);
314 my $page = hex($line);
316 for (my $i = 0; $i < 16; $i++)
319 for (my $j = 0; $j < 16; $j++)
322 my $ech = &{"encode_$type"}($ch,$page);
323 my $val = hex(substr($line,0,4,''));
324 next if $val == 0xFFFD;
325 if ($val || (!$ch && !$page))
327 my $el = length($ech);
328 $max_el = $el if (!defined($max_el) || $el > $max_el);
329 $min_el = $el if (!defined($min_el) || $el < $min_el);
330 my $uch = encode_U($val);
331 if (exists $seen{$uch})
333 warn sprintf("U%04X is %02X%02X and %02X%02X\n",
334 $val,$page,$ch,@{$seen{$uch}});
338 $seen{$uch} = [$page,$ch];
340 enter($e2u,$ech,$uch,$e2u,0);
341 enter($u2e,$uch,$ech,$u2e,0);
345 # No character at this position
346 # enter($e2u,$ech,undef,$e2u);
354 output($ch,$name.'_utf8',$e2u);
355 output($ch,'utf8_'.$name,$u2e);
356 $encoding{$name} = [$e2u->{Cname},$u2e->{Cname},
357 outstring($ch,$e2u->{Cname}.'_def',$rep),length($rep)];
361 output_enc($ch,$name,$e2u);
365 output_ucm($ch,$name,$u2e,$rep,$min_el,$max_el);
371 my ($a,$s,$d,$t,$fb) = @_;
373 my $b = substr($s,0,1);
377 $e = [$b,$b,'',{},length($s),0,$fb];
382 enter($e->[3],substr($s,1),$d,$t,$fb);
388 $e->[5] = length($d);
394 my ($fh,$name,$s) = @_;
395 my $sym = $strings{$s};
398 foreach my $o (keys %strings)
400 my $i = index($o,$s);
404 $sym .= sprintf("+0x%02x",$i) if ($i);
408 $strings{$s} = $sym = $name;
409 printf $fh "\nstatic const U8 %s[%d] =\n",$name,length($s);
410 # Do in chunks of 16 chars to constrain line length
411 # Assumes ANSI C adjacent string litteral concatenation
414 my $c = substr($s,0,16,'');
415 print $fh '"',join('',map(sprintf('\x%02x',ord($_)),split(//,$c))),'"';
416 print $fh "\n" if length($s);
428 my @keys = grep(ref($a->{$_}),sort keys %$a);
431 foreach my $b (@keys)
433 my ($s,$f,$out,$t,$end) = @{$a->{$b}};
435 ord($b) == ord($a->{$l}[1])+1 &&
436 $a->{$l}[3] == $a->{$b}[3] &&
437 $a->{$l}[4] == $a->{$b}[4] &&
438 $a->{$l}[5] == $a->{$b}[5] &&
439 $a->{$l}[6] == $a->{$b}[6]
440 # && length($a->{$l}[2]) < 16
443 my $i = ord($b)-ord($a->{$l}[0]);
445 $a->{$l}[2] .= $a->{$b}[2];
452 if (exists $t->{Cname})
454 $t->{'Forward'} = 1 if $t != $a;
458 process(sprintf("%s_%02x",$name,ord($s)),$t);
461 if (ord($keys[-1]) < 255)
463 my $t = chr(ord($keys[-1])+1);
464 $a->{$t} = [$t,chr(255),undef,$a,0,0];
467 $a->{'Entries'} = \@ent;
473 my $name = $a->{'Cname'};
475 foreach my $b (@{$a->{'Entries'}})
477 next unless $a->{$b}[5];
478 my $s = ord($a->{$b}[0]);
479 my $e = ord($a->{$b}[1]);
480 outstring($fh,sprintf("%s__%02x_%02x",$name,$s,$e),$a->{$b}[2]);
484 print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"];\n";
487 foreach my $b (@{$a->{'Entries'}})
489 my ($s,$e,$out,$t,$end,$l) = @{$a->{$b}};
490 outtable($fh,$t) unless $t->{'Done'};
492 print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"] = {\n";
493 foreach my $b (@{$a->{'Entries'}})
495 my ($s,$e,$out,$t,$end,$l,$fb) = @{$a->{$b}};
502 printf $fh outstring($fh,'',$out);
508 print $fh ",",$t->{Cname};
509 printf $fh ",0x%02x,0x%02x,$l,$end},\n",$sc,$ec;
516 my ($fh,$name,$a) = @_;
524 my ($fh,$name,$a) = @_;
525 foreach my $b (sort keys %$a)
527 my ($s,$e,$out,$t,$end,$l,$fb) = @{$a->{$b}};
538 my ($fh,$a,$t,$pre) = @_;
539 # warn sprintf("Page %x\n",$pre);
540 foreach my $b (sort keys %$t)
542 my ($s,$e,$out,$n,$end,$l,$fb) = @{$t->{$b}};
543 die "oops $s $e" unless $s eq $e;
545 if ($n != $a && $n != $t)
547 output_ucm_page($fh,$a,$n,(($pre|($u &0x3F)) << 6)&0xFFFF);
553 $u = $pre|($u &0x3f);
555 printf $fh "<U%04X> ",$u;
556 foreach my $c (split(//,$out))
558 printf $fh "\\x%02X",ord($c);
560 printf $fh " |%d\n",($fb ? 1 : 0);
564 warn join(',',@{$t->{$b}},$a,$t);
571 my ($fh,$name,$a,$rep,$min_el,$max_el) = @_;
572 print $fh "# Written by $0 @orig_ARGV\n" unless $opt{'q'};
573 print $fh "<code_set_name> \"$name\"\n";
576 print $fh "<mb_cur_min> $min_el\n";
580 print $fh "<mb_cur_max> $max_el\n";
584 print $fh "<subchar> ";
585 foreach my $c (split(//,$rep))
587 printf $fh "\\x%02X",ord($c);
591 print $fh "#\nCHARMAP\n";
592 output_ucm_page($fh,$a,$a,0);
593 print $fh "END CHARMAP\n";