Commit | Line | Data |
017e2add |
1 | #!../../perl -w |
508a30f1 |
2 | BEGIN { |
b0bf7e4c |
3 | unshift @INC, qw(../../lib ../../../lib); |
4 | $ENV{PATH} .= ';../..;../../..' if $^O eq 'MSWin32'; |
508a30f1 |
5 | } |
017e2add |
6 | use strict; |
afdae191 |
7 | use Getopt::Std; |
8 | my @orig_ARGV = @ARGV; |
dcda1f94 |
9 | |
017e2add |
10 | sub encode_U |
11 | { |
9b37254d |
12 | # UTF-8 encode long hand - only covers part of perl's range |
017e2add |
13 | my $uv = shift; |
1b026014 |
14 | # chr() works in native space so convert value from table |
15 | # into that space before using chr(). |
16 | my $ch = chr(utf8::unicode_to_native($uv)); |
17 | # Now get core perl to encode that the way it likes. |
18 | utf8::encode($ch); |
19 | return $ch; |
017e2add |
20 | } |
21 | |
22 | sub encode_S |
23 | { |
14a8264b |
24 | # encode single byte |
017e2add |
25 | my ($ch,$page) = @_; |
26 | return chr($ch); |
27 | } |
28 | |
29 | sub encode_D |
30 | { |
14a8264b |
31 | # encode double byte MS byte first |
017e2add |
32 | my ($ch,$page) = @_; |
33 | return chr($page).chr($ch); |
34 | } |
35 | |
36 | sub encode_M |
37 | { |
14a8264b |
38 | # encode Multi-byte - single for 0..255 otherwise double |
017e2add |
39 | my ($ch,$page) = @_; |
40 | return &encode_D if $page; |
41 | return &encode_S; |
42 | } |
43 | |
14a8264b |
44 | # Win32 does not expand globs on command line |
252a8565 |
45 | eval "\@ARGV = map(glob(\$_),\@ARGV)" if ($^O eq 'MSWin32'); |
18b7339f |
46 | |
afdae191 |
47 | my %opt; |
4cfc977c |
48 | getopts('qOo:f:n:',\%opt); |
afdae191 |
49 | my $cname = (exists $opt{'o'}) ? $opt{'o'} : shift(@ARGV); |
2f2b4ff2 |
50 | chmod(0666,$cname) if -f $cname && !-w $cname; |
017e2add |
51 | open(C,">$cname") || die "Cannot open $cname:$!"; |
afdae191 |
52 | |
53 | |
2f2b4ff2 |
54 | my $dname = $cname; |
023d8852 |
55 | $dname =~ s/(\.[^\.]*)?$/_def.h/; |
e0c49a6b |
56 | |
dcda1f94 |
57 | my ($doC,$doEnc,$doUcm,$doPet); |
e0c49a6b |
58 | |
59 | if ($cname =~ /\.(c|xs)$/) |
60 | { |
61 | $doC = 1; |
62 | chmod(0666,$dname) if -f $cname && !-w $dname; |
63 | open(D,">$dname") || die "Cannot open $dname:$!"; |
64 | my $hname = $cname; |
65 | $hname =~ s/(\.[^\.]*)?$/.h/; |
66 | chmod(0666,$hname) if -f $cname && !-w $hname; |
67 | open(H,">$hname") || die "Cannot open $hname:$!"; |
68 | |
69 | foreach my $fh (\*C,\*D,\*H) |
70 | { |
afdae191 |
71 | print $fh <<"END" unless $opt{'q'}; |
14a8264b |
72 | /* |
73 | !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
74 | This file was autogenerated by: |
023d8852 |
75 | $^X $0 @orig_ARGV |
14a8264b |
76 | */ |
77 | END |
e0c49a6b |
78 | } |
14a8264b |
79 | |
e0c49a6b |
80 | if ($cname =~ /(\w+)\.xs$/) |
81 | { |
82 | print C "#include <EXTERN.h>\n"; |
83 | print C "#include <perl.h>\n"; |
84 | print C "#include <XSUB.h>\n"; |
85 | print C "#define U8 U8\n"; |
86 | } |
87 | print C "#include \"encode.h\"\n"; |
d811239c |
88 | |
e0c49a6b |
89 | } |
90 | elsif ($cname =~ /\.enc$/) |
91 | { |
92 | $doEnc = 1; |
93 | } |
94 | elsif ($cname =~ /\.ucm$/) |
2f2b4ff2 |
95 | { |
e0c49a6b |
96 | $doUcm = 1; |
2f2b4ff2 |
97 | } |
dcda1f94 |
98 | elsif ($cname =~ /\.pet$/) |
99 | { |
100 | $doPet = 1; |
101 | } |
017e2add |
102 | |
afdae191 |
103 | my @encfiles; |
3865e8e1 |
104 | if (exists $opt{'f'}) |
afdae191 |
105 | { |
106 | # -F is followed by name of file containing list of filenames |
3865e8e1 |
107 | my $flist = $opt{'f'}; |
afdae191 |
108 | open(FLIST,$flist) || die "Cannot open $flist:$!"; |
109 | chomp(@encfiles = <FLIST>); |
110 | close(FLIST); |
111 | } |
112 | else |
113 | { |
114 | @encfiles = @ARGV; |
115 | } |
c6fdb90a |
116 | |
017e2add |
117 | my %encoding; |
118 | my %strings; |
b1e7e56f |
119 | my $saved = 0; |
120 | my $subsave = 0; |
121 | my $strings = 0; |
017e2add |
122 | |
2f2b4ff2 |
123 | sub cmp_name |
124 | { |
125 | if ($a =~ /^.*-(\d+)/) |
126 | { |
127 | my $an = $1; |
128 | if ($b =~ /^.*-(\d+)/) |
129 | { |
130 | my $r = $an <=> $1; |
131 | return $r if $r; |
132 | } |
133 | } |
134 | return $a cmp $b; |
135 | } |
136 | |
dcda1f94 |
137 | |
c6fdb90a |
138 | foreach my $enc (sort cmp_name @encfiles) |
017e2add |
139 | { |
9b37254d |
140 | my ($name,$sfx) = $enc =~ /^.*?([\w-]+)\.(enc|ucm)$/; |
51ef4e11 |
141 | $name = $opt{'n'} if exists $opt{'n'}; |
017e2add |
142 | if (open(E,$enc)) |
143 | { |
9b37254d |
144 | if ($sfx eq 'enc') |
145 | { |
dcda1f94 |
146 | compile_enc(\*E,lc($name)); |
9b37254d |
147 | } |
148 | else |
149 | { |
dcda1f94 |
150 | compile_ucm(\*E,lc($name)); |
9b37254d |
151 | } |
017e2add |
152 | } |
153 | else |
154 | { |
155 | warn "Cannot open $enc for $name:$!"; |
156 | } |
157 | } |
158 | |
e0c49a6b |
159 | if ($doC) |
2f2b4ff2 |
160 | { |
b1e7e56f |
161 | print STDERR "Writing compiled form\n"; |
dcda1f94 |
162 | foreach my $name (sort cmp_name keys %encoding) |
163 | { |
164 | my ($e2u,$u2e,$erep,$min_el,$max_el) = @{$encoding{$name}}; |
165 | output(\*C,$name.'_utf8',$e2u); |
166 | output(\*C,'utf8_'.$name,$u2e); |
167 | push(@{$encoding{$name}},outstring(\*C,$e2u->{Cname}.'_def',$erep)); |
168 | } |
e0c49a6b |
169 | foreach my $enc (sort cmp_name keys %encoding) |
170 | { |
dcda1f94 |
171 | my ($e2u,$u2e,$rep,$min_el,$max_el,$rsym) = @{$encoding{$enc}}; |
172 | my @info = ($e2u->{Cname},$u2e->{Cname},$rsym,length($rep),$min_el,$max_el); |
e0c49a6b |
173 | my $sym = "${enc}_encoding"; |
174 | $sym =~ s/\W+/_/g; |
175 | print C "encode_t $sym = \n"; |
dcda1f94 |
176 | print C " {",join(',',@info,"{\"$enc\",(const char *)0}"),"};\n\n"; |
e0c49a6b |
177 | } |
2f2b4ff2 |
178 | |
e0c49a6b |
179 | foreach my $enc (sort cmp_name keys %encoding) |
180 | { |
181 | my $sym = "${enc}_encoding"; |
182 | $sym =~ s/\W+/_/g; |
183 | print H "extern encode_t $sym;\n"; |
d811239c |
184 | print D " Encode_XSEncoding(aTHX_ &$sym);\n"; |
e0c49a6b |
185 | } |
017e2add |
186 | |
e0c49a6b |
187 | if ($cname =~ /(\w+)\.xs$/) |
188 | { |
189 | my $mod = $1; |
d811239c |
190 | print C <<'END'; |
191 | |
023d8852 |
192 | static void |
d811239c |
193 | Encode_XSEncoding(pTHX_ encode_t *enc) |
194 | { |
195 | dSP; |
196 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
197 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))),stash); |
198 | int i = 0; |
199 | PUSHMARK(sp); |
200 | XPUSHs(sv); |
201 | while (enc->name[i]) |
202 | { |
203 | const char *name = enc->name[i++]; |
204 | XPUSHs(sv_2mortal(newSVpvn(name,strlen(name)))); |
205 | } |
206 | PUTBACK; |
207 | call_pv("Encode::define_encoding",G_DISCARD); |
208 | SvREFCNT_dec(sv); |
209 | } |
210 | |
211 | END |
212 | |
e0c49a6b |
213 | print C "\nMODULE = Encode::$mod\tPACKAGE = Encode::$mod\n\n"; |
214 | print C "BOOT:\n{\n"; |
215 | print C "#include \"$dname\"\n"; |
216 | print C "}\n"; |
217 | } |
218 | close(D); |
219 | close(H); |
b1e7e56f |
220 | printf STDERR "%d bytes in string tables\n",$strings; |
221 | printf STDERR "%d bytes (%.3g%%) saved spotting duplicates\n",$saved,100*$saved/$strings if $saved; |
222 | printf STDERR "%d bytes (%.3g%%) saved using substrings\n",$subsave,100*$subsave/$strings if $subsave; |
2f2b4ff2 |
223 | } |
dcda1f94 |
224 | elsif ($doEnc) |
225 | { |
226 | foreach my $name (sort cmp_name keys %encoding) |
227 | { |
228 | my ($e2u,$u2e,$erep,$min_el,$max_el) = @{$encoding{$name}}; |
229 | output_enc(\*C,$name,$e2u); |
230 | } |
231 | } |
232 | elsif ($doUcm) |
233 | { |
234 | foreach my $name (sort cmp_name keys %encoding) |
235 | { |
236 | my ($e2u,$u2e,$erep,$min_el,$max_el) = @{$encoding{$name}}; |
237 | output_ucm(\*C,$name,$u2e,$erep,$min_el,$max_el); |
238 | } |
239 | } |
240 | |
017e2add |
241 | close(C); |
242 | |
9b37254d |
243 | |
b1e7e56f |
244 | |
9b37254d |
245 | sub compile_ucm |
246 | { |
dcda1f94 |
247 | my ($fh,$name) = @_; |
9b37254d |
248 | my $e2u = {}; |
249 | my $u2e = {}; |
250 | my $cs; |
251 | my %attr; |
252 | while (<$fh>) |
253 | { |
254 | s/#.*$//; |
255 | last if /^\s*CHARMAP\s*$/i; |
256 | if (/^\s*<(\w+)>\s+"?([^"]*)"?\s*$/i) |
257 | { |
258 | $attr{$1} = $2; |
259 | } |
260 | } |
261 | if (!defined($cs = $attr{'code_set_name'})) |
262 | { |
263 | warn "No <code_set_name> in $name\n"; |
264 | } |
265 | else |
266 | { |
51ef4e11 |
267 | $name = $cs unless exists $opt{'n'}; |
9b37254d |
268 | } |
269 | my $erep; |
270 | my $urep; |
afdae191 |
271 | my $max_el; |
272 | my $min_el; |
9b37254d |
273 | if (exists $attr{'subchar'}) |
274 | { |
afdae191 |
275 | my @byte; |
276 | $attr{'subchar'} =~ /^\s*/cg; |
277 | push(@byte,$1) while $attr{'subchar'} =~ /\G\\x([0-9a-f]+)/icg; |
278 | $erep = join('',map(chr(hex($_)),@byte)); |
9b37254d |
279 | } |
b1e7e56f |
280 | print "Reading $name ($cs)\n"; |
9b37254d |
281 | my $nfb = 0; |
282 | my $hfb = 0; |
283 | while (<$fh>) |
284 | { |
285 | s/#.*$//; |
286 | last if /^\s*END\s+CHARMAP\s*$/i; |
287 | next if /^\s*$/; |
afdae191 |
288 | my ($u,@byte); |
289 | my $fb = ''; |
290 | $u = $1 if (/^<U([0-9a-f]+)>\s+/igc); |
291 | push(@byte,$1) while /\G\\x([0-9a-f]+)/igc; |
292 | $fb = $1 if /\G\s*(\|[0-3])/gc; |
293 | # warn "$_: $u @byte | $fb\n"; |
294 | die "Bad line:$_" unless /\G\s*(#.*)?$/gc; |
9b37254d |
295 | if (defined($u)) |
296 | { |
297 | my $uch = encode_U(hex($u)); |
e0c49a6b |
298 | my $ech = join('',map(chr(hex($_)),@byte)); |
afdae191 |
299 | my $el = length($ech); |
300 | $max_el = $el if (!defined($max_el) || $el > $max_el); |
301 | $min_el = $el if (!defined($min_el) || $el < $min_el); |
9b37254d |
302 | if (length($fb)) |
303 | { |
304 | $fb = substr($fb,1); |
305 | $hfb++; |
306 | } |
307 | else |
308 | { |
309 | $nfb++; |
310 | $fb = '0'; |
311 | } |
312 | # $fb is fallback flag |
313 | # 0 - round trip safe |
314 | # 1 - fallback for unicode -> enc |
315 | # 2 - skip sub-char mapping |
316 | # 3 - fallback enc -> unicode |
317 | enter($u2e,$uch,$ech,$u2e,$fb+0) if ($fb =~ /[01]/); |
318 | enter($e2u,$ech,$uch,$e2u,$fb+0) if ($fb =~ /[03]/); |
319 | } |
320 | else |
321 | { |
322 | warn $_; |
323 | } |
9b37254d |
324 | } |
325 | if ($nfb && $hfb) |
326 | { |
327 | die "$nfb entries without fallback, $hfb entries with\n"; |
328 | } |
dcda1f94 |
329 | $encoding{$name} = [$e2u,$u2e,$erep,$min_el,$max_el]; |
9b37254d |
330 | } |
331 | |
14a8264b |
332 | sub compile_enc |
017e2add |
333 | { |
dcda1f94 |
334 | my ($fh,$name) = @_; |
017e2add |
335 | my $e2u = {}; |
336 | my $u2e = {}; |
337 | |
338 | my $type; |
339 | while ($type = <$fh>) |
340 | { |
341 | last if $type !~ /^\s*#/; |
342 | } |
343 | chomp($type); |
344 | return if $type eq 'E'; |
345 | my ($def,$sym,$pages) = split(/\s+/,scalar(<$fh>)); |
14a8264b |
346 | warn "$type encoded $name\n"; |
017e2add |
347 | my $rep = ''; |
afdae191 |
348 | my $min_el; |
349 | my $max_el; |
017e2add |
350 | { |
351 | my $v = hex($def); |
352 | no strict 'refs'; |
353 | $rep = &{"encode_$type"}($v & 0xFF, ($v >> 8) & 0xffe); |
354 | } |
e03ac092 |
355 | my %seen; |
017e2add |
356 | while ($pages--) |
357 | { |
358 | my $line = <$fh>; |
359 | chomp($line); |
360 | my $page = hex($line); |
361 | my $ch = 0; |
362 | for (my $i = 0; $i < 16; $i++) |
363 | { |
364 | my $line = <$fh>; |
365 | for (my $j = 0; $j < 16; $j++) |
366 | { |
367 | no strict 'refs'; |
368 | my $ech = &{"encode_$type"}($ch,$page); |
369 | my $val = hex(substr($line,0,4,'')); |
e03ac092 |
370 | next if $val == 0xFFFD; |
017e2add |
371 | if ($val || (!$ch && !$page)) |
372 | { |
afdae191 |
373 | my $el = length($ech); |
374 | $max_el = $el if (!defined($max_el) || $el > $max_el); |
375 | $min_el = $el if (!defined($min_el) || $el < $min_el); |
017e2add |
376 | my $uch = encode_U($val); |
e03ac092 |
377 | if (exists $seen{$uch}) |
378 | { |
379 | warn sprintf("U%04X is %02X%02X and %02X%02X\n", |
380 | $val,$page,$ch,@{$seen{$uch}}); |
381 | } |
382 | else |
383 | { |
384 | $seen{$uch} = [$page,$ch]; |
385 | } |
9b37254d |
386 | enter($e2u,$ech,$uch,$e2u,0); |
387 | enter($u2e,$uch,$ech,$u2e,0); |
017e2add |
388 | } |
389 | else |
390 | { |
391 | # No character at this position |
392 | # enter($e2u,$ech,undef,$e2u); |
393 | } |
394 | $ch++; |
395 | } |
396 | } |
397 | } |
dcda1f94 |
398 | $encoding{$name} = [$e2u,$u2e,$rep,$min_el,$max_el]; |
017e2add |
399 | } |
400 | |
401 | sub enter |
402 | { |
9b37254d |
403 | my ($a,$s,$d,$t,$fb) = @_; |
017e2add |
404 | $t = $a if @_ < 4; |
de745a2e |
405 | |
406 | while (1) { |
407 | $s =~ s/(.)//s; |
408 | my $b = $1; |
409 | my $e = $a->{$b}; |
410 | # 0 1 2 3 4 5 |
411 | $a->{$b} = $e = [$b,$b,'',{},1+length($s),0,$fb] unless $e; |
412 | unless (length($s)) { |
017e2add |
413 | $e->[2] = $d; |
414 | $e->[3] = $t; |
415 | $e->[5] = length($d); |
de745a2e |
416 | return; |
017e2add |
417 | } |
de745a2e |
418 | # Tail recursion. |
419 | $a = $e->[3]; |
420 | } |
017e2add |
421 | } |
422 | |
b1e7e56f |
423 | |
424 | |
017e2add |
425 | sub outstring |
426 | { |
427 | my ($fh,$name,$s) = @_; |
428 | my $sym = $strings{$s}; |
b1e7e56f |
429 | if ($sym) |
430 | { |
431 | $saved += length($s); |
432 | } |
433 | else |
017e2add |
434 | { |
4cfc977c |
435 | if ($opt{'O'}) { |
436 | foreach my $o (keys %strings) |
437 | { |
438 | my $i = index($o,$s); |
439 | if ($i >= 0) |
440 | { |
441 | $sym = $strings{$o}; |
442 | $sym .= sprintf("+0x%02x",$i) if ($i); |
443 | $subsave += length($s); |
444 | $strings{$s} = $sym; |
445 | return $sym; |
446 | } |
447 | } |
448 | } |
017e2add |
449 | $strings{$s} = $sym = $name; |
b1e7e56f |
450 | $strings += length($s); |
de745a2e |
451 | my $definition = sprintf "static const U8 %s[%d] = { ",$name,length($s); |
452 | # Maybe we should assert that these are all <256. |
453 | $definition .= join(',',unpack "C*",$s); |
454 | # We have a single long line. Split it at convenient commas. |
455 | $definition =~ s/(.{74,77},)/$1\n/g; |
456 | print $fh "$definition };\n\n"; |
017e2add |
457 | } |
458 | return $sym; |
459 | } |
460 | |
14a8264b |
461 | sub process |
017e2add |
462 | { |
14a8264b |
463 | my ($name,$a) = @_; |
017e2add |
464 | $name =~ s/\W+/_/g; |
465 | $a->{Cname} = $name; |
466 | my @keys = grep(ref($a->{$_}),sort keys %$a); |
017e2add |
467 | my $l; |
468 | my @ent; |
469 | foreach my $b (@keys) |
470 | { |
471 | my ($s,$f,$out,$t,$end) = @{$a->{$b}}; |
472 | if (defined($l) && |
473 | ord($b) == ord($a->{$l}[1])+1 && |
474 | $a->{$l}[3] == $a->{$b}[3] && |
475 | $a->{$l}[4] == $a->{$b}[4] && |
9b37254d |
476 | $a->{$l}[5] == $a->{$b}[5] && |
477 | $a->{$l}[6] == $a->{$b}[6] |
2f2b4ff2 |
478 | # && length($a->{$l}[2]) < 16 |
479 | ) |
017e2add |
480 | { |
481 | my $i = ord($b)-ord($a->{$l}[0]); |
482 | $a->{$l}[1] = $b; |
483 | $a->{$l}[2] .= $a->{$b}[2]; |
484 | } |
485 | else |
486 | { |
487 | $l = $b; |
488 | push(@ent,$b); |
489 | } |
14a8264b |
490 | if (exists $t->{Cname}) |
017e2add |
491 | { |
14a8264b |
492 | $t->{'Forward'} = 1 if $t != $a; |
493 | } |
494 | else |
495 | { |
496 | process(sprintf("%s_%02x",$name,ord($s)),$t); |
017e2add |
497 | } |
498 | } |
499 | if (ord($keys[-1]) < 255) |
500 | { |
501 | my $t = chr(ord($keys[-1])+1); |
502 | $a->{$t} = [$t,chr(255),undef,$a,0,0]; |
503 | push(@ent,$t); |
504 | } |
14a8264b |
505 | $a->{'Entries'} = \@ent; |
506 | } |
507 | |
508 | sub outtable |
509 | { |
510 | my ($fh,$a) = @_; |
511 | my $name = $a->{'Cname'}; |
017e2add |
512 | # String tables |
14a8264b |
513 | foreach my $b (@{$a->{'Entries'}}) |
017e2add |
514 | { |
515 | next unless $a->{$b}[5]; |
516 | my $s = ord($a->{$b}[0]); |
517 | my $e = ord($a->{$b}[1]); |
518 | outstring($fh,sprintf("%s__%02x_%02x",$name,$s,$e),$a->{$b}[2]); |
519 | } |
14a8264b |
520 | if ($a->{'Forward'}) |
521 | { |
522 | print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"];\n"; |
523 | } |
524 | $a->{'Done'} = 1; |
525 | foreach my $b (@{$a->{'Entries'}}) |
526 | { |
527 | my ($s,$e,$out,$t,$end,$l) = @{$a->{$b}}; |
528 | outtable($fh,$t) unless $t->{'Done'}; |
529 | } |
530 | print $fh "\nstatic encpage_t $name\[",scalar(@{$a->{'Entries'}}),"] = {\n"; |
531 | foreach my $b (@{$a->{'Entries'}}) |
017e2add |
532 | { |
9b37254d |
533 | my ($s,$e,$out,$t,$end,$l,$fb) = @{$a->{$b}}; |
017e2add |
534 | my $sc = ord($s); |
535 | my $ec = ord($e); |
e0c49a6b |
536 | $end |= 0x80 if $fb; |
017e2add |
537 | print $fh "{"; |
538 | if ($l) |
539 | { |
540 | printf $fh outstring($fh,'',$out); |
541 | } |
542 | else |
543 | { |
544 | print $fh "0"; |
545 | } |
546 | print $fh ",",$t->{Cname}; |
2f2b4ff2 |
547 | printf $fh ",0x%02x,0x%02x,$l,$end},\n",$sc,$ec; |
017e2add |
548 | } |
14a8264b |
549 | print $fh "};\n"; |
550 | } |
551 | |
552 | sub output |
553 | { |
554 | my ($fh,$name,$a) = @_; |
555 | process($name,$a); |
556 | # Sub-tables |
557 | outtable($fh,$a); |
017e2add |
558 | } |
559 | |
e0c49a6b |
560 | sub output_enc |
561 | { |
562 | my ($fh,$name,$a) = @_; |
563 | foreach my $b (sort keys %$a) |
564 | { |
565 | my ($s,$e,$out,$t,$end,$l,$fb) = @{$a->{$b}}; |
566 | } |
567 | } |
568 | |
569 | sub decode_U |
570 | { |
571 | my $s = shift; |
e0c49a6b |
572 | } |
573 | |
f39fb8cc |
574 | my @uname; |
575 | sub char_names |
576 | { |
9fb51cbc |
577 | my $s = do "unicore/Name.pl"; |
578 | die "char_names: unicore/Name.pl: $!\n" unless defined $s; |
f39fb8cc |
579 | pos($s) = 0; |
580 | while ($s =~ /\G([0-9a-f]+)\t([0-9a-f]*)\t(.*?)\s*\n/igc) |
581 | { |
582 | my $name = $3; |
583 | my $s = hex($1); |
584 | last if $s >= 0x10000; |
585 | my $e = length($2) ? hex($2) : $s; |
586 | for (my $i = $s; $i <= $e; $i++) |
587 | { |
588 | $uname[$i] = $name; |
589 | # print sprintf("U%04X $name\n",$i); |
590 | } |
591 | } |
592 | } |
593 | |
e0c49a6b |
594 | sub output_ucm_page |
595 | { |
f39fb8cc |
596 | my ($cmap,$a,$t,$pre) = @_; |
e0c49a6b |
597 | # warn sprintf("Page %x\n",$pre); |
598 | foreach my $b (sort keys %$t) |
599 | { |
600 | my ($s,$e,$out,$n,$end,$l,$fb) = @{$t->{$b}}; |
601 | die "oops $s $e" unless $s eq $e; |
602 | my $u = ord($s); |
603 | if ($n != $a && $n != $t) |
604 | { |
f39fb8cc |
605 | output_ucm_page($cmap,$a,$n,(($pre|($u &0x3F)) << 6)&0xFFFF); |
e0c49a6b |
606 | } |
607 | elsif (length($out)) |
608 | { |
609 | if ($pre) |
610 | { |
611 | $u = $pre|($u &0x3f); |
612 | } |
f39fb8cc |
613 | my $s = sprintf "<U%04X> ",$u; |
e0c49a6b |
614 | foreach my $c (split(//,$out)) |
615 | { |
f39fb8cc |
616 | $s .= sprintf "\\x%02X",ord($c); |
e0c49a6b |
617 | } |
f39fb8cc |
618 | $s .= sprintf " |%d # %s\n",($fb ? 1 : 0),$uname[$u]; |
619 | push(@$cmap,$s); |
e0c49a6b |
620 | } |
621 | else |
622 | { |
623 | warn join(',',@{$t->{$b}},$a,$t); |
624 | } |
625 | } |
626 | } |
627 | |
628 | sub output_ucm |
629 | { |
f39fb8cc |
630 | my ($fh,$name,$h,$rep,$min_el,$max_el) = @_; |
df1df145 |
631 | print $fh "# $0 @orig_ARGV\n" unless $opt{'q'}; |
afdae191 |
632 | print $fh "<code_set_name> \"$name\"\n"; |
f39fb8cc |
633 | char_names(); |
afdae191 |
634 | if (defined $min_el) |
635 | { |
636 | print $fh "<mb_cur_min> $min_el\n"; |
637 | } |
638 | if (defined $max_el) |
639 | { |
640 | print $fh "<mb_cur_max> $max_el\n"; |
641 | } |
642 | if (defined $rep) |
643 | { |
644 | print $fh "<subchar> "; |
645 | foreach my $c (split(//,$rep)) |
646 | { |
647 | printf $fh "\\x%02X",ord($c); |
648 | } |
649 | print $fh "\n"; |
650 | } |
f39fb8cc |
651 | my @cmap; |
652 | output_ucm_page(\@cmap,$h,$h,0); |
afdae191 |
653 | print $fh "#\nCHARMAP\n"; |
f39fb8cc |
654 | foreach my $line (sort { substr($a,8) cmp substr($b,8) } @cmap) |
655 | { |
656 | print $fh $line; |
657 | } |
e0c49a6b |
658 | print $fh "END CHARMAP\n"; |
659 | } |
017e2add |
660 | |