?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
beta-0.89.2
[luatex.git] / source / texk / web2c / luatexdir / unilib / src / makeutype.c
blobffa7a20a574ebe66702f87650d97000ea1877bcb
1 /* Copyright (C) 2000-2012 by George Williams */
2 /*
3 * Redistribution and use in source and binary forms, with or without
4 * modification, are permitted provided that the following conditions are met:
5
6 * Redistributions of source code must retain the above copyright notice, this
7 * list of conditions and the following disclaimer.
8
9 * Redistributions in binary form must reproduce the above copyright notice,
10 * this list of conditions and the following disclaimer in the documentation
11 * and/or other materials provided with the distribution.
12
13 * The name of the author may not be used to endorse or promote products
14 * derived from this software without specific prior written permission.
15
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
17 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
18 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
19 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
20 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
21 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
22 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
23 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
24 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
25 * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 /*
29 * This program generates 4 different files based on the latest UnicodeData.txt
30 * obtained from http://unicode.org which is then used to build FontForge
31 *
32 * To generate the latest files, you will first need to go and get these 4 files
33 * and put them in the Unicode subdirectory:
34 * http://unicode.org/Public/UNIDATA/LineBreak.txt
35 * http://unicode.org/Public/UNIDATA/NamesList.txt
36 * http://unicode.org/Public/UNIDATA/PropList.txt
37 * http://unicode.org/Public/UNIDATA/UnicodeData.txt
38 *
39 * Next, you will need to build ./makeutype before you can use it:
40 * Run "make makeutype"
41 * or
42 * Run "gcc -s -I../inc -o makeutype makeutype.c"
43 *
44 * Then run the executable binary "/makeutype".
45 * This will create 4 files in the same directory:
46 * ArabicForms.c, unialt.c, utype.c, utype.h
47 * (please move utype.h into Fontforge's "../inc" subdirectory)
48 *
49 * When done building the updated files, you can clean-up by removing
50 * LineBreak.txt, NamesList.txt, PropList.txt,UnicodeData.txt, and the
51 * binary executable file makeutype as they are no longer needed now.
52 */
53
54
55 /* Build a ctype array out of the UnicodeData.txt and PropList.txt files */
56 #include <stdio.h>
57 #include <stdlib.h>
58 #include <string.h>
59 #include <unibasics.h>
60 #define true 1
61 #define false 0
62
63 /*#define MAXC 0x600 /* Last upper/lower case dicodomy is Armenian 0x580, er, nope. 1fff (greek and latin extended) then full-width 0xff00 */
64 #define MAXC 65536
65 #define MAXA 18
66
67 /* These values get stored within flags[unicodechar={0..MAXC}] */
68 #define _LOWER 1
69 #define _UPPER 2
70 #define _TITLE 4
71 #define _DIGIT 8
72 #define _SPACE 0x10
73 #define _PUNCT 0x20
74 #define _HEX 0x40
75 #define _ZEROWIDTH 0x80
76
77 #define _LEFT_2_RIGHT 0x100
78 #define _RIGHT_2_LEFT 0x200
79 #define _ENUMERIC 0x400
80 #define _ANUMERIC 0x800
81 #define _ENS 0x1000
82 #define _CS 0x2000
83 #define _ENT 0x4000
84 #define _COMBINING 0x8000
85
86 #define _BREAKBEFOREOK 0x10000
87 #define _BREAKAFTEROK 0x20000
88 #define _NONSTART 0x40000 /* small kana, close punct, can't start a line */
89 #define _NONEND 0x80000 /* open punct, can't end a line */
90 /*#define _MUSTBREAK 0x100000 /* newlines, paragraphs, etc. */
91 #define _URLBREAKAFTER 0x100000 /* break after slash not followed by digits (ie. in URLs not fractions or dates) */
92
93 #define _ALPHABETIC 0x200000
94 #define _IDEOGRAPHIC 0x400000
95
96 #define _INITIAL 0x800000
97 #define _MEDIAL 0x1000000
98 #define _FINAL 0x2000000
99 #define _ISOLATED 0x4000000
100
101 #define _NOBREAK 0x8000000
102 #define _DecompositionNormative 0x10000000
103
104 #define _CombiningClass 0xff
105 #define _Above 0x100
106 #define _Below 0x200
107 #define _Overstrike 0x400
108 #define _Left 0x800
109 #define _Right 0x1000
110 #define _Joins2 0x2000
111 #define _CenterLeft 0x4000
112 #define _CenterRight 0x8000
113 #define _CenteredOutside 0x10000
114 #define _Outside 0x20000
115 #define _RightEdge 0x40000
116 #define _LeftEdge 0x80000
117 #define _Touching 0x100000
118
119 #include "combiners.h"
120
121 char *names[MAXC];
122 unsigned short mytolower[MAXC];
123 unsigned short mytoupper[MAXC];
124 unsigned short mytotitle[MAXC];
125 unsigned char mynumericvalue[MAXC];
126 unsigned short mymirror[MAXC];
127 unsigned long flags[MAXC]; /* 32 binary flags for each unicode.org character */
128 unsigned long flags2[MAXC];
129 unichar_t alts[MAXC][MAXA+1];
130 unsigned long assignedcodepoints[0x120000/32]; /* 32 characters represented per each long value */
131
132 const char GeneratedFileMessage[] = "\n/* This file was generated using the program 'makeutype' */\n\n";
133 const char CantReadFile[] = "Can't find or read file %s\n"; /* exit(1) */
134 const char CantSaveFile[] = "Can't open or write to output file %s\n"; /* exit(2) */
135 const char NoMoreMemory[] = "Can't access more memory.\n"; /* exit(3) */
136 const char LineLengthBg[] = "Error with %s. Found line too long: %s\n"; /* exit(4) */
137
138 static void FreeNamesMemorySpace() {
139 long index;
140 for ( index=0; index<MAXC ; ++index ) {
141 if ( names[index]!=NULL ) free( names[index] );
142 }
143 }
144
145 static void FigureAlternates(long index, char *apt, int normative) {
146 int alt, i;
147 char *end;
148 int isisolated=0, iscircled=0;
149
150 for ( i=0; ; ++i ) {
151 while ( *apt=='<' ) {
152 isisolated = strncmp(apt,"<isolated>",strlen("<isolated>"))==0;
153 iscircled = strncmp(apt,"<circle>",strlen("<circle>"))==0;
154 while ( *apt && *apt!='>' ) ++apt;
155 if ( *apt=='>' ) ++apt;
156 while ( *apt==' ' ) ++apt;
157 }
158 alt = strtol(apt,&end,16);
159 if ( end==apt )
160 break;
161 apt = end;
162 if ( i<MAXA ) {
163 alts[index][i] = alt;
164 alts[index][i+1] = 0;
165 }
166 }
167 if ( iscircled && i<MAXA ) {
168 alts[index][i] = 0x20dd;
169 alts[index][++i] = 0;
170 }
171 if ( i>MAXA )
172 fprintf( stderr, "%d is too many alternates for U+%04X\n", i, index );
173 if ( i>0 && normative)
174 flags[index] |= _DecompositionNormative;
175
176 /* arabic isolated forms are alternates for the standard forms */
177 if ( isisolated && alts[index][0]>=0x600 && alts[index][0]<0x6ff && alts[index][1]==0 &&
178 alts[alts[index][0]][0]==0 )
179 alts[alts[index][0]][0] = index;
180 }
181
182 static void processAssignment(long index,char *pt) {
183 static long first=-1;
184 long i;
185
186 if ( index>0x11ffff )
187 return;
188 ++pt; /* move past semicolon */
189 if ( *pt!='<' ) {
190 assignedcodepoints[index/32] |= (1<<(index%32)); /* This Unicode char is visible */
191 } else if ( strstr(pt,", First")!=NULL ) { /* start of an extended charset */
192 first = index;
193 } else if ( strstr(pt,", Last")!=NULL ) { /* end of an extended charset */
194 if ( first==-1 || first > index )
195 fprintf( stderr,"Something went wrong, first isn't defined at last. %x\n", index );
196 else if ( first>=0xd800 && first<=0xdfff )
197 /* surrogate pairs. Not assigned really */;
198 else {
199 /* mark all characters visible in the range of {First...Last} */
200 for ( i=first; i<=index; ++i )
201 assignedcodepoints[i/32] |= (1<<(i%32));
202 }
203 first = -1;
204 }
205 }
206
207 static void readin(void) {
208 char buffer[512+1], buf2[300+1], oldname[301], *pt, *end, *pt1;
209 long index, lc, uc, tc, flg, val, indexend, wasfirst;
210 int cc;
211 FILE *fp;
212 int i,j;
213
214 buffer[512]='\0'; buf2[0] = buf2[300]='\0'; oldname[0]='\0';
215 if ((fp = fopen("UnicodeData.txt","r"))==NULL ) {
216 fprintf( stderr, CantReadFile,"UnicodeData.txt" );
217 exit(1);
218 }
219 while ( fgets(buffer,sizeof(buffer)-1,fp)!=NULL ) {
220 if (strlen(buffer)>=299) { /* previous version was linelength of 300 chars, jul2012 */
221 fprintf( stderr, LineLengthBg,"UnicodeData.txt",buffer );
222 fprintf( stderr, "\n%s\n",buffer );
223
224 fclose(fp);
225 FreeNamesMemorySpace();
226 exit(4);
227 }
228 if ( *buffer=='#' )
229 continue;
230 flg = 0;
231 /* Unicode character value */
232 index = strtol(buffer,&end,16);
233 processAssignment(index,end);
234 if ( index>=MAXC ) /* For now can only deal with BMP !!!! */
235 continue;
236 pt = end;
237 if ( *pt==';' ) {
238 ++pt;
239 /* buf2 = character name */
240 for ( pt1=pt; *pt1!=';' && *pt1!='\0'; ++pt1 );
241 strncpy(buf2,pt,pt1-pt); buf2[pt1-pt] = '\0'; pt = pt1;
242 if ( *pt==';' ) ++pt;
243 /* general category */
244 for ( pt1=pt; *pt1!=';' && *pt1!='\0'; ++pt1 );
245 if ( strncmp(pt,"Lu",pt1-pt)==0 )
246 flg |= _UPPER|_ALPHABETIC;
247 else if ( strncmp(pt,"Ll",pt1-pt)==0 )
248 flg |= _LOWER|_ALPHABETIC;
249 else if ( strncmp(pt,"Lt",pt1-pt)==0 )
250 flg |= _TITLE|_ALPHABETIC;
251 else if ( strncmp(pt,"Lo",pt1-pt)==0 )
252 flg |= _ALPHABETIC;
253 else if ( strncmp(pt,"Nd",pt1-pt)==0 )
254 flg |= _DIGIT;
255 pt = pt1;
256 if ( *pt==';' ) ++pt;
257 /* Unicode combining classes, I do my own version later */
258 cc = strtol(pt,&end,16);
259 pt = end;
260 if ( *pt==';' ) ++pt;
261 /* Bidirectional Category */
262 for ( pt1=pt; *pt1!=';' && *pt1!='\0'; ++pt1 );
263 if ( strncmp(pt,"L",pt1-pt)==0 || strncmp(pt,"LRE",pt1-pt)==0 || strncmp(pt,"LRO",pt1-pt)==0 )
264 flg |= _LEFT_2_RIGHT;
265 if ( strncmp(pt,"R",pt1-pt)==0 || strncmp(pt,"AL",pt1-pt)==0 || strncmp(pt,"RLE",pt1-pt)==0 || strncmp(pt,"RLO",pt1-pt)==0 )
266 flg |= _RIGHT_2_LEFT;
267 else if ( strncmp(pt,"EN",pt1-pt)==0 )
268 flg |= _ENUMERIC;
269 else if ( strncmp(pt,"ES",pt1-pt)==0 )
270 flg |= _ENS;
271 else if ( strncmp(pt,"ET",pt1-pt)==0 )
272 flg |= _ENT;
273 else if ( strncmp(pt,"AN",pt1-pt)==0 )
274 flg |= _ANUMERIC;
275 else if ( strncmp(pt,"CS",pt1-pt)==0 )
276 flg |= _CS;
277 pt = pt1;
278 if ( *pt==';' ) ++pt;
279 /* character decomposition */
280 if ( strncmp(pt,"<initial>",strlen("<initial>"))==0 )
281 flg |= _INITIAL;
282 else if ( strncmp(pt,"<final>",strlen("<final>"))==0 )
283 flg |= _FINAL;
284 else if ( strncmp(pt,"<medial>",strlen("<medial>"))==0 )
285 flg |= _MEDIAL;
286 else if ( strncmp(pt,"<isolated>",strlen("<isolated>"))==0 )
287 flg |= _ISOLATED;
288 FigureAlternates(index,pt, true);
289 while ( *pt!=';' && *pt!='\0' ) ++pt;
290 if ( *pt==';' ) ++pt;
291 /* Don't care about decimal digit value */
292 while ( *pt!=';' && *pt!='\0' ) ++pt;
293 if ( *pt==';' ) ++pt;
294 /* Don't care about digit value */
295 while ( *pt!=';' && *pt!='\0' ) ++pt;
296 if ( *pt==';' ) ++pt;
297 /* numeric value */
298 val = strtol(pt,&end,10);
299 if ( pt==end ) val = -1;
300 pt = end;
301 if ( *pt==';' ) ++pt;
302 /* Don't care about mirrored value */
303 while ( *pt!=';' && *pt!='\0' ) ++pt;
304 if ( *pt==';' ) ++pt;
305 /* Only care about old name (unicode 1.0) for control characters */
306 for ( pt1=pt; *pt1!=';' && *pt1!='\0'; ++pt1 );
307 strncpy(oldname,pt,pt1-pt); oldname[pt1-pt] = '\0';
308 if ( pt1-pt>100 ) oldname[100] = '\0'; pt = pt1;
309 if ( *pt==';' ) ++pt;
310 /* Don't care about 10646 comment field */
311 while ( *pt!=';' && *pt!='\0' ) ++pt;
312 if ( *pt==';' ) ++pt;
313 /* upper-case value */
314 uc = strtol(pt,&end,16);
315 if ( end==pt )
316 uc = index;
317 pt = end;
318 if ( *pt==';' ) ++pt;
319 /* lower-case value */
320 lc = strtol(pt,&end,16);
321 if ( end==pt )
322 lc = index;
323 pt = end;
324 if ( *pt==';' ) ++pt;
325 /* title-case value */
326 tc = strtol(pt,&end,16);
327 if ( end==pt )
328 tc = index;
329 pt = end;
330 if ( *pt==';' ) ++pt;
331 if ( index>=MAXC )
332 break;
333 mytolower[index]= lc;
334 mytoupper[index]= uc;
335 mytotitle[index]= tc;
336 mynumericvalue[index]= val;
337 flags[index] |= flg;
338 flags2[index] = cc&0xff;
339 if ( strstr(buf2," First>")!=NULL )
340 wasfirst = index;
341 else if ( strstr(buf2," Last>")!=NULL ) {
342 for ( ; wasfirst<index; ++wasfirst ) {
343 mytolower[wasfirst]= wasfirst;
344 mytoupper[wasfirst]= wasfirst;
345 mytotitle[wasfirst]= wasfirst;
346 mynumericvalue[wasfirst]= -1;
347 flags[wasfirst] = flg;
348 flags2[wasfirst] = cc&0xff;
349 }
350 } else {
351 if ( strcmp(buf2,"<control>")==0 ) {
352 strcat(buf2, " ");
353 strcat(buf2, oldname);
354 }
355 names[index]= strdup(buf2);
356 }
357 }
358 }
359 fclose(fp);
360
361 if ((fp = fopen("LineBreak.txt","r"))==NULL ) {
362 fprintf( stderr, CantReadFile, "LineBreak.txt" );
363 FreeNamesMemorySpace();
364 exit(1);
365 }
366 while ( fgets(buffer,sizeof(buffer)-1,fp)!=NULL ) {
367 if (strlen(buffer)>=299) { /* previous version was linelength of 300 chars, jul2012 */
368 fprintf( stderr, LineLengthBg,"LineBreak.txt",buffer );
369 fclose(fp);
370 FreeNamesMemorySpace();
371 exit(4);
372 }
373 if ( *buffer=='#' )
374 continue;
375 flg = 0;
376 /* Unicode character value */
377 indexend = index = strtol(buffer,&end,16);
378 if ( index>0xffff ) /* Only BMP now !!!!!! */
379 continue;
380 pt = end;
381 if ( *pt=='.' && pt[1]=='.' ) {
382 indexend = strtol(pt+2,&end,16);
383 if ( indexend>0xffff ) indexend = 0xffff; /* Only BMP now !!!!! */
384 pt = end;
385 }
386 if ( *pt==';' ) {
387 ++pt;
388 for ( pt1=pt; *pt1!=';' && *pt1!=' ' && *pt1!='\0'; ++pt1 );
389 if ( strncmp(pt,"BK",pt1-pt)==0 || strncmp(pt,"CR",pt1-pt)==0 || strncmp(pt,"LF",pt1-pt)==0 )
390 /*flg |= _MUSTBREAK*/;
391 else if ( strncmp(pt,"NS",pt1-pt)==0 || strncmp(pt,"CL",pt1-pt)==0 )
392 flg |= _NONSTART;
393 else if ( strncmp(pt,"OP",pt1-pt)==0 || strncmp(pt,"CM",pt1-pt)==0 )
394 flg |= _NONEND;
395 else if ( strncmp(pt,"GL",pt1-pt)==0 )
396 flg |= _NONEND|_NONSTART;
397 else if ( strncmp(pt,"SP",pt1-pt)==0 || strncmp(pt,"HY",pt1-pt)==0 ||
398 strncmp(pt,"BA",pt1-pt)==0 ||
399 strncmp(pt,"ZW",pt1-pt)==0 )
400 flg |= _BREAKAFTEROK;
401 else if ( strncmp(pt,"BB",pt1-pt)==0 )
402 flg |= _BREAKBEFOREOK;
403 else if ( strncmp(pt,"B2",pt1-pt)==0 )
404 flg |= _BREAKBEFOREOK|_BREAKAFTEROK;
405 else if ( strncmp(pt,"ID",pt1-pt)==0 )
406 flg |= _BREAKBEFOREOK|_BREAKAFTEROK;
407 else if ( strncmp(pt,"SY",pt1-pt)==0 )
408 flg |= _URLBREAKAFTER;
409 pt = pt1;
410 for ( ; index<=indexend; ++index )
411 flags[index] |= flg;
412 #if 0
413 if ( strstr(pt," First>")!=NULL )
414 wasfirst = index;
415 else if ( strstr(pt," Last>")!=NULL ) {
416 for ( ; wasfirst<index; ++wasfirst ) {
417 flags[wasfirst] = flg;
418 }
419 }
420 #endif
421 }
422 }
423 fclose(fp);
424
425 if ((fp = fopen("PropList.txt","r"))==NULL ) {
426 fprintf( stderr, CantReadFile, "PropList.txt" );
427 FreeNamesMemorySpace();
428 exit(1);
429 }
430 while ( fgets(buffer,sizeof(buffer)-1,fp)!=NULL ) {
431 flg = 0;
432 if (strlen(buffer)>=299) { /* previous version was linelength of 300 chars, jul2012 */
433 fprintf( stderr, LineLengthBg,"PropList.txt",buffer );
434 fclose(fp);
435 FreeNamesMemorySpace();
436 exit(4);
437 }
438 if ( true || strncmp(buffer,"Property dump for:", strlen("Property dump for:"))==0 ) {
439 if ( strstr(buffer, "(Zero-width)")!=NULL || strstr(buffer, "ZERO WIDTH")!=NULL )
440 flg = _ZEROWIDTH;
441 else if ( strstr(buffer, "(White space)")!=NULL || strstr(buffer, "White_Space")!=NULL )
442 flg = _SPACE;
443 else if ( strstr(buffer, "(Punctuation)")!=NULL || strstr(buffer, "Punctuation")!=NULL )
444 flg = _PUNCT;
445 else if ( strstr(buffer, "(Alphabetic)")!=NULL || strstr(buffer, "Alphabetic")!=NULL )
446 flg = _ALPHABETIC;
447 else if ( strstr(buffer, "(Ideographic)")!=NULL || strstr(buffer, "Ideographic")!=NULL )
448 flg = _IDEOGRAPHIC;
449 else if ( strstr(buffer, "(Hex Digit)")!=NULL || strstr(buffer, "Hex_Digit")!=NULL )
450 flg = _HEX;
451 else if ( strstr(buffer, "(Combining)")!=NULL || strstr(buffer, "COMBINING")!=NULL )
452 flg = _COMBINING;
453 else if ( strstr(buffer, "(Non-break)")!=NULL )
454 flg = _NOBREAK;
455 if ( flg!=0 ) {
456 if (( buffer[0]>='0' && buffer[0]<='9') || (buffer[0]>='A' && buffer[0]<='F')) {
457 index = wasfirst = strtol(buffer,NULL,16);
458 if ( buffer[4]=='.' && buffer[5]=='.' )
459 index = strtol(buffer+6,NULL,16);
460 for ( ; wasfirst<=index && wasfirst<=0xffff; ++wasfirst ) /* BMP !!!!! */
461 flags[wasfirst] |= flg;
462 }
463 }
464 }
465 }
466 fclose(fp);
467 /* There used to be a zero width property, but no longer */
468 flags[0x200B] |= _ZEROWIDTH;
469 flags[0x200C] |= _ZEROWIDTH;
470 flags[0x200D] |= _ZEROWIDTH;
471 flags[0x2060] |= _ZEROWIDTH;
472 flags[0xFEFF] |= _ZEROWIDTH;
473 /* There used to be a No Break property, but no longer */
474 flags[0x00A0] |= _NOBREAK;
475 flags[0x2011] |= _NOBREAK;
476 flags[0x202F] |= _NOBREAK;
477 flags[0xFEFF] |= _NOBREAK;
478
479 if ((fp = fopen("NamesList.txt","r"))==NULL ) {
480 fprintf( stderr, CantReadFile, "NamesList.txt" );
481 FreeNamesMemorySpace();
482 exit(1);
483 }
484 while ( fgets(buffer,sizeof(buffer)-1,fp)!=NULL ) {
485 flg = 0;
486 if (strlen(buffer)>=511) {
487 fprintf( stderr, LineLengthBg,"NamesList.txt",buffer );
488 fclose(fp);
489 FreeNamesMemorySpace();
490 exit(4);
491 }
492 if ( (index = strtol(buffer,NULL,16))!=0 ) {
493 if ( strstr(buffer, "COMBINING")!=NULL )
494 flg = _COMBINING;
495 else if ( strstr(buffer, "N0-BREAK")!=NULL )
496 flg = _NOBREAK;
497 else if ( strstr(buffer, "ZERO WIDTH")!=NULL )
498 flg = _ZEROWIDTH;
499
500 if ( index<0xffff ) /* !!!!! BMP */
501 flags[wasfirst] |= flg;
502 }
503 }
504 fclose(fp);
505
506 for ( i=0; combiners[i].low!=0; ++i ) {
507 for ( j=combiners[i].low; j<=combiners[i].high; ++j )
508 flags2[j] |= combiners[i].pos[j-combiners[i].low];
509 }
510 }
511
512 static void readcorpfile(char *prefix, char *corp) {
513 char buffer[300+1], buf2[300+1], *pt, *end, *pt1;
514 long index;
515 FILE *fp;
516
517 buffer[300]='\0'; buf2[0] = buf2[300]='\0';
518 if ((fp = fopen(corp,"r"))==NULL ) {
519 fprintf( stderr, CantReadFile, corp ); /* Not essential */
520 return;
521 }
522 while ( fgets(buffer,sizeof(buffer)-1,fp)!=NULL ) {
523 if (strlen(buffer)>=299) {
524 fprintf( stderr, LineLengthBg,corp,buffer );
525 fclose(fp);
526 FreeNamesMemorySpace();
527 exit(4);
528 }
529 if ( *buffer=='#' )
530 continue;
531 /* code */
532 index = strtol(buffer,&end,16);
533 pt = end;
534 if ( *pt==';' ) {
535 ++pt;
536 while ( *pt!=';' && *pt!='\0' ) ++pt;
537 if ( *pt==';' ) ++pt;
538 /* character name */
539 for ( pt1=pt; *pt1!=';' && *pt1!='\0' && *pt1!='\n' && *pt1!='\r'; ++pt1 );
540 strncpy(buf2,pt,pt1-pt); buf2[pt1-pt] = '\0'; pt = pt1;
541 if ( *pt==';' ) ++pt;
542 /* character decomposition */
543 FigureAlternates(index,pt, false);
544 if ( index>=MAXC )
545 break;
546 if ((names[index]= malloc(strlen(buf2)+strlen(prefix)+4)) == NULL) {
547 fprintf( stderr, NoMoreMemory );
548 fclose(fp);
549 FreeNamesMemorySpace();
550 exit(3);
551 }
552 strcpy(names[index],prefix); strcat(names[index],buf2);
553 }
554 }
555 fclose(fp);
556 }
557
558 static int find(char *base, char *suffix) {
559 char name[300+1];
560 int i;
561
562 name[300]='\0';
563 strcpy(name,base);
564 strcat(name,suffix);
565
566 for ( i=0; i<MAXC; ++i )
567 if ( names[i]!=NULL && strcmp(names[i],name)==0 )
568 return( i );
569
570 return( -1 );
571 }
572
573 static void dumparabicdata(FILE *header) {
574 FILE *data;
575 struct arabicforms {
576 unsigned short initial, medial, final, isolated;
577 unsigned int isletter: 1;
578 unsigned int joindual: 1;
579 unsigned int required_lig_with_alef: 1;
580 } forms[256];
581 int i, j, index;
582
583 memset(forms,'\0',sizeof(forms));
584 for ( i=0x600; i<0x700; ++i ) {
585 j = i-0x600;
586 if ( names[i]==NULL )
587 /* No op (not defined) */;
588 else if ( strncmp(names[i],"ARABIC LETTER ",strlen("ARABIC LETTER "))!=0 )
589 /* No op (not a letter, no fancy forms) */
590 forms[j].initial = forms[j].medial = forms[j].final = forms[j].isolated = i;
591 else {
592 forms[j].isletter = 1;
593 forms[j].initial = forms[j].medial = forms[j].final = forms[j].isolated = i;
594 if ( (index = find(names[i]," ISOLATED FORM"))!= -1 )
595 forms[j].isolated = index;
596 if ( (index = find(names[i]," FINAL FORM"))!= -1 )
597 forms[j].final = index;
598 if ( (index = find(names[i]," INITIAL FORM"))!= -1 )
599 forms[j].initial = index;
600 if ( (index = find(names[i]," MEDIAL FORM"))!= -1 )
601 forms[j].medial = index;
602 if ( forms[j].initial!=i && forms[j].medial!=i )
603 forms[j].joindual = 1;
604 }
605 }
606 forms[0x44/* 0x644 == LAM */].required_lig_with_alef = 1;
607
608 fprintf(header,"\nextern struct arabicforms {\n" );
609 fprintf(header," unsigned short initial, medial, final, isolated;\n" );
610 fprintf(header," unsigned int isletter: 1;\n" );
611 fprintf(header," unsigned int joindual: 1;\n" );
612 fprintf(header," unsigned int required_lig_with_alef: 1;\n" );
613 fprintf(header,"} ArabicForms[256];\t/* for chars 0x600-0x6ff, subtract 0x600 to use array */\n" );
614
615 data = fopen( "ArabicForms.c","w");
616 if ( data==NULL || data==NULL ) {
617 fprintf( stderr, CantSaveFile, "ArabicForms.c" );
618 FreeNamesMemorySpace();
619 exit(2);
620 }
621
622 fprintf( data, "#include <utype.h>\n" );
623
624 fprintf( data, GeneratedFileMessage );
625
626 fprintf( data, "struct arabicforms ArabicForms[] = {\n" );
627 fprintf( data, "\t/* initial, medial, final, isolated, isletter, joindual, required_lig_with_alef */\n");
628 for ( i=0; i<256; ++i ) {
629 fprintf( data, "\t{ 0x%04x, 0x%04x, 0x%04x, 0x%04x, %d, %d, %d }",
630 forms[i].initial, forms[i].medial, forms[i].final, forms[i].isolated,
631 forms[i].isletter, forms[i].joindual, forms[i].required_lig_with_alef);
632 if ( i==255 )
633 fprintf( data, "\n");
634 else
635 if ( (i & 31)==0 )
636 fprintf( data, ",\t/* 0x%04x */\n",0x600+i);
637 else
638 fprintf( data, ",\n");
639 }
640 fprintf( data, "};\n" );
641 fclose( data );
642 }
643
644 static void dump() {
645 FILE *header, *data;
646 int i,j;
647
648 header=fopen("utype.h","w");
649 data = fopen("utype.c","w");
650
651 if ( header==NULL || data==NULL ) {
652 fprintf( stderr, CantSaveFile, "(utype.[ch])" );
653 if ( header ) fclose( header );
654 if ( data ) fclose( data );
655 FreeNamesMemorySpace();
656 exit(2);
657 }
658
659 fprintf( header, "#ifndef _UTYPE_H\n" );
660 fprintf( header, "#define _UTYPE_H\n" );
661
662 fprintf( header, GeneratedFileMessage );
663
664 fprintf( header, "#include <ctype.h>\t\t/* Include here so we can control it. If a system header includes it later bad things happen */\n" );
665 fprintf( header, "#include <unibasics.h>\t\t/* Include here so we can use pre-defined int types to correctly size constant data arrays. */\n" );
666 fprintf( header, "#ifdef tolower\n" );
667 fprintf( header, "# undef tolower\n" );
668 fprintf( header, "#endif\n" );
669 fprintf( header, "#ifdef toupper\n" );
670 fprintf( header, "# undef toupper\n" );
671 fprintf( header, "#endif\n" );
672 fprintf( header, "#ifdef islower\n" );
673 fprintf( header, "# undef islower\n" );
674 fprintf( header, "#endif\n" );
675 fprintf( header, "#ifdef isupper\n" );
676 fprintf( header, "# undef isupper\n" );
677 fprintf( header, "#endif\n" );
678 fprintf( header, "#ifdef isalpha\n" );
679 fprintf( header, "# undef isalpha\n" );
680 fprintf( header, "#endif\n" );
681 fprintf( header, "#ifdef isdigit\n" );
682 fprintf( header, "# undef isdigit\n" );
683 fprintf( header, "#endif\n" );
684 fprintf( header, "#ifdef isalnum\n" );
685 fprintf( header, "# undef isalnum\n" );
686 fprintf( header, "#endif\n" );
687 fprintf( header, "#ifdef isspace\n" );
688 fprintf( header, "# undef isspace\n" );
689 fprintf( header, "#endif\n" );
690 fprintf( header, "#ifdef ispunct\n" );
691 fprintf( header, "# undef ispunct\n" );
692 fprintf( header, "#endif\n" );
693 fprintf( header, "#ifdef ishexdigit\n" );
694 fprintf( header, "# undef ishexdigit\n" );
695 fprintf( header, "#endif\n\n" );
696
697 fprintf( header, "#define ____L 0x%0x\n", _LOWER );
698 fprintf( header, "#define ____U 0x%0x\n", _UPPER );
699 fprintf( header, "#define ____TITLE 0x%0x\n", _TITLE );
700 fprintf( header, "#define ____D 0x%0x\n", _DIGIT );
701 fprintf( header, "#define ____S 0x%0x\n", _SPACE );
702 fprintf( header, "#define ____P 0x%0x\n", _PUNCT );
703 fprintf( header, "#define ____X 0x%0x\n", _HEX );
704 fprintf( header, "#define ____ZW 0x%0x\n", _ZEROWIDTH );
705 fprintf( header, "#define ____L2R 0x%0x\n", _LEFT_2_RIGHT );
706 fprintf( header, "#define ____R2L 0x%0x\n", _RIGHT_2_LEFT );
707 fprintf( header, "#define ____ENUM 0x%0x\n", _ENUMERIC );
708 fprintf( header, "#define ____ANUM 0x%0x\n", _ANUMERIC );
709 fprintf( header, "#define ____ENS 0x%0x\n", _ENS );
710 fprintf( header, "#define ____CS 0x%0x\n", _CS );
711 fprintf( header, "#define ____ENT 0x%0x\n", _ENT );
712 fprintf( header, "#define ____COMBINE 0x%0x\n", _COMBINING );
713 fprintf( header, "#define ____BB 0x%0x\n", _BREAKBEFOREOK );
714 fprintf( header, "#define ____BA 0x%0x\n", _BREAKAFTEROK );
715 fprintf( header, "#define ____NS 0x%0x\n", _NONSTART );
716 fprintf( header, "#define ____NE 0x%0x\n", _NONEND );
717 fprintf( header, "#define ____UB 0x%0x\n", _URLBREAKAFTER );
718 fprintf( header, "#define ____NB 0x%0x\n", _NOBREAK );
719 fprintf( header, "#define ____AL 0x%0x\n", _ALPHABETIC );
720 fprintf( header, "#define ____ID 0x%0x\n", _IDEOGRAPHIC );
721 fprintf( header, "#define ____INITIAL 0x%0x\n", _INITIAL );
722 fprintf( header, "#define ____MEDIAL 0x%0x\n", _MEDIAL );
723 fprintf( header, "#define ____FINAL 0x%0x\n", _FINAL );
724 fprintf( header, "#define ____ISOLATED 0x%0x\n", _ISOLATED );
725 fprintf( header, "#define ____DECOMPNORM 0x%0x\n", _DecompositionNormative );
726 fprintf( header, "\n" );
727
728 fprintf( header, "extern const unsigned short ____tolower[];\n" );
729 fprintf( header, "extern const unsigned short ____toupper[];\n" );
730 fprintf( header, "extern const unsigned short ____totitle[];\n" );
731 fprintf( header, "extern const unsigned short ____tomirror[];\n" );
732 fprintf( header, "extern const unsigned char ____digitval[];\n" );
733 fprintf( header, "extern const uint32 ____utype[];\n\n" );
734
735 fprintf( header, "#define ____COMBININGCLASS\t0x%0x\n", _CombiningClass );
736 fprintf( header, "#define ____ABOVE\t\t0x%0x\n", _Above );
737 fprintf( header, "#define ____BELOW\t\t0x%0x\n", _Below );
738 fprintf( header, "#define ____OVERSTRIKE\t\t0x%0x\n", _Overstrike );
739 fprintf( header, "#define ____LEFT\t\t0x%0x\n", _Left );
740 fprintf( header, "#define ____RIGHT\t\t0x%0x\n", _Right );
741 fprintf( header, "#define ____JOINS2\t\t0x%0x\n", _Joins2 );
742 fprintf( header, "#define ____CENTERLEFT\t\t0x%0x\n", _CenterLeft );
743 fprintf( header, "#define ____CENTERRIGHT\t\t0x%0x\n", _CenterRight );
744 fprintf( header, "#define ____CENTEREDOUTSIDE\t0x%0x\n", _CenteredOutside );
745 fprintf( header, "#define ____OUTSIDE\t\t0x%0x\n", _Outside );
746 fprintf( header, "#define ____LEFTEDGE\t\t0x%0x\n", _LeftEdge );
747 fprintf( header, "#define ____RIGHTEDGE\t\t0x%0x\n", _RightEdge );
748 fprintf( header, "#define ____TOUCHING\t\t0x%0x\n", _Touching );
749 fprintf( header, "#define ____COMBININGPOSMASK\t0x%0x\n",
750 _Outside|_CenteredOutside|_CenterRight|_CenterLeft|_Joins2|
751 _Right|_Left|_Overstrike|_Below|_Above|_RightEdge|_LeftEdge|
752 _Touching );
753 fprintf( header, "#define ____NOPOSDATAGIVEN\t(uint32)(-1)\t/* -1 == no position data given */\n\n" );
754
755 fprintf( header, "#define combiningclass(ch)\t(____utype2[(ch)+1]&____COMBININGCLASS)\n" );
756 fprintf( header, "#define combiningposmask(ch)\t(____utype2[(ch)+1]&____COMBININGPOSMASK)\n\n" );
757
758 fprintf( header, "extern const uint32\t____utype2[];\t\t\t/* hold position boolean flags for each Unicode.org defined character */\n\n" );
759
760 fprintf( header, "#define isunicodepointassigned(ch) (____codepointassigned[(ch)/32]&(1<<((ch)%%32)))\n\n" );
761
762 fprintf( header, "extern const uint32\t____codepointassigned[];\t/* 1bit_boolean_flag x 32 = exists in Unicode.org character chart list. */\n\n" );
763
764 fprintf( header, "#define tolower(ch) (____tolower[(ch)+1])\n" );
765 fprintf( header, "#define toupper(ch) (____toupper[(ch)+1])\n" );
766 fprintf( header, "#define totitle(ch) (____totitle[(ch)+1])\n" );
767 fprintf( header, "#define tomirror(ch) (____tomirror[(ch)+1])\n" );
768 fprintf( header, "#define tovalue(ch) (____digitval[(ch)+1])\n" );
769 fprintf( header, "#define islower(ch) (____utype[(ch)+1]&____L)\n" );
770 fprintf( header, "#define isupper(ch) (____utype[(ch)+1]&____U)\n" );
771 fprintf( header, "#define istitle(ch) (____utype[(ch)+1]&____TITLE)\n" );
772 fprintf( header, "#define isalpha(ch) (____utype[(ch)+1]&(____L|____U|____TITLE|____AL))\n" );
773 fprintf( header, "#define isdigit(ch) (____utype[(ch)+1]&____D)\n" );
774 fprintf( header, "#define isalnum(ch) (____utype[(ch)+1]&(____L|____U|____TITLE|____AL|____D))\n" );
775 fprintf( header, "#define isideographic(ch) (____utype[(ch)+1]&____ID)\n" );
776 fprintf( header, "#define isideoalpha(ch) (____utype[(ch)+1]&(____ID|____L|____U|____TITLE|____AL))\n" );
777 fprintf( header, "#define isspace(ch) (____utype[(ch)+1]&____S)\n" );
778 fprintf( header, "#define ispunct(ch) (____utype[(ch)+1]&____P)\n" );
779 fprintf( header, "#define ishexdigit(ch) (____utype[(ch)+1]&____X)\n" );
780 fprintf( header, "#define iszerowidth(ch) (____utype[(ch)+1]&____ZW)\n" );
781 fprintf( header, "#define islefttoright(ch) (____utype[(ch)+1]&____L2R)\n" );
782 fprintf( header, "#define isrighttoleft(ch) (____utype[(ch)+1]&____R2L)\n" );
783 fprintf( header, "#define iseuronumeric(ch) (____utype[(ch)+1]&____ENUM)\n" );
784 fprintf( header, "#define isarabnumeric(ch) (____utype[(ch)+1]&____ANUM)\n" );
785 fprintf( header, "#define iseuronumsep(ch) (____utype[(ch)+1]&____ENS)\n" );
786 fprintf( header, "#define iscommonsep(ch) (____utype[(ch)+1]&____CS)\n" );
787 fprintf( header, "#define iseuronumterm(ch) (____utype[(ch)+1]&____ENT)\n" );
788 fprintf( header, "#define iscombining(ch) (____utype[(ch)+1]&____COMBINE)\n" );
789 fprintf( header, "#define isbreakbetweenok(ch1,ch2) (((____utype[(ch1)+1]&____BA) && !(____utype[(ch2)+1]&____NS)) || ((____utype[(ch2)+1]&____BB) && !(____utype[(ch1)+1]&____NE)) || (!(____utype[(ch2)+1]&____D) && ch1=='/'))\n" );
790 fprintf( header, "#define isnobreak(ch) (____utype[(ch)+1]&____NB)\n" );
791 fprintf( header, "#define isarabinitial(ch) (____utype[(ch)+1]&____INITIAL)\n" );
792 fprintf( header, "#define isarabmedial(ch) (____utype[(ch)+1]&____MEDIAL)\n" );
793 fprintf( header, "#define isarabfinal(ch) (____utype[(ch)+1]&____FINAL)\n" );
794 fprintf( header, "#define isarabisolated(ch) (____utype[(ch)+1]&____ISOLATED)\n\n" );
795 fprintf( header, "#define isdecompositionnormative(ch) (____utype[(ch)+1]&____DECOMPNORM)\n\n" );
796
797 fprintf( header, "\n" );
798
799 fprintf( data, "#include \"utype.h\"\n" );
800 fprintf( data, GeneratedFileMessage );
801 fprintf( data, "const unsigned short ____tolower[]= { 0,\n" );
802 for ( i=0; i<MAXC; i+=j ) {
803 fprintf( data, " " );
804 for ( j=0; j<8 && i+j<MAXC-1; ++j )
805 fprintf(data, " 0x%04x,", mytolower[i+j]);
806 if ( i+j==MAXC-1 ) {
807 fprintf(data, " 0x%04x\n};\n\n", mytolower[i+j]);
808 break;
809 } else
810 if ( (i & 63)==0 )
811 fprintf( data, "\t/* 0x%04x */\n",i);
812 else
813 fprintf( data, "\n");
814 }
815 fprintf( data, "const unsigned short ____toupper[] = { 0,\n" );
816 for ( i=0; i<MAXC; i+=j ) {
817 fprintf( data, " " );
818 for ( j=0; j<8 && i+j<MAXC-1; ++j )
819 fprintf(data, " 0x%04x,", mytoupper[i+j]);
820 if ( i+j==MAXC-1 ) {
821 fprintf(data, " 0x%04x\n};\n\n", mytoupper[i+j]);
822 break;
823 } else
824 if ( (i & 63)==0 )
825 fprintf( data, "\t/* 0x%04x */\n",i);
826 else
827 fprintf( data, "\n");
828 }
829 fprintf( data, "const unsigned short ____totitle[] = { 0,\n" );
830 for ( i=0; i<MAXC; i+=j ) {
831 fprintf( data, " " );
832 for ( j=0; j<8 && i+j<MAXC-1; ++j )
833 fprintf(data, " 0x%04x,", mytotitle[i+j]);
834 if ( i+j==MAXC-1 ) {
835 fprintf(data, " 0x%04x\n};\n\n", mytotitle[i+j]);
836 break;
837 } else
838 if ( (i & 63)==0 )
839 fprintf( data, "\t/* 0x%04x */\n",i);
840 else
841 fprintf( data, "\n");
842 }
843 fprintf( data, "const unsigned short ____tomirror[] = { 0,\n" );
844 for ( i=0; i<MAXC; i+=j ) {
845 fprintf( data, " " );
846 for ( j=0; j<8 && i+j<MAXC-1; ++j )
847 fprintf(data, " 0x%04x,", mymirror[i+j]);
848 if ( i+j==MAXC-1 ) {
849 fprintf(data, " 0x%04x\n};\n\n", mymirror[i+j]);
850 break;
851 } else
852 if ( (i & 63)==0 )
853 fprintf( data, "\t/* 0x%04x */\n",i);
854 else
855 fprintf( data, "\n");
856 }
857 fprintf( data, "const unsigned char ____digitval[] = { 0,\n" );
858 for ( i=0; i<MAXC; i+=j ) {
859 fprintf( data, " " );
860 for ( j=0; j<8 && i+j<MAXC-1; ++j )
861 fprintf(data, " 0x%02x,", mynumericvalue[i+j]);
862 if ( i+j==MAXC-1 ) {
863 fprintf(data, " 0x%02x\n};\n\n", mynumericvalue[i+j]);
864 break;
865 } else
866 if ( (i & 63)==0 )
867 fprintf( data, "\t/* 0x%04x */\n",i);
868 else
869 fprintf( data, "\n");
870 }
871 fprintf( data, "const uint32 ____utype[] = { 0,\n" );
872 for ( i=0; i<MAXC; i+=j ) {
873 fprintf( data, " " );
874 for ( j=0; j<8 && i+j<MAXC-1; ++j )
875 fprintf(data, " 0x%08x,", flags[i+j]);
876 if ( i+j==MAXC-1 ) {
877 fprintf(data, " 0x%08x\n};\n\n", flags[i+j]);
878 break;
879 } else
880 if ( (i & 63)==0 )
881 fprintf( data, "\t/* 0x%04x */\n",i);
882 else
883 fprintf( data, "\n");
884 }
885 fprintf( data, "const uint32 ____utype2[] = { 0,\n" );
886 fprintf( data, " /* binary flags used for physical layout of each unicode.org character */\n" );
887 for ( i=0; i<MAXC; i+=j ) {
888 fprintf( data, " " );
889 for ( j=0; j<8 && i+j<MAXC-1; ++j )
890 fprintf(data, " 0x%08x,", flags2[i+j]);
891 if ( i+j==MAXC-1 ) {
892 fprintf(data, " 0x%08x\n};\n\n", flags2[i+j]);
893 break;
894 } else
895 if ( (i & 63)==0 )
896 fprintf( data, "\t/* 0x%04x */\n",i);
897 else
898 fprintf( data, "\n");
899 }
900
901 fprintf( data, "const uint32 ____codepointassigned[] = {\n" );
902 fprintf( data, " /* 32 unicode.org characters represented for each data value in array */\n" );
903 for ( i=0; i<0x120000/32; i+=j ) {
904 fprintf( data, " " );
905 for ( j=0; j<8 && i+j<0x120000/32-1; ++j )
906 fprintf(data, " 0x%08x,", assignedcodepoints[i+j]);
907 if ( i+j==0x120000/32-1 ) {
908 fprintf(data, " 0x%08x\n};\n\n", assignedcodepoints[i+j]);
909 break;
910 } else
911 if ( (i & 63)==0 )
912 fprintf( data, "\t/* 0x%04x */\n",i);
913 else
914 fprintf( data, "\n");
915 }
916
917 fclose( data );
918
919 #if 0
920 data = fopen( "uninames.c", "w");
921 if ( data==NULL ) {
922 fprintf( stderr, CantSaveFile, "uninames.c" );
923 FreeNamesMemorySpace();
924 exit( 1 );
925 }
926
927 fprintf( data, "#include <stdio.h>\n" );
928 fprintf( data, "#include <utype.h>\n" );
929 fprintf( data, GeneratedFileMessage );
930 for ( i=0; i<MAXC; ++i ) if ( names[i]!=NULL && *names[i]!='<' ) {
931 fprintf( data, "static const unsigned short _%04x[] = { ", i );
932 for ( j=0; names[i][j]!='\0'; ++j )
933 fprintf( data, "'%c', ", names[i][j]);
934 fprintf( data, "0 };\n" );
935 }
936 for ( i=0; i<MAXC; i+=256 ) { int any=0;
937 for ( j=0; j<256 && !any; ++j )
938 if ( names[i+j]!=NULL && *names[i+j]!='<' ) any = 1;
939 if ( any ) {
940 fprintf( data, "static const unsigned short * const __%04x[] = {\n", i );
941 for ( j=0; j<256; ++j )
942 if ( names[i+j]!=NULL && *names[i+j]!='<' )
943 fprintf( data, "\t_%04x%s\n", i+j, j==255?"":"," );
944 else
945 fprintf( data, "\tNULL%s\n", j==255?"":"," );
946 fprintf(data, "};\n" );
947 }
948 }
949 fprintf( data, "static const unsigned short *const __NULLALL[] = {\n" );
950 for ( j=0; j<255; ++j ) fprintf( data, "\tNULL,\n" );
951 fprintf( data, "\tNULL\n};\n" );
952 fprintf( data, "\nunsigned const short *const*const UnicodeCharacterNames[] = {\n" );
953 for ( i=0; i<MAXC; i+=256 ) { int any=0;
954 for ( j=0; j<256 && !any; ++j )
955 if ( names[i+j]!=NULL && *names[i+j]!='<' ) any = 1;
956 if ( any )
957 fprintf( data, "\t__%04x%s\n", i, i==0xff00?"":"," );
958 else
959 fprintf( data, "\t__NULLALL%s\n", i==0xff00?"":"," );
960 }
961 fprintf( data, "};\n" );
962 fclose( data );
963
964 fprintf( header, "\nextern const unsigned short *const *const UnicodeCharacterNames[];\n" );
965 fprintf( header, "\t/* An array of 256 arrays of 256 strings of 2byte unicode chars */\n" );
966 fprintf( header, "\t/* containing the names of all non-algorithmically named unicode chars */\n" );
967 #endif
968
969 dumparabicdata(header);
970 fprintf( header, "\n#define _SOFT_HYPHEN\t0xad\n" );
971 fprintf( header, "\n#define _DOUBLE_S\t0xdf\n" );
972 fprintf( header, "\n#endif\n" );
973 fclose( header );
974 }
975
976 static int AnyAlts(int i ) {
977 int j;
978
979 for ( j=(i<<8); j< (i<<8)+0x100; ++j )
980 if ( alts[j][0]!=0 )
981 return( 1 );
982
983 return( 0 );
984 }
985
986 static void dump_alttable() {
987 int i,j;
988 FILE *file;
989
990 if (( file = fopen("unialt.c","w" ))==NULL ) {
991 fprintf(stderr, CantSaveFile, "unialt.c" );
992 return;
993 }
994
995 fprintf(file, "#include <chardata.h>\n" );
996
997 fprintf(file, GeneratedFileMessage );
998
999 for ( i=32; i<MAXC; ++i ) {
1000 if ( alts[i][0]!=0 ) {
1001 fprintf( file, "static const unichar_t str_%x[] = { 0x%04x, ", i, alts[i][0] );
1002 for ( j=1; j<MAXA && alts[i][j]!=0; ++j )
1003 fprintf( file, "0x%04x, ", alts[i][j] );
1004 fprintf( file, "0 };\n" );
1005 }
1006 }
1007 fprintf( file, "\n" );
1008
1009 fprintf( file, "static const unichar_t *const up_allzeros[256] = { NULL };\n\n" );
1010
1011 for ( i=32; i<0xffff; ++i ) {
1012 if ( alts[i][0]!=0 ) {
1013 fprintf( file, "static const unichar_t * const tab_%x[] = {\n", i>>8 );
1014 for ( j=(i&0xff00); j<=(i&0xff00)+0xff; ++j ) {
1015 if ( alts[j][0]==0 )
1016 fprintf( file, "0, " );
1017 else
1018 fprintf( file, "str_%x,\n", j );
1019 }
1020 fprintf(file, "0};\n\n" );
1021 i = j;
1022 }
1023 }
1024
1025 fprintf( file, "const unichar_t *const * const unicode_alternates[] = {\n" );
1026 for ( i=0; i<=0xff; ++i ) {
1027 if ( AnyAlts(i) )
1028 fprintf(file, "tab_%x,\n", i );
1029 else
1030 fprintf(file, "up_allzeros, " );
1031 }
1032 fprintf(file, "0};\n" );
1033 fclose(file);
1034
1035 #if 0 /* moved to dump.c to make build process easier */
1036 /* Now only one thing touches chardata. Even though it's our entry */
1037 /* best if done elsewhere */
1038 file = fopen("chardata.h","a");
1039 fprintf( file, GeneratedFileMessage );
1040 fprintf( file,"\nextern const unichar_t *const * const unicode_alternates[];\n" );
1041 fclose(file);
1042 #endif
1043 }
1044
1045 static void visualalts(void) {
1046
1047 /* These non-normative decompositions allow display algorithems to */
1048 /* pick something that looks right, even if the character doesn't mean */
1049 /* what it should. For example Alpha LOOKS LIKE A so if we don't have */
1050 /* an Alpha character available we can draw it with an A. But this decomp */
1051 /* is not normative and should not be used for ordering purposes */
1052
1053 /* ligatures */
1054 alts[0x152][0] = 'O'; alts[0x152][1] = 'E';
1055 alts[0x153][0] = 'o'; alts[0x153][1] = 'e';
1056 /* I don't bother with AE, ae because they are in latin1 and so common */
1057
1058 /* Things which look alike to my eyes */
1059 alts[0x110][0] = 0xD0; alts[0x110][1] = '\0';
1060 alts[0x138][0] = 0x3ba;
1061 alts[0x182][0] = 0x402;
1062 alts[0x189][0] = 0xD0;
1063 alts[0x19e][0] = 0x3b7;
1064 alts[0x19f][0] = 0x398;
1065 alts[0x1a9][0] = 0x3a3;
1066 alts[0x1c0][0] = '|';
1067 alts[0x1c1][0] = '|'; alts[0x1c1][1] = '|';
1068
1069 alts[0x269][0] = 0x3b9;
1070
1071 alts[0x278][0] = 0x3a6; /* IPA */
1072 alts[0x299][0] = 0x432;
1073 alts[0x292][0] = 0x1b7;
1074 alts[0x29c][0] = 0x43d;
1075
1076 alts[0x2b9][0] = '\'';
1077 alts[0x2ba][0] = '"';
1078 alts[0x2bc][0] = '\'';
1079 alts[0x2c4][0] = '^';
1080 alts[0x2c6][0] = '^';
1081 alts[0x2c8][0] = '\'';
1082 alts[0x2dc][0] = '~';
1083 alts[0x2e0][0] = 0x263;
1084 alts[0x2e1][0] = 'l';
1085 alts[0x2e2][0] = 's';
1086 alts[0x2e3][0] = 'x';
1087 alts[0x2e4][0] = 0x2e4;
1088
1089 alts[0x301][0] = 0xb4;
1090 alts[0x302][0] = '^';
1091 alts[0x303][0] = '~';
1092 alts[0x308][0] = 0xa8;
1093 alts[0x30a][0] = 0xb0;
1094 alts[0x30b][0] = '"';
1095 alts[0x30e][0] = '"';
1096 alts[0x327][0] = 0xb8;
1097
1098 alts[0x374][0] = '\''; /* Greek */
1099 alts[0x375][0] = 0x2cf;
1100 alts[0x37a][0] = 0x345;
1101 alts[0x37e][0] = ';';
1102 alts[0x391][0] = 'A';
1103 alts[0x392][0] = 'B';
1104 alts[0x393][0] = 0x413;
1105 alts[0x395][0] = 'E';
1106 alts[0x396][0] = 'Z';
1107 alts[0x397][0] = 'H';
1108 alts[0x399][0] = 'I';
1109 alts[0x39a][0] = 'K';
1110 alts[0x39c][0] = 'M';
1111 alts[0x39d][0] = 'N';
1112 alts[0x39f][0] = 'O';
1113 alts[0x3A1][0] = 'P';
1114 alts[0x3A4][0] = 'T';
1115 alts[0x3A5][0] = 'Y';
1116 alts[0x3A7][0] = 'X';
1117 alts[0x3ba][0] = 0x138;
1118 alts[0x3bf][0] = 'o';
1119 alts[0x3c1][0] = 'p';
1120 alts[0x3c7][0] = 'x';
1121
1122 alts[0x405][0] = 'S'; /* Cyrillic */
1123 alts[0x406][0] = 'I';
1124 alts[0x408][0] = 'J';
1125 alts[0x410][0] = 'A';
1126 alts[0x412][0] = 'B';
1127 alts[0x413][0] = 0x393;
1128 alts[0x415][0] = 'E';
1129 alts[0x41a][0] = 'K';
1130 alts[0x41c][0] = 'M';
1131 alts[0x41d][0] = 'H';
1132 alts[0x41e][0] = 'O';
1133 alts[0x41f][0] = 0x3a0;
1134 alts[0x420][0] = 'P';
1135 alts[0x421][0] = 'C';
1136 alts[0x422][0] = 'T';
1137 alts[0x424][0] = 0x3a6;
1138 alts[0x425][0] = 'X';
1139 alts[0x430][0] = 'a';
1140 alts[0x435][0] = 'e';
1141 alts[0x43a][0] = 0x3ba;
1142 alts[0x43e][0] = 'o';
1143 alts[0x43f][0] = 0x3c0; /* Not quite right, but close */
1144 alts[0x440][0] = 'p';
1145 alts[0x441][0] = 'c';
1146 alts[0x443][0] = 'y';
1147 alts[0x445][0] = 'x';
1148 alts[0x455][0] = 's';
1149 alts[0x456][0] = 'i';
1150 alts[0x458][0] = 'j';
1151
1152 alts[0x470][0] = 0x3a8; /* extended Cyrillic */
1153 alts[0x471][0] = 0x3c8;
1154 alts[0x4ae][0] = 'Y';
1155 alts[0x4c0][0] = 'I';
1156 alts[0x4d4][0] = 0xc6;
1157 alts[0x4d5][0] = 0xe6;
1158 alts[0x4e0][0] = 0x1b7;
1159 alts[0x4e1][0] = 0x292;
1160 alts[0x4e8][0] = 0x398;
1161 alts[0x4e9][0] = 0x3b8;
1162
1163 alts[0x54f][0] = 'S'; /* Armenian */
1164 alts[0x555][0] = 'O';
1165 alts[0x570][0] = 0x26f;
1166 alts[0x570][0] = 'h';
1167 alts[0x578][0] = 'n';
1168 alts[0x57a][0] = 0x270;
1169 alts[0x57d][0] = 'u';
1170 alts[0x581][0] = 0x261;
1171 alts[0x582][0] = 0x269;
1172 alts[0x584][0] = 'f';
1173 alts[0x585][0] = 'o';
1174 alts[0x589][0] = ':';
1175
1176 alts[0x5f0][0] = 0x5d5; alts[0x5f0][1] = 0x5d5; /* Yiddish ligs */
1177 alts[0x5f1][0] = 0x5d5; alts[0x5f1][1] = 0x5d9; /* 0x5d9 should be drawn first (r to l) */
1178 alts[0x5f2][0] = 0x5d9; alts[0x5f2][1] = 0x5d9;
1179
1180 alts[0x60c][0] = 0x2018; /* Arabic */
1181 alts[0x66a][0] = '%';
1182 alts[0x66c][0] = ',';
1183 alts[0x66d][0] = 0x22c6;
1184 alts[0x6d4][0] = 0xb7;
1185
1186 /* Many of the Korean Jamo are ligatures of other Jamo */
1187 /* 0x110b often, but not always, rides underneath (0x1135 it's left) */
1188 /* Chosung */
1189 alts[0x1101][0] = 0x1100; alts[0x1101][1] = 0x1100;
1190 alts[0x1104][0] = 0x1103; alts[0x1104][1] = 0x1103;
1191 alts[0x1108][0] = 0x1107; alts[0x1108][1] = 0x1107;
1192 alts[0x110a][0] = 0x1109; alts[0x110a][1] = 0x1109;
1193 alts[0x110d][0] = 0x110c; alts[0x110d][1] = 0x110c;
1194 alts[0x1113][0] = 0x1102; alts[0x1113][1] = 0x1100;
1195 alts[0x1114][0] = 0x1102; alts[0x1114][1] = 0x1102;
1196 alts[0x1115][0] = 0x1102; alts[0x1115][1] = 0x1103;
1197 alts[0x1116][0] = 0x1102; alts[0x1116][1] = 0x1107;
1198 alts[0x1117][0] = 0x1103; alts[0x1117][1] = 0x1100;
1199 alts[0x1118][0] = 0x1105; alts[0x1118][1] = 0x1102;
1200 alts[0x1119][0] = 0x1105; alts[0x1119][1] = 0x1105;
1201 alts[0x111a][0] = 0x1105; alts[0x111a][1] = 0x1112;
1202 alts[0x111b][0] = 0x1105; alts[0x111b][1] = 0x110b;
1203 alts[0x111c][0] = 0x1106; alts[0x111c][1] = 0x1107;
1204 alts[0x111d][0] = 0x1106; alts[0x111d][1] = 0x110b;
1205 alts[0x111e][0] = 0x1107; alts[0x111e][1] = 0x1100;
1206 alts[0x111f][0] = 0x1107; alts[0x111f][1] = 0x1102;
1207 alts[0x1120][0] = 0x1107; alts[0x1120][1] = 0x1103;
1208 alts[0x1121][0] = 0x1107; alts[0x1121][1] = 0x1109;
1209 alts[0x1122][0] = 0x1107; alts[0x1122][1] = 0x1109; alts[0x1122][2] = 0x1100;
1210 alts[0x1123][0] = 0x1107; alts[0x1123][1] = 0x1109; alts[0x1123][2] = 0x1103;
1211 alts[0x1124][0] = 0x1107; alts[0x1124][1] = 0x1109; alts[0x1124][2] = 0x1107;
1212 alts[0x1125][0] = 0x1107; alts[0x1125][1] = 0x1109; alts[0x1125][2] = 0x1109;
1213 alts[0x1126][0] = 0x1107; alts[0x1126][1] = 0x1109; alts[0x1126][2] = 0x110c;
1214 alts[0x1127][0] = 0x1107; alts[0x1127][1] = 0x110c;
1215 alts[0x1128][0] = 0x1107; alts[0x1128][1] = 0x110e;
1216 alts[0x1129][0] = 0x1107; alts[0x1129][1] = 0x1110;
1217 alts[0x112a][0] = 0x1107; alts[0x112a][1] = 0x1111;
1218 alts[0x112b][0] = 0x1107; alts[0x112b][1] = 0x110b;
1219 alts[0x112c][0] = 0x1107; alts[0x112c][1] = 0x1107; alts[0x112c][2] = 0x110b;
1220 alts[0x112d][0] = 0x1109; alts[0x112d][1] = 0x1100;
1221 alts[0x112e][0] = 0x1109; alts[0x112e][1] = 0x1102;
1222 alts[0x112f][0] = 0x1109; alts[0x112f][1] = 0x1103;
1223 alts[0x1130][0] = 0x1109; alts[0x1130][1] = 0x1105;
1224 alts[0x1131][0] = 0x1109; alts[0x1131][1] = 0x1106;
1225 alts[0x1132][0] = 0x1109; alts[0x1132][1] = 0x1107;
1226 alts[0x1133][0] = 0x1109; alts[0x1133][1] = 0x1107; alts[0x1133][2] = 0x1100;
1227 alts[0x1134][0] = 0x1109; alts[0x1134][1] = 0x1109; alts[0x1134][2] = 0x1109;
1228 alts[0x1135][0] = 0x1109; alts[0x1135][1] = 0x110b;
1229 alts[0x1136][0] = 0x1109; alts[0x1136][1] = 0x110c;
1230 alts[0x1137][0] = 0x1109; alts[0x1137][1] = 0x110e;
1231 alts[0x1138][0] = 0x1109; alts[0x1138][1] = 0x110f;
1232 alts[0x1139][0] = 0x1109; alts[0x1139][1] = 0x1110;
1233 alts[0x113a][0] = 0x1109; alts[0x113a][1] = 0x1111;
1234 alts[0x113b][0] = 0x1109; alts[0x113b][1] = 0x1112;
1235 alts[0x113d][0] = 0x113c; alts[0x113d][1] = 0x113c;
1236 alts[0x113f][0] = 0x113e; alts[0x113f][1] = 0x113e;
1237 alts[0x1141][0] = 0x110b; alts[0x1141][1] = 0x1100;
1238 alts[0x1142][0] = 0x110b; alts[0x1142][1] = 0x1103;
1239 alts[0x1143][0] = 0x110b; alts[0x1143][1] = 0x1106;
1240 alts[0x1144][0] = 0x110b; alts[0x1144][1] = 0x1107;
1241 alts[0x1145][0] = 0x110b; alts[0x1145][1] = 0x1109;
1242 alts[0x1146][0] = 0x110b; alts[0x1146][1] = 0x1140;
1243 alts[0x1147][0] = 0x110b; alts[0x1147][1] = 0x110b;
1244 alts[0x1148][0] = 0x110b; alts[0x1148][1] = 0x110c;
1245 alts[0x1149][0] = 0x110b; alts[0x1149][1] = 0x110e;
1246 alts[0x114a][0] = 0x110b; alts[0x114a][1] = 0x1110;
1247 alts[0x114b][0] = 0x110b; alts[0x114b][1] = 0x1111;
1248 alts[0x114d][0] = 0x110c; alts[0x114d][1] = 0x110b;
1249 alts[0x114f][0] = 0x114e; alts[0x114f][1] = 0x114e;
1250 alts[0x1151][0] = 0x1150; alts[0x1151][1] = 0x1150;
1251 alts[0x1152][0] = 0x110e; alts[0x1152][1] = 0x110f;
1252 alts[0x1153][0] = 0x110e; alts[0x1153][1] = 0x1112;
1253 alts[0x1156][0] = 0x1111; alts[0x1156][1] = 0x1107;
1254 alts[0x1157][0] = 0x1111; alts[0x1157][1] = 0x110b;
1255 alts[0x1158][0] = 0x1112; alts[0x1158][1] = 0x1112;
1256 /* Jungsung */
1257 alts[0x1162][0] = 0x1161; alts[0x1162][1] = 0x1175;
1258 alts[0x1164][0] = 0x1163; alts[0x1164][1] = 0x1175;
1259 alts[0x1166][0] = 0x1165; alts[0x1166][1] = 0x1175;
1260 alts[0x1168][0] = 0x1167; alts[0x1168][1] = 0x1175;
1261 alts[0x116a][0] = 0x1169; alts[0x116a][1] = 0x1161;
1262 alts[0x116b][0] = 0x1169; alts[0x116b][1] = 0x1162;
1263 alts[0x116c][0] = 0x1169; alts[0x116c][1] = 0x1175;
1264 alts[0x116f][0] = 0x116e; alts[0x116f][1] = 0x1165;
1265 alts[0x1170][0] = 0x116e; alts[0x1170][1] = 0x1166;
1266 alts[0x1171][0] = 0x116e; alts[0x1171][1] = 0x1175;
1267 alts[0x1174][0] = 0x1173; alts[0x1174][1] = 0x1175;
1268 alts[0x1176][0] = 0x1161; alts[0x1176][1] = 0x1169;
1269 alts[0x1177][0] = 0x1161; alts[0x1177][1] = 0x116e;
1270 alts[0x1178][0] = 0x1163; alts[0x1178][1] = 0x1169;
1271 alts[0x1179][0] = 0x1163; alts[0x1179][1] = 0x116d;
1272 alts[0x117a][0] = 0x1165; alts[0x117a][1] = 0x1169;
1273 alts[0x117b][0] = 0x1165; alts[0x117b][1] = 0x116e;
1274 alts[0x117c][0] = 0x1165; alts[0x117c][1] = 0x1173;
1275 alts[0x117d][0] = 0x1167; alts[0x117d][1] = 0x1169;
1276 alts[0x117e][0] = 0x1167; alts[0x117e][1] = 0x116e;
1277 alts[0x117f][0] = 0x1169; alts[0x117f][1] = 0x1165;
1278 alts[0x1180][0] = 0x1169; alts[0x1180][1] = 0x1166;
1279 alts[0x1181][0] = 0x1169; alts[0x1181][1] = 0x1168;
1280 alts[0x1182][0] = 0x1169; alts[0x1182][1] = 0x1169;
1281 alts[0x1183][0] = 0x1169; alts[0x1183][1] = 0x116e;
1282 alts[0x1184][0] = 0x116d; alts[0x1184][1] = 0x1163;
1283 alts[0x1185][0] = 0x116d; alts[0x1185][1] = 0x1164;
1284 alts[0x1186][0] = 0x116d; alts[0x1186][1] = 0x1167;
1285 alts[0x1187][0] = 0x116d; alts[0x1187][1] = 0x1169;
1286 alts[0x1188][0] = 0x116d; alts[0x1188][1] = 0x1175;
1287 alts[0x1189][0] = 0x116e; alts[0x1189][1] = 0x1161;
1288 alts[0x118a][0] = 0x116e; alts[0x118a][1] = 0x1162;
1289 alts[0x118b][0] = 0x116e; alts[0x118b][1] = 0x1165; alts[0x118b][2] = 0x1173;
1290 alts[0x118c][0] = 0x116e; alts[0x118c][1] = 0x1168;
1291 alts[0x118d][0] = 0x116e; alts[0x118d][1] = 0x116e;
1292 alts[0x118e][0] = 0x1172; alts[0x118e][1] = 0x1161;
1293 alts[0x118f][0] = 0x1172; alts[0x118f][1] = 0x1165;
1294 alts[0x1190][0] = 0x1172; alts[0x1190][1] = 0x1166;
1295 alts[0x1191][0] = 0x1172; alts[0x1191][1] = 0x1167;
1296 alts[0x1192][0] = 0x1172; alts[0x1192][1] = 0x1168;
1297 alts[0x1193][0] = 0x1172; alts[0x1193][1] = 0x116e;
1298 alts[0x1194][0] = 0x1172; alts[0x1194][1] = 0x1175;
1299 alts[0x1195][0] = 0x1173; alts[0x1195][1] = 0x116e;
1300 alts[0x1196][0] = 0x1173; alts[0x1196][1] = 0x1173;
1301 alts[0x1197][0] = 0x1174; alts[0x1197][1] = 0x116e;
1302 alts[0x1198][0] = 0x1175; alts[0x1198][1] = 0x1161;
1303 alts[0x1199][0] = 0x1175; alts[0x1199][1] = 0x1163;
1304 alts[0x119a][0] = 0x1175; alts[0x119a][1] = 0x1169;
1305 alts[0x119b][0] = 0x1175; alts[0x119b][1] = 0x116e;
1306 alts[0x119c][0] = 0x1175; alts[0x119c][1] = 0x1173;
1307 alts[0x119d][0] = 0x1175; alts[0x119d][1] = 0x119e;
1308 alts[0x119f][0] = 0x119e; alts[0x119f][1] = 0x1165;
1309 alts[0x11a0][0] = 0x119e; alts[0x11a0][1] = 0x116e;
1310 alts[0x11a1][0] = 0x119e; alts[0x11a1][1] = 0x1175;
1311 alts[0x11a2][0] = 0x119e; alts[0x11a2][1] = 0x119e;
1312 /* Jongsung */
1313 alts[0x11a8][0] = 0x1100;
1314 alts[0x11a9][0] = 0x11a8; alts[0x11a9][1] = 0x11a8;
1315 alts[0x11aa][0] = 0x11a8; alts[0x11aa][1] = 0x11ba;
1316 alts[0x11ab][0] = 0x1102;
1317 alts[0x11ac][0] = 0x11ab; alts[0x11ac][1] = 0x11bd;
1318 alts[0x11ad][0] = 0x11ab; alts[0x11ad][1] = 0x11c2;
1319 alts[0x11ae][0] = 0x1103;
1320 alts[0x11af][0] = 0x1105;
1321 alts[0x11b0][0] = 0x11af; alts[0x11b0][1] = 0x11a8;
1322 alts[0x11b1][0] = 0x11af; alts[0x11b1][1] = 0x11b7;
1323 alts[0x11b2][0] = 0x11af; alts[0x11b2][1] = 0x11b8;
1324 alts[0x11b3][0] = 0x11af; alts[0x11b3][1] = 0x11ba;
1325 alts[0x11b4][0] = 0x11af; alts[0x11b4][1] = 0x11c0;
1326 alts[0x11b5][0] = 0x11af; alts[0x11b5][1] = 0x11c1;
1327 alts[0x11b6][0] = 0x11af; alts[0x11b6][1] = 0x11c2;
1328 alts[0x11b7][0] = 0x1106;
1329 alts[0x11b8][0] = 0x1107;
1330 alts[0x11b9][0] = 0x11b8; alts[0x11b9][1] = 0x11ba;
1331 alts[0x11ba][0] = 0x1109;
1332 alts[0x11bb][0] = 0x11ba; alts[0x11bb][1] = 0x11ba;
1333 alts[0x11bc][0] = 0x110b;
1334 alts[0x11bd][0] = 0x110c;
1335 alts[0x11be][0] = 0x110e;
1336 alts[0x11bf][0] = 0x110f;
1337 alts[0x11c0][0] = 0x1110;
1338 alts[0x11c1][0] = 0x1111;
1339 alts[0x11c2][0] = 0x1112;
1340 alts[0x11c3][0] = 0x11a8; alts[0x11c3][1] = 0x11af;
1341 alts[0x11c4][0] = 0x11a8; alts[0x11c4][1] = 0x11ba; alts[0x11c4][2] = 0x11a8;
1342 alts[0x11c5][0] = 0x11ab; alts[0x11c5][1] = 0x11a8;
1343 alts[0x11c6][0] = 0x11ab; alts[0x11c6][1] = 0x11ae;
1344 alts[0x11c7][0] = 0x11ab; alts[0x11c7][1] = 0x11ba;
1345 alts[0x11c8][0] = 0x11ab; alts[0x11c8][1] = 0x11eb;
1346 alts[0x11c9][0] = 0x11ab; alts[0x11c9][1] = 0x11c0;
1347 alts[0x11ca][0] = 0x11ae; alts[0x11ca][1] = 0x11a8;
1348 alts[0x11cb][0] = 0x11ae; alts[0x11cb][1] = 0x11af;
1349 alts[0x11cc][0] = 0x11af; alts[0x11cc][1] = 0x11a8; alts[0x11cc][2] = 0x11ba;
1350 alts[0x11cd][0] = 0x11af; alts[0x11cd][1] = 0x11ab;
1351 alts[0x11ce][0] = 0x11af; alts[0x11ce][1] = 0x11ae;
1352 alts[0x11cf][0] = 0x11af; alts[0x11cf][1] = 0x11ae; alts[0x11cf][2] = 0x11c2;
1353 alts[0x11d0][0] = 0x11af; alts[0x11d0][1] = 0x11af;
1354 alts[0x11d1][0] = 0x11af; alts[0x11d1][1] = 0x11b7; alts[0x11d1][2] = 0x11a8;
1355 alts[0x11d2][0] = 0x11af; alts[0x11d2][1] = 0x11b7; alts[0x11d2][2] = 0x11ba;
1356 alts[0x11d3][0] = 0x11af; alts[0x11d3][1] = 0x11b8; alts[0x11d3][2] = 0x11ba;
1357 alts[0x11d4][0] = 0x11af; alts[0x11d4][1] = 0x11b8; alts[0x11d4][2] = 0x11c2;
1358 /*alts[0x11d5][0] = 0x11af; alts[0x11d5][1] = 0x11b8; alts[0x11d5][2] = 0x11bc;*/
1359 alts[0x11d5][0] = 0x11af; alts[0x11d5][1] = 0x11e6;
1360 alts[0x11d6][0] = 0x11af; alts[0x11d6][1] = 0x11ba; alts[0x11d6][2] = 0x11ba;
1361 alts[0x11d7][0] = 0x11af; alts[0x11d7][1] = 0x11eb;
1362 alts[0x11d8][0] = 0x11af; alts[0x11d8][1] = 0x11bf;
1363 alts[0x11d9][0] = 0x11af; alts[0x11d9][1] = 0x11f9;
1364 alts[0x11da][0] = 0x11b7; alts[0x11da][1] = 0x11a8;
1365 alts[0x11db][0] = 0x11b7; alts[0x11db][1] = 0x11af;
1366 alts[0x11dc][0] = 0x11b7; alts[0x11dc][1] = 0x11b8;
1367 alts[0x11dd][0] = 0x11b7; alts[0x11dd][1] = 0x11ba;
1368 alts[0x11de][0] = 0x11b7; alts[0x11de][1] = 0x11ba; alts[0x11de][2] = 0x11ba;
1369 alts[0x11df][0] = 0x11b7; alts[0x11df][1] = 0x11eb;
1370 alts[0x11e0][0] = 0x11b7; alts[0x11e0][1] = 0x11be;
1371 alts[0x11e1][0] = 0x11b7; alts[0x11e1][1] = 0x11c2;
1372 alts[0x11e2][0] = 0x11b7; alts[0x11e2][1] = 0x11bc;
1373 alts[0x11e3][0] = 0x11b8; alts[0x11e3][1] = 0x11af;
1374 alts[0x11e4][0] = 0x11b8; alts[0x11e4][1] = 0x11c1;
1375 alts[0x11e5][0] = 0x11b8; alts[0x11e5][1] = 0x11c2;
1376 alts[0x11e6][0] = 0x11b8; alts[0x11e6][1] = 0x11bc;
1377 alts[0x11e7][0] = 0x11ba; alts[0x11e7][1] = 0x11a8;
1378 alts[0x11e8][0] = 0x11ba; alts[0x11e8][1] = 0x11ae;
1379 alts[0x11e9][0] = 0x11ba; alts[0x11e9][1] = 0x11af;
1380 alts[0x11ea][0] = 0x11ba; alts[0x11ea][1] = 0x11b8;
1381 alts[0x11eb][0] = 0x1140;
1382 alts[0x11ec][0] = 0x11bc; alts[0x11ec][1] = 0x11a8;
1383 alts[0x11ed][0] = 0x11bc; alts[0x11ed][1] = 0x11a8; alts[0x11ed][2] = 0x11a8;
1384 alts[0x11ee][0] = 0x11bc; alts[0x11ee][1] = 0x11bc;
1385 alts[0x11ef][0] = 0x11bc; alts[0x11ef][1] = 0x11bf;
1386 alts[0x11f0][0] = 0x114c;
1387 alts[0x11f1][0] = 0x11f0; alts[0x11f1][1] = 0x11ba;
1388 alts[0x11f2][0] = 0x11f0; alts[0x11f2][1] = 0x11eb;
1389 alts[0x11f3][0] = 0x11c1; alts[0x11f3][1] = 0x11b8;
1390 alts[0x11f4][0] = 0x11c1; alts[0x11f4][1] = 0x11bc;
1391 alts[0x11f5][0] = 0x11c2; alts[0x11f5][1] = 0x11ab;
1392 alts[0x11f6][0] = 0x11c2; alts[0x11f6][1] = 0x11af;
1393 alts[0x11f7][0] = 0x11c2; alts[0x11f7][1] = 0x11b7;
1394 alts[0x11f8][0] = 0x11c2; alts[0x11f8][1] = 0x11b8;
1395 alts[0x11f9][0] = 0x1159;
1396
1397 alts[0x13a0][0] = 'D'; /* Cherokee */
1398 alts[0x13a1][0] = 'R';
1399 alts[0x13a2][0] = 'T';
1400 alts[0x13a2][0] = 'T';
1401 alts[0x13a9][0] = 0x423;
1402 alts[0x13aa][0] = 'A';
1403 alts[0x13ab][0] = 'J';
1404 alts[0x13ac][0] = 'E';
1405 alts[0x13b1][0] = 0x393;
1406 alts[0x13b3][0] = 'W';
1407 alts[0x13b7][0] = 'M';
1408 alts[0x13bb][0] = 'H';
1409 alts[0x13be][0] = 0x398;
1410 alts[0x13c0][0] = 'G';
1411 alts[0x13c2][0] = 'h';
1412 alts[0x13c3][0] = 'Z';
1413 alts[0x13cf][0] = 0x42c;
1414 alts[0x13d9][0] = 'V';
1415 alts[0x13da][0] = 'S';
1416 alts[0x13de][0] = 'L';
1417 alts[0x13df][0] = 'C';
1418 alts[0x13e2][0] = 'P';
1419 alts[0x13e6][0] = 'K';
1420 alts[0x13f4][0] = 'B';
1421
1422 alts[0x2000][0] = ' '; /* punctuation */
1423 alts[0x2001][0] = ' ';
1424 alts[0x2010][0] = '-';
1425 alts[0x2011][0] = '-';
1426 alts[0x2012][0] = '-';
1427 alts[0x2013][0] = '-';
1428 alts[0x2014][0] = '-';
1429 alts[0x2015][0] = '-';
1430 alts[0x2016][0] = '|'; alts[0x2016][1] = '|';
1431 alts[0x2018][0] = '`';
1432 alts[0x2019][0] = '\'';
1433 alts[0x201c][0] = '"';
1434 alts[0x201d][0] = '"';
1435 alts[0x2024][0] = '.';
1436 alts[0x2025][0] = '.'; alts[0x2025][1] = '.';
1437 alts[0x2026][0] = '.'; alts[0x2026][1] = '.'; alts[0x2026][2] = '.';
1438 alts[0x2032][0] = '\'';
1439 alts[0x2033][0] = '"';
1440 alts[0x2035][0] = '`';
1441 alts[0x2036][0] = '"';
1442 alts[0x2039][0] = '<';
1443 alts[0x203a][0] = '>';
1444 alts[0x203c][0] = '!'; alts[0x203c][1] = '!';
1445 alts[0x2048][0] = '?'; alts[0x2048][1] = '!';
1446 alts[0x2049][0] = '!'; alts[0x2049][1] = '?';
1447
1448 alts[0x2126][0] = 0x3a9;
1449
1450 alts[0x2205][0] = 0xd8; /* Mathematical operators */
1451 alts[0x2206][0] = 0x394;
1452 alts[0x220f][0] = 0x3a0;
1453 alts[0x2211][0] = 0x3a3;
1454 alts[0x2212][0] = '-';
1455 alts[0x2215][0] = '/';
1456 alts[0x2216][0] = '\\';
1457 alts[0x2217][0] = '*';
1458 alts[0x2218][0] = 0xb0;
1459 alts[0x2219][0] = 0xb7;
1460 alts[0x2223][0] = '|';
1461 alts[0x2225][0] = '|'; alts[0x2225][1] = '|';
1462 alts[0x2236][0] = ':';
1463 alts[0x223c][0] = '~';
1464 alts[0x226a][0] = 0xab;
1465 alts[0x226b][0] = 0xbb;
1466 alts[0x2299][0] = 0x298;
1467 alts[0x22c4][0] = 0x25ca;
1468 alts[0x22c5][0] = 0xb7;
1469 alts[0x22ef][0] = 0xb7; alts[0x22ef][1] = 0xb7; alts[0x22ef][2] = 0xb7;
1470
1471 alts[0x2303][0] = '^'; /* Misc Technical */
1472
1473 alts[0x2373][0] = 0x3b9; /* APL greek */
1474 alts[0x2374][0] = 0x3c1;
1475 alts[0x2375][0] = 0x3c9;
1476 alts[0x237a][0] = 0x3b1;
1477
1478 /* names of control chars */
1479 alts[0x2400][0] = 'N'; alts[0x2400][1] = 'U'; alts[0x2400][2] = 'L';
1480 alts[0x2401][0] = 'S'; alts[0x2401][1] = 'O'; alts[0x2401][2] = 'H';
1481 alts[0x2402][0] = 'S'; alts[0x2402][1] = 'T'; alts[0x2402][2] = 'X';
1482 alts[0x2403][0] = 'E'; alts[0x2403][1] = 'T'; alts[0x2403][2] = 'X';
1483 alts[0x2404][0] = 'E'; alts[0x2404][1] = 'O'; alts[0x2404][2] = 'T';
1484 alts[0x2405][0] = 'E'; alts[0x2405][1] = 'N'; alts[0x2405][2] = 'A';
1485 alts[0x2406][0] = 'A'; alts[0x2406][1] = 'C'; alts[0x2406][2] = 'K';
1486 alts[0x2407][0] = 'B'; alts[0x2407][1] = 'E'; alts[0x2407][2] = 'L';
1487 alts[0x2408][0] = 'B'; alts[0x2408][1] = 'S';
1488 alts[0x2409][0] = 'H'; alts[0x2409][1] = 'T';
1489 alts[0x240A][0] = 'L'; alts[0x240a][1] = 'F';
1490 alts[0x240b][0] = 'V'; alts[0x240b][1] = 'T';
1491 alts[0x240C][0] = 'F'; alts[0x240c][1] = 'F';
1492 alts[0x240d][0] = 'C'; alts[0x240d][1] = 'R';
1493 alts[0x240e][0] = 'S'; alts[0x240e][1] = 'O';
1494 alts[0x240f][0] = 'S'; alts[0x240f][1] = 'I';
1495 alts[0x2410][0] = 'D'; alts[0x2410][1] = 'L'; alts[0x2410][2] = 'E';
1496 alts[0x2411][0] = 'D'; alts[0x2411][1] = 'C'; alts[0x2411][2] = '1';
1497 alts[0x2412][0] = 'D'; alts[0x2412][1] = 'C'; alts[0x2412][2] = '2';
1498 alts[0x2413][0] = 'D'; alts[0x2413][1] = 'C'; alts[0x2413][2] = '3';
1499 alts[0x2414][0] = 'D'; alts[0x2414][1] = 'C'; alts[0x2414][2] = '4';
1500 alts[0x2415][0] = 'N'; alts[0x2415][1] = 'A'; alts[0x2415][2] = 'K';
1501 alts[0x2416][0] = 'S'; alts[0x2416][1] = 'Y'; alts[0x2416][2] = 'N';
1502 alts[0x2417][0] = 'E'; alts[0x2417][1] = 'T'; alts[0x2417][2] = 'B';
1503 alts[0x2418][0] = 'C'; alts[0x2418][1] = 'A'; alts[0x2418][2] = 'N';
1504 alts[0x2419][0] = 'E'; alts[0x2419][1] = 'M';
1505 alts[0x241a][0] = 'S'; alts[0x241a][1] = 'U'; alts[0x241a][2] = 'B';
1506 alts[0x241b][0] = 'E'; alts[0x241b][1] = 'S'; alts[0x241b][2] = 'C';
1507 alts[0x241c][0] = 'F'; alts[0x241c][1] = 'S';
1508 alts[0x241d][0] = 'G'; alts[0x241d][1] = 'S';
1509 alts[0x241e][0] = 'R'; alts[0x241e][1] = 'S';
1510 alts[0x241f][0] = 'U'; alts[0x241f][1] = 'S';
1511 alts[0x2420][0] = 'S'; alts[0x2420][1] = 'P';
1512 alts[0x2421][0] = 'D'; alts[0x2421][1] = 'E'; alts[0x2421][2] = 'L';
1513 alts[0x2422][0] = 0x180;
1514
1515 alts[0x2500][0] = 0x2014;
1516 alts[0x2502][0] = '|';
1517 alts[0x25b3][0] = 0x2206;
1518 alts[0x25b8][0] = 0x2023;
1519 alts[0x25bd][0] = 0x2207;
1520 alts[0x25c7][0] = 0x25ca;
1521 alts[0x25e6][0] = 0xb0;
1522
1523 alts[0x2662][0] = 0x25ca;
1524
1525 alts[0x2731][0] = '*';
1526 alts[0x2758][0] = '|';
1527 alts[0x2762][0] = '!';
1528
1529 alts[0x3001][0] = ','; /* Idiographic symbols */
1530 alts[0x3008][0] = '<';
1531 alts[0x3009][0] = '>';
1532 alts[0x300a][0] = 0xab;
1533 alts[0x300b][0] = 0xbb;
1534
1535 /* The Hangul Compatibility Jamo are just copies of the real Jamo */
1536 /* (different spacing semantics though) */
1537 alts[0x3131][0] = 0x1100;
1538 alts[0x3132][0] = 0x1101;
1539 alts[0x3133][0] = 0x11aa;
1540 alts[0x3134][0] = 0x1102;
1541 alts[0x3135][0] = 0x11ac;
1542 alts[0x3136][0] = 0x11ad;
1543 alts[0x3137][0] = 0x1103;
1544 alts[0x3138][0] = 0x1104;
1545 alts[0x3139][0] = 0x1105;
1546 alts[0x313a][0] = 0x11b0;
1547 alts[0x313b][0] = 0x11b1;
1548 alts[0x313c][0] = 0x11b2;
1549 alts[0x313d][0] = 0x11b3;
1550 alts[0x313e][0] = 0x11b4;
1551 alts[0x313f][0] = 0x11b5;
1552 alts[0x3140][0] = 0x111a;
1553 alts[0x3141][0] = 0x1106;
1554 alts[0x3142][0] = 0x1107;
1555 alts[0x3143][0] = 0x1108;
1556 alts[0x3144][0] = 0x1121;
1557 alts[0x3145][0] = 0x1109;
1558 alts[0x3146][0] = 0x110a;
1559 alts[0x3147][0] = 0x110b;
1560 alts[0x3148][0] = 0x110c;
1561 alts[0x3149][0] = 0x110d;
1562 alts[0x314a][0] = 0x110e;
1563 alts[0x314b][0] = 0x110f;
1564 alts[0x314c][0] = 0x1110;
1565 alts[0x314d][0] = 0x1111;
1566 alts[0x314e][0] = 0x1112;
1567 alts[0x314f][0] = 0x1161;
1568 alts[0x3150][0] = 0x1162;
1569 alts[0x3151][0] = 0x1163;
1570 alts[0x3152][0] = 0x1164;
1571 alts[0x3153][0] = 0x1165;
1572 alts[0x3154][0] = 0x1166;
1573 alts[0x3155][0] = 0x1167;
1574 alts[0x3156][0] = 0x1168;
1575 alts[0x3157][0] = 0x1169;
1576 alts[0x3158][0] = 0x116a;
1577 alts[0x3159][0] = 0x116b;
1578 alts[0x315a][0] = 0x116c;
1579 alts[0x315b][0] = 0x116d;
1580 alts[0x315c][0] = 0x116e;
1581 alts[0x315d][0] = 0x116f;
1582 alts[0x315e][0] = 0x1170;
1583 alts[0x315f][0] = 0x1171;
1584 alts[0x3160][0] = 0x1172;
1585 alts[0x3161][0] = 0x1173;
1586 alts[0x3162][0] = 0x1174;
1587 alts[0x3163][0] = 0x1175;
1588 alts[0x3164][0] = 0x1160;
1589 alts[0x3165][0] = 0x1114;
1590 alts[0x3166][0] = 0x1115;
1591 alts[0x3167][0] = 0x11c7;
1592 alts[0x3168][0] = 0x11c8;
1593 alts[0x3169][0] = 0x11cc;
1594 alts[0x316a][0] = 0x11ce;
1595 alts[0x316b][0] = 0x11d3;
1596 alts[0x316c][0] = 0x11d7;
1597 alts[0x316d][0] = 0x11d9;
1598 alts[0x316e][0] = 0x111c;
1599 alts[0x316f][0] = 0x11dd;
1600 alts[0x3170][0] = 0x11df;
1601 alts[0x3171][0] = 0x111d;
1602 alts[0x3172][0] = 0x111e;
1603 alts[0x3173][0] = 0x1120;
1604 alts[0x3174][0] = 0x1122;
1605 alts[0x3175][0] = 0x1123;
1606 alts[0x3176][0] = 0x1127;
1607 alts[0x3177][0] = 0x1129;
1608 alts[0x3178][0] = 0x112b;
1609 alts[0x3179][0] = 0x112c;
1610 alts[0x317a][0] = 0x112d;
1611 alts[0x317b][0] = 0x112e;
1612 alts[0x317c][0] = 0x112f;
1613 alts[0x317d][0] = 0x1132;
1614 alts[0x317e][0] = 0x1136;
1615 alts[0x317f][0] = 0x1140;
1616 alts[0x3180][0] = 0x1147;
1617 alts[0x3181][0] = 0x114c;
1618 alts[0x3182][0] = 0x11f1;
1619 alts[0x3183][0] = 0x11f2;
1620 alts[0x3184][0] = 0x1157;
1621 alts[0x3185][0] = 0x1158;
1622 alts[0x3186][0] = 0x1159;
1623 alts[0x3187][0] = 0x1184;
1624 alts[0x3188][0] = 0x1185;
1625 alts[0x3189][0] = 0x1188;
1626 alts[0x318a][0] = 0x1191;
1627 alts[0x318b][0] = 0x1192;
1628 alts[0x318c][0] = 0x1194;
1629 alts[0x318d][0] = 0x119e;
1630 alts[0x318e][0] = 0x11a1;
1631
1632 alts[0xff5f][0] = 0x2e28; alts[0x2e28][0] = 0xff5f; /* similar double brackets*/
1633 alts[0xff60][0] = 0x2e29; alts[0x2e29][0] = 0xff60;
1634 }
1635
1636
1637 static void cheat(void) {
1638 #if 0
1639 /* Adobe's private use symbols (some) */
1640 alts[0xf6de][0] = '-'; /* 3/4 em dash */
1641
1642 alts[0xf66d][0] = 0xf761; alts[0xf66d][1] = 0x306; /* A breve */
1643 alts[0xf66e][0] = 0xf761; alts[0xf66e][1] = 0x304; /* A macron */
1644 alts[0xf66f][0] = 0xf761; alts[0xf66f][1] = 0x328; /* A ogonek */
1645 alts[0xf670][0] = 0xf7e6; alts[0xf670][1] = 0x301; /* AE acute */
1646 alts[0xf671][0] = 0xf763; alts[0xf671][1] = 0x301;
1647 alts[0xf672][0] = 0xf763; alts[0xf672][1] = 0x30c; /* C caron */
1648 alts[0xf673][0] = 0xf763; alts[0xf673][1] = 0x302; /* C circumflex */
1649 alts[0xf674][0] = 0xf763; alts[0xf674][1] = 0x307; /* C dot above */
1650 alts[0xf675][0] = 0xf764; alts[0xf675][1] = 0x30c; /* D caron */
1651 alts[0xf677][0] = 0xf765; alts[0xf677][1] = 0x306; /* E breve */
1652 alts[0xf678][0] = 0xf765; alts[0xf678][1] = 0x30c;
1653 alts[0xf679][0] = 0xf765; alts[0xf679][1] = 0x307;
1654 alts[0xf67a][0] = 0xf765; alts[0xf67a][1] = 0x304;
1655 alts[0xf67c][0] = 0xf765; alts[0xf67c][1] = 0x328;
1656 alts[0xf67d][0] = 0xf767; alts[0xf67d][1] = 0x306; /* G breve */
1657 alts[0xf67e][0] = 0xf767; alts[0xf67e][1] = 0x302;
1658 alts[0xf67f][0] = 0xf767; alts[0xf67f][1] = 0x326; /* G comma below */
1659 alts[0xf680][0] = 0xf767; alts[0xf680][1] = 0x307;
1660 alts[0xf682][0] = 0xf768; alts[0xf682][1] = 0x302; /* H circum */
1661 alts[0xf683][0] = 0xf769; alts[0xf683][1] = 0x306; /* I breve */
1662 alts[0xf684][0] = 0xf769; alts[0xf684][1] = 0xf76a; /* I J */
1663 alts[0xf685][0] = 0xf769; alts[0xf685][1] = 0x304;
1664 alts[0xf686][0] = 0xf769; alts[0xf686][1] = 0x328;
1665 alts[0xf687][0] = 0xf769; alts[0xf687][1] = 0x303; /* I tilde */
1666 alts[0xf688][0] = 0xf76a; alts[0xf688][1] = 0x302; /* J circum */
1667 alts[0xf689][0] = 0xf76b; alts[0xf689][1] = 0x326; /* K comma below */
1668 alts[0xf68a][0] = 0xf76c; alts[0xf68a][1] = 0x301; /* L accute */
1669 alts[0xf68b][0] = 0xf76c; alts[0xf68b][1] = 0x30c;
1670 alts[0xf68c][0] = 0xf76c; alts[0xf68c][1] = 0x326;
1671 alts[0xf68d][0] = 0xf76c; alts[0xf68d][1] = 0xB7; /* L middle dot */
1672 alts[0xf68e][0] = 0xf76e; alts[0xf68e][1] = 0x301; /* N accute */
1673 alts[0xf68f][0] = 0xf76e; alts[0xf68f][1] = 0x30c;
1674 alts[0xf690][0] = 0xf76e; alts[0xf690][1] = 0x326;
1675 alts[0xf691][0] = 0xf76f; alts[0xf691][1] = 0x306; /* O breve */
1676 alts[0xf692][0] = 0xf76f; alts[0xf692][1] = 0x30b; /* O double accute */
1677 alts[0xf693][0] = 0xf76f; alts[0xf693][1] = 0x304;
1678 alts[0xf694][0] = 0xf7f8; alts[0xf694][1] = 0x301;
1679 alts[0xf695][0] = 0xf772; alts[0xf695][1] = 0x301; /* R acute */
1680 alts[0xf696][0] = 0xf772; alts[0xf696][1] = 0x30c;
1681 alts[0xf697][0] = 0xf772; alts[0xf697][1] = 0x326;
1682 alts[0xf698][0] = 0xf773; alts[0xf698][1] = 0x301; /* S acute */
1683 alts[0xf699][0] = 0xf773; alts[0xf699][1] = 0x327; /* S cedilla */
1684 alts[0xf69a][0] = 0xf773; alts[0xf69a][1] = 0x302;
1685 alts[0xf69b][0] = 0xf773; alts[0xf69b][1] = 0x326;
1686 alts[0xf69d][0] = 0xf774; alts[0xf69d][1] = 0x30c; /* T caron */
1687 alts[0xf69e][0] = 0xf774; alts[0xf69e][1] = 0x326;
1688 alts[0xf69f][0] = 0xf775; alts[0xf69f][1] = 0x306; /* U breve */
1689 alts[0xf6a0][0] = 0xf775; alts[0xf6a0][1] = 0x30b;
1690 alts[0xf6a1][0] = 0xf775; alts[0xf6a1][1] = 0x304;
1691 alts[0xf6a2][0] = 0xf775; alts[0xf6a2][1] = 0x328;
1692 alts[0xf6a3][0] = 0xf775; alts[0xf6a3][1] = 0x30a; /* U ring */
1693 alts[0xf6a4][0] = 0xf775; alts[0xf6a4][1] = 0x303;
1694 alts[0xf6a5][0] = 0xf777; alts[0xf6a5][1] = 0x301; /* W accute */
1695 alts[0xf6a6][0] = 0xf777; alts[0xf6a6][1] = 0x302;
1696 alts[0xf6a7][0] = 0xf777; alts[0xf6a7][1] = 0x308;
1697 alts[0xf6a8][0] = 0xf777; alts[0xf6a8][1] = 0x300;
1698 alts[0xf6a9][0] = 0xf779; alts[0xf6a9][1] = 0x302;
1699 alts[0xf6aa][0] = 0xf779; alts[0xf6aa][1] = 0x300;
1700 alts[0xf6ab][0] = 0xf77a; alts[0xf6ab][1] = 0x301;
1701 alts[0xf6ac][0] = 0xf77a; alts[0xf6ac][1] = 0x307;
1702 alts[0xf6ad][0] = 0xf769; alts[0xf6ad][1] = 0x307;
1703
1704 alts[0xf6be][0] = 'j';
1705 alts[0xf6bf][0] = 'L'; alts[0xf6bf][1] = 'L';
1706 alts[0xf6c0][0] = 'l'; alts[0xf6c0][1] = 'l';
1707 alts[0xf6c3][0] = 0x313;
1708 alts[0xf6c9][0] = 0x2ca;
1709 alts[0xf6ca][0] = 0x2c7;
1710 alts[0xf6cb][0] = 0xa8;
1711 alts[0xf6cc][0] = 0xa8; alts[0xf6cc][1] = 0x2ca;
1712 alts[0xf6cd][0] = 0xa8; alts[0xf6cd][1] = 0x2cb;
1713 alts[0xf6ce][0] = 0x2cb;
1714 alts[0xf6cf][0] = 0x2dd;
1715 alts[0xf6d0][0] = 0x2c9;
1716 alts[0xf6d1][0] = 0x2d8;
1717 alts[0xf6d3][0] = 0x30f;
1718 alts[0xf6d4][0] = 0x2d8;
1719 alts[0xf6d6][0] = 0x30f;
1720 alts[0xf6d7][0] = 0xa8; alts[0xf6d7][1] = 0x2ca;
1721 alts[0xf6d8][0] = 0xa8; alts[0xf6d8][1] = 0x2cb;
1722 alts[0xf6d9][0] = 0xa9;
1723 alts[0xf6da][0] = 0xae;
1724 alts[0xf6db][0] = 0x2122;
1725 alts[0xf6de][0] = '-';
1726
1727 alts[0xf6f4][0] = 0x2d8;
1728 alts[0xf6f5][0] = 0x2c7;
1729 alts[0xf6f6][0] = 0x2c6;
1730 alts[0xf6f7][0] = 0x2d9;
1731 alts[0xf6f8][0] = 0x2dd;
1732 alts[0xf6fa][0] = 0xf76f; alts[0xf6fa][1] = 0xf765;
1733 alts[0xf6fb][0] = 0xf76f; alts[0xf6fb][1] = 0x328;
1734 alts[0xf6fc][0] = 0xf772; alts[0xf6fc][1] = 0x30a;
1735 alts[0xf6fd][0] = 0xf773; alts[0xf6fd][1] = 0x30c;
1736 alts[0xf6fe][0] = 0xf774; alts[0xf6fe][1] = 0x303;
1737 alts[0xf6ff][0] = 0xf77a; alts[0xf6ff][1] = 0x30c;
1738 alts[0xf760][0] = '`';
1739 alts[0xf7a8][0] = 0xa8;
1740 alts[0xf7af][0] = 0xaf;
1741 alts[0xf7b4][0] = 0xb4;
1742 alts[0xf7b8][0] = 0xb8;
1743 alts[0xf7e0][0] = 0xf761; alts[0xf7e0][1] = 0x300;
1744 alts[0xf7e1][0] = 0xf761; alts[0xf7e1][1] = 0x301;
1745 alts[0xf7e2][0] = 0xf761; alts[0xf7e2][1] = 0x302;
1746 alts[0xf7e3][0] = 0xf761; alts[0xf7e3][1] = 0x303;
1747 alts[0xf7e4][0] = 0xf761; alts[0xf7e4][1] = 0x308;
1748 alts[0xf7e5][0] = 0xf761; alts[0xf7e5][1] = 0x30a;
1749 alts[0xf7e6][0] = 0xf761; alts[0xf7e6][1] = 0xf765;
1750 alts[0xf7e7][0] = 0xf763; alts[0xf7e7][1] = 0x327;
1751 alts[0xf7e8][0] = 0xf765; alts[0xf7e8][1] = 0x300;
1752 alts[0xf7e9][0] = 0xf765; alts[0xf7e9][1] = 0x301;
1753 alts[0xf7ea][0] = 0xf765; alts[0xf7ea][1] = 0x302;
1754 alts[0xf7eb][0] = 0xf765; alts[0xf7eb][1] = 0x308;
1755 alts[0xf7ec][0] = 0xf769; alts[0xf7ec][1] = 0x300;
1756 alts[0xf7ed][0] = 0xf769; alts[0xf7ed][1] = 0x301;
1757 alts[0xf7ee][0] = 0xf769; alts[0xf7ee][1] = 0x302;
1758 alts[0xf7ef][0] = 0xf769; alts[0xf7ef][1] = 0x308;
1759 alts[0xf7f1][0] = 0xf76e; alts[0xf7f1][1] = 0x303;
1760 alts[0xf7f2][0] = 0xf76f; alts[0xf7f2][1] = 0x300;
1761 alts[0xf7f3][0] = 0xf76f; alts[0xf7f3][1] = 0x301;
1762 alts[0xf7f4][0] = 0xf76f; alts[0xf7f4][1] = 0x302;
1763 alts[0xf7f5][0] = 0xf76f; alts[0xf7f5][1] = 0x303;
1764 alts[0xf7f6][0] = 0xf76f; alts[0xf7f6][1] = 0x308;
1765 alts[0xf7f9][0] = 0xf775; alts[0xf7f9][1] = 0x300;
1766 alts[0xf7fa][0] = 0xf775; alts[0xf7fa][1] = 0x301;
1767 alts[0xf7fb][0] = 0xf775; alts[0xf7fb][1] = 0x302;
1768 alts[0xf7fc][0] = 0xf775; alts[0xf7fc][1] = 0x308;
1769
1770 alts[0xf7fd][0] = 0xf779; alts[0xf7fd][1] = 0x301;
1771 alts[0xf7ff][0] = 0xf779; alts[0xf7ff][1] = 0x308;
1772
1773 mytoupper[0xf6be] = 'J'; /* Adobe's dotlessj character */
1774 mytotitle[0xf6be] = 'J';
1775 mytolower[0xf6bf] = 0xf6c0; /* Adobe's LL character */
1776 mytoupper[0xf6c0] = 0xf6bf; /* Adobe's ll character */
1777 mytotitle[0xf6c0] = 0xf6bf;
1778 #endif
1779
1780 mymirror['('] = ')';
1781 mymirror[')'] = '(';
1782 mymirror['>'] = '<';
1783 mymirror['<'] = '>';
1784 mymirror['['] = ']';
1785 mymirror[']'] = '[';
1786 mymirror['{'] = '}';
1787 mymirror['}'] = '{';
1788 mymirror[0xab] = 0xbb; /* double Guillemet */
1789 mymirror[0xbb] = 0xab;
1790 mymirror[0x2039] = 0x203A; /* single Guillemet */
1791 mymirror[0x203A] = 0x2039;
1792 mymirror[0x2045] = 0x2046; /* square bracket with quill */
1793 mymirror[0x2046] = 0x2045;
1794 mymirror[0x207D] = 0x207E; /* superscript paren */
1795 mymirror[0x207E] = 0x207D;
1796 mymirror[0x208D] = 0x208E; /* subscript paren */
1797 mymirror[0x208E] = 0x208D;
1798 /* mathematical symbols not mirrorred!!!! some tech symbols missing too */
1799 /* no code points */
1800 mymirror[0x2308] = 0x2309; /* ceiling */
1801 mymirror[0x2309] = 0x2308;
1802 mymirror[0x230a] = 0x230b; /* floor */
1803 mymirror[0x230b] = 0x230a;
1804 mymirror[0x2329] = 0x232a; /* bra/ket */
1805 mymirror[0x232a] = 0x2329;
1806 mymirror[0x3008] = 0x3009; /* CJK symbols */
1807 mymirror[0x3009] = 0x3008;
1808 mymirror[0x300a] = 0x300b;
1809 mymirror[0x300b] = 0x300a;
1810 mymirror[0x300c] = 0x300d;
1811 mymirror[0x300d] = 0x300c;
1812 mymirror[0x300e] = 0x300f;
1813 mymirror[0x300f] = 0x300e;
1814 mymirror[0x3010] = 0x3011;
1815 mymirror[0x3011] = 0x3010;
1816 mymirror[0x3014] = 0x3015;
1817 mymirror[0x3015] = 0x3014;
1818 mymirror[0x3016] = 0x3017;
1819 mymirror[0x3017] = 0x3016;
1820 mymirror[0x3018] = 0x3019;
1821 mymirror[0x3019] = 0x3018;
1822 mymirror[0x301a] = 0x301b;
1823 mymirror[0x301b] = 0x301a;
1824 }
1825
1826 int main() {
1827 visualalts(); /* pre-populate matrix with visual alternative fonts */
1828 readin(); /* load the "official" Unicode data from unicode.org */
1829 /* Apple's file contains no interesting information that I can see */
1830 /* Adobe's file is interesting, but should only be used conditionally */
1831 /* so apply at a different level */
1832 /* readcorpfile("ADOBE ", "AdobeCorporateuse.txt"); */
1833 cheat(); /* over-ride with these mods after reading input files */
1834 dump(); /* create utype.h, utype.c and ArabicForms.c */
1835 dump_alttable(); /* create unialt.c */
1836 FreeNamesMemorySpace(); /* cleanup alloc of memory */
1837 return( 0 );
1838 }
LuaTeX git mirror