| 1 |
/* Copyright 1994 Pubdic Project. |
| 2 |
* |
| 3 |
* Permission to use, copy, modify, distribute and sell this software |
| 4 |
* and its documentation for any purpose is hereby granted without |
| 5 |
* fee, provided that the above copyright notice appear in all copies |
| 6 |
* and that both that copyright notice and this permission notice |
| 7 |
* appear in supporting documentation, and that the name of Pubdic |
| 8 |
* Project not be used in advertising or publicity pertaining to |
| 9 |
* distribution of the software without specific, written prior |
| 10 |
* permission. Pubdic Project makes no representations about the |
| 11 |
* suitability of this software for any purpose. It is provided "as |
| 12 |
* is" without express or implied warranty. |
| 13 |
* |
| 14 |
* PUBDIC PROJECT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, |
| 15 |
* INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN |
| 16 |
* NO EVENT SHALL PUBDIC PROJECT BE LIABLE FOR ANY SPECIAL, INDIRECT OR |
| 17 |
* CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF |
| 18 |
* USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR |
| 19 |
* OTHER TORTUOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR |
| 20 |
* PERFORMANCE OF THIS SOFTWARE. |
| 21 |
*/ |
| 22 |
|
| 23 |
#ifndef lint |
| 24 |
static char rcsid[] = "$Id: pod.c,v 1.6 2005/11/29 17:00:02 aonoto Exp $"; |
| 25 |
#endif |
| 26 |
|
| 27 |
#ifdef HAVE_CONFIG_H |
| 28 |
# include <config.h> |
| 29 |
#endif |
| 30 |
|
| 31 |
#include <stdio.h> |
| 32 |
#if STDC_HEADERS |
| 33 |
# include <stdlib.h> |
| 34 |
# include <stddef.h> |
| 35 |
# include <string.h> |
| 36 |
#else |
| 37 |
# if HAVE_MALLOC_H |
| 38 |
# include <malloc.h> |
| 39 |
# endif |
| 40 |
# if HAVE_STRINGS_H |
| 41 |
# include <strings.h> |
| 42 |
# endif |
| 43 |
#endif /* STDC_HEADERS */ |
| 44 |
|
| 45 |
#define POD_WCHAR |
| 46 |
#ifdef POD_WCHAR |
| 47 |
typedef unsigned short Wchar; |
| 48 |
#else |
| 49 |
#include <locale.h> |
| 50 |
#include <widec.h> |
| 51 |
#define Wchar wchar_t |
| 52 |
#endif |
| 53 |
|
| 54 |
#if !(HAVE_BZERO) && (HAVE_MEMSET) |
| 55 |
# define bzero(a, c) memset(a, 0, c) |
| 56 |
#endif |
| 57 |
|
| 58 |
static char *program; |
| 59 |
static int compare, ignore_hinshi_to_compare, sort_by_frequency, merge_sj3; |
| 60 |
static int merge_kind, wnn_type_output, canna_type_output, sj3_type_output; |
| 61 |
static int list_kinds; |
| 62 |
static int copy_frequency, extract_kana = 0; |
| 63 |
static long specific_kind; |
| 64 |
static FILE *in1, *in2; |
| 65 |
static char *common_out, *old_out, *new_out, *hinshi_table, *bunrui; |
| 66 |
static char *description_table; |
| 67 |
static int selhinshi = 0; |
| 68 |
|
| 69 |
/* hinshi_direction */ |
| 70 |
#define INORDER 0 |
| 71 |
#define REVERSE 1 |
| 72 |
|
| 73 |
static int hinshi_direction = INORDER; /* see above */ |
| 74 |
|
| 75 |
#define READBUFSIZE 128 |
| 76 |
#define DICBUFSIZE (2 << 13) |
| 77 |
#define DICBUFINDEXMASK (DICBUFSIZE - 1) |
| 78 |
#define HINSHIBUFSIZE (2 << 13) |
| 79 |
#define HINSHIBUFINDEXMASK (HINSHIBUFSIZE - 1) |
| 80 |
|
| 81 |
/* status of intern() */ |
| 82 |
#define FOUND 0 |
| 83 |
#define CREATE 1 |
| 84 |
|
| 85 |
/* ���������������� */ |
| 86 |
|
| 87 |
struct hinshipack |
| 88 |
{ |
| 89 |
int nhinshis; |
| 90 |
Wchar *hinshi; |
| 91 |
unsigned flags; /* see below */ |
| 92 |
struct hinshipack *next; |
| 93 |
}; |
| 94 |
|
| 95 |
/* values of (struct hinshipack.)flags */ |
| 96 |
#define REPLACED 1 |
| 97 |
|
| 98 |
/* ����������������������������������������������(��������) */ |
| 99 |
|
| 100 |
struct descpack |
| 101 |
{ |
| 102 |
Wchar *hinshi, *tandesc, *yomdesc; |
| 103 |
struct descpack *next; |
| 104 |
}; |
| 105 |
|
| 106 |
/* �������������������������������� */ |
| 107 |
|
| 108 |
struct kindpack |
| 109 |
{ |
| 110 |
Wchar *kind; |
| 111 |
long kindbit; |
| 112 |
}; |
| 113 |
|
| 114 |
/* ���������������� */ |
| 115 |
|
| 116 |
struct dicpack |
| 117 |
{ |
| 118 |
Wchar *yomi, *tango; |
| 119 |
struct hinshipack *hinshi; |
| 120 |
int hindo; |
| 121 |
long kind; |
| 122 |
Wchar *extdata; |
| 123 |
unsigned flags; /* SEE BELOW */ |
| 124 |
struct dicpack *next; |
| 125 |
}; |
| 126 |
|
| 127 |
/* values of (struct dicpack.)flags */ |
| 128 |
#define COMMON 001 |
| 129 |
#define NEW 002 |
| 130 |
|
| 131 |
#if defined (__STDC__) || defined (_AIX) || (defined (__mips) && defined (_SYSTYPE_SVR4)) || defined(_WIN32) |
| 132 |
/* Prototype for C89 (or later) */ |
| 133 |
#ifdef POD_WCHAR |
| 134 |
size_t Mbstowcs (Wchar *d, char *ss, int n); |
| 135 |
size_t Wcstombs (char *d, Wchar *s, int n); |
| 136 |
int Wscmp (register Wchar *s1, register Wchar *s2); |
| 137 |
Wchar *Wscpy (Wchar *d, register Wchar *s); |
| 138 |
int Wslen (Wchar *s); |
| 139 |
int Watoi (Wchar *s); |
| 140 |
static void Fputws (Wchar *s, FILE *f); |
| 141 |
Wchar *Fgetws (Wchar *buf, int siz, FILE *f); |
| 142 |
#endif /* POD_WCHAR */ |
| 143 |
|
| 144 |
static int all_kana (Wchar *s); |
| 145 |
static Wchar *findslash (Wchar *s); |
| 146 |
static Wchar *extstr (Wchar *p, Wchar **pp, int *key_return); |
| 147 |
static void malloc_failed (void); |
| 148 |
static struct hinshipack *internhinshi (Wchar *str, int flag); |
| 149 |
static void replace_hinshi (void); |
| 150 |
static void select_hinshi (int n); |
| 151 |
static void freedesc (struct descpack *p); |
| 152 |
static struct descpack *interndesc (Wchar *hin, Wchar *tan, Wchar *yom); |
| 153 |
static struct descpack *searchdesc (Wchar *hin); |
| 154 |
static void store_description (void); |
| 155 |
static long internkind (Wchar *s); |
| 156 |
static void listkinds (void); |
| 157 |
static int kindcompar (struct kindpack *k1, struct kindpack *k2); |
| 158 |
static void sortkind (void); |
| 159 |
static struct dicpack *intern (int key, Wchar *yomi, Wchar *kouho, Wchar *hinshi, int hindo, long kind, int *stat, long flags); |
| 160 |
static void storepd (FILE *file); |
| 161 |
static void comparepd (FILE *file); |
| 162 |
static void canna_output (FILE *cf, struct dicpack *p, Wchar *h, int n); |
| 163 |
static void entry_out (FILE *cf, struct dicpack *p, Wchar *h, int n, Wchar *ex); |
| 164 |
static void printentry (FILE *cf, struct dicpack *p); |
| 165 |
static void showentry (struct dicpack **pd, int n); |
| 166 |
static int diccompar (struct dicpack **p1, struct dicpack **p2); |
| 167 |
static int dichindocompar (struct dicpack **p1, struct dicpack **p2); |
| 168 |
void shrinkargs (char **argv, int n, int count); |
| 169 |
static void parseargs (int argc, char *argv[]); |
| 170 |
#endif |
| 171 |
|
| 172 |
#ifndef POD_WCHAR |
| 173 |
# define Mbstowcs mbstowcs |
| 174 |
# define Wcstombs wcstombs |
| 175 |
# define Wscmp wscmp |
| 176 |
# define Wscpy wscpy |
| 177 |
# define Wslen wslen |
| 178 |
# define Fgetws fgetws |
| 179 |
# define Fputws fputws |
| 180 |
#else |
| 181 |
# define SS2 0x8e |
| 182 |
# define SS3 0x8f |
| 183 |
# define MSB 0x80 |
| 184 |
# define MSK 0x7f |
| 185 |
|
| 186 |
# define WCG0 0x0000 |
| 187 |
# define WCG1 0x8080 |
| 188 |
# define WCG2 0x0080 |
| 189 |
# define WCG3 0x8000 |
| 190 |
# define WCMSK 0x8080 |
| 191 |
|
| 192 |
size_t |
| 193 |
Mbstowcs (d, ss, n) |
| 194 |
Wchar *d; |
| 195 |
char *ss; |
| 196 |
int n; |
| 197 |
{ |
| 198 |
register Wchar *p = d; |
| 199 |
register int ch; |
| 200 |
register unsigned char *s = (unsigned char *) ss; |
| 201 |
|
| 202 |
while ((ch = *s++) && (p - d < n)) |
| 203 |
{ |
| 204 |
if (ch & MSB) |
| 205 |
{ |
| 206 |
if (ch == SS2) |
| 207 |
{ /* kana */ |
| 208 |
*p++ = (Wchar) * s++; |
| 209 |
} |
| 210 |
else if (ch == SS3) |
| 211 |
{ |
| 212 |
*p++ = (Wchar) ((*s << 8) | (*(s + 1) & MSK)); |
| 213 |
s += 2; |
| 214 |
} |
| 215 |
else |
| 216 |
{ |
| 217 |
*p++ = (Wchar) ((ch << 8) | (*s++ & 0xff)); |
| 218 |
} |
| 219 |
} |
| 220 |
else |
| 221 |
{ |
| 222 |
*p++ = (Wchar) ch; |
| 223 |
} |
| 224 |
} |
| 225 |
*p = (Wchar) 0; |
| 226 |
return p - d; |
| 227 |
} |
| 228 |
|
| 229 |
size_t |
| 230 |
Wcstombs (d, s, n) |
| 231 |
char *d; |
| 232 |
Wchar *s; |
| 233 |
int n; |
| 234 |
{ |
| 235 |
register char *p = d; |
| 236 |
register Wchar ch; |
| 237 |
|
| 238 |
while ((ch = *s++) && (p - d + 2 < n)) |
| 239 |
{ |
| 240 |
switch (ch & WCMSK) |
| 241 |
{ |
| 242 |
case WCG0: |
| 243 |
*p++ = (char) ch; |
| 244 |
break; |
| 245 |
|
| 246 |
case WCG1: |
| 247 |
*p++ = (char) ((ch >> 8) & 0xff); |
| 248 |
*p++ = (char) (ch & 0xff); |
| 249 |
break; |
| 250 |
|
| 251 |
case WCG2: |
| 252 |
*p++ = SS2; |
| 253 |
*p++ = (char) ch; |
| 254 |
break; |
| 255 |
|
| 256 |
case WCG3: |
| 257 |
*p++ = SS3; |
| 258 |
*p++ = (char) ((ch >> 8) & 0xff); |
| 259 |
*p++ = (char) ((ch & 0xff) | MSB); |
| 260 |
break; |
| 261 |
} |
| 262 |
} |
| 263 |
*p = '\0'; |
| 264 |
return p - d; |
| 265 |
} |
| 266 |
|
| 267 |
int |
| 268 |
Wscmp (s1, s2) |
| 269 |
register Wchar *s1, *s2; |
| 270 |
{ |
| 271 |
register int res; |
| 272 |
|
| 273 |
/* ������������������������������������������������������������ |
| 274 |
���������������������������������������������������������� |
| 275 |
|
| 276 |
while (*s1 && *s2 && && *s1 == *s2) { |
| 277 |
s1++; s2++; |
| 278 |
} |
| 279 |
return *s1 - *s2; |
| 280 |
|
| 281 |
����������s1 �� s2 �� EOS ('\0') ���������������������������� |
| 282 |
������������������������������������������������ EOS ���������� |
| 283 |
������������������������*s1 - *s2 �������� |
| 284 |
*/ |
| 285 |
|
| 286 |
while (!(res = *s1 - *s2++) && *s1++) |
| 287 |
; |
| 288 |
return res; |
| 289 |
} |
| 290 |
|
| 291 |
Wchar * |
| 292 |
Wscpy (d, s) |
| 293 |
Wchar *d; |
| 294 |
register Wchar *s; |
| 295 |
{ |
| 296 |
register Wchar *p = d, ch; |
| 297 |
|
| 298 |
while (ch = *s++) |
| 299 |
{ |
| 300 |
*p++ = ch; |
| 301 |
} |
| 302 |
*p = (Wchar) 0; |
| 303 |
return d; |
| 304 |
} |
| 305 |
|
| 306 |
int |
| 307 |
Wslen (s) |
| 308 |
Wchar *s; |
| 309 |
{ |
| 310 |
register Wchar *p = s; |
| 311 |
|
| 312 |
while (*p) |
| 313 |
p++; |
| 314 |
return p - s; |
| 315 |
} |
| 316 |
|
| 317 |
int |
| 318 |
Watoi (s) |
| 319 |
Wchar *s; |
| 320 |
{ |
| 321 |
register int res = 0; |
| 322 |
register Wchar ch; |
| 323 |
|
| 324 |
while ((ch = *s++) && ((Wchar) '0' <= ch) && (ch <= (Wchar) '9')) |
| 325 |
{ |
| 326 |
res *= 10; |
| 327 |
res += ch - (Wchar) '0'; |
| 328 |
} |
| 329 |
return res; |
| 330 |
} |
| 331 |
|
| 332 |
static void |
| 333 |
Fputws (s, f) |
| 334 |
Wchar *s; |
| 335 |
FILE *f; |
| 336 |
{ |
| 337 |
char buf[READBUFSIZE]; |
| 338 |
|
| 339 |
if (Wcstombs (buf, s, READBUFSIZE)) |
| 340 |
{ |
| 341 |
(void) fputs (buf, f); |
| 342 |
} |
| 343 |
} |
| 344 |
|
| 345 |
Wchar * |
| 346 |
Fgetws (buf, siz, f) |
| 347 |
Wchar *buf; |
| 348 |
int siz; |
| 349 |
FILE *f; |
| 350 |
{ |
| 351 |
char mbuf[READBUFSIZE], *p; |
| 352 |
|
| 353 |
p = fgets (mbuf, READBUFSIZE, f); |
| 354 |
if (p) |
| 355 |
{ |
| 356 |
if (Mbstowcs (buf, mbuf, siz)) |
| 357 |
{ |
| 358 |
return buf; |
| 359 |
} |
| 360 |
} |
| 361 |
return (Wchar *) 0; |
| 362 |
} |
| 363 |
#endif |
| 364 |
|
| 365 |
/* s �������������������������������������������������� */ |
| 366 |
|
| 367 |
static int |
| 368 |
all_kana (s) |
| 369 |
Wchar *s; |
| 370 |
{ |
| 371 |
static Wchar xa = 0, xke, aa, *p; |
| 372 |
|
| 373 |
if (!xa) |
| 374 |
{ |
| 375 |
Mbstowcs (&xa, "\045\041", 1); |
| 376 |
Mbstowcs (&xke, "\045\166", 1); |
| 377 |
Mbstowcs (&aa, "\041\074", 1); |
| 378 |
} |
| 379 |
|
| 380 |
for (p = s; *p; p++) |
| 381 |
{ |
| 382 |
if (!(*p == aa || (xa <= *p && *p <= xke))) |
| 383 |
{ |
| 384 |
return 0; |
| 385 |
} |
| 386 |
} |
| 387 |
|
| 388 |
return 1; |
| 389 |
} |
| 390 |
|
| 391 |
/* ���������������� */ |
| 392 |
|
| 393 |
static Wchar * |
| 394 |
findslash (s) |
| 395 |
Wchar *s; |
| 396 |
{ |
| 397 |
while (*s) |
| 398 |
{ |
| 399 |
if (*s == (Wchar) '/') |
| 400 |
{ |
| 401 |
return s; |
| 402 |
} |
| 403 |
s++; |
| 404 |
} |
| 405 |
return (Wchar *) 0; |
| 406 |
} |
| 407 |
|
| 408 |
/* ���������������������� */ |
| 409 |
|
| 410 |
static Wchar * |
| 411 |
extstr (p, pp, key_return) |
| 412 |
Wchar *p, **pp; |
| 413 |
int *key_return; |
| 414 |
{ |
| 415 |
Wchar *res; |
| 416 |
int key = 0; |
| 417 |
|
| 418 |
while (*p == (Wchar) ' ' || *p == (Wchar) '\t') |
| 419 |
p++; |
| 420 |
res = p; |
| 421 |
while (*p && *p != (Wchar) ' ' && *p != (Wchar) '\t' && *p != (Wchar) '\n') |
| 422 |
{ |
| 423 |
key += (int) *p++; |
| 424 |
} |
| 425 |
*p++ = (Wchar) '\0'; |
| 426 |
if (pp) |
| 427 |
*pp = p; |
| 428 |
if (key_return) |
| 429 |
*key_return = key; |
| 430 |
return res; |
| 431 |
} |
| 432 |
|
| 433 |
static struct hinshipack *partsofspeech[HINSHIBUFSIZE]; |
| 434 |
|
| 435 |
static void |
| 436 |
malloc_failed () |
| 437 |
{ |
| 438 |
(void) fprintf (stderr, "%s: malloc failed.\n", program); |
| 439 |
} |
| 440 |
|
| 441 |
/* �������������������������������� */ |
| 442 |
|
| 443 |
static struct hinshipack * |
| 444 |
internhinshi (str, flag) |
| 445 |
Wchar *str; |
| 446 |
int flag; |
| 447 |
{ |
| 448 |
struct hinshipack *p, **pp; |
| 449 |
Wchar *s; |
| 450 |
int key = 0; |
| 451 |
|
| 452 |
for (s = str; *s; s++) |
| 453 |
key += (int) *s; |
| 454 |
key = ((unsigned) key & HINSHIBUFINDEXMASK); |
| 455 |
for (pp = partsofspeech + key; p = *pp; pp = &(p->next)) |
| 456 |
{ |
| 457 |
if (!Wscmp (p->hinshi, str)) |
| 458 |
{ |
| 459 |
return p; |
| 460 |
} |
| 461 |
} |
| 462 |
if (flag) |
| 463 |
{ |
| 464 |
p = (struct hinshipack *) malloc (sizeof (struct hinshipack)); |
| 465 |
if (p) |
| 466 |
{ |
| 467 |
*pp = p; |
| 468 |
(void) bzero (p, sizeof (struct hinshipack)); |
| 469 |
p->hinshi = (Wchar *) malloc ((Wslen (str) + 1) * sizeof (Wchar)); |
| 470 |
if (p->hinshi) |
| 471 |
{ |
| 472 |
(void) Wscpy (p->hinshi, str); |
| 473 |
p->nhinshis = 1; |
| 474 |
return p; |
| 475 |
} |
| 476 |
free (p); |
| 477 |
} |
| 478 |
malloc_failed (); |
| 479 |
} |
| 480 |
return (struct hinshipack *) 0; |
| 481 |
} |
| 482 |
|
| 483 |
/* ������������������ */ |
| 484 |
|
| 485 |
static void |
| 486 |
replace_hinshi () |
| 487 |
{ |
| 488 |
FILE *f; |
| 489 |
Wchar readbuf[READBUFSIZE], *to, *from, *s; |
| 490 |
struct hinshipack *hinshientry, *p; |
| 491 |
int i, err = 0; |
| 492 |
|
| 493 |
f = fopen (hinshi_table, "r"); |
| 494 |
if (!f) |
| 495 |
{ |
| 496 |
(void) fprintf (stderr, "%s: can not open the table file of parts of speech \"%s\".\n", program, hinshi_table); |
| 497 |
exit (1); |
| 498 |
} |
| 499 |
while (s = Fgetws (readbuf, READBUFSIZE, f)) |
| 500 |
{ |
| 501 |
from = extstr (s, &s, 0); |
| 502 |
to = extstr (s, &s, 0); |
| 503 |
if (hinshi_direction == REVERSE) |
| 504 |
{ |
| 505 |
Wchar *xx = from; |
| 506 |
from = to; |
| 507 |
to = xx; |
| 508 |
} |
| 509 |
|
| 510 |
hinshientry = internhinshi (from, 0); |
| 511 |
if (hinshientry) |
| 512 |
{ |
| 513 |
Wchar *xx; |
| 514 |
|
| 515 |
xx = (Wchar *) malloc ((Wslen (to) + 1) * sizeof (Wchar)); |
| 516 |
if (xx) |
| 517 |
{ |
| 518 |
Wchar *cp; |
| 519 |
int n = 1; |
| 520 |
|
| 521 |
(void) Wscpy (xx, to); |
| 522 |
free (hinshientry->hinshi); |
| 523 |
hinshientry->hinshi = xx; |
| 524 |
for (cp = xx; *cp; cp++) |
| 525 |
{ |
| 526 |
if (*cp == (Wchar) '/') |
| 527 |
{ |
| 528 |
*cp = (Wchar) 0; |
| 529 |
n++; |
| 530 |
} |
| 531 |
} |
| 532 |
hinshientry->nhinshis = n; |
| 533 |
hinshientry->flags |= REPLACED; |
| 534 |
} |
| 535 |
else |
| 536 |
{ |
| 537 |
malloc_failed (); |
| 538 |
} |
| 539 |
} |
| 540 |
} |
| 541 |
(void) fclose (f); |
| 542 |
|
| 543 |
for (i = 0; i < HINSHIBUFSIZE; i++) |
| 544 |
{ |
| 545 |
for (p = partsofspeech[i]; p; p = p->next) |
| 546 |
{ |
| 547 |
if (!(p->flags & REPLACED)) |
| 548 |
{ |
| 549 |
(void) fprintf (stderr, "%s: The replacement for \"", program); |
| 550 |
Fputws (p->hinshi, stderr); |
| 551 |
(void) fprintf (stderr, "\" is not mentioned in the table.\n"); |
| 552 |
err = 1; |
| 553 |
} |
| 554 |
} |
| 555 |
} |
| 556 |
if (err) |
| 557 |
{ |
| 558 |
exit (1); |
| 559 |
} |
| 560 |
} |
| 561 |
|
| 562 |
static void |
| 563 |
select_hinshi (n) |
| 564 |
int n; |
| 565 |
{ |
| 566 |
Wchar *s, *t, *xx; |
| 567 |
struct hinshipack *p; |
| 568 |
int i; |
| 569 |
|
| 570 |
if (!n) |
| 571 |
return; |
| 572 |
|
| 573 |
for (i = 0; i < HINSHIBUFSIZE; i++) |
| 574 |
{ |
| 575 |
for (p = partsofspeech[i]; p; p = p->next) |
| 576 |
{ |
| 577 |
switch (n) |
| 578 |
{ |
| 579 |
case 1: |
| 580 |
s = findslash (p->hinshi); |
| 581 |
if (s) |
| 582 |
{ |
| 583 |
*s = (Wchar) 0; |
| 584 |
} |
| 585 |
break; |
| 586 |
|
| 587 |
case 2: |
| 588 |
s = findslash (p->hinshi); |
| 589 |
if (s) |
| 590 |
{ |
| 591 |
s++; |
| 592 |
t = findslash (s); |
| 593 |
if (t) |
| 594 |
{ |
| 595 |
xx = (Wchar *) malloc ((t - s + 1) * sizeof (Wchar)); |
| 596 |
if (xx) |
| 597 |
{ |
| 598 |
*t = (Wchar) 0; |
| 599 |
Wscpy (xx, s); |
| 600 |
t = p->hinshi; |
| 601 |
p->hinshi = xx; |
| 602 |
(void) free ((char *) t); |
| 603 |
} |
| 604 |
} |
| 605 |
} |
| 606 |
break; |
| 607 |
|
| 608 |
case 3: |
| 609 |
s = findslash (p->hinshi); |
| 610 |
if (s) |
| 611 |
{ |
| 612 |
t = findslash (s + 1); |
| 613 |
if (t) |
| 614 |
{ |
| 615 |
t++; |
| 616 |
xx = (Wchar *) malloc ((Wslen (t) + 1) * sizeof (Wchar)); |
| 617 |
if (xx) |
| 618 |
{ |
| 619 |
Wscpy (xx, t); |
| 620 |
t = p->hinshi; |
| 621 |
p->hinshi = xx; |
| 622 |
(void) free ((char *) t); |
| 623 |
} |
| 624 |
} |
| 625 |
} |
| 626 |
break; |
| 627 |
|
| 628 |
default: |
| 629 |
break; |
| 630 |
} |
| 631 |
} |
| 632 |
} |
| 633 |
} |
| 634 |
|
| 635 |
static void |
| 636 |
freedesc (p) |
| 637 |
struct descpack *p; |
| 638 |
{ |
| 639 |
free (p->hinshi); |
| 640 |
free (p->tandesc); |
| 641 |
free (p->yomdesc); |
| 642 |
free (p); |
| 643 |
} |
| 644 |
|
| 645 |
static struct descpack *description[HINSHIBUFSIZE]; |
| 646 |
|
| 647 |
/* ������������ */ |
| 648 |
|
| 649 |
static struct descpack * |
| 650 |
interndesc (hin, tan, yom) |
| 651 |
Wchar *hin, *tan, *yom; |
| 652 |
{ |
| 653 |
struct descpack *p, **pp, *next = (struct descpack *) 0; |
| 654 |
Wchar *s; |
| 655 |
int key = 0; |
| 656 |
|
| 657 |
for (s = hin; *s; s++) |
| 658 |
key += (int) *s; |
| 659 |
key = ((unsigned) key & HINSHIBUFINDEXMASK); |
| 660 |
for (pp = description + key; p = *pp; pp = &(p->next)) |
| 661 |
{ |
| 662 |
if (!Wscmp (p->hinshi, hin)) |
| 663 |
{ |
| 664 |
if (!Wscmp (p->tandesc, tan) && !Wscmp (p->yomdesc, yom)) |
| 665 |
{ |
| 666 |
return p; |
| 667 |
} |
| 668 |
else |
| 669 |
{ |
| 670 |
*pp = next = p->next; |
| 671 |
freedesc (p); |
| 672 |
break; |
| 673 |
} |
| 674 |
} |
| 675 |
} |
| 676 |
p = (struct descpack *) malloc (sizeof (struct descpack)); |
| 677 |
if (p) |
| 678 |
{ |
| 679 |
*pp = p; |
| 680 |
(void) bzero (p, sizeof (struct descpack)); |
| 681 |
p->next = next; |
| 682 |
p->hinshi = (Wchar *) malloc ((Wslen (hin) + 1) * sizeof (Wchar)); |
| 683 |
if (p->hinshi) |
| 684 |
{ |
| 685 |
(void) Wscpy (p->hinshi, hin); |
| 686 |
p->tandesc = (Wchar *) malloc ((Wslen (tan) + 1) * sizeof (Wchar)); |
| 687 |
if (p->tandesc) |
| 688 |
{ |
| 689 |
(void) Wscpy (p->tandesc, tan); |
| 690 |
p->yomdesc = (Wchar *) malloc ((Wslen (yom) + 1) * sizeof (Wchar)); |
| 691 |
if (p->yomdesc) |
| 692 |
{ |
| 693 |
(void) Wscpy (p->yomdesc, yom); |
| 694 |
return p; |
| 695 |
} |
| 696 |
free (p->tandesc); |
| 697 |
} |
| 698 |
free (p->hinshi); |
| 699 |
} |
| 700 |
free (p); |
| 701 |
} |
| 702 |
malloc_failed (); |
| 703 |
return (struct descpack *) 0; |
| 704 |
} |
| 705 |
|
| 706 |
/* ������������ */ |
| 707 |
|
| 708 |
static struct descpack * |
| 709 |
searchdesc (hin) |
| 710 |
Wchar *hin; |
| 711 |
{ |
| 712 |
struct descpack *p, **pp; |
| 713 |
Wchar *s; |
| 714 |
int key = 0; |
| 715 |
|
| 716 |
for (s = hin; *s; s++) |
| 717 |
key += (int) *s; |
| 718 |
key = ((unsigned) key & HINSHIBUFINDEXMASK); |
| 719 |
for (pp = description + key; p = *pp; pp = &(p->next)) |
| 720 |
{ |
| 721 |
if (!Wscmp (p->hinshi, hin)) |
| 722 |
{ |
| 723 |
return p; |
| 724 |
} |
| 725 |
} |
| 726 |
return (struct descpack *) 0; |
| 727 |
} |
| 728 |
|
| 729 |
static void |
| 730 |
store_description () |
| 731 |
{ |
| 732 |
FILE *f; |
| 733 |
Wchar readbuf[READBUFSIZE], *hin, *tan, *yom, *s; |
| 734 |
|
| 735 |
if (!description_table) |
| 736 |
{ |
| 737 |
return; |
| 738 |
} |
| 739 |
|
| 740 |
f = fopen (description_table, "r"); |
| 741 |
if (!f) |
| 742 |
{ |
| 743 |
(void) fprintf (stderr, "%s: can not open the table file of parts of speech \"%s\".\n", program, description_table); |
| 744 |
exit (1); |
| 745 |
} |
| 746 |
while (s = Fgetws (readbuf, READBUFSIZE, f)) |
| 747 |
{ |
| 748 |
Wchar nl[1]; |
| 749 |
|
| 750 |
nl[0] = (Wchar) 0; |
| 751 |
hin = tan = yom = nl; |
| 752 |
hin = extstr (s, &s, 0); |
| 753 |
if (*hin) |
| 754 |
{ |
| 755 |
tan = extstr (s, &s, 0); |
| 756 |
if (*tan) |
| 757 |
{ |
| 758 |
yom = extstr (s, &s, 0); |
| 759 |
} |
| 760 |
} |
| 761 |
|
| 762 |
interndesc (hin, tan, yom); |
| 763 |
} |
| 764 |
(void) fclose (f); |
| 765 |
} |
| 766 |
|
| 767 |
struct kindpack kinds[sizeof (long) * 8]; |
| 768 |
static int nkinds; |
| 769 |
|
| 770 |
#define KIHONBIT 1L |
| 771 |
|
| 772 |
/* ���������� */ |
| 773 |
|
| 774 |
static long |
| 775 |
internkind (s) |
| 776 |
Wchar *s; |
| 777 |
{ |
| 778 |
int i; |
| 779 |
Wchar *p; |
| 780 |
|
| 781 |
p = findslash (s); |
| 782 |
if (p) |
| 783 |
{ |
| 784 |
long res; |
| 785 |
|
| 786 |
*p = (Wchar) '\0'; |
| 787 |
res = internkind (s); |
| 788 |
res |= internkind (p + 1); |
| 789 |
return res; |
| 790 |
} |
| 791 |
else |
| 792 |
{ |
| 793 |
for (i = 0; i < nkinds; i++) |
| 794 |
{ |
| 795 |
if (!Wscmp (s, kinds[i].kind)) |
| 796 |
{ |
| 797 |
return kinds[i].kindbit; |
| 798 |
} |
| 799 |
} |
| 800 |
if (nkinds < (sizeof (long) * 8) && (kinds[nkinds].kind = (Wchar *) malloc ((Wslen (s) + 1) * sizeof (Wchar)))) |
| 801 |
{ |
| 802 |
(void) Wscpy (kinds[nkinds].kind, s); |
| 803 |
kinds[nkinds].kindbit = 1 << nkinds; |
| 804 |
return kinds[nkinds++].kindbit; |
| 805 |
} |
| 806 |
return 0; |
| 807 |
} |
| 808 |
} |
| 809 |
|
| 810 |
/* ���������������� */ |
| 811 |
|
| 812 |
static void |
| 813 |
listkinds () |
| 814 |
{ |
| 815 |
int i; |
| 816 |
|
| 817 |
for (i = 0; i < nkinds; i++) |
| 818 |
{ |
| 819 |
Fputws (kinds[i].kind, stdout); |
| 820 |
putchar ('\n'); |
| 821 |
} |
| 822 |
} |
| 823 |
|
| 824 |
static int |
| 825 |
kindcompar (k1, k2) |
| 826 |
struct kindpack *k1, *k2; |
| 827 |
{ |
| 828 |
return Wscmp (k1->kind, k2->kind); |
| 829 |
} |
| 830 |
|
| 831 |
static void |
| 832 |
sortkind () |
| 833 |
{ |
| 834 |
qsort (kinds, nkinds, sizeof (struct kindpack), kindcompar); |
| 835 |
} |
| 836 |
|
| 837 |
static struct dicpack *dic[DICBUFSIZE], **pdic; |
| 838 |
static int ndicentries = 0; |
| 839 |
|
| 840 |
/* |
| 841 |
|
| 842 |
intern -- ������������������/���� |
| 843 |
|
| 844 |
���������� stat ���������������������������������������������������������� |
| 845 |
�������������������������������������������������������������������������� |
| 846 |
|
| 847 |
flags ������������������������������(������������)�� |
| 848 |
|
| 849 |
hinshi �� 0 ��������������������kind �� 0 ������������������-m ������������ |
| 850 |
������������������������ |
| 851 |
|
| 852 |
*/ |
| 853 |
|
| 854 |
/* flags */ |
| 855 |
#define IGNORE_HINSHI 1L |
| 856 |
#define IGNORE_KIND 2L |
| 857 |
|
| 858 |
static struct dicpack * |
| 859 |
intern (key, yomi, kouho, hinshi, hindo, kind, stat, flags) |
| 860 |
int key, hindo, *stat; |
| 861 |
Wchar *yomi, *kouho, *hinshi; |
| 862 |
long kind, flags; |
| 863 |
{ |
| 864 |
struct dicpack *p, **pp; |
| 865 |
struct descpack *dp; |
| 866 |
Wchar nl[1], *yomdesc = nl, *tandesc = nl; |
| 867 |
Wchar *yom = (Wchar *) 0, *tan = (Wchar *) 0, *dhinshi, *dh; |
| 868 |
|
| 869 |
nl[0] = (Wchar) '\0'; |
| 870 |
|
| 871 |
if (description_table) |
| 872 |
{ |
| 873 |
dhinshi = dh = hinshi; /* ������������������ */ |
| 874 |
while (*dh) |
| 875 |
{ |
| 876 |
if (*dh++ == (Wchar) '/') |
| 877 |
{ |
| 878 |
dhinshi = dh; |
| 879 |
} |
| 880 |
} |
| 881 |
dp = searchdesc (dhinshi); |
| 882 |
if (dp) |
| 883 |
{ |
| 884 |
yomdesc = dp->yomdesc; |
| 885 |
tandesc = dp->tandesc; |
| 886 |
if (Wslen (yomdesc)) |
| 887 |
{ |
| 888 |
Wchar *t; |
| 889 |
t = (Wchar *) malloc ((Wslen (yomi) + Wslen (yomdesc) + 1) * sizeof (Wchar)); |
| 890 |
if (t) |
| 891 |
{ |
| 892 |
Wscpy (t, yomi); |
| 893 |
yom = yomi = t; |
| 894 |
Wscpy (yomi + Wslen (yomi), yomdesc); |
| 895 |
} |
| 896 |
} |
| 897 |
if (Wslen (tandesc)) |
| 898 |
{ |
| 899 |
Wchar *t; |
| 900 |
t = (Wchar *) malloc ((Wslen (kouho) + Wslen (tandesc) + 1) * sizeof (Wchar)); |
| 901 |
if (t) |
| 902 |
{ |
| 903 |
Wscpy (t, kouho); |
| 904 |
tan = kouho = t; |
| 905 |
Wscpy (kouho + Wslen (kouho), tandesc); |
| 906 |
} |
| 907 |
} |
| 908 |
} |
| 909 |
else |
| 910 |
{ |
| 911 |
char foo[64]; |
| 912 |
|
| 913 |
fprintf (stderr, "no description rule for "); |
| 914 |
Wcstombs (foo, dhinshi, 64); |
| 915 |
fprintf (stderr, "%s.\n", foo); |
| 916 |
} |
| 917 |
} |
| 918 |
|
| 919 |
key = ((unsigned) key & DICBUFINDEXMASK); |
| 920 |
for (pp = dic + key; p = *pp; pp = &(p->next)) |
| 921 |
{ |
| 922 |
if (!Wscmp (p->yomi, yomi) && !Wscmp (p->tango, kouho) && ((flags & IGNORE_HINSHI) || !Wscmp (p->hinshi->hinshi, hinshi)) && ((flags & IGNORE_KIND) || ((p->kind & kind) == kind))) |
| 923 |
{ |
| 924 |
/* match */ |
| 925 |
if (stat) |
| 926 |
*stat = FOUND; |
| 927 |
if (yom) |
| 928 |
free (yom); |
| 929 |
if (tan) |
| 930 |
free (tan); |
| 931 |
return p; |
| 932 |
} |
| 933 |
} |
| 934 |
if (stat) |
| 935 |
{ |
| 936 |
p = (struct dicpack *) malloc (sizeof (struct dicpack)); |
| 937 |
if (p) |
| 938 |
{ |
| 939 |
*pp = p; |
| 940 |
(void) bzero (p, sizeof (struct dicpack)); |
| 941 |
p->yomi = (Wchar *) malloc ((Wslen (yomi) + 1) * sizeof (Wchar)); |
| 942 |
if (p->yomi) |
| 943 |
{ |
| 944 |
(void) Wscpy (p->yomi, yomi); |
| 945 |
p->tango = (Wchar *) malloc ((Wslen (kouho) + 1) * sizeof (Wchar)); |
| 946 |
if (p->tango) |
| 947 |
{ |
| 948 |
(void) Wscpy (p->tango, kouho); |
| 949 |
p->hinshi = internhinshi (hinshi, 1); |
| 950 |
if (p->hinshi) |
| 951 |
{ |
| 952 |
p->hindo = hindo; |
| 953 |
*stat = CREATE; |
| 954 |
ndicentries++; |
| 955 |
p->kind = kind; |
| 956 |
p->extdata = (Wchar *) 0; |
| 957 |
if (yom) |
| 958 |
free (yom); |
| 959 |
if (tan) |
| 960 |
free (tan); |
| 961 |
return p; |
| 962 |
} |
| 963 |
free (p->tango); |
| 964 |
} |
| 965 |
free (p->yomi); |
| 966 |
} |
| 967 |
free (p); |
| 968 |
} |
| 969 |
malloc_failed (); |
| 970 |
} |
| 971 |
if (yom) |
| 972 |
free (yom); |
| 973 |
if (tan) |
| 974 |
free (tan); |
| 975 |
return (struct dicpack *) 0; |
| 976 |
} |
| 977 |
|
| 978 |
/* ������������������������������ fn ���������� */ |
| 979 |
|
| 980 |
static void |
| 981 |
for_all_interned (fn) |
| 982 |
void (*fn) (); |
| 983 |
{ |
| 984 |
int i; |
| 985 |
struct dicpack *p; |
| 986 |
|
| 987 |
for (i = 0; i < DICBUFSIZE; i++) |
| 988 |
{ |
| 989 |
for (p = dic[i]; p; p = p->next) |
| 990 |
{ |
| 991 |
(*fn) (p); |
| 992 |
} |
| 993 |
} |
| 994 |
} |
| 995 |
|
| 996 |
static void |
| 997 |
storepd (file) |
| 998 |
FILE *file; |
| 999 |
{ |
| 1000 |
Wchar readbuf[READBUFSIZE], *p, *yomi, *hinshi, *kouho, *hindo, *kind; |
| 1001 |
int nhindo, key, tkey, stat; |
| 1002 |
long kindbit; |
| 1003 |
struct dicpack *dicentry; |
| 1004 |
|
| 1005 |
while (p = Fgetws (readbuf, READBUFSIZE, file)) |
| 1006 |
{ |
| 1007 |
key = 0; |
| 1008 |
yomi = extstr (p, &p, &tkey); |
| 1009 |
key += tkey; |
| 1010 |
kouho = extstr (p, &p, &tkey); |
| 1011 |
key += tkey; |
| 1012 |
hinshi = extstr (p, &p, 0); |
| 1013 |
hindo = extstr (p, &p, 0); |
| 1014 |
nhindo = Watoi (hindo); |
| 1015 |
|
| 1016 |
kind = extstr (p, 0, 0); |
| 1017 |
if (*kind) |
| 1018 |
{ |
| 1019 |
kindbit = internkind (kind); |
| 1020 |
} |
| 1021 |
else |
| 1022 |
{ |
| 1023 |
kindbit = KIHONBIT; |
| 1024 |
} |
| 1025 |
|
| 1026 |
dicentry = intern (key, yomi, kouho, hinshi, nhindo, kindbit, &stat, IGNORE_KIND); |
| 1027 |
if (dicentry) |
| 1028 |
{ |
| 1029 |
dicentry->kind |= kindbit; |
| 1030 |
} |
| 1031 |
} |
| 1032 |
} |
| 1033 |
|
| 1034 |
static void |
| 1035 |
comparepd (file) |
| 1036 |
FILE *file; |
| 1037 |
{ |
| 1038 |
Wchar readbuf[READBUFSIZE], *p, *yomi, *hinshi, *kouho, *hindo, *kind; |
| 1039 |
int nhindo, key, tkey, stat, *statp = &stat; |
| 1040 |
struct dicpack *dicentry; |
| 1041 |
long kindbit, flags = 0L; |
| 1042 |
|
| 1043 |
while (p = Fgetws (readbuf, READBUFSIZE, file)) |
| 1044 |
{ |
| 1045 |
key = 0; |
| 1046 |
yomi = extstr (p, &p, &tkey); |
| 1047 |
key += tkey; |
| 1048 |
kouho = extstr (p, &p, &tkey); |
| 1049 |
key += tkey; |
| 1050 |
hinshi = extstr (p, &p, 0); |
| 1051 |
if (ignore_hinshi_to_compare) |
| 1052 |
{ |
| 1053 |
flags |= IGNORE_HINSHI; |
| 1054 |
} |
| 1055 |
hindo = extstr (p, &p, 0); |
| 1056 |
nhindo = Watoi (hindo); |
| 1057 |
|
| 1058 |
kind = extstr (p, 0, 0); |
| 1059 |
if (*kind) |
| 1060 |
{ |
| 1061 |
kindbit = internkind (kind); |
| 1062 |
} |
| 1063 |
else |
| 1064 |
{ |
| 1065 |
kindbit = KIHONBIT; |
| 1066 |
} |
| 1067 |
if (merge_kind || merge_sj3) |
| 1068 |
{ |
| 1069 |
flags |= IGNORE_KIND; |
| 1070 |
} |
| 1071 |
if (copy_frequency) |
| 1072 |
{ |
| 1073 |
statp = (int *) 0; |
| 1074 |
} |
| 1075 |
|
| 1076 |
dicentry = intern (key, yomi, kouho, hinshi, nhindo, kindbit, statp, flags); |
| 1077 |
|
| 1078 |
if (dicentry) |
| 1079 |
{ |
| 1080 |
if (copy_frequency) |
| 1081 |
{ |
| 1082 |
dicentry->hindo = nhindo; |
| 1083 |
dicentry->flags &= ~COMMON; |
| 1084 |
} |
| 1085 |
else if (ignore_hinshi_to_compare && stat == FOUND) |
| 1086 |
{ |
| 1087 |
/* ���������������������������������� */ |
| 1088 |
struct dicpack *pd; |
| 1089 |
|
| 1090 |
for (pd = dicentry; pd; pd = pd->next) |
| 1091 |
{ |
| 1092 |
if (!Wscmp (pd->yomi, yomi) && !Wscmp (pd->tango, kouho)) |
| 1093 |
{ |
| 1094 |
pd->flags |= COMMON; |
| 1095 |
if (!merge_sj3) |
| 1096 |
{ |
| 1097 |
pd->kind |= kindbit; |
| 1098 |
} |
| 1099 |
|
| 1100 |
if (merge_sj3) |
| 1101 |
{ |
| 1102 |
int len = 0; |
| 1103 |
Wchar *dat; |
| 1104 |
|
| 1105 |
if (pd->extdata) |
| 1106 |
{ |
| 1107 |
len = Wslen (pd->extdata); |
| 1108 |
} |
| 1109 |
dat = (Wchar *) malloc ((Wslen (hinshi) + 1 + len) * sizeof (Wchar)); |
| 1110 |
if (dat) |
| 1111 |
{ |
| 1112 |
if (len) |
| 1113 |
{ |
| 1114 |
(void) Wscpy (dat, pd->extdata); |
| 1115 |
(void) free ((char *) pd->extdata); |
| 1116 |
} |
| 1117 |
(void) Wscpy (dat + len, hinshi); |
| 1118 |
pd->extdata = dat; |
| 1119 |
} |
| 1120 |
} |
| 1121 |
} |
| 1122 |
} |
| 1123 |
} |
| 1124 |
else |
| 1125 |
{ |
| 1126 |
dicentry->kind |= kindbit; |
| 1127 |
if (stat == FOUND) |
| 1128 |
{ |
| 1129 |
dicentry->flags |= COMMON; |
| 1130 |
} |
| 1131 |
else |
| 1132 |
{ /* CREATE */ |
| 1133 |
dicentry->flags |= NEW; |
| 1134 |
} |
| 1135 |
} |
| 1136 |
} |
| 1137 |
} |
| 1138 |
} |
| 1139 |
|
| 1140 |
static void |
| 1141 |
canna_output (cf, p, h, n) |
| 1142 |
FILE *cf; |
| 1143 |
struct dicpack *p; |
| 1144 |
Wchar *h; |
| 1145 |
int n; |
| 1146 |
{ |
| 1147 |
for (; n-- > 0; h += Wslen (h) + 1) |
| 1148 |
{ |
| 1149 |
Fputws (p->yomi, cf); |
| 1150 |
(void) putc (' ', cf); |
| 1151 |
Fputws (h, cf); |
| 1152 |
if (p->hindo) |
| 1153 |
{ |
| 1154 |
(void) fprintf (cf, "*%d", p->hindo); |
| 1155 |
} |
| 1156 |
(void) putc (' ', cf); |
| 1157 |
Fputws (p->tango, cf); |
| 1158 |
(void) putc ('\n', cf); |
| 1159 |
} |
| 1160 |
} |
| 1161 |
|
| 1162 |
static void |
| 1163 |
entry_out (cf, p, h, n, ex) |
| 1164 |
FILE *cf; |
| 1165 |
struct dicpack *p; |
| 1166 |
Wchar *h; |
| 1167 |
int n; |
| 1168 |
Wchar *ex; |
| 1169 |
{ |
| 1170 |
int i, f = 1; |
| 1171 |
long b; |
| 1172 |
|
| 1173 |
for (; n-- > 0; h += Wslen (h) + 1) |
| 1174 |
{ |
| 1175 |
Fputws (p->yomi, cf); |
| 1176 |
(void) putc (' ', cf); |
| 1177 |
Fputws (p->tango, cf); |
| 1178 |
(void) putc (' ', cf); |
| 1179 |
if (merge_sj3 && ex) |
| 1180 |
{ |
| 1181 |
Fputws (ex, cf); |
| 1182 |
(void) putc ('/', cf); |
| 1183 |
} |
| 1184 |
Fputws (h, cf); |
| 1185 |
if (!sj3_type_output) |
| 1186 |
{ |
| 1187 |
(void) fprintf (cf, " %d", p->hindo); |
| 1188 |
} |
| 1189 |
|
| 1190 |
if (!wnn_type_output) |
| 1191 |
{ |
| 1192 |
if (bunrui) |
| 1193 |
{ |
| 1194 |
(void) printf (" %s", bunrui); |
| 1195 |
} |
| 1196 |
else |
| 1197 |
{ |
| 1198 |
if (specific_kind) |
| 1199 |
{ |
| 1200 |
b = (specific_kind & p->kind); |
| 1201 |
} |
| 1202 |
else |
| 1203 |
{ |
| 1204 |
b = p->kind; |
| 1205 |
} |
| 1206 |
if (b != KIHONBIT) |
| 1207 |
{ /* ���������������������������� */ |
| 1208 |
for (i = 0; i < nkinds; i++) |
| 1209 |
{ |
| 1210 |
if (b & kinds[i].kindbit) |
| 1211 |
{ |
| 1212 |
if (f) |
| 1213 |
{ |
| 1214 |
(void) putc (' ', cf); |
| 1215 |
f = 0; |
| 1216 |
} |
| 1217 |
else |
| 1218 |
{ |
| 1219 |
(void) putc ('/', cf); |
| 1220 |
} |
| 1221 |
Fputws (kinds[i].kind, cf); |
| 1222 |
} |
| 1223 |
} |
| 1224 |
} |
| 1225 |
} |
| 1226 |
} |
| 1227 |
(void) putc ('\n', cf); |
| 1228 |
} |
| 1229 |
} |
| 1230 |
|
| 1231 |
/* p ���������������������������� cf ���������� */ |
| 1232 |
|
| 1233 |
static void |
| 1234 |
printentry (cf, p) |
| 1235 |
FILE *cf; |
| 1236 |
struct dicpack *p; |
| 1237 |
{ |
| 1238 |
if (specific_kind && !(p->kind & specific_kind)) |
| 1239 |
{ |
| 1240 |
return; |
| 1241 |
} |
| 1242 |
|
| 1243 |
if (extract_kana && !all_kana (p->tango)) |
| 1244 |
{ |
| 1245 |
return; |
| 1246 |
} |
| 1247 |
|
| 1248 |
if (selhinshi && !p->hinshi->hinshi[0]) |
| 1249 |
{ |
| 1250 |
return; |
| 1251 |
} |
| 1252 |
|
| 1253 |
if (canna_type_output) |
| 1254 |
{ |
| 1255 |
canna_output (cf, p, p->hinshi->hinshi, p->hinshi->nhinshis); |
| 1256 |
} |
| 1257 |
else |
| 1258 |
{ |
| 1259 |
entry_out (cf, p, p->hinshi->hinshi, p->hinshi->nhinshis, p->extdata); |
| 1260 |
} |
| 1261 |
} |
| 1262 |
|
| 1263 |
static void |
| 1264 |
showdeleted (p) |
| 1265 |
struct dicpack *p; |
| 1266 |
{ |
| 1267 |
if (!(p->flags & COMMON)) |
| 1268 |
{ |
| 1269 |
(void) printf ("- "); |
| 1270 |
printentry (stdout, p); |
| 1271 |
} |
| 1272 |
} |
| 1273 |
|
| 1274 |
static void |
| 1275 |
showentry (pd, n) |
| 1276 |
struct dicpack **pd; |
| 1277 |
int n; |
| 1278 |
{ |
| 1279 |
FILE *cf = (FILE *) 0, *of = (FILE *) 0, *nf = (FILE *) 0; |
| 1280 |
struct dicpack *p; |
| 1281 |
int i; |
| 1282 |
|
| 1283 |
if (common_out) |
| 1284 |
{ |
| 1285 |
if (common_out[0] != '-' || common_out[1]) |
| 1286 |
{ |
| 1287 |
cf = fopen (common_out, "w"); |
| 1288 |
if (!cf) |
| 1289 |
{ |
| 1290 |
(void) fprintf (stderr, "%s: can not open file \"%s\".\n", program, common_out); |
| 1291 |
exit (1); |
| 1292 |
} |
| 1293 |
} |
| 1294 |
else |
| 1295 |
{ |
| 1296 |
cf = stdout; |
| 1297 |
} |
| 1298 |
} |
| 1299 |
if (old_out) |
| 1300 |
{ |
| 1301 |
if (old_out[0] != '-' || old_out[1]) |
| 1302 |
{ |
| 1303 |
of = fopen (old_out, "w"); |
| 1304 |
if (!of) |
| 1305 |
{ |
| 1306 |
(void) fprintf (stderr, "%s: can not open file \"%s\".\n", program, old_out); |
| 1307 |
exit (1); |
| 1308 |
} |
| 1309 |
} |
| 1310 |
else |
| 1311 |
{ |
| 1312 |
of = stdout; |
| 1313 |
} |
| 1314 |
} |
| 1315 |
if (new_out) |
| 1316 |
{ |
| 1317 |
if (new_out[0] != '-' || new_out[1]) |
| 1318 |
{ |
| 1319 |
nf = fopen (new_out, "w"); |
| 1320 |
if (!nf) |
| 1321 |
{ |
| 1322 |
(void) fprintf (stderr, "%s: can not open file \"%s\".\n", program, new_out); |
| 1323 |
exit (1); |
| 1324 |
} |
| 1325 |
} |
| 1326 |
else |
| 1327 |
{ |
| 1328 |
nf = stdout; |
| 1329 |
} |
| 1330 |
} |
| 1331 |
|
| 1332 |
for (i = 0; i < n; i++) |
| 1333 |
{ |
| 1334 |
p = pd[i]; |
| 1335 |
if (compare) |
| 1336 |
{ |
| 1337 |
if (p->flags & COMMON) |
| 1338 |
{ |
| 1339 |
if (cf) |
| 1340 |
{ |
| 1341 |
printentry (cf, p); |
| 1342 |
} |
| 1343 |
} |
| 1344 |
else if (p->flags & NEW) |
| 1345 |
{ |
| 1346 |
if (nf) |
| 1347 |
{ |
| 1348 |
printentry (nf, p); |
| 1349 |
} |
| 1350 |
} |
| 1351 |
else |
| 1352 |
{ |
| 1353 |
if (of) |
| 1354 |
{ |
| 1355 |
printentry (of, p); |
| 1356 |
} |
| 1357 |
} |
| 1358 |
} |
| 1359 |
else |
| 1360 |
{ /* just print the normalized dictionary */ |
| 1361 |
printentry (stdout, p); |
| 1362 |
} |
| 1363 |
} |
| 1364 |
} |
| 1365 |
|
| 1366 |
static int |
| 1367 |
diccompar (p1, p2) |
| 1368 |
struct dicpack **p1, **p2; |
| 1369 |
{ |
| 1370 |
int n; |
| 1371 |
if (n = Wscmp ((*p1)->yomi, (*p2)->yomi)) |
| 1372 |
{ |
| 1373 |
return n; |
| 1374 |
} |
| 1375 |
else if (n = Wscmp ((*p1)->tango, (*p2)->tango)) |
| 1376 |
{ |
| 1377 |
return n; |
| 1378 |
} |
| 1379 |
else if (n = Wscmp ((*p1)->hinshi->hinshi, (*p2)->hinshi->hinshi)) |
| 1380 |
{ |
| 1381 |
return n; |
| 1382 |
} |
| 1383 |
else |
| 1384 |
{ /* impossible */ |
| 1385 |
return 0; |
| 1386 |
} |
| 1387 |
} |
| 1388 |
|
| 1389 |
static int |
| 1390 |
dichindocompar (p1, p2) |
| 1391 |
struct dicpack **p1, **p2; |
| 1392 |
{ |
| 1393 |
int n; |
| 1394 |
if (n = Wscmp ((*p1)->yomi, (*p2)->yomi)) |
| 1395 |
{ |
| 1396 |
return n; |
| 1397 |
} |
| 1398 |
else if (n = ((*p2)->hindo - (*p1)->hindo)) |
| 1399 |
{ |
| 1400 |
return n; |
| 1401 |
} |
| 1402 |
else if (n = Wscmp ((*p1)->tango, (*p2)->tango)) |
| 1403 |
{ |
| 1404 |
return n; |
| 1405 |
} |
| 1406 |
else if (n = Wscmp ((*p1)->hinshi->hinshi, (*p2)->hinshi->hinshi)) |
| 1407 |
{ |
| 1408 |
return n; |
| 1409 |
} |
| 1410 |
else |
| 1411 |
{ /* impossible */ |
| 1412 |
return 0; |
| 1413 |
} |
| 1414 |
} |
| 1415 |
|
| 1416 |
void |
| 1417 |
shrinkargs (argv, n, count) |
| 1418 |
char **argv; |
| 1419 |
int n, count; |
| 1420 |
{ |
| 1421 |
int i; |
| 1422 |
|
| 1423 |
for (i = 0; i + n < count; i++) |
| 1424 |
{ |
| 1425 |
argv[i] = argv[i + n]; |
| 1426 |
} |
| 1427 |
} |
| 1428 |
|
| 1429 |
static void |
| 1430 |
parseargs (argc, argv) |
| 1431 |
int argc; |
| 1432 |
char *argv[]; |
| 1433 |
{ |
| 1434 |
int i; |
| 1435 |
|
| 1436 |
for (program = argv[0] + strlen (argv[0]); argv[0] < program; program--) |
| 1437 |
{ |
| 1438 |
if (program[0] == '/') |
| 1439 |
{ |
| 1440 |
program++; |
| 1441 |
break; |
| 1442 |
} |
| 1443 |
} |
| 1444 |
|
| 1445 |
for (i = 1; i < argc;) |
| 1446 |
{ |
| 1447 |
if (argv[i][0] == '-' && argv[i][2] == '\0') |
| 1448 |
{ |
| 1449 |
switch (argv[i][1]) |
| 1450 |
{ |
| 1451 |
case '1': |
| 1452 |
case '2': |
| 1453 |
case '3': |
| 1454 |
selhinshi = argv[i][1] - '0'; |
| 1455 |
shrinkargs (argv + i, 1, argc - i); |
| 1456 |
argc -= 1; |
| 1457 |
break; |
| 1458 |
|
| 1459 |
case 'b': |
| 1460 |
bunrui = argv[i + 1]; |
| 1461 |
shrinkargs (argv + i, 2, argc - i); |
| 1462 |
argc -= 2; |
| 1463 |
break; |
| 1464 |
|
| 1465 |
case 'c': |
| 1466 |
common_out = argv[i + 1]; |
| 1467 |
shrinkargs (argv + i, 2, argc - i); |
| 1468 |
argc -= 2; |
| 1469 |
break; |
| 1470 |
|
| 1471 |
case 'd': |
| 1472 |
description_table = argv[i + 1]; |
| 1473 |
shrinkargs (argv + i, 2, argc - i); |
| 1474 |
argc -= 2; |
| 1475 |
break; |
| 1476 |
|
| 1477 |
case 'f': |
| 1478 |
copy_frequency = 1; |
| 1479 |
shrinkargs (argv + i, 1, argc - i); |
| 1480 |
argc -= 1; |
| 1481 |
break; |
| 1482 |
|
| 1483 |
case 'h': |
| 1484 |
ignore_hinshi_to_compare = 1; |
| 1485 |
shrinkargs (argv + i, 1, argc - i); |
| 1486 |
argc -= 1; |
| 1487 |
break; |
| 1488 |
|
| 1489 |
case 'i': |
| 1490 |
canna_type_output = 1; |
| 1491 |
wnn_type_output = 0; |
| 1492 |
shrinkargs (argv + i, 1, argc - i); |
| 1493 |
argc -= 1; |
| 1494 |
break; |
| 1495 |
|
| 1496 |
case 'j': |
| 1497 |
extract_kana = 1; |
| 1498 |
shrinkargs (argv + i, 1, argc - i); |
| 1499 |
argc -= 1; |
| 1500 |
break; |
| 1501 |
|
| 1502 |
case 'k': |
| 1503 |
{ |
| 1504 |
Wchar buf[READBUFSIZE]; |
| 1505 |
|
| 1506 |
(void) Mbstowcs (buf, argv[i + 1], READBUFSIZE); |
| 1507 |
specific_kind |= internkind (buf); |
| 1508 |
} |
| 1509 |
shrinkargs (argv + i, 2, argc - i); |
| 1510 |
argc -= 2; |
| 1511 |
break; |
| 1512 |
|
| 1513 |
case 'l': |
| 1514 |
list_kinds = 1; |
| 1515 |
shrinkargs (argv + i, 1, argc - i); |
| 1516 |
argc -= 1; |
| 1517 |
break; |
| 1518 |
|
| 1519 |
case 'm': |
| 1520 |
merge_kind = 1; |
| 1521 |
shrinkargs (argv + i, 1, argc - 1); |
| 1522 |
argc -= 1; |
| 1523 |
break; |
| 1524 |
|
| 1525 |
case 'n': |
| 1526 |
new_out = argv[i + 1]; |
| 1527 |
shrinkargs (argv + i, 2, argc - i); |
| 1528 |
argc -= 2; |
| 1529 |
break; |
| 1530 |
|
| 1531 |
case 'o': |
| 1532 |
old_out = argv[i + 1]; |
| 1533 |
shrinkargs (argv + i, 2, argc - i); |
| 1534 |
argc -= 2; |
| 1535 |
break; |
| 1536 |
|
| 1537 |
case 'p': |
| 1538 |
sort_by_frequency = 1; |
| 1539 |
shrinkargs (argv + i, 1, argc - i); |
| 1540 |
argc -= 1; |
| 1541 |
break; |
| 1542 |
|
| 1543 |
case 'r': |
| 1544 |
hinshi_table = argv[i + 1]; |
| 1545 |
shrinkargs (argv + i, 2, argc - i); |
| 1546 |
argc -= 2; |
| 1547 |
hinshi_direction = REVERSE; |
| 1548 |
break; |
| 1549 |
|
| 1550 |
case 's': |
| 1551 |
hinshi_table = argv[i + 1]; |
| 1552 |
shrinkargs (argv + i, 2, argc - i); |
| 1553 |
argc -= 2; |
| 1554 |
break; |
| 1555 |
|
| 1556 |
case 'v': |
| 1557 |
sj3_type_output = 1; |
| 1558 |
wnn_type_output = 1; /* Wnn ������������������������ */ |
| 1559 |
shrinkargs (argv + i, 1, argc - i); |
| 1560 |
argc -= 1; |
| 1561 |
break; |
| 1562 |
|
| 1563 |
case 'w': |
| 1564 |
canna_type_output = 0; |
| 1565 |
sj3_type_output = 0; |
| 1566 |
wnn_type_output = 1; |
| 1567 |
shrinkargs (argv + i, 1, argc - i); |
| 1568 |
argc -= 1; |
| 1569 |
break; |
| 1570 |
|
| 1571 |
case 'x': |
| 1572 |
merge_sj3 = 1; |
| 1573 |
ignore_hinshi_to_compare = 1; |
| 1574 |
shrinkargs (argv + i, 1, argc - i); |
| 1575 |
argc -= 1; |
| 1576 |
break; |
| 1577 |
|
| 1578 |
default: |
| 1579 |
i++; |
| 1580 |
break; |
| 1581 |
} |
| 1582 |
} |
| 1583 |
else |
| 1584 |
{ |
| 1585 |
i++; |
| 1586 |
} |
| 1587 |
} |
| 1588 |
|
| 1589 |
if (argc < 2) |
| 1590 |
{ |
| 1591 |
(void) fprintf (stderr, "Usage: %s dic1 [dic2] [-c filecommon] ...\n", program); |
| 1592 |
exit (1); |
| 1593 |
} |
| 1594 |
|
| 1595 |
if (argv[1][0] != '-' || argv[1][1]) |
| 1596 |
{ |
| 1597 |
in1 = fopen (argv[1], "r"); |
| 1598 |
if (!in1) |
| 1599 |
{ |
| 1600 |
(void) fprintf (stderr, "%s: can not open file \"%s\".\n", program, argv[1]); |
| 1601 |
exit (1); |
| 1602 |
} |
| 1603 |
} |
| 1604 |
if (argc == 3) |
| 1605 |
{ |
| 1606 |
if (argv[2][0] != '-' || argv[2][1]) |
| 1607 |
{ |
| 1608 |
in2 = fopen (argv[2], "r"); |
| 1609 |
if (!in2) |
| 1610 |
{ |
| 1611 |
(void) fprintf (stderr, "%s: can not open file \"%s\".\n", program, argv[2]); |
| 1612 |
exit (1); |
| 1613 |
} |
| 1614 |
} |
| 1615 |
} |
| 1616 |
else |
| 1617 |
{ |
| 1618 |
in2 = (FILE *) 0; |
| 1619 |
} |
| 1620 |
if (description_table) |
| 1621 |
{ |
| 1622 |
store_description (); |
| 1623 |
} |
| 1624 |
} |
| 1625 |
|
| 1626 |
static Wchar kihonh[] = { |
| 1627 |
(Wchar) 'k', (Wchar) 'i', (Wchar) 'h', (Wchar) 'o', (Wchar) 'n', (Wchar) 0, |
| 1628 |
}; |
| 1629 |
|
| 1630 |
int |
| 1631 |
main (argc, argv) |
| 1632 |
int argc; |
| 1633 |
char *argv[]; |
| 1634 |
{ |
| 1635 |
#ifndef POD_WCHAR |
| 1636 |
setlocale (LC_ALL, ""); |
| 1637 |
#endif |
| 1638 |
|
| 1639 |
in1 = in2 = stdin; |
| 1640 |
(void) internkind (kihonh); /* ������������1L ���������� */ |
| 1641 |
parseargs (argc, argv); |
| 1642 |
storepd (in1); |
| 1643 |
(void) fclose (in1); |
| 1644 |
|
| 1645 |
if (in2) |
| 1646 |
{ |
| 1647 |
compare = 1; |
| 1648 |
comparepd (in2); |
| 1649 |
(void) fclose (in2); |
| 1650 |
} |
| 1651 |
|
| 1652 |
if (list_kinds) |
| 1653 |
{ |
| 1654 |
listkinds (); |
| 1655 |
exit (0); |
| 1656 |
} |
| 1657 |
|
| 1658 |
if (selhinshi) |
| 1659 |
{ |
| 1660 |
select_hinshi (selhinshi); |
| 1661 |
} |
| 1662 |
else if (hinshi_table) |
| 1663 |
{ |
| 1664 |
replace_hinshi (); |
| 1665 |
} |
| 1666 |
|
| 1667 |
pdic = (struct dicpack **) malloc (ndicentries * sizeof (struct dicpack *)); |
| 1668 |
if (pdic) |
| 1669 |
{ |
| 1670 |
int i, j; |
| 1671 |
struct dicpack *p; |
| 1672 |
|
| 1673 |
for (i = 0, j = 0; i < DICBUFSIZE; i++) |
| 1674 |
{ |
| 1675 |
for (p = dic[i]; p; p = p->next) |
| 1676 |
{ |
| 1677 |
pdic[j++] = p; |
| 1678 |
} |
| 1679 |
} |
| 1680 |
if (sort_by_frequency) |
| 1681 |
{ |
| 1682 |
qsort (pdic, ndicentries, sizeof (struct dicpack *), dichindocompar); |
| 1683 |
} |
| 1684 |
else |
| 1685 |
{ |
| 1686 |
qsort (pdic, ndicentries, sizeof (struct dicpack *), diccompar); |
| 1687 |
} |
| 1688 |
sortkind (); |
| 1689 |
showentry (pdic, ndicentries); |
| 1690 |
} |
| 1691 |
else |
| 1692 |
{ |
| 1693 |
malloc_failed (); |
| 1694 |
} |
| 1695 |
exit (0); |
| 1696 |
} |