projects / unix-history / blame
| Commit | Line | Data |
|---|---|---|
| a36a91d2 SL |
1 | #ifndef lint |
| 2 | static char sccsid[] = "@(#)spell.c 4.1 %G%"; | |
| 3 | #endif | |
| 4 | ||
| 5 | #include "spell.h" | |
| 6 | #define DLEV 2 | |
| 7 | ||
| 8 | char *strcat(); | |
| 9 | int strip(); | |
| 10 | char *skipv(); | |
| 11 | int an(); | |
| 12 | int s(); | |
| 13 | int es(); | |
| 14 | int ily(); | |
| 15 | int ncy(); | |
| 16 | int CCe(); | |
| 17 | int VCe(); | |
| 18 | int bility(); | |
| 19 | int tion(); | |
| 20 | int ize(); | |
| 21 | int y_to_e(); | |
| 22 | int i_to_y(); | |
| 23 | int nop(); | |
| 24 | int metry(); | |
| 25 | ||
| 26 | struct suftab { | |
| 27 | char *suf; | |
| 28 | int (*p1)(); | |
| 29 | int n1; | |
| 30 | char *d1; | |
| 31 | char *a1; | |
| 32 | int (*p2)(); | |
| 33 | int n2; | |
| 34 | char *d2; | |
| 35 | char *a2; | |
| 36 | } suftab[] = { | |
| 37 | {"ssen",ily,4,"-y+iness","+ness" }, | |
| 38 | {"ssel",ily,4,"-y+i+less","+less" }, | |
| 39 | {"se",s,1,"","+s", es,2,"-y+ies","+es" }, | |
| 40 | {"s'",s,2,"","+'s"}, | |
| 41 | {"s",s,1,"","+s"}, | |
| 42 | {"ecn",ncy,1,"","-t+ce"}, | |
| 43 | {"ycn",ncy,1,"","-cy+t"}, | |
| 44 | {"ytilb",nop,0,"",""}, | |
| 45 | {"ytilib",bility,5,"-le+ility",""}, | |
| 46 | {"elbaif",i_to_y,4,"-y+iable",""}, | |
| 47 | {"elba",CCe,4,"-e+able","+able"}, | |
| 48 | {"yti",CCe,3,"-e+ity","+ity"}, | |
| 49 | {"ylb",y_to_e,1,"-e+y",""}, | |
| 50 | {"yl",ily,2,"-y+ily","+ly"}, | |
| 51 | {"laci",strip,2,"","+al"}, | |
| 52 | {"latnem",strip,2,"","+al"}, | |
| 53 | {"lanoi",strip,2,"","+al"}, | |
| 54 | {"tnem",strip,4,"","+ment"}, | |
| 55 | {"gni",CCe,3,"-e+ing","+ing"}, | |
| 56 | {"reta",nop,0,"",""}, | |
| 57 | {"re",strip,1,"","+r", i_to_y,2,"-y+ier","+er"}, | |
| 58 | {"de",strip,1,"","+d", i_to_y,2,"-y+ied","+ed"}, | |
| 59 | {"citsi",strip,2,"","+ic"}, | |
| 60 | {"cihparg",i_to_y,1,"-y+ic",""}, | |
| 61 | {"tse",strip,2,"","+st", i_to_y,3,"-y+iest","+est"}, | |
| 62 | {"cirtem",i_to_y,1,"-y+ic",""}, | |
| 63 | {"yrtem",metry,0,"-ry+er",""}, | |
| 64 | {"cigol",i_to_y,1,"-y+ic",""}, | |
| 65 | {"tsigol",i_to_y,2,"-y+ist",""}, | |
| 66 | {"tsi",VCe,3,"-e+ist","+ist"}, | |
| 67 | {"msi",VCe,3,"-e+ism","+ist"}, | |
| 68 | {"noitacif",i_to_y,6,"-y+ication",""}, | |
| 69 | {"noitazi",ize,5,"-e+ation",""}, | |
| 70 | {"rota",tion,2,"-e+or",""}, | |
| 71 | {"noit",tion,3,"-e+ion","+ion"}, | |
| 72 | {"naino",an,3,"","+ian"}, | |
| 73 | {"na",an,1,"","+n"}, | |
| 74 | {"evit",tion,3,"-e+ive","+ive"}, | |
| 75 | {"ezi",CCe,3,"-e+ize","+ize"}, | |
| 76 | {"pihs",strip,4,"","+ship"}, | |
| 77 | {"dooh",ily,4,"-y+hood","+hood"}, | |
| 78 | {"ekil",strip,4,"","+like"}, | |
| 79 | 0 | |
| 80 | }; | |
| 81 | ||
| 82 | char *preftab[] = { | |
| 83 | "anti", | |
| 84 | "bio", | |
| 85 | "dis", | |
| 86 | "electro", | |
| 87 | "en", | |
| 88 | "fore", | |
| 89 | "hyper", | |
| 90 | "intra", | |
| 91 | "inter", | |
| 92 | "iso", | |
| 93 | "kilo", | |
| 94 | "magneto", | |
| 95 | "meta", | |
| 96 | "micro", | |
| 97 | "milli", | |
| 98 | "mis", | |
| 99 | "mono", | |
| 100 | "multi", | |
| 101 | "non", | |
| 102 | "out", | |
| 103 | "over", | |
| 104 | "photo", | |
| 105 | "poly", | |
| 106 | "pre", | |
| 107 | "pseudo", | |
| 108 | "re", | |
| 109 | "semi", | |
| 110 | "stereo", | |
| 111 | "sub", | |
| 112 | "super", | |
| 113 | "thermo", | |
| 114 | "ultra", | |
| 115 | "under", /*must precede un*/ | |
| 116 | "un", | |
| 117 | 0 | |
| 118 | }; | |
| 119 | ||
| 120 | int vflag; | |
| 121 | int xflag; | |
| 122 | char word[100]; | |
| 123 | char original[100]; | |
| 124 | char *deriv[40]; | |
| 125 | char affix[40]; | |
| 126 | ||
| 127 | main(argc,argv) | |
| 128 | char **argv; | |
| 129 | { | |
| 130 | register char *ep, *cp; | |
| 131 | register char *dp; | |
| 132 | int fold; | |
| 133 | int j; | |
| 134 | FILE *file, *found; | |
| 135 | if(!prime(argc,argv)) { | |
| 136 | fprintf(stderr, | |
| 137 | "spell: cannot initialize hash table\n"); | |
| 138 | exit(1); | |
| 139 | } | |
| 140 | found = fopen(argv[2],"w"); | |
| 141 | for(argc-=3,argv+=3; argc>0 && argv[0][0]=='-'; argc--,argv++) | |
| 142 | switch(argv[0][1]) { | |
| 143 | case 'b': | |
| 144 | ise(); | |
| 145 | break; | |
| 146 | case 'v': | |
| 147 | vflag++; | |
| 148 | break; | |
| 149 | case 'x': | |
| 150 | xflag++; | |
| 151 | break; | |
| 152 | } | |
| 153 | for(;; fprintf(file,"%s%s\n",affix,original)) { | |
| 154 | affix[0] = 0; | |
| 155 | file = found; | |
| 156 | for(ep=word;(*ep=j=getchar())!='\n';ep++) | |
| 157 | if(j == EOF) | |
| 158 | exit(0); | |
| 159 | for(cp=word,dp=original; cp<ep; ) | |
| 160 | *dp++ = *cp++; | |
| 161 | *dp = 0; | |
| 162 | fold = 0; | |
| 163 | for(cp=word;cp<ep;cp++) | |
| 164 | if(islower(*cp)) | |
| 165 | goto lcase; | |
| 166 | if(putsuf(ep,".",0)) | |
| 167 | continue; | |
| 168 | ++fold; | |
| 169 | for(cp=original+1,dp=word+1;dp<ep;dp++,cp++) | |
| 170 | *dp = Tolower(*cp); | |
| 171 | lcase: | |
| 172 | if(putsuf(ep,".",0)||suffix(ep,0)) | |
| 173 | continue; | |
| 174 | if(isupper(word[0])) { | |
| 175 | for(cp=original,dp=word; *dp = *cp++; dp++) | |
| 176 | if (fold) *dp = Tolower(*dp); | |
| 177 | word[0] = Tolower(word[0]); | |
| 178 | goto lcase; | |
| 179 | } | |
| 180 | file = stdout; | |
| 181 | } | |
| 182 | } | |
| 183 | ||
| 184 | suffix(ep,lev) | |
| 185 | char *ep; | |
| 186 | { | |
| 187 | register struct suftab *t; | |
| 188 | register char *cp, *sp; | |
| 189 | lev += DLEV; | |
| 190 | deriv[lev] = deriv[lev-1] = 0; | |
| 191 | for(t= &suftab[0];sp=t->suf;t++) { | |
| 192 | cp = ep; | |
| 193 | while(*sp) | |
| 194 | if(*--cp!=*sp++) | |
| 195 | goto next; | |
| 196 | for(sp=cp; --sp>=word&&!vowel(*sp); ) ; | |
| 197 | if(sp<word) | |
| 198 | return(0); | |
| 199 | if((*t->p1)(ep-t->n1,t->d1,t->a1,lev+1)) | |
| 200 | return(1); | |
| 201 | if(t->p2!=0) { | |
| 202 | deriv[lev] = deriv[lev+1] = 0; | |
| 203 | return((*t->p2)(ep-t->n2,t->d2,t->a2,lev)); | |
| 204 | } | |
| 205 | return(0); | |
| 206 | next: ; | |
| 207 | } | |
| 208 | return(0); | |
| 209 | } | |
| 210 | ||
| 211 | nop() | |
| 212 | { | |
| 213 | return(0); | |
| 214 | } | |
| 215 | ||
| 216 | strip(ep,d,a,lev) | |
| 217 | char *ep,*d,*a; | |
| 218 | { | |
| 219 | return(putsuf(ep,a,lev)||suffix(ep,lev)); | |
| 220 | } | |
| 221 | ||
| 222 | s(ep,d,a,lev) | |
| 223 | char *ep,*d,*a; | |
| 224 | { | |
| 225 | if(lev>DLEV+1) | |
| 226 | return(0); | |
| 227 | if(*ep=='s'&&ep[-1]=='s') | |
| 228 | return(0); | |
| 229 | return(strip(ep,d,a,lev)); | |
| 230 | } | |
| 231 | ||
| 232 | an(ep,d,a,lev) | |
| 233 | char *ep,*d,*a; | |
| 234 | { | |
| 235 | if(!isupper(*word)) /*must be proper name*/ | |
| 236 | return(0); | |
| 237 | return(putsuf(ep,a,lev)); | |
| 238 | } | |
| 239 | ||
| 240 | ize(ep,d,a,lev) | |
| 241 | char *ep,*d,*a; | |
| 242 | { | |
| 243 | *ep++ = 'e'; | |
| 244 | return(strip(ep,"",d,lev)); | |
| 245 | } | |
| 246 | ||
| 247 | y_to_e(ep,d,a,lev) | |
| 248 | char *ep,*d,*a; | |
| 249 | { | |
| 250 | *ep++ = 'e'; | |
| 251 | return(strip(ep,"",d,lev)); | |
| 252 | } | |
| 253 | ||
| 254 | ily(ep,d,a,lev) | |
| 255 | char *ep,*d,*a; | |
| 256 | { | |
| 257 | if(ep[-1]=='i') | |
| 258 | return(i_to_y(ep,d,a,lev)); | |
| 259 | else | |
| 260 | return(strip(ep,d,a,lev)); | |
| 261 | } | |
| 262 | ||
| 263 | ncy(ep,d,a,lev) | |
| 264 | char *ep, *d, *a; | |
| 265 | { | |
| 266 | if(skipv(skipv(ep-1))<word) | |
| 267 | return(0); | |
| 268 | ep[-1] = 't'; | |
| 269 | return(strip(ep,d,a,lev)); | |
| 270 | } | |
| 271 | ||
| 272 | bility(ep,d,a,lev) | |
| 273 | char *ep,*d,*a; | |
| 274 | { | |
| 275 | *ep++ = 'l'; | |
| 276 | return(y_to_e(ep,d,a,lev)); | |
| 277 | } | |
| 278 | ||
| 279 | i_to_y(ep,d,a,lev) | |
| 280 | char *ep,*d,*a; | |
| 281 | { | |
| 282 | if(ep[-1]=='i') { | |
| 283 | ep[-1] = 'y'; | |
| 284 | a = d; | |
| 285 | } | |
| 286 | return(strip(ep,"",a,lev)); | |
| 287 | } | |
| 288 | ||
| 289 | es(ep,d,a,lev) | |
| 290 | char *ep,*d,*a; | |
| 291 | { | |
| 292 | if(lev>DLEV) | |
| 293 | return(0); | |
| 294 | switch(ep[-1]) { | |
| 295 | default: | |
| 296 | return(0); | |
| 297 | case 'i': | |
| 298 | return(i_to_y(ep,d,a,lev)); | |
| 299 | case 's': | |
| 300 | case 'h': | |
| 301 | case 'z': | |
| 302 | case 'x': | |
| 303 | return(strip(ep,d,a,lev)); | |
| 304 | } | |
| 305 | } | |
| 306 | ||
| 307 | metry(ep,d,a,lev) | |
| 308 | char *ep, *d,*a; | |
| 309 | { | |
| 310 | ep[-2] = 'e'; | |
| 311 | ep[-1] = 'r'; | |
| 312 | return(strip(ep,d,a,lev)); | |
| 313 | } | |
| 314 | ||
| 315 | tion(ep,d,a,lev) | |
| 316 | char *ep,*d,*a; | |
| 317 | { | |
| 318 | switch(ep[-2]) { | |
| 319 | case 'c': | |
| 320 | case 'r': | |
| 321 | return(putsuf(ep,a,lev)); | |
| 322 | case 'a': | |
| 323 | return(y_to_e(ep,d,a,lev)); | |
| 324 | } | |
| 325 | return(0); | |
| 326 | } | |
| 327 | ||
| 328 | /* possible consonant-consonant-e ending*/ | |
| 329 | CCe(ep,d,a,lev) | |
| 330 | char *ep,*d,*a; | |
| 331 | { | |
| 332 | switch(ep[-1]) { | |
| 333 | case 'l': | |
| 334 | if(vowel(ep[-2])) | |
| 335 | break; | |
| 336 | switch(ep[-2]) { | |
| 337 | case 'l': | |
| 338 | case 'r': | |
| 339 | case 'w': | |
| 340 | break; | |
| 341 | default: | |
| 342 | return(y_to_e(ep,d,a,lev)); | |
| 343 | } | |
| 344 | break; | |
| 345 | case 's': | |
| 346 | if(ep[-2]=='s') | |
| 347 | break; | |
| 348 | case 'c': | |
| 349 | case 'g': | |
| 350 | if(*ep=='a') | |
| 351 | return(0); | |
| 352 | case 'v': | |
| 353 | case 'z': | |
| 354 | if(vowel(ep[-2])) | |
| 355 | break; | |
| 356 | case 'u': | |
| 357 | if(y_to_e(ep,d,a,lev)) | |
| 358 | return(1); | |
| 359 | if(!(ep[-2]=='n'&&ep[-1]=='g')) | |
| 360 | return(0); | |
| 361 | } | |
| 362 | return(VCe(ep,d,a,lev)); | |
| 363 | } | |
| 364 | ||
| 365 | /* possible consonant-vowel-consonant-e ending*/ | |
| 366 | VCe(ep,d,a,lev) | |
| 367 | char *ep,*d,*a; | |
| 368 | { | |
| 369 | char c; | |
| 370 | c = ep[-1]; | |
| 371 | if(c=='e') | |
| 372 | return(0); | |
| 373 | if(!vowel(c) && vowel(ep[-2])) { | |
| 374 | c = *ep; | |
| 375 | *ep++ = 'e'; | |
| 376 | if(putsuf(ep,d,lev)||suffix(ep,lev)) | |
| 377 | return(1); | |
| 378 | ep--; | |
| 379 | *ep = c; | |
| 380 | } | |
| 381 | return(strip(ep,d,a,lev)); | |
| 382 | } | |
| 383 | ||
| 384 | char *lookuppref(wp,ep) | |
| 385 | char **wp; | |
| 386 | char *ep; | |
| 387 | { | |
| 388 | register char **sp; | |
| 389 | register char *bp,*cp; | |
| 390 | for(sp=preftab;*sp;sp++) { | |
| 391 | bp = *wp; | |
| 392 | for(cp= *sp;*cp;cp++,bp++) | |
| 393 | if(Tolower(*bp)!=*cp) | |
| 394 | goto next; | |
| 395 | for(cp=bp;cp<ep;cp++) | |
| 396 | if(vowel(*cp)) { | |
| 397 | *wp = bp; | |
| 398 | return(*sp); | |
| 399 | } | |
| 400 | next: ; | |
| 401 | } | |
| 402 | return(0); | |
| 403 | } | |
| 404 | ||
| 405 | putsuf(ep,a,lev) | |
| 406 | char *ep,*a; | |
| 407 | { | |
| 408 | register char *cp; | |
| 409 | char *bp; | |
| 410 | register char *pp; | |
| 411 | int val = 0; | |
| 412 | char space[20]; | |
| 413 | deriv[lev] = a; | |
| 414 | if(putw(word,ep,lev)) | |
| 415 | return(1); | |
| 416 | bp = word; | |
| 417 | pp = space; | |
| 418 | deriv[lev+1] = pp; | |
| 419 | while(cp=lookuppref(&bp,ep)) { | |
| 420 | *pp++ = '+'; | |
| 421 | while(*pp = *cp++) | |
| 422 | pp++; | |
| 423 | if(putw(bp,ep,lev+1)) { | |
| 424 | val = 1; | |
| 425 | break; | |
| 426 | } | |
| 427 | } | |
| 428 | deriv[lev+1] = deriv[lev+2] = 0; | |
| 429 | return(val); | |
| 430 | } | |
| 431 | ||
| 432 | putw(bp,ep,lev) | |
| 433 | char *bp,*ep; | |
| 434 | { | |
| 435 | register i, j; | |
| 436 | char duple[3]; | |
| 437 | if(ep-bp<=1) | |
| 438 | return(0); | |
| 439 | if(vowel(*ep)) { | |
| 440 | if(monosyl(bp,ep)) | |
| 441 | return(0); | |
| 442 | } | |
| 443 | i = dict(bp,ep); | |
| 444 | if(i==0&&vowel(*ep)&&ep[-1]==ep[-2]&&monosyl(bp,ep-1)) { | |
| 445 | ep--; | |
| 446 | deriv[++lev] = duple; | |
| 447 | duple[0] = '+'; | |
| 448 | duple[1] = *ep; | |
| 449 | duple[2] = 0; | |
| 450 | i = dict(bp,ep); | |
| 451 | } | |
| 452 | if(vflag==0||i==0) | |
| 453 | return(i); | |
| 454 | j = lev; | |
| 455 | do { | |
| 456 | if(deriv[j]) | |
| 457 | strcat(affix,deriv[j]); | |
| 458 | } while(--j>0); | |
| 459 | strcat(affix,"\t"); | |
| 460 | return(i); | |
| 461 | } | |
| 462 | ||
| 463 | ||
| 464 | monosyl(bp,ep) | |
| 465 | char *bp, *ep; | |
| 466 | { | |
| 467 | if(ep<bp+2) | |
| 468 | return(0); | |
| 469 | if(vowel(*--ep)||!vowel(*--ep) | |
| 470 | ||ep[1]=='x'||ep[1]=='w') | |
| 471 | return(0); | |
| 472 | while(--ep>=bp) | |
| 473 | if(vowel(*ep)) | |
| 474 | return(0); | |
| 475 | return(1); | |
| 476 | } | |
| 477 | ||
| 478 | char * | |
| 479 | skipv(s) | |
| 480 | char *s; | |
| 481 | { | |
| 482 | if(s>=word&&vowel(*s)) | |
| 483 | s--; | |
| 484 | while(s>=word&&!vowel(*s)) | |
| 485 | s--; | |
| 486 | return(s); | |
| 487 | } | |
| 488 | ||
| 489 | vowel(c) | |
| 490 | { | |
| 491 | switch(Tolower(c)) { | |
| 492 | case 'a': | |
| 493 | case 'e': | |
| 494 | case 'i': | |
| 495 | case 'o': | |
| 496 | case 'u': | |
| 497 | case 'y': | |
| 498 | return(1); | |
| 499 | } | |
| 500 | return(0); | |
| 501 | } | |
| 502 | ||
| 503 | /* crummy way to Britishise */ | |
| 504 | ise() | |
| 505 | { | |
| 506 | register struct suftab *p; | |
| 507 | for(p = suftab;p->suf;p++) { | |
| 508 | ztos(p->suf); | |
| 509 | ztos(p->d1); | |
| 510 | ztos(p->a1); | |
| 511 | } | |
| 512 | } | |
| 513 | ztos(s) | |
| 514 | char *s; | |
| 515 | { | |
| 516 | for(;*s;s++) | |
| 517 | if(*s=='z') | |
| 518 | *s = 's'; | |
| 519 | } | |
| 520 | ||
| 521 | dict(bp,ep) | |
| 522 | char *bp, *ep; | |
| 523 | { | |
| 524 | register char *wp; | |
| 525 | long h; | |
| 526 | register long *lp; | |
| 527 | register i; | |
| 528 | if(xflag) | |
| 529 | printf("=%.*s\n",ep-bp,bp); | |
| 530 | for(i=0; i<NP; i++) { | |
| 531 | for (wp = bp, h = 0, lp = pow2[i]; wp < ep; ++wp, ++lp) | |
| 532 | h += *wp * *lp; | |
| 533 | h += '\n' * *lp; | |
| 534 | h %= p[i]; | |
| 535 | if(get(h)==0) | |
| 536 | return(0); | |
| 537 | } | |
| 538 | return(1); | |
| 539 | } |