mdw@git.distorted.org.uk Git - anag/blob - anag.c

   1 /* -*-c-*-
   2  *
   3  * Main driver for anag
   4  *
   5  * (c) 2001 Mark Wooding
   6  */
   7
   8 /*----- Licensing notice --------------------------------------------------*
   9  *
  10  * This file is part of Anag: a simple wordgame helper.
  11  *
  12  * Anag is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * Anag is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with Anag; if not, write to the Free Software Foundation,
  24  * Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  25  */
  26
  27 /*----- Header files ------------------------------------------------------*/
  28
  29 #include "anag.h"
  30
  31 /*----- Static variables --------------------------------------------------*/
  32
  33 static const char *file = DICTIONARY;
  34
  35 /*----- Help text functions -----------------------------------------------*/
  36
  37 static void usage(FILE *fp)
  38   { pquis(fp, "Usage: $ [-f file] expression\n"); }
  39
  40 static void version(FILE *fp)
  41   { pquis(fp, "$, version " VERSION "\n"); }
  42
  43 static void help(FILE *fp)
  44 {
  45   version(fp);
  46   fputc('\n', fp);
  47   usage(fp);
  48   fputs("\n\
  49 Searches a wordlist, printing all of the words which match an expression.\n\
  50 \n\
  51 Options supported are:\n\
  52 \n\
  53 -h, --help              display this help text\n\
  54 -v, --version           display the program's version number\n\
  55 -u, --usage             display a very brief usage message\n\
  56 -f, --file FILE         read wordlist from FILE, not `" DICTIONARY "'\n\
  57 \n\
  58 The basic tests in the expression are:\n\
  59 \n\
  60 -anagram WORD           matches a full-length anagram\n\
  61 -subgram WORD           matches words which only use letters in WORD\n\
  62 -wildcard PATTERN       matches with wildcards `*' and `?'\n\
  63 -trackword WORD         matches words which can be found in a trackword\n\
  64 -mono PATTERN           matches words isomorphic to the given PATTERN\n\
  65 "
  66 #ifdef HAVE_REGCOMP
  67 "\
  68 -regexp REGEXP          matches with an (extended) regular expression\n\
  69 "
  70 #endif
  71 #if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
  72 "\
  73 -pcre REGEXP            matches with a Perl-like regular expression\n\
  74 "
  75 #endif
  76 "\
  77 -length [+|-]N          matches if length is [at least|at most] N\n\
  78 -longest                output longest matches found here\n\
  79 -shortest               output shortest matches found here\n\
  80 \n\
  81 These simple tests can be combined using the operators `-a', `-o' and `-n'\n\
  82 (for `and', `or' and `not'; they may also be written `&', `|' and `!' if\n\
  83 you like), and grouped using parentheses `(' and `)'.\n\
  84 ", fp); /*"*/
  85 }
  86
  87 /*----- The options parser ------------------------------------------------*/
  88
  89 /* --- Options table structure --- */
  90
  91 struct opt {
  92   const char *name;
  93   unsigned nargs;
  94   unsigned f;
  95   unsigned tag;
  96 };
  97
  98 enum {
  99   O_HELP, O_VERSION, O_USAGE,
 100   O_FILE,
 101   O_AND, O_OR, O_NOT, O_LPAREN, O_RPAREN,
 102   O_ANAG, O_SUBG, O_WILD, O_TRACK, O_REGEXP, O_PCRE, O_MONO, O_LENGTH,
 103   O_LONGEST, O_SHORTEST,
 104   O_EOF
 105 };
 106
 107 #define OF_SHORT 1u
 108
 109 static const struct opt opttab[] = {
 110
 111   /* --- Options -- don't form part of the language --- */
 112
 113   { "help",             0,      OF_SHORT,       O_HELP },
 114   { "version",          0,      OF_SHORT,       O_VERSION },
 115   { "usage",            0,      OF_SHORT,       O_USAGE },
 116   { "file",             1,      OF_SHORT,       O_FILE },
 117
 118   /* --- Operators -- provide the basic structure of the language --- *
 119    *
 120    * These are also given magical names by the parser.
 121    */
 122
 123   { "and",              0,      OF_SHORT,       O_AND },
 124   { "or",               0,      OF_SHORT,       O_OR },
 125   { "not",              0,      OF_SHORT,       O_NOT },
 126
 127   /* --- Actual matching operations -- do something useful --- */
 128
 129   { "anagram",          1,      0,              O_ANAG },
 130   { "subgram",          1,      0,              O_SUBG },
 131   { "wildcard",         1,      0,              O_WILD },
 132   { "trackword",        1,      0,              O_TRACK },
 133   { "mono",             1,      0,              O_MONO },
 134 #ifdef HAVE_REGCOMP
 135   { "regexp",           1,      0,              O_REGEXP },
 136 #endif
 137 #if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
 138   { "pcre",             1,      0,              O_PCRE },
 139 #endif
 140   { "length",           1,      0,              O_LENGTH },
 141   { "longest",          0,      0,              O_LONGEST },
 142   { "shortest",         0,      0,              O_SHORTEST },
 143
 144   /* --- End marker --- */
 145
 146   { 0,                  0,      0,              0 }
 147 };
 148
 149 static int ac;
 150 static const char *const *av;
 151 static int ai;
 152
 153 /* --- @nextopt@ --- *
 154  *
 155  * Arguments:   @const char ***arg@ = where to store the arg pointer
 156  *
 157  * Returns:     The tag of the next option.
 158  *
 159  * Use:         Scans the next option off the command line.  If the option
 160  *              doesn't form part of the language, it's processed internally,
 161  *              and you'll never see it from here.  On exit, the @arg@
 162  *              pointer is set to contain the address of the option scanned,
 163  *              followed by its arguments if any.  You're expected to know
 164  *              how many arguments there are for your option.
 165  */
 166
 167 static unsigned nextopt(const char *const **arg)
 168 {
 169   for (;;) {
 170     const struct opt *o, *oo;
 171     size_t sz;
 172     const char *p;
 173
 174     /* Pick the next option off the front. */
 175     *arg = av + ai;
 176     if (ai >= ac) return (O_EOF);
 177     p = av[ai++];
 178
 179     /* Cope with various forms of magic. */
 180     if (p[0] != '-') {
 181       if (!p[1]) switch (*p) {
 182         case '&': return (O_AND);
 183         case '|': return (O_OR);
 184         case '!': return (O_NOT);
 185         case '(': return (O_LPAREN);
 186         case ')': return (O_RPAREN);
 187       }
 188       goto bad;
 189     }
 190
 191     /* Now cope with other sorts of weirdies.  By the end of this, a leading
 192      * `-' or `--' will have been stripped.
 193      */
 194     p++;
 195     if (!*p) goto bad;
 196     if (*p == '-') p++;
 197     if (!*p) {
 198       if (ai < ac) die("syntax error near `--': rubbish at end of line");
 199       return (O_EOF);
 200     }
 201
 202     /*Now look the word up in my table. */
 203     sz = strlen(p);
 204     oo = 0;
 205     for (o = opttab; o->name; o++) {
 206       if (strncmp(p, o->name, sz) == 0) {
 207         if (strlen(o->name) == sz || ((o->f & OF_SHORT) && sz == 1))
 208           { oo = o; break; }
 209         if (oo)
 210           die("ambiguous option name `-%s' (could match `-%s' or `-%s')",
 211               p, oo->name, o->name);
 212         oo = o;
 213       }
 214     }
 215     if (!oo) die("unrecognized option name `-%s'", p);
 216
 217     /* Sort out the arguments. */
 218     if (ai + oo->nargs > ac)
 219       die("too few arguments for `-%s' (need %u)", oo->name, oo->nargs);
 220     ai += oo->nargs;
 221
 222     /* Now process the option. */
 223     switch (oo->tag) {
 224       case O_HELP: help(stdout); exit(0);
 225       case O_VERSION: version(stdout); exit(0);
 226       case O_USAGE: usage(stdout); exit(0);
 227       case O_FILE: file = (*arg)[1]; break;
 228       default: return (oo->tag);
 229     }
 230     continue;
 231   bad:
 232     die("syntax error near `%s': unknown token type", av[ai - 1]);
 233   }
 234 }
 235
 236 /*----- Node types for operators ------------------------------------------*/
 237
 238 /* --- Node structures --- */
 239
 240 typedef struct node_bin {
 241   node n;
 242   node *left;
 243   node *right;
 244 } node_bin;
 245
 246 typedef struct node_un {
 247   node n;
 248   node *arg;
 249 } node_un;
 250
 251 /* --- Node functions --- */
 252
 253 static int n_or(node *nn, const char *p, size_t sz)
 254 {
 255   node_bin *n = (node_bin *)nn;
 256   return (n->left->func(n->left, p, sz) || n->right->func(n->right, p, sz));
 257 }
 258
 259 static int n_and(node *nn, const char *p, size_t sz)
 260 {
 261   node_bin *n = (node_bin *)nn;
 262   return (n->left->func(n->left, p, sz) && n->right->func(n->right, p, sz));
 263 }
 264
 265 static int n_not(node *nn, const char *p, size_t sz)
 266 {
 267   node_un *n = (node_un *)nn;
 268   return (!n->arg->func(n->arg, p, sz));
 269 }
 270
 271 /*----- Other simple node types -------------------------------------------*/
 272
 273 enum { LESS = -1, EQUAL = 0, GREATER = 1 };
 274
 275 typedef struct node_numeric {
 276   node n;
 277   int dir;
 278   int i;
 279 } node_numeric;
 280
 281 static void parse_numeric(const char *p, int *dir, int *i)
 282 {
 283   long l;
 284   const char *pp = p;
 285   char *q;
 286
 287   switch (*pp) {
 288     case '-': *dir = LESS; pp++; break;
 289     case '+': *dir = GREATER; pp++; break;
 290     default: *dir = EQUAL; break;
 291   }
 292   errno = 0;
 293   l = strtol(pp, &q, 0);
 294   if (*q || errno || l < INT_MIN || l > INT_MAX)
 295     die("bad numeric parameter `%s'", p);
 296   *i = l;
 297 }
 298
 299 static node *make_numeric(const char *const *av,
 300                           int (*func)(struct node *, const char *, size_t))
 301 {
 302   node_numeric *n = xmalloc(sizeof(*n));
 303   parse_numeric(av[0], &n->dir, &n->i);
 304   n->n.func = func;
 305   return (&n->n);
 306 }
 307
 308 static int cmp_numeric(int x, int dir, int n)
 309 {
 310   switch (dir) {
 311     case LESS: return (x <= n);
 312     case EQUAL: return (x == n);
 313     case GREATER: return (x >= n);
 314   }
 315   abort();
 316 }
 317
 318 static int n_length(node *nn, const char *p, size_t sz)
 319 {
 320   node_numeric *n = (node_numeric *)nn;
 321   return (cmp_numeric(sz, n->dir, n->i));
 322 }
 323
 324 /*----- Parser for the expression syntax ----------------------------------*/
 325
 326 /* --- A parser context --- */
 327
 328 typedef struct p_ctx {
 329   unsigned t;
 330   const char *const *a;
 331 } p_ctx;
 332
 333 /* --- Parser structure --- *
 334  *
 335  * This is a simple recursive descent parser.  The context retains
 336  * information about the current token.  Each function is passed the address
 337  * of a node pointer to fill in.  This simplifies the binary operator code
 338  * somewhat, relative to returning pointers to node trees.
 339  */
 340
 341 static void p_expr(p_ctx *p, node **/*nn*/);
 342
 343 static void p_next(p_ctx *p)
 344 {
 345   static const char *const eof[] = { "<end>", 0 };
 346   p->t = nextopt(&p->a);
 347   if (p->t == O_EOF) p->a = eof;
 348 }
 349
 350 static void p_factor(p_ctx *p, node **nn)
 351 {
 352   node_un *n;
 353   if (p->t == O_LPAREN) {
 354     p_next(p);
 355     p_expr(p, nn);
 356     if (p->t != O_RPAREN) die("syntax error near `%s': missing `)'", *p->a);
 357     p_next(p);
 358   } else if (p->t == O_NOT) {
 359     n = xmalloc(sizeof(node_un));
 360     n->n.func = n_not;
 361     *nn = &n->n;
 362     p_next(p);
 363     p_factor(p, &n->arg);
 364   } else {
 365     switch (p->t) {
 366       case O_ANAG: *nn = anagram(p->a + 1); break;
 367       case O_SUBG: *nn = subgram(p->a + 1); break;
 368       case O_WILD: *nn = wildcard(p->a + 1); break;
 369       case O_TRACK: *nn = trackword(p->a + 1); break;
 370 #ifdef HAVE_REGCOMP
 371       case O_REGEXP: *nn = regexp(p->a + 1); break;
 372 #endif
 373 #if defined(HAVE_PCRE) || defined(HAVE_PCRE2)
 374       case O_PCRE: *nn = pcrenode(p->a + 1); break;
 375 #endif
 376       case O_MONO: *nn = mono(p->a + 1); break;
 377       case O_LENGTH: *nn = make_numeric(p->a + 1, n_length); break;
 378       case O_LONGEST: *nn = longest(p->a + 1); break;
 379       case O_SHORTEST: *nn = shortest(p->a + 1); break;
 380       default: die("syntax error near `%s': unexpected token", *p->a);
 381     }
 382     p_next(p);
 383   }
 384 }
 385
 386 static void p_term(p_ctx *p, node **nn)
 387 {
 388   node_bin *n;
 389   for (;;) {
 390     p_factor(p, nn);
 391     switch (p->t) {
 392       case O_AND: p_next(p); break;
 393       default: break;
 394       case O_RPAREN: case O_OR: case O_EOF: return;
 395     }
 396     n = xmalloc(sizeof(node_bin));
 397     n->left = *nn;
 398     n->n.func = n_and;
 399     *nn = &n->n;
 400     nn = &n->right;
 401   }
 402 }
 403
 404 static void p_expr(p_ctx *p, node **nn)
 405 {
 406   node_bin *n;
 407   for (;;) {
 408     p_term(p, nn);
 409     if (p->t != O_OR) break;
 410     p_next(p);
 411     n = xmalloc(sizeof(node_bin));
 412     n->left = *nn;
 413     n->n.func = n_or;
 414     *nn = &n->n;
 415     nn = &n->right;
 416   }
 417 }
 418
 419 /* --- @p_argv@ --- *
 420  *
 421  * Arguments:   @int argc@ = number of command-line arguments
 422  *              @const char *const argv[]@ = vectoor of arguments
 423  *
 424  * Returns:     A compiled node, parsed from the arguments.
 425  *
 426  * Use:         Does the donkey-work of parsing a command-line.
 427  */
 428
 429 static node *p_argv(int argc, const char *const argv[])
 430 {
 431   p_ctx p;
 432   node *n;
 433
 434   av = argv;
 435   ac = argc;
 436   ai = 1;
 437   p_next(&p);
 438   if (p.t == O_EOF) {
 439     usage(stderr);
 440     pquis(stderr, "(Run `$ --help' for more detail.)\n");
 441     exit(EXIT_FAILURE);
 442   }
 443   p_expr(&p, &n);
 444   if (p.t != O_EOF)
 445     die("syntax error near `%s': rubbish at end of line (too many `)'s?)",
 446         *p.a);
 447   return (n);
 448 }
 449
 450 /*----- At-end stuff ------------------------------------------------------*/
 451
 452 /* --- @atend_register@ --- *
 453  *
 454  * Arguments:   @int (*func)(void *)@ = function to call
 455  *              @void *p@ = handle to pass to it
 456  *
 457  * Returns:     ---
 458  *
 459  * Use:         Adds a function to the list of things to do at the end of the
 460  *              program.  The function should return nonzero if it produced
 461  *              any output.
 462  */
 463
 464 typedef struct atend {
 465   struct atend *next;
 466   int (*func)(void */*p*/);
 467   void *p;
 468 } atend;
 469
 470 static atend *aa_head = 0, **aa_tail = &aa_head;
 471
 472 void atend_register(int (*func)(void */*p*/), void *p)
 473 {
 474   atend *a = xmalloc(sizeof(*a));
 475   a->next = 0;
 476   a->func = func;
 477   a->p = p;
 478   *aa_tail = a;
 479   aa_tail = &a->next;
 480 }
 481
 482 /*----- Main code ---------------------------------------------------------*/
 483
 484 /* --- @main@ --- *
 485  *
 486  * Arguments:   @int argc@ = number of command-line arguments
 487  *              @char *argv[]@ = vector of argument words
 488  *
 489  * Returns:     Zero on success, nonzero on failure.
 490  *
 491  * Use:         Picks entries from a word list which match particular
 492  *              expressions.  This might be of assistance to word-game types.
 493  */
 494
 495 int main(int argc, char *argv[])
 496 {
 497   node *n;
 498   FILE *fp;
 499   dstr d = DSTR_INIT;
 500   int ok = 0;
 501   char *p, *q, *l;
 502   atend *a;
 503
 504   ego(argv[0]);
 505   n = p_argv(argc, (const char *const *)argv);
 506
 507   if ((fp = fopen(file, "r")) == 0)
 508     die("error opening `%s': %s", file, strerror(errno));
 509   for (;;) {
 510     dstr_reset(&d);
 511     if (dstr_putline(&d, fp) < 0) break;
 512     l = d.buf + d.len;
 513     for (p = q = d.buf; p < l; p++) {
 514       if (!isalnum((unsigned char)*p)) continue;
 515       *q++ = tolower((unsigned char)*p);
 516     }
 517     *q = 0;
 518     d.len = q - d.buf;
 519     if (n->func(n, d.buf, d.len)) {
 520       fwrite(d.buf, 1, d.len, stdout);
 521       fputc('\n', stdout);
 522       ok = 1;
 523     }
 524   }
 525   if (ferror(fp) || fclose(fp))
 526     die("error reading `%s': %s", file, strerror(errno));
 527   for (a = aa_head; a; a = a->next)
 528     if (a->func(a->p)) ok = 1;
 529   if (fflush(stdout) || ferror(stdout) || fclose(stdout))
 530     die("error writing output: %s", strerror(errno));
 531   if (!ok) pquis(stderr, "$: no matches found\n");
 532   return (ok ? EX_OK : EX_NONE);
 533 }
 534
 535 /*----- That's all, folks -------------------------------------------------*/