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

   1 /* -*-c-*-
   2  *
   3  * $Id$
   4  *
   5  * Main driver for anag
   6  *
   7  * (c) 2001 Mark Wooding
   8  */
   9
  10 /*----- Licensing notice --------------------------------------------------*
  11  *
  12  * This file is part of Anag: a simple wordgame helper.
  13  *
  14  * Anag is free software; you can redistribute it and/or modify
  15  * it under the terms of the GNU General Public License as published by
  16  * the Free Software Foundation; either version 2 of the License, or
  17  * (at your option) any later version.
  18  *
  19  * Anag is distributed in the hope that it will be useful,
  20  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  21  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  22  * GNU General Public License for more details.
  23  *
  24  * You should have received a copy of the GNU General Public License
  25  * along with Anag; if not, write to the Free Software Foundation,
  26  * Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
  27  */
  28
  29 /*----- Header files ------------------------------------------------------*/
  30
  31 #include "anag.h"
  32
  33 /*----- Static variables --------------------------------------------------*/
  34
  35 static const char *file = DICTIONARY;
  36
  37 /*----- Help text functions -----------------------------------------------*/
  38
  39 static void usage(FILE *fp)
  40 {
  41   pquis(fp, "Usage: $ [-f file] expression\n");
  42 }
  43
  44 static void version(FILE *fp)
  45 {
  46   pquis(fp, "$, version " VERSION "\n");
  47 }
  48
  49 static void help(FILE *fp)
  50 {
  51   version(fp);
  52   fputc('\n', fp);
  53   usage(fp);
  54   fputs("\n\
  55 Searches a wordlist, printing all of the words which match an expression.\n\
  56 \n\
  57 Options supported are:\n\
  58 \n\
  59 -h, --help              display this help text\n\
  60 -v, --version           display the program's version number\n\
  61 -u, --usage             display a very brief usage message\n\
  62 -f, --file FILE         read wordlist from FILE, not `" DICTIONARY "'\n\
  63 \n\
  64 The basic tests in the expression are:\n\
  65 \n\
  66 -anagram WORD           matches a full-length anagram\n\
  67 -subgram WORD           matches words which only use letters in WORD\n\
  68 -wildcard PATTERN       matches with wildcards `*' and `?'\n\
  69 -trackword WORD         matches words which can be found in a trackword\n\
  70 -mono PATTERN           matches words isomorphic to the given PATTERN\n\
  71 "
  72 #ifdef HAVE_REGCOMP
  73 "\
  74 -regexp REGEXP          matches with an (extended) regular expression\n\
  75 "
  76 #endif
  77 #ifdef HAVE_PCRE
  78 "\
  79 -pcre REGEXP            matches with a Perl-like regular expression\n\
  80 "
  81 #endif
  82 "\
  83 -length [+|-]N          matches if length is [at least|at most] N\n\
  84 -longest                output longest matches found here\n\
  85 -shortest               output shortest matches found here\n\
  86 \n\
  87 These simple tests can be combined using the operators `-a', `-o' and `-n'\n\
  88 (for `and', `or' and `not'; they may also be written `&', `|' and `!' if\n\
  89 you like), and grouped using parentheses `(' and `)'.\n\
  90 ", fp); /*"*/
  91 }
  92
  93 /*----- The options parser ------------------------------------------------*/
  94
  95 /* --- Options table structure --- */
  96
  97 struct opt {
  98   const char *name;
  99   unsigned nargs;
 100   unsigned f;
 101   unsigned tag;
 102 };
 103
 104 enum {
 105   O_HELP, O_VERSION, O_USAGE,
 106   O_FILE,
 107   O_AND, O_OR, O_NOT, O_LPAREN, O_RPAREN,
 108   O_ANAG, O_SUBG, O_WILD, O_TRACK, O_REGEXP, O_PCRE, O_MONO, O_LENGTH,
 109   O_LONGEST, O_SHORTEST,
 110   O_EOF
 111 };
 112
 113 #define OF_SHORT 1u
 114
 115 static const struct opt opttab[] = {
 116
 117   /* --- Options -- don't form part of the language --- */
 118
 119   { "help",             0,      OF_SHORT,       O_HELP },
 120   { "version",          0,      OF_SHORT,       O_VERSION },
 121   { "usage",            0,      OF_SHORT,       O_USAGE },
 122   { "file",             1,      OF_SHORT,       O_FILE },
 123
 124   /* --- Operators -- provide the basic structure of the language --- *
 125    *
 126    * These are also given magical names by the parser.
 127    */
 128
 129   { "and",              0,      OF_SHORT,       O_AND },
 130   { "or",               0,      OF_SHORT,       O_OR },
 131   { "not",              0,      OF_SHORT,       O_NOT },
 132
 133   /* --- Actual matching operations -- do something useful --- */
 134
 135   { "anagram",          1,      0,              O_ANAG },
 136   { "subgram",          1,      0,              O_SUBG },
 137   { "wildcard",         1,      0,              O_WILD },
 138   { "trackword",        1,      0,              O_TRACK },
 139   { "mono",             1,      0,              O_MONO },
 140 #ifdef HAVE_REGCOMP
 141   { "regexp",           1,      0,              O_REGEXP },
 142 #endif
 143 #ifdef HAVE_PCRE
 144   { "pcre",             1,      0,              O_PCRE },
 145 #endif
 146   { "length",           1,      0,              O_LENGTH },
 147   { "longest",          0,      0,              O_LONGEST },
 148   { "shortest",         0,      0,              O_SHORTEST },
 149
 150   /* --- End marker --- */
 151
 152   { 0,                  0,      0,              0 }
 153 };
 154
 155 static int ac;
 156 static const char *const *av;
 157 static int ai;
 158
 159 /* --- @nextopt@ --- *
 160  *
 161  * Arguments:   @const char ***arg@ = where to store the arg pointer
 162  *
 163  * Returns:     The tag of the next option.
 164  *
 165  * Use:         Scans the next option off the command line.  If the option
 166  *              doesn't form part of the language, it's processed internally,
 167  *              and you'll never see it from here.  On exit, the @arg@
 168  *              pointer is set to contain the address of the option scanned,
 169  *              followed by its arguments if any.  You're expected to know
 170  *              how many arguments there are for your option.
 171  */
 172
 173 static unsigned nextopt(const char *const **arg)
 174 {
 175   for (;;) {
 176     const struct opt *o, *oo;
 177     size_t sz;
 178     const char *p;
 179
 180     /* --- Pick the next option off the front --- */
 181
 182     *arg = av + ai;
 183     if (ai >= ac)
 184       return (O_EOF);
 185     p = av[ai++];
 186
 187     /* --- Cope with various forms of magic --- */
 188
 189     if (p[0] != '-') {
 190       if (!p[1]) switch (*p) {
 191         case '&': return (O_AND);
 192         case '|': return (O_OR);
 193         case '!': return (O_NOT);
 194         case '(': return (O_LPAREN);
 195         case ')': return (O_RPAREN);
 196       }
 197       goto bad;
 198     }
 199
 200     /* --- Now cope with other sorts of weirdies --- *
 201      *
 202      * By the end of this, a leading `-' or `--' will have been stripped.
 203      */
 204
 205     p++;
 206     if (!*p)
 207       goto bad;
 208     if (*p == '-')
 209       p++;
 210     if (!*p) {
 211       if (ai < ac)
 212         die("syntax error near `--': rubbish at end of line");
 213       return (O_EOF);
 214     }
 215
 216     /* --- Now look the word up in my table --- */
 217
 218     sz = strlen(p);
 219     oo = 0;
 220     for (o = opttab; o->name; o++) {
 221       if (strncmp(p, o->name, sz) == 0) {
 222         if (strlen(o->name) == sz || ((o->f & OF_SHORT) && sz == 1)) {
 223           oo = o;
 224           break;
 225         }
 226         if (oo) {
 227           die("ambiguous option name `-%s' (could match `-%s' or `-%s')",
 228               p, oo->name, o->name);
 229         }
 230         oo = o;
 231       }
 232     }
 233     if (!oo)
 234       die("unrecognized option name `-%s'", p);
 235
 236     /* --- Sort out the arguments --- */
 237
 238     if (ai + oo->nargs > ac)
 239       die("too few arguments for `-%s' (need %u)", oo->name, oo->nargs);
 240     ai += oo->nargs;
 241
 242     /* --- Now process the option --- */
 243
 244     switch (oo->tag) {
 245       case O_HELP:
 246         help(stdout);
 247         exit(0);
 248       case O_VERSION:
 249         version(stdout);
 250         exit(0);
 251       case O_USAGE:
 252         usage(stdout);
 253         exit(0);
 254       case O_FILE:
 255         file = (*arg)[1];
 256         break;
 257       default:
 258         return (oo->tag);
 259     }
 260     continue;
 261   bad:
 262     die("syntax error near `%s': unknown token type", av[ai - 1]);
 263   }
 264 }
 265
 266 /*----- Node types for operators ------------------------------------------*/
 267
 268 /* --- Node structures --- */
 269
 270 typedef struct node_bin {
 271   node n;
 272   node *left;
 273   node *right;
 274 } node_bin;
 275
 276 typedef struct node_un {
 277   node n;
 278   node *arg;
 279 } node_un;
 280
 281 /* --- Node functions --- */
 282
 283 static int n_or(node *nn, const char *p, size_t sz)
 284 {
 285   node_bin *n = (node_bin *)nn;
 286   return (n->left->func(n->left, p, sz) || n->right->func(n->right, p, sz));
 287 }
 288
 289 static int n_and(node *nn, const char *p, size_t sz)
 290 {
 291   node_bin *n = (node_bin *)nn;
 292   return (n->left->func(n->left, p, sz) && n->right->func(n->right, p, sz));
 293 }
 294
 295 static int n_not(node *nn, const char *p, size_t sz)
 296 {
 297   node_un *n = (node_un *)nn;
 298   return (!n->arg->func(n->arg, p, sz));
 299 }
 300
 301 /*----- Other simple node types -------------------------------------------*/
 302
 303 enum { LESS = -1, EQUAL = 0, GREATER = 1 };
 304
 305 typedef struct node_numeric {
 306   node n;
 307   int dir;
 308   int i;
 309 } node_numeric;
 310
 311 static void parse_numeric(const char *p, int *dir, int *i)
 312 {
 313   long l;
 314   const char *pp = p;
 315   char *q;
 316
 317   switch (*pp) {
 318     case '-': *dir = LESS; pp++; break;
 319     case '+': *dir = GREATER; pp++; break;
 320     default: *dir = EQUAL; break;
 321   }
 322   errno = 0;
 323   l = strtol(pp, &q, 0);
 324   if (*q || errno || l < INT_MIN || l > INT_MAX)
 325     die("bad numeric parameter `%s'", p);
 326   *i = l;
 327 }
 328
 329 static node *make_numeric(const char *const *av,
 330                           int (*func)(struct node *, const char *, size_t))
 331 {
 332   node_numeric *n = xmalloc(sizeof(*n));
 333   parse_numeric(av[0], &n->dir, &n->i);
 334   n->n.func = func;
 335   return (&n->n);
 336 }
 337
 338 static int cmp_numeric(int x, int dir, int n)
 339 {
 340   switch (dir) {
 341     case LESS: return (x <= n);
 342     case EQUAL: return (x == n);
 343     case GREATER: return (x >= n);
 344   }
 345   abort();
 346 }
 347
 348 static int n_length(node *nn, const char *p, size_t sz)
 349 {
 350   node_numeric *n = (node_numeric *)nn;
 351   return (cmp_numeric(sz, n->dir, n->i));
 352 }
 353
 354 /*----- Parser for the expression syntax ----------------------------------*/
 355
 356 /* --- A parser context --- */
 357
 358 typedef struct p_ctx {
 359   unsigned t;
 360   const char *const *a;
 361 } p_ctx;
 362
 363 /* --- Parser structure --- *
 364  *
 365  * This is a simple recursive descent parser.  The context retains
 366  * information about the current token.  Each function is passed the address
 367  * of a node pointer to fill in.  This simplifies the binary operator code
 368  * somewhat, relative to returning pointers to node trees.
 369  */
 370
 371 static void p_expr(p_ctx *p, node **/*nn*/);
 372
 373 static void p_next(p_ctx *p)
 374 {
 375   static const char *const eof[] = { "<end>", 0 };
 376   p->t = nextopt(&p->a);
 377   if (p->t == O_EOF)
 378     p->a = eof;
 379 }
 380
 381 static void p_factor(p_ctx *p, node **nn)
 382 {
 383   node_un *n;
 384   if (p->t == O_LPAREN) {
 385     p_next(p);
 386     p_expr(p, nn);
 387     if (p->t != O_RPAREN)
 388       die("syntax error near `%s': missing `)'", *p->a);
 389     p_next(p);
 390   } else if (p->t == O_NOT) {
 391     n = xmalloc(sizeof(node_un));
 392     n->n.func = n_not;
 393     *nn = &n->n;
 394     p_next(p);
 395     p_factor(p, &n->arg);
 396   } else {
 397     switch (p->t) {
 398       case O_ANAG: *nn = anagram(p->a + 1); break;
 399       case O_SUBG: *nn = subgram(p->a + 1); break;
 400       case O_WILD: *nn = wildcard(p->a + 1); break;
 401       case O_TRACK: *nn = trackword(p->a + 1); break;
 402 #ifdef HAVE_REGCOMP
 403       case O_REGEXP: *nn = regexp(p->a + 1); break;
 404 #endif
 405 #ifdef HAVE_PCRE
 406       case O_PCRE: *nn = pcrenode(p->a + 1); break;
 407 #endif
 408       case O_MONO: *nn = mono(p->a + 1); break;
 409       case O_LENGTH: *nn = make_numeric(p->a + 1, n_length); break;
 410       case O_LONGEST: *nn = longest(p->a + 1); break;
 411       case O_SHORTEST: *nn = shortest(p->a + 1); break;
 412       default: die("syntax error near `%s': unexpected token", *p->a);
 413     }
 414     p_next(p);
 415   }
 416 }
 417
 418 static void p_term(p_ctx *p, node **nn)
 419 {
 420   node_bin *n;
 421   for (;;) {
 422     p_factor(p, nn);
 423     switch (p->t) {
 424       case O_AND:
 425         p_next(p);
 426       default:
 427         break;
 428       case O_RPAREN:
 429       case O_OR:
 430       case O_EOF:
 431         return;
 432     }
 433     n = xmalloc(sizeof(node_bin));
 434     n->left = *nn;
 435     n->n.func = n_and;
 436     *nn = &n->n;
 437     nn = &n->right;
 438   }
 439 }
 440
 441 static void p_expr(p_ctx *p, node **nn)
 442 {
 443   node_bin *n;
 444   for (;;) {
 445     p_term(p, nn);
 446     if (p->t != O_OR)
 447       break;
 448     p_next(p);
 449     n = xmalloc(sizeof(node_bin));
 450     n->left = *nn;
 451     n->n.func = n_or;
 452     *nn = &n->n;
 453     nn = &n->right;
 454   }
 455 }
 456
 457 /* --- @p_argv@ --- *
 458  *
 459  * Arguments:   @int argc@ = number of command-line arguments
 460  *              @const char *const argv[]@ = vectoor of arguments
 461  *
 462  * Returns:     A compiled node, parsed from the arguments.
 463  *
 464  * Use:         Does the donkey-work of parsing a command-line.
 465  */
 466
 467 static node *p_argv(int argc, const char *const argv[])
 468 {
 469   p_ctx p;
 470   node *n;
 471
 472   av = argv;
 473   ac = argc;
 474   ai = 1;
 475   p_next(&p);
 476   if (p.t == O_EOF) {
 477     usage(stderr);
 478     pquis(stderr, "(Run `$ --help' for more detail.)\n");
 479     exit(EXIT_FAILURE);
 480   }
 481   p_expr(&p, &n);
 482   if (p.t != O_EOF) {
 483     die("syntax error near `%s': rubbish at end of line (too many `)'s?)",
 484         *p.a);
 485   }
 486   return (n);
 487 }
 488
 489 /*----- At-end stuff ------------------------------------------------------*/
 490
 491 /* --- @atend_register@ --- *
 492  *
 493  * Arguments:   @int (*func)(void *)@ = function to call
 494  *              @void *p@ = handle to pass to it
 495  *
 496  * Returns:     ---
 497  *
 498  * Use:         Adds a function to the list of things to do at the end of the
 499  *              program.  The function should return nonzero if it produced
 500  *              any output.
 501  */
 502
 503 typedef struct atend {
 504   struct atend *next;
 505   int (*func)(void */*p*/);
 506   void *p;
 507 } atend;
 508
 509 static atend *aa_head = 0, **aa_tail = &aa_head;
 510
 511 void atend_register(int (*func)(void */*p*/), void *p)
 512 {
 513   atend *a = xmalloc(sizeof(*a));
 514   a->next = 0;
 515   a->func = func;
 516   a->p = p;
 517   *aa_tail = a;
 518   aa_tail = &a->next;
 519 }
 520
 521 /*----- Main code ---------------------------------------------------------*/
 522
 523 /* --- @main@ --- *
 524  *
 525  * Arguments:   @int argc@ = number of command-line arguments
 526  *              @char *argv[]@ = vector of argument words
 527  *
 528  * Returns:     Zero on success, nonzero on failure.
 529  *
 530  * Use:         Picks entries from a word list which match particular
 531  *              expressions.  This might be of assistance to word-game types.
 532  */
 533
 534 int main(int argc, char *argv[])
 535 {
 536   node *n;
 537   FILE *fp;
 538   dstr d = DSTR_INIT;
 539   int ok = 0;
 540   char *p, *q, *l;
 541   atend *a;
 542
 543   ego(argv[0]);
 544   n = p_argv(argc, (const char *const *)argv);
 545
 546   if ((fp = fopen(file, "r")) == 0)
 547     die("error opening `%s': %s", file, strerror(errno));
 548   for (;;) {
 549     dstr_reset(&d);
 550     if (dstr_putline(&d, fp) < 0)
 551       break;
 552     l = d.buf + d.len;
 553     for (p = q = d.buf; p < l; p++) {
 554       if (!isalnum((unsigned char)*p))
 555         continue;
 556       *q++ = tolower((unsigned char)*p);
 557     }
 558     *q = 0;
 559     d.len = q - d.buf;
 560     if (n->func(n, d.buf, d.len)) {
 561       fwrite(d.buf, 1, d.len, stdout);
 562       fputc('\n', stdout);
 563       ok = 1;
 564     }
 565   }
 566   if (ferror(fp) || fclose(fp))
 567     die("error reading `%s': %s", file, strerror(errno));
 568   for (a = aa_head; a; a = a->next) {
 569     if (a->func(a->p))
 570       ok = 1;
 571   }
 572   if (fflush(stdout) || ferror(stdout) || fclose(stdout))
 573     die("error writing output: %s", strerror(errno));
 574   if (!ok) pquis(stderr, "$: no matches found\n");
 575   return (ok ? EX_OK : EX_NONE);
 576 }
 577
 578 /*----- That's all, folks -------------------------------------------------*/