Minor big fixes to parser.

[anag] / anag.c

/* -*-c-*-
 *
 * $Id: anag.c,v 1.4 2001/02/19 19:18:50 mdw Exp $
 *
 * Main driver for anag
 *
 * (c) 2001 Mark Wooding
 */

/*----- Licensing notice --------------------------------------------------* 
 *
 * This file is part of Anag: a simple wordgame helper.
 *
 * Anag is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 * 
 * Anag is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 * 
 * You should have received a copy of the GNU General Public License
 * along with Anag; if not, write to the Free Software Foundation,
 * Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
 */

/*----- Revision history --------------------------------------------------* 
 *
 * $Log: anag.c,v $
 * Revision 1.4  2001/02/19 19:18:50  mdw
 * Minor big fixes to parser.
 *
 * Revision 1.3  2001/02/16 21:45:19  mdw
 * Be more helpful.  Improve full help message.  Special-case error for
 * empty command strings.
 *
 * Revision 1.2  2001/02/07 09:09:11  mdw
 * Fix spurious error when `-file' is used.
 *
 * Revision 1.1  2001/02/04 17:14:42  mdw
 * Initial checkin
 *
 */

/*----- Header files ------------------------------------------------------*/

#include "anag.h"

/*----- Static variables --------------------------------------------------*/

static const char *file = DICTIONARY;

/*----- Help text functions -----------------------------------------------*/

static void usage(FILE *fp)
{
  pquis(fp, "Usage: $ [-f file] expression\n");
}

static void version(FILE *fp)
{
  pquis(fp, "$, version " VERSION "\n");
}

static void help(FILE *fp)
{
  version(fp);
  fputc('\n', fp);
  usage(fp);
  fputs("\n\
Searches a wordlist, printing all of the words which match an expression.\n\
\n\
Options supported are:\n\
\n\
-h, --help		display this help text\n\
-v, --version		display the program's version number\n\
-u, --usage		display a very brief usage message\n\
-f, --file FILE		read wordlist from FILE, not `" DICTIONARY "'\n\
\n\
The basic tests in the expression are:\n\
\n\
-anagram WORD		matches a full-length anagram\n\
-subgram WORD		matches words which only use letters in WORD\n\
-wildcard PATTERN	matches with wildcards `*' and `?'\n\
-trackword WORD		matches words which can be found in a trackword\n\
\n\
These simple tests can be combined using the operators `-a', `-o' and `-n'\n\
(for `and', `or' and `not'; they may also be written `&', `|' and `!' if\n\
you like), and grouped using parentheses `(' and `)'.\n\
", fp);
}

/*----- The options parser ------------------------------------------------*/

/* --- Options table structure --- */

struct opt {
  const char *name;
  unsigned nargs;
  unsigned f;
  unsigned tag;
};

enum {
  O_HELP, O_VERSION, O_USAGE,
  O_FILE,
  O_AND, O_OR, O_NOT, O_LPAREN, O_RPAREN,
  O_ANAG, O_SUBG, O_WILD, O_TRACK,
  O_EOF
};

#define OF_SHORT 1u

static const struct opt opttab[] = {

  /* --- Options -- don't form part of the language --- */

  { "help",		0,	OF_SHORT,	O_HELP },
  { "version",		0,	OF_SHORT,	O_VERSION },
  { "usage",		0,	OF_SHORT,	O_USAGE },
  { "file",		1,	OF_SHORT,	O_FILE },

  /* --- Operators -- provide the basic structure of the language --- *
   *
   * These are also given magical names by the parser.
   */

  { "and",		0,	OF_SHORT,	O_AND },
  { "or",		0,	OF_SHORT,	O_OR },
  { "not",		0,	OF_SHORT,	O_NOT },

  /* --- Actual matching oeprations -- do something useful --- */

  { "anagram",		1,	0,		O_ANAG },
  { "subgram",		1,	0,		O_SUBG },
  { "wildcard",		1,	0,		O_WILD },
  { "trackword",	1,	0,		O_TRACK },

  /* --- End marker --- */

  { 0,			0,	0,		0 }
};

static int ac;
static const char *const *av;
static int ai;

/* --- @nextopt@ --- *
 *
 * Arguments:	@const char ***arg@ = where to store the arg pointer
 *
 * Returns:	The tag of the next option.
 *
 * Use:		Scans the next option off the command line.  If the option
 *		doesn't form part of the language, it's processed internally,
 *		and you'll never see it from here.  On exit, the @arg@
 *		pointer is set to contain the address of the option scanned,
 *		followed by its arguments if any.  You're expected to know
 *		how many arguments there are for your option.
 */

static unsigned nextopt(const char *const **arg)
{
  for (;;) {
    const struct opt *o, *oo;
    size_t sz;
    const char *p;

    /* --- Pick the next option off the front --- */

    *arg = av + ai;
    if (ai >= ac)
      return (O_EOF);
    p = av[ai++];

    /* --- Cope with various forms of magic --- */

    if (p[0] != '-') {
      if (!p[1]) switch (*p) {
	case '&': return (O_AND);
	case '|': return (O_OR);
	case '!': return (O_NOT);
	case '(': return (O_LPAREN);
	case ')': return (O_RPAREN);
      }
      goto bad;
    }

    /* --- Now cope with other sorts of weirdies --- *
     *
     * By the end of this, a leading `-' or `--' will have been stripped.
     */

    p++;
    if (!*p)
      goto bad;
    if (*p == '-')
      p++;
    if (!*p) {
      if (ai < ac)
	die("syntax error near `--': rubbish at end of line");
      return (O_EOF);
    }

    /* --- Now look the word up in my table --- */

    sz = strlen(p);
    oo = 0;
    for (o = opttab; o->name; o++) {
      if (strncmp(p, o->name, sz) == 0) {
	if (strlen(o->name) == sz || ((o->f & OF_SHORT) && sz == 1)) {
	  oo = o;
	  break;
	}
	if (oo) {
	  die("ambiguous option name `-%s' (could match `-%s' or `-%s')",
	      p, oo->name, o->name);
	}
	oo = o;
      }
    }
    if (!oo)
      die("unrecognized option name `-%s'", p);

    /* --- Sort out the arguments --- */

    if (ai + oo->nargs > ac)
      die("too few arguments for `-%s' (need %u)", oo->name, oo->nargs);
    ai += oo->nargs;

    /* --- Now process the option --- */

    switch (oo->tag) {
      case O_HELP:
	help(stdout);
	exit(0);
      case O_VERSION:
	version(stdout);
	exit(0);
      case O_USAGE:
	usage(stdout);
	exit(0);
      case O_FILE:
	file = (*arg)[1];
	break;
      default:
	return (oo->tag);
    }
    continue;
  bad:
    die("syntax error near `%s': unknown token type", av[ai - 1]);
  }
}

/*----- Node types for operators ------------------------------------------*/

/* --- Node structures --- */

typedef struct node_bin {
  node n;
  node *left;
  node *right;
} node_bin;

typedef struct node_un {
  node n;
  node *arg;
} node_un;

/* --- Node functions --- */

static int n_or(node *nn, const char *p, size_t sz)
{
  node_bin *n = (node_bin *)nn;
  return (n->left->func(n->left, p, sz) || n->right->func(n->right, p, sz));
}

static int n_and(node *nn, const char *p, size_t sz)
{
  node_bin *n = (node_bin *)nn;
  return (n->left->func(n->left, p, sz) && n->right->func(n->right, p, sz));
}

static int n_not(node *nn, const char *p, size_t sz)
{
  node_un *n = (node_un *)nn;
  return (!n->arg->func(n->arg, p, sz));
}

/*----- Parser for the expression syntax ----------------------------------*/

/* --- A parser context --- */

typedef struct p_ctx {
  unsigned t;
  const char *const *a;
} p_ctx;

/* --- Parser structure --- *
 *
 * This is a simple recursive descent parser.  The context retains
 * information about the current token.  Each function is passed the address
 * of a node pointer to fill in.  This simplifies the binary operator code
 * somewhat, relative to returning pointers to node trees.
 */

static void p_expr(p_ctx *p, node **/*nn*/);

static void p_next(p_ctx *p)
{
  static const char *const eof[] = { "<end>", 0 };
  p->t = nextopt(&p->a);
  if (p->t == O_EOF)
    p->a = eof;
}

static void p_factor(p_ctx *p, node **nn)
{
  node_un *n;
  if (p->t == O_LPAREN) {
    p_next(p);
    p_expr(p, nn);
    if (p->t != O_RPAREN)
      die("syntax error near `%s': missing `)'", *p->a);
    p_next(p);
  } else if (p->t == O_NOT) {
    n = xmalloc(sizeof(node_un));
    n->n.func = n_not;
    *nn = &n->n;
    p_next(p);
    p_factor(p, &n->arg);
  } else {
    switch (p->t) {
      case O_ANAG: *nn = anagram(p->a + 1); break;
      case O_SUBG: *nn = subgram(p->a + 1); break;
      case O_WILD: *nn = wildcard(p->a + 1); break;
      case O_TRACK: *nn = trackword(p->a + 1); break;
      default: die("syntax error near `%s': unexpected token", *p->a);
    }
    p_next(p);
  }
}

static void p_term(p_ctx *p, node **nn)
{
  node_bin *n;
  for (;;) {
    p_factor(p, nn);
    switch (p->t) {
      case O_AND:
	p_next(p);
      default:
	break;
      case O_RPAREN:
      case O_OR:
      case O_EOF:
	return;
    }
    n = xmalloc(sizeof(node_bin));
    n->left = *nn;
    n->n.func = n_and;
    *nn = &n->n;
    nn = &n->right;
  }
}

static void p_expr(p_ctx *p, node **nn)
{
  node_bin *n;
  for (;;) {
    p_term(p, nn);
    if (p->t != O_OR)
      break;
    p_next(p);
    n = xmalloc(sizeof(node_bin));
    n->left = *nn;
    n->n.func = n_or;
    *nn = &n->n;
    nn = &n->right;
  }
}

/* --- @p_argv@ --- *
 *
 * Arguments:	@int argc@ = number of command-line arguments
 *		@const char *const argv[]@ = vectoor of arguments
 *
 * Returns:	A compiled node, parsed from the arguments.
 *
 * Use:		Does the donkey-work of parsing a command-line.
 */

static node *p_argv(int argc, const char *const argv[])
{
  p_ctx p;
  node *n;

  av = argv;
  ac = argc;
  ai = 1;
  p_next(&p);
  if (p.t == O_EOF) {
    usage(stderr);
    pquis(stderr, "(Run `$ --help' for more detail.)\n");
    exit(EXIT_FAILURE);
  }
  p_expr(&p, &n);
  if (p.t != O_EOF) {
    die("syntax error near `%s': rubbish at end of line (too many `)'s?)",
	*p.a);
  }
  return (n);
}

/*----- Main code ---------------------------------------------------------*/

/* --- @main@ --- *
 *
 * Arguments:	@int argc@ = number of command-line arguments
 *		@char *argv[]@ = vector of argument words
 *
 * Returns:	Zero on success, nonzero on failure.
 *
 * Use:		Picks entries from a word list which match particular
 *		expressions.  This might be of assistance to word-game types.
 */

int main(int argc, char *argv[])
{
  node *n;
  FILE *fp;
  dstr d = DSTR_INIT;
  char *p, *q, *l;

  ego(argv[0]);
  n = p_argv(argc, (const char *const *)argv);

  if ((fp = fopen(file, "r")) == 0)
    die("error opening `%s': %s", file, strerror(errno));
  for (;;) {
    dstr_reset(&d);
    if (dstr_putline(&d, fp) < 0)
      break;
    l = d.buf + d.len;
    for (p = q = d.buf; p < l; p++) {
      if (!isalnum((unsigned char)*p))
	continue;
      *q++ = tolower((unsigned char)*p);
    }
    *q = 0;
    d.len = q - d.buf;
    if (n->func(n, d.buf, d.len)) {
      fwrite(d.buf, 1, d.len, stdout);
      fputc('\n', stdout);
    }
  }
  if (!feof(fp))
    die("error reading `%s': %s", file, strerror(errno));
  fclose(fp);
  return (0);
}

/*----- That's all, folks -------------------------------------------------*/