mansearch.c

   1 /*      $Id: mansearch.c,v 1.40 2014/08/05 12:34:08 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2012 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #ifdef HAVE_CONFIG_H
  19 #include "config.h"
  20 #endif
  21
  22 #include <sys/mman.h>
  23 #include <assert.h>
  24 #include <fcntl.h>
  25 #include <getopt.h>
  26 #include <limits.h>
  27 #include <regex.h>
  28 #include <stdio.h>
  29 #include <stdint.h>
  30 #include <stddef.h>
  31 #include <stdlib.h>
  32 #include <string.h>
  33 #include <unistd.h>
  34
  35 #ifdef HAVE_OHASH
  36 #include <ohash.h>
  37 #else
  38 #include "compat_ohash.h"
  39 #endif
  40 #include <sqlite3.h>
  41 #ifndef SQLITE_DETERMINISTIC
  42 #define SQLITE_DETERMINISTIC 0
  43 #endif
  44
  45 #include "mandoc.h"
  46 #include "mandoc_aux.h"
  47 #include "manpath.h"
  48 #include "mansearch.h"
  49
  50 extern int mansearch_keymax;
  51 extern const char *const mansearch_keynames[];
  52
  53 #define SQL_BIND_TEXT(_db, _s, _i, _v) \
  54         do { if (SQLITE_OK != sqlite3_bind_text \
  55                 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
  56                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  57         } while (0)
  58 #define SQL_BIND_INT64(_db, _s, _i, _v) \
  59         do { if (SQLITE_OK != sqlite3_bind_int64 \
  60                 ((_s), (_i)++, (_v))) \
  61                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  62         } while (0)
  63 #define SQL_BIND_BLOB(_db, _s, _i, _v) \
  64         do { if (SQLITE_OK != sqlite3_bind_blob \
  65                 ((_s), (_i)++, (&_v), sizeof(_v), SQLITE_STATIC)) \
  66                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  67         } while (0)
  68
  69 struct  expr {
  70         regex_t          regexp;  /* compiled regexp, if applicable */
  71         const char      *substr;  /* to search for, if applicable */
  72         struct expr     *next;    /* next in sequence */
  73         uint64_t         bits;    /* type-mask */
  74         int              equal;   /* equality, not subsring match */
  75         int              open;    /* opening parentheses before */
  76         int              and;     /* logical AND before */
  77         int              close;   /* closing parentheses after */
  78 };
  79
  80 struct  match {
  81         uint64_t         pageid; /* identifier in database */
  82         char            *desc; /* manual page description */
  83         int              form; /* 0 == catpage */
  84 };
  85
  86 static  void             buildnames(struct manpage *, sqlite3 *,
  87                                 sqlite3_stmt *, uint64_t,
  88                                 const char *, int form);
  89 static  char            *buildoutput(sqlite3 *, sqlite3_stmt *,
  90                                  uint64_t, uint64_t);
  91 static  void            *hash_alloc(size_t, void *);
  92 static  void             hash_free(void *, void *);
  93 static  void            *hash_calloc(size_t, size_t, void *);
  94 static  struct expr     *exprcomp(const struct mansearch *,
  95                                 int, char *[]);
  96 static  void             exprfree(struct expr *);
  97 static  struct expr     *exprspec(struct expr *, uint64_t,
  98                                  const char *, const char *);
  99 static  struct expr     *exprterm(const struct mansearch *, char *, int);
 100 static  int              manpage_compare(const void *, const void *);
 101 static  void             sql_append(char **sql, size_t *sz,
 102                                 const char *newstr, int count);
 103 static  void             sql_match(sqlite3_context *context,
 104                                 int argc, sqlite3_value **argv);
 105 static  void             sql_regexp(sqlite3_context *context,
 106                                 int argc, sqlite3_value **argv);
 107 static  char            *sql_statement(const struct expr *);
 108
 109
 110 int
 111 mansearch_setup(int start)
 112 {
 113         static void     *pagecache;
 114         int              c;
 115
 116 #define PC_PAGESIZE     1280
 117 #define PC_NUMPAGES     256
 118
 119         if (start) {
 120                 if (NULL != pagecache) {
 121                         fprintf(stderr, "pagecache already enabled\n");
 122                         return((int)MANDOCLEVEL_BADARG);
 123                 }
 124
 125                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 126                     PROT_READ | PROT_WRITE, MAP_ANON, -1, 0);
 127
 128                 if (MAP_FAILED == pagecache) {
 129                         perror("mmap");
 130                         pagecache = NULL;
 131                         return((int)MANDOCLEVEL_SYSERR);
 132                 }
 133
 134                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 135                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 136
 137                 if (SQLITE_OK == c)
 138                         return((int)MANDOCLEVEL_OK);
 139
 140                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 141
 142         } else if (NULL == pagecache) {
 143                 fprintf(stderr, "pagecache missing\n");
 144                 return((int)MANDOCLEVEL_BADARG);
 145         }
 146
 147         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 148                 perror("munmap");
 149                 pagecache = NULL;
 150                 return((int)MANDOCLEVEL_SYSERR);
 151         }
 152
 153         pagecache = NULL;
 154         return((int)MANDOCLEVEL_OK);
 155 }
 156
 157 int
 158 mansearch(const struct mansearch *search,
 159                 const struct manpaths *paths,
 160                 int argc, char *argv[],
 161                 const char *outkey,
 162                 struct manpage **res, size_t *sz)
 163 {
 164         int              fd, rc, c, indexbit;
 165         int64_t          pageid;
 166         uint64_t         outbit, iterbit;
 167         char             buf[PATH_MAX];
 168         char            *sql;
 169         struct manpage  *mpage;
 170         struct expr     *e, *ep;
 171         sqlite3         *db;
 172         sqlite3_stmt    *s, *s2;
 173         struct match    *mp;
 174         struct ohash_info info;
 175         struct ohash     htab;
 176         unsigned int     idx;
 177         size_t           i, j, cur, maxres;
 178
 179         info.calloc = hash_calloc;
 180         info.alloc = hash_alloc;
 181         info.free = hash_free;
 182         info.key_offset = offsetof(struct match, pageid);
 183
 184         *sz = cur = maxres = 0;
 185         sql = NULL;
 186         *res = NULL;
 187         fd = -1;
 188         e = NULL;
 189         rc = 0;
 190
 191         if (0 == argc)
 192                 goto out;
 193         if (NULL == (e = exprcomp(search, argc, argv)))
 194                 goto out;
 195
 196         outbit = 0;
 197         if (NULL != outkey) {
 198                 for (indexbit = 0, iterbit = 1;
 199                      indexbit < mansearch_keymax;
 200                      indexbit++, iterbit <<= 1) {
 201                         if (0 == strcasecmp(outkey,
 202                             mansearch_keynames[indexbit])) {
 203                                 outbit = iterbit;
 204                                 break;
 205                         }
 206                 }
 207         }
 208
 209         /*
 210          * Save a descriptor to the current working directory.
 211          * Since pathnames in the "paths" variable might be relative,
 212          * and we'll be chdir()ing into them, we need to keep a handle
 213          * on our current directory from which to start the chdir().
 214          */
 215
 216         if (NULL == getcwd(buf, PATH_MAX)) {
 217                 perror("getcwd");
 218                 goto out;
 219         } else if (-1 == (fd = open(buf, O_RDONLY, 0))) {
 220                 perror(buf);
 221                 goto out;
 222         }
 223
 224         sql = sql_statement(e);
 225
 226         /*
 227          * Loop over the directories (containing databases) for us to
 228          * search.
 229          * Don't let missing/bad databases/directories phase us.
 230          * In each, try to open the resident database and, if it opens,
 231          * scan it for our match expression.
 232          */
 233
 234         for (i = 0; i < paths->sz; i++) {
 235                 if (-1 == fchdir(fd)) {
 236                         perror(buf);
 237                         free(*res);
 238                         break;
 239                 } else if (-1 == chdir(paths->paths[i])) {
 240                         perror(paths->paths[i]);
 241                         continue;
 242                 }
 243
 244                 c = sqlite3_open_v2(MANDOC_DB, &db,
 245                     SQLITE_OPEN_READONLY, NULL);
 246
 247                 if (SQLITE_OK != c) {
 248                         perror(MANDOC_DB);
 249                         sqlite3_close(db);
 250                         continue;
 251                 }
 252
 253                 /*
 254                  * Define the SQL functions for substring
 255                  * and regular expression matching.
 256                  */
 257
 258                 c = sqlite3_create_function(db, "match", 2,
 259                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 260                     NULL, sql_match, NULL, NULL);
 261                 assert(SQLITE_OK == c);
 262                 c = sqlite3_create_function(db, "regexp", 2,
 263                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 264                     NULL, sql_regexp, NULL, NULL);
 265                 assert(SQLITE_OK == c);
 266
 267                 j = 1;
 268                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 269                 if (SQLITE_OK != c)
 270                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 271
 272                 for (ep = e; NULL != ep; ep = ep->next) {
 273                         if (NULL == ep->substr) {
 274                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 275                         } else
 276                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 277                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 278                                 SQL_BIND_INT64(db, s, j, ep->bits);
 279                 }
 280
 281                 memset(&htab, 0, sizeof(struct ohash));
 282                 ohash_init(&htab, 4, &info);
 283
 284                 /*
 285                  * Hash each entry on its [unique] document identifier.
 286                  * This is a uint64_t.
 287                  * Instead of using a hash function, simply convert the
 288                  * uint64_t to a uint32_t, the hash value's type.
 289                  * This gives good performance and preserves the
 290                  * distribution of buckets in the table.
 291                  */
 292                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 293                         pageid = sqlite3_column_int64(s, 2);
 294                         idx = ohash_lookup_memory(&htab,
 295                             (char *)&pageid, sizeof(uint64_t),
 296                             (uint32_t)pageid);
 297
 298                         if (NULL != ohash_find(&htab, idx))
 299                                 continue;
 300
 301                         mp = mandoc_calloc(1, sizeof(struct match));
 302                         mp->pageid = pageid;
 303                         mp->form = sqlite3_column_int(s, 1);
 304                         if (TYPE_Nd == outbit)
 305                                 mp->desc = mandoc_strdup(
 306                                     sqlite3_column_text(s, 0));
 307                         ohash_insert(&htab, idx, mp);
 308                 }
 309
 310                 if (SQLITE_DONE != c)
 311                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 312
 313                 sqlite3_finalize(s);
 314
 315                 c = sqlite3_prepare_v2(db,
 316                     "SELECT sec, arch, name, pageid FROM mlinks "
 317                     "WHERE pageid=? ORDER BY sec, arch, name",
 318                     -1, &s, NULL);
 319                 if (SQLITE_OK != c)
 320                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 321
 322                 c = sqlite3_prepare_v2(db,
 323                     "SELECT bits, key, pageid FROM keys "
 324                     "WHERE pageid=? AND bits & ?",
 325                     -1, &s2, NULL);
 326                 if (SQLITE_OK != c)
 327                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 328
 329                 for (mp = ohash_first(&htab, &idx);
 330                                 NULL != mp;
 331                                 mp = ohash_next(&htab, &idx)) {
 332                         if (cur + 1 > maxres) {
 333                                 maxres += 1024;
 334                                 *res = mandoc_reallocarray(*res,
 335                                     maxres, sizeof(struct manpage));
 336                         }
 337                         mpage = *res + cur;
 338                         mpage->sec = 10;
 339                         mpage->form = mp->form;
 340                         buildnames(mpage, db, s, mp->pageid,
 341                             paths->paths[i], mp->form);
 342                         mpage->output = TYPE_Nd & outbit ?
 343                             mp->desc : outbit ?
 344                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 345
 346                         free(mp);
 347                         cur++;
 348                 }
 349
 350                 sqlite3_finalize(s);
 351                 sqlite3_finalize(s2);
 352                 sqlite3_close(db);
 353                 ohash_delete(&htab);
 354         }
 355         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 356         rc = 1;
 357 out:
 358         if (-1 != fd) {
 359                 if (-1 == fchdir(fd))
 360                         perror(buf);
 361                 close(fd);
 362         }
 363         exprfree(e);
 364         free(sql);
 365         *sz = cur;
 366         return(rc);
 367 }
 368
 369 static int
 370 manpage_compare(const void *vp1, const void *vp2)
 371 {
 372         const struct manpage    *mp1, *mp2;
 373         int                      diff;
 374
 375         mp1 = vp1;
 376         mp2 = vp2;
 377         diff = mp1->sec - mp2->sec;
 378         return(diff ? diff : strcasecmp(mp1->names, mp2->names));
 379 }
 380
 381 static void
 382 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 383                 uint64_t pageid, const char *path, int form)
 384 {
 385         char            *newnames, *prevsec, *prevarch;
 386         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 387         size_t           i;
 388         int              c;
 389
 390         mpage->file = NULL;
 391         mpage->names = NULL;
 392         prevsec = prevarch = NULL;
 393         i = 1;
 394         SQL_BIND_INT64(db, s, i, pageid);
 395         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 396
 397                 /* Decide whether we already have some names. */
 398
 399                 if (NULL == mpage->names) {
 400                         oldnames = "";
 401                         sep1 = "";
 402                 } else {
 403                         oldnames = mpage->names;
 404                         sep1 = ", ";
 405                 }
 406
 407                 /* Fetch the next name. */
 408
 409                 sec = sqlite3_column_text(s, 0);
 410                 arch = sqlite3_column_text(s, 1);
 411                 name = sqlite3_column_text(s, 2);
 412
 413                 /* Remember the first section found. */
 414
 415                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 416                         mpage->sec = (*sec - '1') + 1;
 417
 418                 /* If the section changed, append the old one. */
 419
 420                 if (NULL != prevsec &&
 421                     (strcmp(sec, prevsec) ||
 422                      strcmp(arch, prevarch))) {
 423                         sep2 = '\0' == *prevarch ? "" : "/";
 424                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 425                             oldnames, prevsec, sep2, prevarch);
 426                         free(mpage->names);
 427                         oldnames = mpage->names = newnames;
 428                         free(prevsec);
 429                         free(prevarch);
 430                         prevsec = prevarch = NULL;
 431                 }
 432
 433                 /* Save the new section, to append it later. */
 434
 435                 if (NULL == prevsec) {
 436                         prevsec = mandoc_strdup(sec);
 437                         prevarch = mandoc_strdup(arch);
 438                 }
 439
 440                 /* Append the new name. */
 441
 442                 mandoc_asprintf(&newnames, "%s%s%s",
 443                     oldnames, sep1, name);
 444                 free(mpage->names);
 445                 mpage->names = newnames;
 446
 447                 /* Also save the first file name encountered. */
 448
 449                 if (NULL != mpage->file)
 450                         continue;
 451
 452                 if (form) {
 453                         sep1 = "man";
 454                         fsec = sec;
 455                 } else {
 456                         sep1 = "cat";
 457                         fsec = "0";
 458                 }
 459                 sep2 = '\0' == *arch ? "" : "/";
 460                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 461                     path, sep1, sec, sep2, arch, name, fsec);
 462         }
 463         if (SQLITE_DONE != c)
 464                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 465         sqlite3_reset(s);
 466
 467         /* Append one final section to the names. */
 468
 469         if (NULL != prevsec) {
 470                 sep2 = '\0' == *prevarch ? "" : "/";
 471                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 472                     mpage->names, prevsec, sep2, prevarch);
 473                 free(mpage->names);
 474                 mpage->names = newnames;
 475                 free(prevsec);
 476                 free(prevarch);
 477         }
 478 }
 479
 480 static char *
 481 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 482 {
 483         char            *output, *newoutput;
 484         const char      *oldoutput, *sep1, *data;
 485         size_t           i;
 486         int              c;
 487
 488         output = NULL;
 489         i = 1;
 490         SQL_BIND_INT64(db, s, i, pageid);
 491         SQL_BIND_INT64(db, s, i, outbit);
 492         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 493                 if (NULL == output) {
 494                         oldoutput = "";
 495                         sep1 = "";
 496                 } else {
 497                         oldoutput = output;
 498                         sep1 = " # ";
 499                 }
 500                 data = sqlite3_column_text(s, 1);
 501                 mandoc_asprintf(&newoutput, "%s%s%s",
 502                     oldoutput, sep1, data);
 503                 free(output);
 504                 output = newoutput;
 505         }
 506         if (SQLITE_DONE != c)
 507                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 508         sqlite3_reset(s);
 509         return(output);
 510 }
 511
 512 /*
 513  * Implement substring match as an application-defined SQL function.
 514  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 515  * because that would require escaping metacharacters in the string
 516  * being searched for.
 517  */
 518 static void
 519 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 520 {
 521
 522         assert(2 == argc);
 523         sqlite3_result_int(context, NULL != strcasestr(
 524             (const char *)sqlite3_value_text(argv[1]),
 525             (const char *)sqlite3_value_text(argv[0])));
 526 }
 527
 528 /*
 529  * Implement regular expression match
 530  * as an application-defined SQL function.
 531  */
 532 static void
 533 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 534 {
 535
 536         assert(2 == argc);
 537         sqlite3_result_int(context, !regexec(
 538             (regex_t *)sqlite3_value_blob(argv[0]),
 539             (const char *)sqlite3_value_text(argv[1]),
 540             0, NULL, 0));
 541 }
 542
 543 static void
 544 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 545 {
 546         size_t           newsz;
 547
 548         newsz = 1 < count ? (size_t)count : strlen(newstr);
 549         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 550         if (1 < count)
 551                 memset(*sql + *sz, *newstr, (size_t)count);
 552         else
 553                 memcpy(*sql + *sz, newstr, newsz);
 554         *sz += newsz;
 555         (*sql)[*sz] = '\0';
 556 }
 557
 558 /*
 559  * Prepare the search SQL statement.
 560  */
 561 static char *
 562 sql_statement(const struct expr *e)
 563 {
 564         char            *sql;
 565         size_t           sz;
 566         int              needop;
 567
 568         sql = mandoc_strdup(
 569             "SELECT desc, form, pageid FROM mpages WHERE ");
 570         sz = strlen(sql);
 571
 572         for (needop = 0; NULL != e; e = e->next) {
 573                 if (e->and)
 574                         sql_append(&sql, &sz, " AND ", 1);
 575                 else if (needop)
 576                         sql_append(&sql, &sz, " OR ", 1);
 577                 if (e->open)
 578                         sql_append(&sql, &sz, "(", e->open);
 579                 sql_append(&sql, &sz,
 580                     TYPE_Nd & e->bits
 581                     ? (NULL == e->substr
 582                         ? "desc REGEXP ?"
 583                         : "desc MATCH ?")
 584                     : TYPE_Nm == e->bits
 585                     ? (NULL == e->substr
 586                         ? "pageid IN (SELECT pageid FROM names "
 587                           "WHERE name REGEXP ?)"
 588                         : e->equal
 589                         ? "pageid IN (SELECT pageid FROM names "
 590                           "WHERE name = ?)"
 591                         : "pageid IN (SELECT pageid FROM names "
 592                           "WHERE name MATCH ?)")
 593                     : (NULL == e->substr
 594                         ? "pageid IN (SELECT pageid FROM keys "
 595                           "WHERE key REGEXP ? AND bits & ?)"
 596                         : "pageid IN (SELECT pageid FROM keys "
 597                           "WHERE key MATCH ? AND bits & ?)"), 1);
 598                 if (e->close)
 599                         sql_append(&sql, &sz, ")", e->close);
 600                 needop = 1;
 601         }
 602
 603         return(sql);
 604 }
 605
 606 /*
 607  * Compile a set of string tokens into an expression.
 608  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 609  * "(", "foo=bar", etc.).
 610  */
 611 static struct expr *
 612 exprcomp(const struct mansearch *search, int argc, char *argv[])
 613 {
 614         uint64_t         mask;
 615         int              i, toopen, logic, igncase, toclose;
 616         struct expr     *first, *prev, *cur, *next;
 617
 618         first = cur = NULL;
 619         logic = igncase = toclose = 0;
 620         toopen = NULL != search->sec || NULL != search->arch;
 621
 622         for (i = 0; i < argc; i++) {
 623                 if (0 == strcmp("(", argv[i])) {
 624                         if (igncase)
 625                                 goto fail;
 626                         toopen++;
 627                         toclose++;
 628                         continue;
 629                 } else if (0 == strcmp(")", argv[i])) {
 630                         if (toopen || logic || igncase || NULL == cur)
 631                                 goto fail;
 632                         cur->close++;
 633                         if (0 > --toclose)
 634                                 goto fail;
 635                         continue;
 636                 } else if (0 == strcmp("-a", argv[i])) {
 637                         if (toopen || logic || igncase || NULL == cur)
 638                                 goto fail;
 639                         logic = 1;
 640                         continue;
 641                 } else if (0 == strcmp("-o", argv[i])) {
 642                         if (toopen || logic || igncase || NULL == cur)
 643                                 goto fail;
 644                         logic = 2;
 645                         continue;
 646                 } else if (0 == strcmp("-i", argv[i])) {
 647                         if (igncase)
 648                                 goto fail;
 649                         igncase = 1;
 650                         continue;
 651                 }
 652                 next = exprterm(search, argv[i], !igncase);
 653                 if (NULL == next)
 654                         goto fail;
 655                 if (NULL == first)
 656                         first = next;
 657                 else
 658                         cur->next = next;
 659                 prev = cur = next;
 660
 661                 /*
 662                  * Searching for descriptions must be split out
 663                  * because they are stored in the mpages table,
 664                  * not in the keys table.
 665                  */
 666
 667                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 668                         if (mask & cur->bits && ~mask & cur->bits) {
 669                                 next = mandoc_calloc(1,
 670                                     sizeof(struct expr));
 671                                 memcpy(next, cur, sizeof(struct expr));
 672                                 prev->open = 1;
 673                                 cur->bits = mask;
 674                                 cur->next = next;
 675                                 cur = next;
 676                                 cur->bits &= ~mask;
 677                         }
 678                 }
 679                 prev->and = (1 == logic);
 680                 prev->open += toopen;
 681                 if (cur != prev)
 682                         cur->close = 1;
 683
 684                 toopen = logic = igncase = 0;
 685         }
 686         if (toopen || logic || igncase || toclose)
 687                 goto fail;
 688
 689         if (NULL != search->sec || NULL != search->arch)
 690                 cur->close++;
 691         if (NULL != search->arch)
 692                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 693         if (NULL != search->sec)
 694                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 695
 696         return(first);
 697
 698 fail:
 699         if (NULL != first)
 700                 exprfree(first);
 701         return(NULL);
 702 }
 703
 704 static struct expr *
 705 exprspec(struct expr *cur, uint64_t key, const char *value,
 706                 const char *format)
 707 {
 708         char     errbuf[BUFSIZ];
 709         char    *cp;
 710         int      irc;
 711
 712         mandoc_asprintf(&cp, format, value);
 713         cur->next = mandoc_calloc(1, sizeof(struct expr));
 714         cur = cur->next;
 715         cur->and = 1;
 716         cur->bits = key;
 717         if (0 != (irc = regcomp(&cur->regexp, cp,
 718             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 719                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 720                 fprintf(stderr, "regcomp: %s\n", errbuf);
 721                 cur->substr = value;
 722         }
 723         free(cp);
 724         return(cur);
 725 }
 726
 727 static struct expr *
 728 exprterm(const struct mansearch *search, char *buf, int cs)
 729 {
 730         char             errbuf[BUFSIZ];
 731         struct expr     *e;
 732         char            *key, *val;
 733         uint64_t         iterbit;
 734         int              i, irc;
 735
 736         if ('\0' == *buf)
 737                 return(NULL);
 738
 739         e = mandoc_calloc(1, sizeof(struct expr));
 740
 741         if (MANSEARCH_MAN & search->flags) {
 742                 e->bits = search->deftype;
 743                 e->substr = buf;
 744                 e->equal = 1;
 745                 return(e);
 746         }
 747
 748         /*
 749          * Look for an '=' or '~' operator,
 750          * unless forced to some fixed macro keys.
 751          */
 752
 753         if (MANSEARCH_WHATIS & search->flags)
 754                 val = NULL;
 755         else
 756                 val = strpbrk(buf, "=~");
 757
 758         if (NULL == val) {
 759                 e->bits = search->deftype;
 760                 e->substr = buf;
 761
 762         /*
 763          * Found an operator.
 764          * Regexp search is requested by !e->substr.
 765          */
 766
 767         } else {
 768                 if (val == buf)
 769                         e->bits = search->deftype;
 770                 if ('=' == *val)
 771                         e->substr = val + 1;
 772                 *val++ = '\0';
 773                 if (NULL != strstr(buf, "arch"))
 774                         cs = 0;
 775         }
 776
 777         /* Compile regular expressions. */
 778
 779         if (MANSEARCH_WHATIS & search->flags) {
 780                 e->substr = NULL;
 781                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 782         }
 783
 784         if (NULL == e->substr) {
 785                 irc = regcomp(&e->regexp, val,
 786                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 787                 if (MANSEARCH_WHATIS & search->flags)
 788                         free(val);
 789                 if (irc) {
 790                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 791                         fprintf(stderr, "regcomp: %s\n", errbuf);
 792                         free(e);
 793                         return(NULL);
 794                 }
 795         }
 796
 797         if (e->bits)
 798                 return(e);
 799
 800         /*
 801          * Parse out all possible fields.
 802          * If the field doesn't resolve, bail.
 803          */
 804
 805         while (NULL != (key = strsep(&buf, ","))) {
 806                 if ('\0' == *key)
 807                         continue;
 808                 for (i = 0, iterbit = 1;
 809                      i < mansearch_keymax;
 810                      i++, iterbit <<= 1) {
 811                         if (0 == strcasecmp(key,
 812                             mansearch_keynames[i])) {
 813                                 e->bits |= iterbit;
 814                                 break;
 815                         }
 816                 }
 817                 if (i == mansearch_keymax) {
 818                         if (strcasecmp(key, "any")) {
 819                                 free(e);
 820                                 return(NULL);
 821                         }
 822                         e->bits |= ~0ULL;
 823                 }
 824         }
 825
 826         return(e);
 827 }
 828
 829 static void
 830 exprfree(struct expr *p)
 831 {
 832         struct expr     *pp;
 833
 834         while (NULL != p) {
 835                 pp = p->next;
 836                 free(p);
 837                 p = pp;
 838         }
 839 }
 840
 841 static void *
 842 hash_calloc(size_t nmemb, size_t sz, void *arg)
 843 {
 844
 845         return(mandoc_calloc(nmemb, sz));
 846 }
 847
 848 static void *
 849 hash_alloc(size_t sz, void *arg)
 850 {
 851
 852         return(mandoc_malloc(sz));
 853 }
 854
 855 static void
 856 hash_free(void *p, void *arg)
 857 {
 858
 859         free(p);
 860 }