mansearch.c

   1 /*      $Id: mansearch.c,v 1.39 2014/07/24 20:30:45 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
  42 #include "mandoc.h"
  43 #include "mandoc_aux.h"
  44 #include "manpath.h"
  45 #include "mansearch.h"
  46
  47 extern int mansearch_keymax;
  48 extern const char *const mansearch_keynames[];
  49
  50 #define SQL_BIND_TEXT(_db, _s, _i, _v) \
  51         do { if (SQLITE_OK != sqlite3_bind_text \
  52                 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
  53                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  54         } while (0)
  55 #define SQL_BIND_INT64(_db, _s, _i, _v) \
  56         do { if (SQLITE_OK != sqlite3_bind_int64 \
  57                 ((_s), (_i)++, (_v))) \
  58                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  59         } while (0)
  60 #define SQL_BIND_BLOB(_db, _s, _i, _v) \
  61         do { if (SQLITE_OK != sqlite3_bind_blob \
  62                 ((_s), (_i)++, (&_v), sizeof(_v), SQLITE_STATIC)) \
  63                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  64         } while (0)
  65
  66 struct  expr {
  67         regex_t          regexp;  /* compiled regexp, if applicable */
  68         const char      *substr;  /* to search for, if applicable */
  69         struct expr     *next;    /* next in sequence */
  70         uint64_t         bits;    /* type-mask */
  71         int              equal;   /* equality, not subsring match */
  72         int              open;    /* opening parentheses before */
  73         int              and;     /* logical AND before */
  74         int              close;   /* closing parentheses after */
  75 };
  76
  77 struct  match {
  78         uint64_t         pageid; /* identifier in database */
  79         char            *desc; /* manual page description */
  80         int              form; /* 0 == catpage */
  81 };
  82
  83 static  void             buildnames(struct manpage *, sqlite3 *,
  84                                 sqlite3_stmt *, uint64_t,
  85                                 const char *, int form);
  86 static  char            *buildoutput(sqlite3 *, sqlite3_stmt *,
  87                                  uint64_t, uint64_t);
  88 static  void            *hash_alloc(size_t, void *);
  89 static  void             hash_free(void *, void *);
  90 static  void            *hash_calloc(size_t, size_t, void *);
  91 static  struct expr     *exprcomp(const struct mansearch *,
  92                                 int, char *[]);
  93 static  void             exprfree(struct expr *);
  94 static  struct expr     *exprspec(struct expr *, uint64_t,
  95                                  const char *, const char *);
  96 static  struct expr     *exprterm(const struct mansearch *, char *, int);
  97 static  int              manpage_compare(const void *, const void *);
  98 static  void             sql_append(char **sql, size_t *sz,
  99                                 const char *newstr, int count);
 100 static  void             sql_match(sqlite3_context *context,
 101                                 int argc, sqlite3_value **argv);
 102 static  void             sql_regexp(sqlite3_context *context,
 103                                 int argc, sqlite3_value **argv);
 104 static  char            *sql_statement(const struct expr *);
 105
 106
 107 int
 108 mansearch_setup(int start)
 109 {
 110         static void     *pagecache;
 111         int              c;
 112
 113 #define PC_PAGESIZE     1280
 114 #define PC_NUMPAGES     256
 115
 116         if (start) {
 117                 if (NULL != pagecache) {
 118                         fprintf(stderr, "pagecache already enabled\n");
 119                         return((int)MANDOCLEVEL_BADARG);
 120                 }
 121
 122                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 123                     PROT_READ | PROT_WRITE, MAP_ANON, -1, 0);
 124
 125                 if (MAP_FAILED == pagecache) {
 126                         perror("mmap");
 127                         pagecache = NULL;
 128                         return((int)MANDOCLEVEL_SYSERR);
 129                 }
 130
 131                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 132                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 133
 134                 if (SQLITE_OK == c)
 135                         return((int)MANDOCLEVEL_OK);
 136
 137                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 138
 139         } else if (NULL == pagecache) {
 140                 fprintf(stderr, "pagecache missing\n");
 141                 return((int)MANDOCLEVEL_BADARG);
 142         }
 143
 144         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 145                 perror("munmap");
 146                 pagecache = NULL;
 147                 return((int)MANDOCLEVEL_SYSERR);
 148         }
 149
 150         pagecache = NULL;
 151         return((int)MANDOCLEVEL_OK);
 152 }
 153
 154 int
 155 mansearch(const struct mansearch *search,
 156                 const struct manpaths *paths,
 157                 int argc, char *argv[],
 158                 const char *outkey,
 159                 struct manpage **res, size_t *sz)
 160 {
 161         int              fd, rc, c, indexbit;
 162         int64_t          pageid;
 163         uint64_t         outbit, iterbit;
 164         char             buf[PATH_MAX];
 165         char            *sql;
 166         struct manpage  *mpage;
 167         struct expr     *e, *ep;
 168         sqlite3         *db;
 169         sqlite3_stmt    *s, *s2;
 170         struct match    *mp;
 171         struct ohash_info info;
 172         struct ohash     htab;
 173         unsigned int     idx;
 174         size_t           i, j, cur, maxres;
 175
 176         info.calloc = hash_calloc;
 177         info.alloc = hash_alloc;
 178         info.free = hash_free;
 179         info.key_offset = offsetof(struct match, pageid);
 180
 181         *sz = cur = maxres = 0;
 182         sql = NULL;
 183         *res = NULL;
 184         fd = -1;
 185         e = NULL;
 186         rc = 0;
 187
 188         if (0 == argc)
 189                 goto out;
 190         if (NULL == (e = exprcomp(search, argc, argv)))
 191                 goto out;
 192
 193         outbit = 0;
 194         if (NULL != outkey) {
 195                 for (indexbit = 0, iterbit = 1;
 196                      indexbit < mansearch_keymax;
 197                      indexbit++, iterbit <<= 1) {
 198                         if (0 == strcasecmp(outkey,
 199                             mansearch_keynames[indexbit])) {
 200                                 outbit = iterbit;
 201                                 break;
 202                         }
 203                 }
 204         }
 205
 206         /*
 207          * Save a descriptor to the current working directory.
 208          * Since pathnames in the "paths" variable might be relative,
 209          * and we'll be chdir()ing into them, we need to keep a handle
 210          * on our current directory from which to start the chdir().
 211          */
 212
 213         if (NULL == getcwd(buf, PATH_MAX)) {
 214                 perror("getcwd");
 215                 goto out;
 216         } else if (-1 == (fd = open(buf, O_RDONLY, 0))) {
 217                 perror(buf);
 218                 goto out;
 219         }
 220
 221         sql = sql_statement(e);
 222
 223         /*
 224          * Loop over the directories (containing databases) for us to
 225          * search.
 226          * Don't let missing/bad databases/directories phase us.
 227          * In each, try to open the resident database and, if it opens,
 228          * scan it for our match expression.
 229          */
 230
 231         for (i = 0; i < paths->sz; i++) {
 232                 if (-1 == fchdir(fd)) {
 233                         perror(buf);
 234                         free(*res);
 235                         break;
 236                 } else if (-1 == chdir(paths->paths[i])) {
 237                         perror(paths->paths[i]);
 238                         continue;
 239                 }
 240
 241                 c = sqlite3_open_v2(MANDOC_DB, &db,
 242                     SQLITE_OPEN_READONLY, NULL);
 243
 244                 if (SQLITE_OK != c) {
 245                         perror(MANDOC_DB);
 246                         sqlite3_close(db);
 247                         continue;
 248                 }
 249
 250                 /*
 251                  * Define the SQL functions for substring
 252                  * and regular expression matching.
 253                  */
 254
 255                 c = sqlite3_create_function(db, "match", 2,
 256                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 257                     NULL, sql_match, NULL, NULL);
 258                 assert(SQLITE_OK == c);
 259                 c = sqlite3_create_function(db, "regexp", 2,
 260                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 261                     NULL, sql_regexp, NULL, NULL);
 262                 assert(SQLITE_OK == c);
 263
 264                 j = 1;
 265                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 266                 if (SQLITE_OK != c)
 267                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 268
 269                 for (ep = e; NULL != ep; ep = ep->next) {
 270                         if (NULL == ep->substr) {
 271                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 272                         } else
 273                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 274                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 275                                 SQL_BIND_INT64(db, s, j, ep->bits);
 276                 }
 277
 278                 memset(&htab, 0, sizeof(struct ohash));
 279                 ohash_init(&htab, 4, &info);
 280
 281                 /*
 282                  * Hash each entry on its [unique] document identifier.
 283                  * This is a uint64_t.
 284                  * Instead of using a hash function, simply convert the
 285                  * uint64_t to a uint32_t, the hash value's type.
 286                  * This gives good performance and preserves the
 287                  * distribution of buckets in the table.
 288                  */
 289                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 290                         pageid = sqlite3_column_int64(s, 2);
 291                         idx = ohash_lookup_memory(&htab,
 292                             (char *)&pageid, sizeof(uint64_t),
 293                             (uint32_t)pageid);
 294
 295                         if (NULL != ohash_find(&htab, idx))
 296                                 continue;
 297
 298                         mp = mandoc_calloc(1, sizeof(struct match));
 299                         mp->pageid = pageid;
 300                         mp->form = sqlite3_column_int(s, 1);
 301                         if (TYPE_Nd == outbit)
 302                                 mp->desc = mandoc_strdup(
 303                                     sqlite3_column_text(s, 0));
 304                         ohash_insert(&htab, idx, mp);
 305                 }
 306
 307                 if (SQLITE_DONE != c)
 308                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 309
 310                 sqlite3_finalize(s);
 311
 312                 c = sqlite3_prepare_v2(db,
 313                     "SELECT sec, arch, name, pageid FROM mlinks "
 314                     "WHERE pageid=? ORDER BY sec, arch, name",
 315                     -1, &s, NULL);
 316                 if (SQLITE_OK != c)
 317                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 318
 319                 c = sqlite3_prepare_v2(db,
 320                     "SELECT bits, key, pageid FROM keys "
 321                     "WHERE pageid=? AND bits & ?",
 322                     -1, &s2, NULL);
 323                 if (SQLITE_OK != c)
 324                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 325
 326                 for (mp = ohash_first(&htab, &idx);
 327                                 NULL != mp;
 328                                 mp = ohash_next(&htab, &idx)) {
 329                         if (cur + 1 > maxres) {
 330                                 maxres += 1024;
 331                                 *res = mandoc_reallocarray(*res,
 332                                     maxres, sizeof(struct manpage));
 333                         }
 334                         mpage = *res + cur;
 335                         mpage->sec = 10;
 336                         mpage->form = mp->form;
 337                         buildnames(mpage, db, s, mp->pageid,
 338                             paths->paths[i], mp->form);
 339                         mpage->output = TYPE_Nd & outbit ?
 340                             mp->desc : outbit ?
 341                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 342
 343                         free(mp);
 344                         cur++;
 345                 }
 346
 347                 sqlite3_finalize(s);
 348                 sqlite3_finalize(s2);
 349                 sqlite3_close(db);
 350                 ohash_delete(&htab);
 351         }
 352         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 353         rc = 1;
 354 out:
 355         if (-1 != fd) {
 356                 if (-1 == fchdir(fd))
 357                         perror(buf);
 358                 close(fd);
 359         }
 360         exprfree(e);
 361         free(sql);
 362         *sz = cur;
 363         return(rc);
 364 }
 365
 366 static int
 367 manpage_compare(const void *vp1, const void *vp2)
 368 {
 369         const struct manpage    *mp1, *mp2;
 370         int                      diff;
 371
 372         mp1 = vp1;
 373         mp2 = vp2;
 374         diff = mp1->sec - mp2->sec;
 375         return(diff ? diff : strcasecmp(mp1->names, mp2->names));
 376 }
 377
 378 static void
 379 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 380                 uint64_t pageid, const char *path, int form)
 381 {
 382         char            *newnames, *prevsec, *prevarch;
 383         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 384         size_t           i;
 385         int              c;
 386
 387         mpage->file = NULL;
 388         mpage->names = NULL;
 389         prevsec = prevarch = NULL;
 390         i = 1;
 391         SQL_BIND_INT64(db, s, i, pageid);
 392         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 393
 394                 /* Decide whether we already have some names. */
 395
 396                 if (NULL == mpage->names) {
 397                         oldnames = "";
 398                         sep1 = "";
 399                 } else {
 400                         oldnames = mpage->names;
 401                         sep1 = ", ";
 402                 }
 403
 404                 /* Fetch the next name. */
 405
 406                 sec = sqlite3_column_text(s, 0);
 407                 arch = sqlite3_column_text(s, 1);
 408                 name = sqlite3_column_text(s, 2);
 409
 410                 /* Remember the first section found. */
 411
 412                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 413                         mpage->sec = (*sec - '1') + 1;
 414
 415                 /* If the section changed, append the old one. */
 416
 417                 if (NULL != prevsec &&
 418                     (strcmp(sec, prevsec) ||
 419                      strcmp(arch, prevarch))) {
 420                         sep2 = '\0' == *prevarch ? "" : "/";
 421                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 422                             oldnames, prevsec, sep2, prevarch);
 423                         free(mpage->names);
 424                         oldnames = mpage->names = newnames;
 425                         free(prevsec);
 426                         free(prevarch);
 427                         prevsec = prevarch = NULL;
 428                 }
 429
 430                 /* Save the new section, to append it later. */
 431
 432                 if (NULL == prevsec) {
 433                         prevsec = mandoc_strdup(sec);
 434                         prevarch = mandoc_strdup(arch);
 435                 }
 436
 437                 /* Append the new name. */
 438
 439                 mandoc_asprintf(&newnames, "%s%s%s",
 440                     oldnames, sep1, name);
 441                 free(mpage->names);
 442                 mpage->names = newnames;
 443
 444                 /* Also save the first file name encountered. */
 445
 446                 if (NULL != mpage->file)
 447                         continue;
 448
 449                 if (form) {
 450                         sep1 = "man";
 451                         fsec = sec;
 452                 } else {
 453                         sep1 = "cat";
 454                         fsec = "0";
 455                 }
 456                 sep2 = '\0' == *arch ? "" : "/";
 457                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 458                     path, sep1, sec, sep2, arch, name, fsec);
 459         }
 460         if (SQLITE_DONE != c)
 461                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 462         sqlite3_reset(s);
 463
 464         /* Append one final section to the names. */
 465
 466         if (NULL != prevsec) {
 467                 sep2 = '\0' == *prevarch ? "" : "/";
 468                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 469                     mpage->names, prevsec, sep2, prevarch);
 470                 free(mpage->names);
 471                 mpage->names = newnames;
 472                 free(prevsec);
 473                 free(prevarch);
 474         }
 475 }
 476
 477 static char *
 478 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 479 {
 480         char            *output, *newoutput;
 481         const char      *oldoutput, *sep1, *data;
 482         size_t           i;
 483         int              c;
 484
 485         output = NULL;
 486         i = 1;
 487         SQL_BIND_INT64(db, s, i, pageid);
 488         SQL_BIND_INT64(db, s, i, outbit);
 489         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 490                 if (NULL == output) {
 491                         oldoutput = "";
 492                         sep1 = "";
 493                 } else {
 494                         oldoutput = output;
 495                         sep1 = " # ";
 496                 }
 497                 data = sqlite3_column_text(s, 1);
 498                 mandoc_asprintf(&newoutput, "%s%s%s",
 499                     oldoutput, sep1, data);
 500                 free(output);
 501                 output = newoutput;
 502         }
 503         if (SQLITE_DONE != c)
 504                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 505         sqlite3_reset(s);
 506         return(output);
 507 }
 508
 509 /*
 510  * Implement substring match as an application-defined SQL function.
 511  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 512  * because that would require escaping metacharacters in the string
 513  * being searched for.
 514  */
 515 static void
 516 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 517 {
 518
 519         assert(2 == argc);
 520         sqlite3_result_int(context, NULL != strcasestr(
 521             (const char *)sqlite3_value_text(argv[1]),
 522             (const char *)sqlite3_value_text(argv[0])));
 523 }
 524
 525 /*
 526  * Implement regular expression match
 527  * as an application-defined SQL function.
 528  */
 529 static void
 530 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 531 {
 532
 533         assert(2 == argc);
 534         sqlite3_result_int(context, !regexec(
 535             (regex_t *)sqlite3_value_blob(argv[0]),
 536             (const char *)sqlite3_value_text(argv[1]),
 537             0, NULL, 0));
 538 }
 539
 540 static void
 541 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 542 {
 543         size_t           newsz;
 544
 545         newsz = 1 < count ? (size_t)count : strlen(newstr);
 546         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 547         if (1 < count)
 548                 memset(*sql + *sz, *newstr, (size_t)count);
 549         else
 550                 memcpy(*sql + *sz, newstr, newsz);
 551         *sz += newsz;
 552         (*sql)[*sz] = '\0';
 553 }
 554
 555 /*
 556  * Prepare the search SQL statement.
 557  */
 558 static char *
 559 sql_statement(const struct expr *e)
 560 {
 561         char            *sql;
 562         size_t           sz;
 563         int              needop;
 564
 565         sql = mandoc_strdup(
 566             "SELECT desc, form, pageid FROM mpages WHERE ");
 567         sz = strlen(sql);
 568
 569         for (needop = 0; NULL != e; e = e->next) {
 570                 if (e->and)
 571                         sql_append(&sql, &sz, " AND ", 1);
 572                 else if (needop)
 573                         sql_append(&sql, &sz, " OR ", 1);
 574                 if (e->open)
 575                         sql_append(&sql, &sz, "(", e->open);
 576                 sql_append(&sql, &sz,
 577                     TYPE_Nd & e->bits
 578                     ? (NULL == e->substr
 579                         ? "desc REGEXP ?"
 580                         : "desc MATCH ?")
 581                     : TYPE_Nm == e->bits
 582                     ? (NULL == e->substr
 583                         ? "pageid IN (SELECT pageid FROM names "
 584                           "WHERE name REGEXP ?)"
 585                         : e->equal
 586                         ? "pageid IN (SELECT pageid FROM names "
 587                           "WHERE name = ?)"
 588                         : "pageid IN (SELECT pageid FROM names "
 589                           "WHERE name MATCH ?)")
 590                     : (NULL == e->substr
 591                         ? "pageid IN (SELECT pageid FROM keys "
 592                           "WHERE key REGEXP ? AND bits & ?)"
 593                         : "pageid IN (SELECT pageid FROM keys "
 594                           "WHERE key MATCH ? AND bits & ?)"), 1);
 595                 if (e->close)
 596                         sql_append(&sql, &sz, ")", e->close);
 597                 needop = 1;
 598         }
 599
 600         return(sql);
 601 }
 602
 603 /*
 604  * Compile a set of string tokens into an expression.
 605  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 606  * "(", "foo=bar", etc.).
 607  */
 608 static struct expr *
 609 exprcomp(const struct mansearch *search, int argc, char *argv[])
 610 {
 611         uint64_t         mask;
 612         int              i, toopen, logic, igncase, toclose;
 613         struct expr     *first, *prev, *cur, *next;
 614
 615         first = cur = NULL;
 616         logic = igncase = toclose = 0;
 617         toopen = NULL != search->sec || NULL != search->arch;
 618
 619         for (i = 0; i < argc; i++) {
 620                 if (0 == strcmp("(", argv[i])) {
 621                         if (igncase)
 622                                 goto fail;
 623                         toopen++;
 624                         toclose++;
 625                         continue;
 626                 } else if (0 == strcmp(")", argv[i])) {
 627                         if (toopen || logic || igncase || NULL == cur)
 628                                 goto fail;
 629                         cur->close++;
 630                         if (0 > --toclose)
 631                                 goto fail;
 632                         continue;
 633                 } else if (0 == strcmp("-a", argv[i])) {
 634                         if (toopen || logic || igncase || NULL == cur)
 635                                 goto fail;
 636                         logic = 1;
 637                         continue;
 638                 } else if (0 == strcmp("-o", argv[i])) {
 639                         if (toopen || logic || igncase || NULL == cur)
 640                                 goto fail;
 641                         logic = 2;
 642                         continue;
 643                 } else if (0 == strcmp("-i", argv[i])) {
 644                         if (igncase)
 645                                 goto fail;
 646                         igncase = 1;
 647                         continue;
 648                 }
 649                 next = exprterm(search, argv[i], !igncase);
 650                 if (NULL == next)
 651                         goto fail;
 652                 if (NULL == first)
 653                         first = next;
 654                 else
 655                         cur->next = next;
 656                 prev = cur = next;
 657
 658                 /*
 659                  * Searching for descriptions must be split out
 660                  * because they are stored in the mpages table,
 661                  * not in the keys table.
 662                  */
 663
 664                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 665                         if (mask & cur->bits && ~mask & cur->bits) {
 666                                 next = mandoc_calloc(1,
 667                                     sizeof(struct expr));
 668                                 memcpy(next, cur, sizeof(struct expr));
 669                                 prev->open = 1;
 670                                 cur->bits = mask;
 671                                 cur->next = next;
 672                                 cur = next;
 673                                 cur->bits &= ~mask;
 674                         }
 675                 }
 676                 prev->and = (1 == logic);
 677                 prev->open += toopen;
 678                 if (cur != prev)
 679                         cur->close = 1;
 680
 681                 toopen = logic = igncase = 0;
 682         }
 683         if (toopen || logic || igncase || toclose)
 684                 goto fail;
 685
 686         if (NULL != search->sec || NULL != search->arch)
 687                 cur->close++;
 688         if (NULL != search->arch)
 689                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 690         if (NULL != search->sec)
 691                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 692
 693         return(first);
 694
 695 fail:
 696         if (NULL != first)
 697                 exprfree(first);
 698         return(NULL);
 699 }
 700
 701 static struct expr *
 702 exprspec(struct expr *cur, uint64_t key, const char *value,
 703                 const char *format)
 704 {
 705         char     errbuf[BUFSIZ];
 706         char    *cp;
 707         int      irc;
 708
 709         mandoc_asprintf(&cp, format, value);
 710         cur->next = mandoc_calloc(1, sizeof(struct expr));
 711         cur = cur->next;
 712         cur->and = 1;
 713         cur->bits = key;
 714         if (0 != (irc = regcomp(&cur->regexp, cp,
 715             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 716                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 717                 fprintf(stderr, "regcomp: %s\n", errbuf);
 718                 cur->substr = value;
 719         }
 720         free(cp);
 721         return(cur);
 722 }
 723
 724 static struct expr *
 725 exprterm(const struct mansearch *search, char *buf, int cs)
 726 {
 727         char             errbuf[BUFSIZ];
 728         struct expr     *e;
 729         char            *key, *val;
 730         uint64_t         iterbit;
 731         int              i, irc;
 732
 733         if ('\0' == *buf)
 734                 return(NULL);
 735
 736         e = mandoc_calloc(1, sizeof(struct expr));
 737
 738         if (MANSEARCH_MAN & search->flags) {
 739                 e->bits = search->deftype;
 740                 e->substr = buf;
 741                 e->equal = 1;
 742                 return(e);
 743         }
 744
 745         /*
 746          * Look for an '=' or '~' operator,
 747          * unless forced to some fixed macro keys.
 748          */
 749
 750         if (MANSEARCH_WHATIS & search->flags)
 751                 val = NULL;
 752         else
 753                 val = strpbrk(buf, "=~");
 754
 755         if (NULL == val) {
 756                 e->bits = search->deftype;
 757                 e->substr = buf;
 758
 759         /*
 760          * Found an operator.
 761          * Regexp search is requested by !e->substr.
 762          */
 763
 764         } else {
 765                 if (val == buf)
 766                         e->bits = search->deftype;
 767                 if ('=' == *val)
 768                         e->substr = val + 1;
 769                 *val++ = '\0';
 770                 if (NULL != strstr(buf, "arch"))
 771                         cs = 0;
 772         }
 773
 774         /* Compile regular expressions. */
 775
 776         if (MANSEARCH_WHATIS & search->flags) {
 777                 e->substr = NULL;
 778                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 779         }
 780
 781         if (NULL == e->substr) {
 782                 irc = regcomp(&e->regexp, val,
 783                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 784                 if (MANSEARCH_WHATIS & search->flags)
 785                         free(val);
 786                 if (irc) {
 787                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 788                         fprintf(stderr, "regcomp: %s\n", errbuf);
 789                         free(e);
 790                         return(NULL);
 791                 }
 792         }
 793
 794         if (e->bits)
 795                 return(e);
 796
 797         /*
 798          * Parse out all possible fields.
 799          * If the field doesn't resolve, bail.
 800          */
 801
 802         while (NULL != (key = strsep(&buf, ","))) {
 803                 if ('\0' == *key)
 804                         continue;
 805                 for (i = 0, iterbit = 1;
 806                      i < mansearch_keymax;
 807                      i++, iterbit <<= 1) {
 808                         if (0 == strcasecmp(key,
 809                             mansearch_keynames[i])) {
 810                                 e->bits |= iterbit;
 811                                 break;
 812                         }
 813                 }
 814                 if (i == mansearch_keymax) {
 815                         if (strcasecmp(key, "any")) {
 816                                 free(e);
 817                                 return(NULL);
 818                         }
 819                         e->bits |= ~0ULL;
 820                 }
 821         }
 822
 823         return(e);
 824 }
 825
 826 static void
 827 exprfree(struct expr *p)
 828 {
 829         struct expr     *pp;
 830
 831         while (NULL != p) {
 832                 pp = p->next;
 833                 free(p);
 834                 p = pp;
 835         }
 836 }
 837
 838 static void *
 839 hash_calloc(size_t nmemb, size_t sz, void *arg)
 840 {
 841
 842         return(mandoc_calloc(nmemb, sz));
 843 }
 844
 845 static void *
 846 hash_alloc(size_t sz, void *arg)
 847 {
 848
 849         return(mandoc_malloc(sz));
 850 }
 851
 852 static void
 853 hash_free(void *p, void *arg)
 854 {
 855
 856         free(p);
 857 }