mansearch.c

   1 /*      $Id: mansearch.c,v 1.36 2014/04/23 21:06:41 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         uint64_t         bits;    /* type-mask */
  68         const char      *substr;  /* to search for, if applicable */
  69         regex_t          regexp;  /* compiled regexp, if applicable */
  70         int              open;    /* opening parentheses before */
  71         int              and;     /* logical AND before */
  72         int              close;   /* closing parentheses after */
  73         struct expr     *next;    /* next in sequence */
  74 };
  75
  76 struct  match {
  77         uint64_t         pageid; /* identifier in database */
  78         char            *desc; /* manual page description */
  79         int              form; /* 0 == catpage */
  80 };
  81
  82 static  void             buildnames(struct manpage *, sqlite3 *,
  83                                 sqlite3_stmt *, uint64_t,
  84                                 const char *, int form);
  85 static  char            *buildoutput(sqlite3 *, sqlite3_stmt *,
  86                                  uint64_t, uint64_t);
  87 static  void            *hash_alloc(size_t, void *);
  88 static  void             hash_free(void *, size_t, void *);
  89 static  void            *hash_halloc(size_t, void *);
  90 static  struct expr     *exprcomp(const struct mansearch *,
  91                                 int, char *[]);
  92 static  void             exprfree(struct expr *);
  93 static  struct expr     *exprspec(struct expr *, uint64_t,
  94                                  const char *, const char *);
  95 static  struct expr     *exprterm(const struct mansearch *, char *, int);
  96 static  void             sql_append(char **sql, size_t *sz,
  97                                 const char *newstr, int count);
  98 static  void             sql_match(sqlite3_context *context,
  99                                 int argc, sqlite3_value **argv);
 100 static  void             sql_regexp(sqlite3_context *context,
 101                                 int argc, sqlite3_value **argv);
 102 static  char            *sql_statement(const struct expr *);
 103
 104
 105 int
 106 mansearch_setup(int start)
 107 {
 108         static void     *pagecache;
 109         int              c;
 110
 111 #define PC_PAGESIZE     1280
 112 #define PC_NUMPAGES     256
 113
 114         if (start) {
 115                 if (NULL != pagecache) {
 116                         fprintf(stderr, "pagecache already enabled\n");
 117                         return((int)MANDOCLEVEL_BADARG);
 118                 }
 119
 120                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 121                     PROT_READ | PROT_WRITE, MAP_ANON, -1, 0);
 122
 123                 if (MAP_FAILED == pagecache) {
 124                         perror("mmap");
 125                         pagecache = NULL;
 126                         return((int)MANDOCLEVEL_SYSERR);
 127                 }
 128
 129                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 130                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 131
 132                 if (SQLITE_OK == c)
 133                         return((int)MANDOCLEVEL_OK);
 134
 135                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 136
 137         } else if (NULL == pagecache) {
 138                 fprintf(stderr, "pagecache missing\n");
 139                 return((int)MANDOCLEVEL_BADARG);
 140         }
 141
 142         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 143                 perror("munmap");
 144                 pagecache = NULL;
 145                 return((int)MANDOCLEVEL_SYSERR);
 146         }
 147
 148         pagecache = NULL;
 149         return((int)MANDOCLEVEL_OK);
 150 }
 151
 152 int
 153 mansearch(const struct mansearch *search,
 154                 const struct manpaths *paths,
 155                 int argc, char *argv[],
 156                 const char *outkey,
 157                 struct manpage **res, size_t *sz)
 158 {
 159         int              fd, rc, c, indexbit;
 160         int64_t          pageid;
 161         uint64_t         outbit, iterbit;
 162         char             buf[PATH_MAX];
 163         char            *sql;
 164         struct manpage  *mpage;
 165         struct expr     *e, *ep;
 166         sqlite3         *db;
 167         sqlite3_stmt    *s, *s2;
 168         struct match    *mp;
 169         struct ohash_info info;
 170         struct ohash     htab;
 171         unsigned int     idx;
 172         size_t           i, j, cur, maxres;
 173
 174         memset(&info, 0, sizeof(struct ohash_info));
 175
 176         info.halloc = hash_halloc;
 177         info.alloc = hash_alloc;
 178         info.hfree = 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->form = mp->form;
 336                         buildnames(mpage, db, s, mp->pageid,
 337                             paths->paths[i], mp->form);
 338                         mpage->output = TYPE_Nd & outbit ?
 339                             mp->desc : outbit ?
 340                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 341
 342                         free(mp);
 343                         cur++;
 344                 }
 345
 346                 sqlite3_finalize(s);
 347                 sqlite3_finalize(s2);
 348                 sqlite3_close(db);
 349                 ohash_delete(&htab);
 350         }
 351         rc = 1;
 352 out:
 353         if (-1 != fd) {
 354                 if (-1 == fchdir(fd))
 355                         perror(buf);
 356                 close(fd);
 357         }
 358         exprfree(e);
 359         free(sql);
 360         *sz = cur;
 361         return(rc);
 362 }
 363
 364 static void
 365 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 366                 uint64_t pageid, const char *path, int form)
 367 {
 368         char            *newnames, *prevsec, *prevarch;
 369         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 370         size_t           i;
 371         int              c;
 372
 373         mpage->file = NULL;
 374         mpage->names = NULL;
 375         prevsec = prevarch = NULL;
 376         i = 1;
 377         SQL_BIND_INT64(db, s, i, pageid);
 378         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 379
 380                 /* Decide whether we already have some names. */
 381
 382                 if (NULL == mpage->names) {
 383                         oldnames = "";
 384                         sep1 = "";
 385                 } else {
 386                         oldnames = mpage->names;
 387                         sep1 = ", ";
 388                 }
 389
 390                 /* Fetch the next name. */
 391
 392                 sec = sqlite3_column_text(s, 0);
 393                 arch = sqlite3_column_text(s, 1);
 394                 name = sqlite3_column_text(s, 2);
 395
 396                 /* If the section changed, append the old one. */
 397
 398                 if (NULL != prevsec &&
 399                     (strcmp(sec, prevsec) ||
 400                      strcmp(arch, prevarch))) {
 401                         sep2 = '\0' == *prevarch ? "" : "/";
 402                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 403                             oldnames, prevsec, sep2, prevarch);
 404                         free(mpage->names);
 405                         oldnames = mpage->names = newnames;
 406                         free(prevsec);
 407                         free(prevarch);
 408                         prevsec = prevarch = NULL;
 409                 }
 410
 411                 /* Save the new section, to append it later. */
 412
 413                 if (NULL == prevsec) {
 414                         prevsec = mandoc_strdup(sec);
 415                         prevarch = mandoc_strdup(arch);
 416                 }
 417
 418                 /* Append the new name. */
 419
 420                 mandoc_asprintf(&newnames, "%s%s%s",
 421                     oldnames, sep1, name);
 422                 free(mpage->names);
 423                 mpage->names = newnames;
 424
 425                 /* Also save the first file name encountered. */
 426
 427                 if (NULL != mpage->file)
 428                         continue;
 429
 430                 if (form) {
 431                         sep1 = "man";
 432                         fsec = sec;
 433                 } else {
 434                         sep1 = "cat";
 435                         fsec = "0";
 436                 }
 437                 sep2 = '\0' == *arch ? "" : "/";
 438                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 439                     path, sep1, sec, sep2, arch, name, fsec);
 440         }
 441         if (SQLITE_DONE != c)
 442                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 443         sqlite3_reset(s);
 444
 445         /* Append one final section to the names. */
 446
 447         if (NULL != prevsec) {
 448                 sep2 = '\0' == *prevarch ? "" : "/";
 449                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 450                     mpage->names, prevsec, sep2, prevarch);
 451                 free(mpage->names);
 452                 mpage->names = newnames;
 453                 free(prevsec);
 454                 free(prevarch);
 455         }
 456 }
 457
 458 static char *
 459 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 460 {
 461         char            *output, *newoutput;
 462         const char      *oldoutput, *sep1, *data;
 463         size_t           i;
 464         int              c;
 465
 466         output = NULL;
 467         i = 1;
 468         SQL_BIND_INT64(db, s, i, pageid);
 469         SQL_BIND_INT64(db, s, i, outbit);
 470         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 471                 if (NULL == output) {
 472                         oldoutput = "";
 473                         sep1 = "";
 474                 } else {
 475                         oldoutput = output;
 476                         sep1 = " # ";
 477                 }
 478                 data = sqlite3_column_text(s, 1);
 479                 mandoc_asprintf(&newoutput, "%s%s%s",
 480                     oldoutput, sep1, data);
 481                 free(output);
 482                 output = newoutput;
 483         }
 484         if (SQLITE_DONE != c)
 485                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 486         sqlite3_reset(s);
 487         return(output);
 488 }
 489
 490 /*
 491  * Implement substring match as an application-defined SQL function.
 492  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 493  * because that would require escaping metacharacters in the string
 494  * being searched for.
 495  */
 496 static void
 497 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 498 {
 499
 500         assert(2 == argc);
 501         sqlite3_result_int(context, NULL != strcasestr(
 502             (const char *)sqlite3_value_text(argv[1]),
 503             (const char *)sqlite3_value_text(argv[0])));
 504 }
 505
 506 /*
 507  * Implement regular expression match
 508  * as an application-defined SQL function.
 509  */
 510 static void
 511 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 512 {
 513
 514         assert(2 == argc);
 515         sqlite3_result_int(context, !regexec(
 516             (regex_t *)sqlite3_value_blob(argv[0]),
 517             (const char *)sqlite3_value_text(argv[1]),
 518             0, NULL, 0));
 519 }
 520
 521 static void
 522 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 523 {
 524         size_t           newsz;
 525
 526         newsz = 1 < count ? (size_t)count : strlen(newstr);
 527         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 528         if (1 < count)
 529                 memset(*sql + *sz, *newstr, (size_t)count);
 530         else
 531                 memcpy(*sql + *sz, newstr, newsz);
 532         *sz += newsz;
 533         (*sql)[*sz] = '\0';
 534 }
 535
 536 /*
 537  * Prepare the search SQL statement.
 538  */
 539 static char *
 540 sql_statement(const struct expr *e)
 541 {
 542         char            *sql;
 543         size_t           sz;
 544         int              needop;
 545
 546         sql = mandoc_strdup(
 547             "SELECT desc, form, pageid FROM mpages WHERE ");
 548         sz = strlen(sql);
 549
 550         for (needop = 0; NULL != e; e = e->next) {
 551                 if (e->and)
 552                         sql_append(&sql, &sz, " AND ", 1);
 553                 else if (needop)
 554                         sql_append(&sql, &sz, " OR ", 1);
 555                 if (e->open)
 556                         sql_append(&sql, &sz, "(", e->open);
 557                 sql_append(&sql, &sz,
 558                     TYPE_Nd & e->bits
 559                     ? (NULL == e->substr
 560                         ? "desc REGEXP ?"
 561                         : "desc MATCH ?")
 562                     : TYPE_Nm == e->bits
 563                     ? (NULL == e->substr
 564                         ? "pageid IN (SELECT pageid FROM names "
 565                           "WHERE name REGEXP ?)"
 566                         : "pageid IN (SELECT pageid FROM names "
 567                           "WHERE name MATCH ?)")
 568                     : (NULL == e->substr
 569                         ? "pageid IN (SELECT pageid FROM keys "
 570                           "WHERE key REGEXP ? AND bits & ?)"
 571                         : "pageid IN (SELECT pageid FROM keys "
 572                           "WHERE key MATCH ? AND bits & ?)"), 1);
 573                 if (e->close)
 574                         sql_append(&sql, &sz, ")", e->close);
 575                 needop = 1;
 576         }
 577
 578         return(sql);
 579 }
 580
 581 /*
 582  * Compile a set of string tokens into an expression.
 583  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 584  * "(", "foo=bar", etc.).
 585  */
 586 static struct expr *
 587 exprcomp(const struct mansearch *search, int argc, char *argv[])
 588 {
 589         uint64_t         mask;
 590         int              i, toopen, logic, igncase, toclose;
 591         struct expr     *first, *prev, *cur, *next;
 592
 593         first = cur = NULL;
 594         logic = igncase = toclose = 0;
 595         toopen = NULL != search->sec || NULL != search->arch;
 596
 597         for (i = 0; i < argc; i++) {
 598                 if (0 == strcmp("(", argv[i])) {
 599                         if (igncase)
 600                                 goto fail;
 601                         toopen++;
 602                         toclose++;
 603                         continue;
 604                 } else if (0 == strcmp(")", argv[i])) {
 605                         if (toopen || logic || igncase || NULL == cur)
 606                                 goto fail;
 607                         cur->close++;
 608                         if (0 > --toclose)
 609                                 goto fail;
 610                         continue;
 611                 } else if (0 == strcmp("-a", argv[i])) {
 612                         if (toopen || logic || igncase || NULL == cur)
 613                                 goto fail;
 614                         logic = 1;
 615                         continue;
 616                 } else if (0 == strcmp("-o", argv[i])) {
 617                         if (toopen || logic || igncase || NULL == cur)
 618                                 goto fail;
 619                         logic = 2;
 620                         continue;
 621                 } else if (0 == strcmp("-i", argv[i])) {
 622                         if (igncase)
 623                                 goto fail;
 624                         igncase = 1;
 625                         continue;
 626                 }
 627                 next = exprterm(search, argv[i], !igncase);
 628                 if (NULL == next)
 629                         goto fail;
 630                 if (NULL == first)
 631                         first = next;
 632                 else
 633                         cur->next = next;
 634                 prev = cur = next;
 635
 636                 /*
 637                  * Searching for descriptions must be split out
 638                  * because they are stored in the mpages table,
 639                  * not in the keys table.
 640                  */
 641
 642                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 643                         if (mask & cur->bits && ~mask & cur->bits) {
 644                                 next = mandoc_calloc(1,
 645                                     sizeof(struct expr));
 646                                 memcpy(next, cur, sizeof(struct expr));
 647                                 prev->open = 1;
 648                                 cur->bits = mask;
 649                                 cur->next = next;
 650                                 cur = next;
 651                                 cur->bits &= ~mask;
 652                         }
 653                 }
 654                 prev->and = (1 == logic);
 655                 prev->open += toopen;
 656                 if (cur != prev)
 657                         cur->close = 1;
 658
 659                 toopen = logic = igncase = 0;
 660         }
 661         if (toopen || logic || igncase || toclose)
 662                 goto fail;
 663
 664         if (NULL != search->sec || NULL != search->arch)
 665                 cur->close++;
 666         if (NULL != search->arch)
 667                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 668         if (NULL != search->sec)
 669                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 670
 671         return(first);
 672
 673 fail:
 674         if (NULL != first)
 675                 exprfree(first);
 676         return(NULL);
 677 }
 678
 679 static struct expr *
 680 exprspec(struct expr *cur, uint64_t key, const char *value,
 681                 const char *format)
 682 {
 683         char     errbuf[BUFSIZ];
 684         char    *cp;
 685         int      irc;
 686
 687         mandoc_asprintf(&cp, format, value);
 688         cur->next = mandoc_calloc(1, sizeof(struct expr));
 689         cur = cur->next;
 690         cur->and = 1;
 691         cur->bits = key;
 692         if (0 != (irc = regcomp(&cur->regexp, cp,
 693             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 694                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 695                 fprintf(stderr, "regcomp: %s\n", errbuf);
 696                 cur->substr = value;
 697         }
 698         free(cp);
 699         return(cur);
 700 }
 701
 702 static struct expr *
 703 exprterm(const struct mansearch *search, char *buf, int cs)
 704 {
 705         char             errbuf[BUFSIZ];
 706         struct expr     *e;
 707         char            *key, *v;
 708         uint64_t         iterbit;
 709         int              i, irc;
 710
 711         if ('\0' == *buf)
 712                 return(NULL);
 713
 714         e = mandoc_calloc(1, sizeof(struct expr));
 715
 716         /*"whatis" mode uses an opaque string and default fields. */
 717
 718         if (MANSEARCH_WHATIS & search->flags) {
 719                 e->substr = buf;
 720                 e->bits = search->deftype;
 721                 return(e);
 722         }
 723
 724         /*
 725          * If no =~ is specified, search with equality over names and
 726          * descriptions.
 727          * If =~ begins the phrase, use name and description fields.
 728          */
 729
 730         if (NULL == (v = strpbrk(buf, "=~"))) {
 731                 e->substr = buf;
 732                 e->bits = search->deftype;
 733                 return(e);
 734         } else if (v == buf)
 735                 e->bits = search->deftype;
 736
 737         if ('~' == *v++) {
 738                 if (NULL != strstr(buf, "arch"))
 739                         cs = 0;
 740                 if (0 != (irc = regcomp(&e->regexp, v,
 741                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE)))) {
 742                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 743                         fprintf(stderr, "regcomp: %s\n", errbuf);
 744                         free(e);
 745                         return(NULL);
 746                 }
 747         } else
 748                 e->substr = v;
 749         v[-1] = '\0';
 750
 751         /*
 752          * Parse out all possible fields.
 753          * If the field doesn't resolve, bail.
 754          */
 755
 756         while (NULL != (key = strsep(&buf, ","))) {
 757                 if ('\0' == *key)
 758                         continue;
 759                 for (i = 0, iterbit = 1;
 760                      i < mansearch_keymax;
 761                      i++, iterbit <<= 1) {
 762                         if (0 == strcasecmp(key,
 763                             mansearch_keynames[i])) {
 764                                 e->bits |= iterbit;
 765                                 break;
 766                         }
 767                 }
 768                 if (i == mansearch_keymax) {
 769                         if (strcasecmp(key, "any")) {
 770                                 free(e);
 771                                 return(NULL);
 772                         }
 773                         e->bits |= ~0ULL;
 774                 }
 775         }
 776
 777         return(e);
 778 }
 779
 780 static void
 781 exprfree(struct expr *p)
 782 {
 783         struct expr     *pp;
 784
 785         while (NULL != p) {
 786                 pp = p->next;
 787                 free(p);
 788                 p = pp;
 789         }
 790 }
 791
 792 static void *
 793 hash_halloc(size_t sz, void *arg)
 794 {
 795
 796         return(mandoc_calloc(1, sz));
 797 }
 798
 799 static void *
 800 hash_alloc(size_t sz, void *arg)
 801 {
 802
 803         return(mandoc_malloc(sz));
 804 }
 805
 806 static void
 807 hash_free(void *p, size_t sz, void *arg)
 808 {
 809
 810         free(p);
 811 }