contrib/mdocml/mansearch.c

   1 /*      $Id: mansearch.c,v 1.42 2014/08/09 14:24:53 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,
 127                     MAP_SHARED | MAP_ANON, -1, 0);
 128
 129                 if (MAP_FAILED == pagecache) {
 130                         perror("mmap");
 131                         pagecache = NULL;
 132                         return((int)MANDOCLEVEL_SYSERR);
 133                 }
 134
 135                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 136                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 137
 138                 if (SQLITE_OK == c)
 139                         return((int)MANDOCLEVEL_OK);
 140
 141                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 142
 143         } else if (NULL == pagecache) {
 144                 fprintf(stderr, "pagecache missing\n");
 145                 return((int)MANDOCLEVEL_BADARG);
 146         }
 147
 148         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 149                 perror("munmap");
 150                 pagecache = NULL;
 151                 return((int)MANDOCLEVEL_SYSERR);
 152         }
 153
 154         pagecache = NULL;
 155         return((int)MANDOCLEVEL_OK);
 156 }
 157
 158 int
 159 mansearch(const struct mansearch *search,
 160                 const struct manpaths *paths,
 161                 int argc, char *argv[],
 162                 const char *outkey,
 163                 struct manpage **res, size_t *sz)
 164 {
 165         int              fd, rc, c, indexbit;
 166         int64_t          pageid;
 167         uint64_t         outbit, iterbit;
 168         char             buf[PATH_MAX];
 169         char            *sql;
 170         struct manpage  *mpage;
 171         struct expr     *e, *ep;
 172         sqlite3         *db;
 173         sqlite3_stmt    *s, *s2;
 174         struct match    *mp;
 175         struct ohash_info info;
 176         struct ohash     htab;
 177         unsigned int     idx;
 178         size_t           i, j, cur, maxres;
 179
 180         info.calloc = hash_calloc;
 181         info.alloc = hash_alloc;
 182         info.free = hash_free;
 183         info.key_offset = offsetof(struct match, pageid);
 184
 185         *sz = cur = maxres = 0;
 186         sql = NULL;
 187         *res = NULL;
 188         fd = -1;
 189         e = NULL;
 190         rc = 0;
 191
 192         if (0 == argc)
 193                 goto out;
 194         if (NULL == (e = exprcomp(search, argc, argv)))
 195                 goto out;
 196
 197         outbit = 0;
 198         if (NULL != outkey) {
 199                 for (indexbit = 0, iterbit = 1;
 200                      indexbit < mansearch_keymax;
 201                      indexbit++, iterbit <<= 1) {
 202                         if (0 == strcasecmp(outkey,
 203                             mansearch_keynames[indexbit])) {
 204                                 outbit = iterbit;
 205                                 break;
 206                         }
 207                 }
 208         }
 209
 210         /*
 211          * Save a descriptor to the current working directory.
 212          * Since pathnames in the "paths" variable might be relative,
 213          * and we'll be chdir()ing into them, we need to keep a handle
 214          * on our current directory from which to start the chdir().
 215          */
 216
 217         if (NULL == getcwd(buf, PATH_MAX)) {
 218                 perror("getcwd");
 219                 goto out;
 220         } else if (-1 == (fd = open(buf, O_RDONLY, 0))) {
 221                 perror(buf);
 222                 goto out;
 223         }
 224
 225         sql = sql_statement(e);
 226
 227         /*
 228          * Loop over the directories (containing databases) for us to
 229          * search.
 230          * Don't let missing/bad databases/directories phase us.
 231          * In each, try to open the resident database and, if it opens,
 232          * scan it for our match expression.
 233          */
 234
 235         for (i = 0; i < paths->sz; i++) {
 236                 if (-1 == fchdir(fd)) {
 237                         perror(buf);
 238                         free(*res);
 239                         break;
 240                 } else if (-1 == chdir(paths->paths[i])) {
 241                         perror(paths->paths[i]);
 242                         continue;
 243                 }
 244
 245                 c = sqlite3_open_v2(MANDOC_DB, &db,
 246                     SQLITE_OPEN_READONLY, NULL);
 247
 248                 if (SQLITE_OK != c) {
 249                         perror(MANDOC_DB);
 250                         sqlite3_close(db);
 251                         continue;
 252                 }
 253
 254                 /*
 255                  * Define the SQL functions for substring
 256                  * and regular expression matching.
 257                  */
 258
 259                 c = sqlite3_create_function(db, "match", 2,
 260                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 261                     NULL, sql_match, NULL, NULL);
 262                 assert(SQLITE_OK == c);
 263                 c = sqlite3_create_function(db, "regexp", 2,
 264                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 265                     NULL, sql_regexp, NULL, NULL);
 266                 assert(SQLITE_OK == c);
 267
 268                 j = 1;
 269                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 270                 if (SQLITE_OK != c)
 271                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 272
 273                 for (ep = e; NULL != ep; ep = ep->next) {
 274                         if (NULL == ep->substr) {
 275                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 276                         } else
 277                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 278                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 279                                 SQL_BIND_INT64(db, s, j, ep->bits);
 280                 }
 281
 282                 memset(&htab, 0, sizeof(struct ohash));
 283                 ohash_init(&htab, 4, &info);
 284
 285                 /*
 286                  * Hash each entry on its [unique] document identifier.
 287                  * This is a uint64_t.
 288                  * Instead of using a hash function, simply convert the
 289                  * uint64_t to a uint32_t, the hash value's type.
 290                  * This gives good performance and preserves the
 291                  * distribution of buckets in the table.
 292                  */
 293                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 294                         pageid = sqlite3_column_int64(s, 2);
 295                         idx = ohash_lookup_memory(&htab,
 296                             (char *)&pageid, sizeof(uint64_t),
 297                             (uint32_t)pageid);
 298
 299                         if (NULL != ohash_find(&htab, idx))
 300                                 continue;
 301
 302                         mp = mandoc_calloc(1, sizeof(struct match));
 303                         mp->pageid = pageid;
 304                         mp->form = sqlite3_column_int(s, 1);
 305                         if (TYPE_Nd == outbit)
 306                                 mp->desc = mandoc_strdup((const char *)
 307                                     sqlite3_column_text(s, 0));
 308                         ohash_insert(&htab, idx, mp);
 309                 }
 310
 311                 if (SQLITE_DONE != c)
 312                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 313
 314                 sqlite3_finalize(s);
 315
 316                 c = sqlite3_prepare_v2(db,
 317                     "SELECT sec, arch, name, pageid FROM mlinks "
 318                     "WHERE pageid=? ORDER BY sec, arch, name",
 319                     -1, &s, NULL);
 320                 if (SQLITE_OK != c)
 321                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 322
 323                 c = sqlite3_prepare_v2(db,
 324                     "SELECT bits, key, pageid FROM keys "
 325                     "WHERE pageid=? AND bits & ?",
 326                     -1, &s2, NULL);
 327                 if (SQLITE_OK != c)
 328                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 329
 330                 for (mp = ohash_first(&htab, &idx);
 331                                 NULL != mp;
 332                                 mp = ohash_next(&htab, &idx)) {
 333                         if (cur + 1 > maxres) {
 334                                 maxres += 1024;
 335                                 *res = mandoc_reallocarray(*res,
 336                                     maxres, sizeof(struct manpage));
 337                         }
 338                         mpage = *res + cur;
 339                         mpage->sec = 10;
 340                         mpage->form = mp->form;
 341                         buildnames(mpage, db, s, mp->pageid,
 342                             paths->paths[i], mp->form);
 343                         mpage->output = TYPE_Nd & outbit ?
 344                             mp->desc : outbit ?
 345                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 346
 347                         free(mp);
 348                         cur++;
 349                 }
 350
 351                 sqlite3_finalize(s);
 352                 sqlite3_finalize(s2);
 353                 sqlite3_close(db);
 354                 ohash_delete(&htab);
 355         }
 356         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 357         rc = 1;
 358 out:
 359         if (-1 != fd) {
 360                 if (-1 == fchdir(fd))
 361                         perror(buf);
 362                 close(fd);
 363         }
 364         exprfree(e);
 365         free(sql);
 366         *sz = cur;
 367         return(rc);
 368 }
 369
 370 static int
 371 manpage_compare(const void *vp1, const void *vp2)
 372 {
 373         const struct manpage    *mp1, *mp2;
 374         int                      diff;
 375
 376         mp1 = vp1;
 377         mp2 = vp2;
 378         diff = mp1->sec - mp2->sec;
 379         return(diff ? diff : strcasecmp(mp1->names, mp2->names));
 380 }
 381
 382 static void
 383 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 384                 uint64_t pageid, const char *path, int form)
 385 {
 386         char            *newnames, *prevsec, *prevarch;
 387         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 388         size_t           i;
 389         int              c;
 390
 391         mpage->file = NULL;
 392         mpage->names = NULL;
 393         prevsec = prevarch = NULL;
 394         i = 1;
 395         SQL_BIND_INT64(db, s, i, pageid);
 396         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 397
 398                 /* Decide whether we already have some names. */
 399
 400                 if (NULL == mpage->names) {
 401                         oldnames = "";
 402                         sep1 = "";
 403                 } else {
 404                         oldnames = mpage->names;
 405                         sep1 = ", ";
 406                 }
 407
 408                 /* Fetch the next name. */
 409
 410                 sec = (const char *)sqlite3_column_text(s, 0);
 411                 arch = (const char *)sqlite3_column_text(s, 1);
 412                 name = (const char *)sqlite3_column_text(s, 2);
 413
 414                 /* Remember the first section found. */
 415
 416                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 417                         mpage->sec = (*sec - '1') + 1;
 418
 419                 /* If the section changed, append the old one. */
 420
 421                 if (NULL != prevsec &&
 422                     (strcmp(sec, prevsec) ||
 423                      strcmp(arch, prevarch))) {
 424                         sep2 = '\0' == *prevarch ? "" : "/";
 425                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 426                             oldnames, prevsec, sep2, prevarch);
 427                         free(mpage->names);
 428                         oldnames = mpage->names = newnames;
 429                         free(prevsec);
 430                         free(prevarch);
 431                         prevsec = prevarch = NULL;
 432                 }
 433
 434                 /* Save the new section, to append it later. */
 435
 436                 if (NULL == prevsec) {
 437                         prevsec = mandoc_strdup(sec);
 438                         prevarch = mandoc_strdup(arch);
 439                 }
 440
 441                 /* Append the new name. */
 442
 443                 mandoc_asprintf(&newnames, "%s%s%s",
 444                     oldnames, sep1, name);
 445                 free(mpage->names);
 446                 mpage->names = newnames;
 447
 448                 /* Also save the first file name encountered. */
 449
 450                 if (NULL != mpage->file)
 451                         continue;
 452
 453                 if (form) {
 454                         sep1 = "man";
 455                         fsec = sec;
 456                 } else {
 457                         sep1 = "cat";
 458                         fsec = "0";
 459                 }
 460                 sep2 = '\0' == *arch ? "" : "/";
 461                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 462                     path, sep1, sec, sep2, arch, name, fsec);
 463         }
 464         if (SQLITE_DONE != c)
 465                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 466         sqlite3_reset(s);
 467
 468         /* Append one final section to the names. */
 469
 470         if (NULL != prevsec) {
 471                 sep2 = '\0' == *prevarch ? "" : "/";
 472                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 473                     mpage->names, prevsec, sep2, prevarch);
 474                 free(mpage->names);
 475                 mpage->names = newnames;
 476                 free(prevsec);
 477                 free(prevarch);
 478         }
 479 }
 480
 481 static char *
 482 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 483 {
 484         char            *output, *newoutput;
 485         const char      *oldoutput, *sep1, *data;
 486         size_t           i;
 487         int              c;
 488
 489         output = NULL;
 490         i = 1;
 491         SQL_BIND_INT64(db, s, i, pageid);
 492         SQL_BIND_INT64(db, s, i, outbit);
 493         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 494                 if (NULL == output) {
 495                         oldoutput = "";
 496                         sep1 = "";
 497                 } else {
 498                         oldoutput = output;
 499                         sep1 = " # ";
 500                 }
 501                 data = (const char *)sqlite3_column_text(s, 1);
 502                 mandoc_asprintf(&newoutput, "%s%s%s",
 503                     oldoutput, sep1, data);
 504                 free(output);
 505                 output = newoutput;
 506         }
 507         if (SQLITE_DONE != c)
 508                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 509         sqlite3_reset(s);
 510         return(output);
 511 }
 512
 513 /*
 514  * Implement substring match as an application-defined SQL function.
 515  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 516  * because that would require escaping metacharacters in the string
 517  * being searched for.
 518  */
 519 static void
 520 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 521 {
 522
 523         assert(2 == argc);
 524         sqlite3_result_int(context, NULL != strcasestr(
 525             (const char *)sqlite3_value_text(argv[1]),
 526             (const char *)sqlite3_value_text(argv[0])));
 527 }
 528
 529 /*
 530  * Implement regular expression match
 531  * as an application-defined SQL function.
 532  */
 533 static void
 534 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 535 {
 536
 537         assert(2 == argc);
 538         sqlite3_result_int(context, !regexec(
 539             (regex_t *)sqlite3_value_blob(argv[0]),
 540             (const char *)sqlite3_value_text(argv[1]),
 541             0, NULL, 0));
 542 }
 543
 544 static void
 545 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 546 {
 547         size_t           newsz;
 548
 549         newsz = 1 < count ? (size_t)count : strlen(newstr);
 550         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 551         if (1 < count)
 552                 memset(*sql + *sz, *newstr, (size_t)count);
 553         else
 554                 memcpy(*sql + *sz, newstr, newsz);
 555         *sz += newsz;
 556         (*sql)[*sz] = '\0';
 557 }
 558
 559 /*
 560  * Prepare the search SQL statement.
 561  */
 562 static char *
 563 sql_statement(const struct expr *e)
 564 {
 565         char            *sql;
 566         size_t           sz;
 567         int              needop;
 568
 569         sql = mandoc_strdup(
 570             "SELECT desc, form, pageid FROM mpages WHERE ");
 571         sz = strlen(sql);
 572
 573         for (needop = 0; NULL != e; e = e->next) {
 574                 if (e->and)
 575                         sql_append(&sql, &sz, " AND ", 1);
 576                 else if (needop)
 577                         sql_append(&sql, &sz, " OR ", 1);
 578                 if (e->open)
 579                         sql_append(&sql, &sz, "(", e->open);
 580                 sql_append(&sql, &sz,
 581                     TYPE_Nd & e->bits
 582                     ? (NULL == e->substr
 583                         ? "desc REGEXP ?"
 584                         : "desc MATCH ?")
 585                     : TYPE_Nm == e->bits
 586                     ? (NULL == e->substr
 587                         ? "pageid IN (SELECT pageid FROM names "
 588                           "WHERE name REGEXP ?)"
 589                         : e->equal
 590                         ? "pageid IN (SELECT pageid FROM names "
 591                           "WHERE name = ?)"
 592                         : "pageid IN (SELECT pageid FROM names "
 593                           "WHERE name MATCH ?)")
 594                     : (NULL == e->substr
 595                         ? "pageid IN (SELECT pageid FROM keys "
 596                           "WHERE key REGEXP ? AND bits & ?)"
 597                         : "pageid IN (SELECT pageid FROM keys "
 598                           "WHERE key MATCH ? AND bits & ?)"), 1);
 599                 if (e->close)
 600                         sql_append(&sql, &sz, ")", e->close);
 601                 needop = 1;
 602         }
 603
 604         return(sql);
 605 }
 606
 607 /*
 608  * Compile a set of string tokens into an expression.
 609  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 610  * "(", "foo=bar", etc.).
 611  */
 612 static struct expr *
 613 exprcomp(const struct mansearch *search, int argc, char *argv[])
 614 {
 615         uint64_t         mask;
 616         int              i, toopen, logic, igncase, toclose;
 617         struct expr     *first, *prev, *cur, *next;
 618
 619         first = cur = NULL;
 620         logic = igncase = toclose = 0;
 621         toopen = NULL != search->sec || NULL != search->arch;
 622
 623         for (i = 0; i < argc; i++) {
 624                 if (0 == strcmp("(", argv[i])) {
 625                         if (igncase)
 626                                 goto fail;
 627                         toopen++;
 628                         toclose++;
 629                         continue;
 630                 } else if (0 == strcmp(")", argv[i])) {
 631                         if (toopen || logic || igncase || NULL == cur)
 632                                 goto fail;
 633                         cur->close++;
 634                         if (0 > --toclose)
 635                                 goto fail;
 636                         continue;
 637                 } else if (0 == strcmp("-a", argv[i])) {
 638                         if (toopen || logic || igncase || NULL == cur)
 639                                 goto fail;
 640                         logic = 1;
 641                         continue;
 642                 } else if (0 == strcmp("-o", argv[i])) {
 643                         if (toopen || logic || igncase || NULL == cur)
 644                                 goto fail;
 645                         logic = 2;
 646                         continue;
 647                 } else if (0 == strcmp("-i", argv[i])) {
 648                         if (igncase)
 649                                 goto fail;
 650                         igncase = 1;
 651                         continue;
 652                 }
 653                 next = exprterm(search, argv[i], !igncase);
 654                 if (NULL == next)
 655                         goto fail;
 656                 if (NULL == first)
 657                         first = next;
 658                 else
 659                         cur->next = next;
 660                 prev = cur = next;
 661
 662                 /*
 663                  * Searching for descriptions must be split out
 664                  * because they are stored in the mpages table,
 665                  * not in the keys table.
 666                  */
 667
 668                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 669                         if (mask & cur->bits && ~mask & cur->bits) {
 670                                 next = mandoc_calloc(1,
 671                                     sizeof(struct expr));
 672                                 memcpy(next, cur, sizeof(struct expr));
 673                                 prev->open = 1;
 674                                 cur->bits = mask;
 675                                 cur->next = next;
 676                                 cur = next;
 677                                 cur->bits &= ~mask;
 678                         }
 679                 }
 680                 prev->and = (1 == logic);
 681                 prev->open += toopen;
 682                 if (cur != prev)
 683                         cur->close = 1;
 684
 685                 toopen = logic = igncase = 0;
 686         }
 687         if (toopen || logic || igncase || toclose)
 688                 goto fail;
 689
 690         if (NULL != search->sec || NULL != search->arch)
 691                 cur->close++;
 692         if (NULL != search->arch)
 693                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 694         if (NULL != search->sec)
 695                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 696
 697         return(first);
 698
 699 fail:
 700         if (NULL != first)
 701                 exprfree(first);
 702         return(NULL);
 703 }
 704
 705 static struct expr *
 706 exprspec(struct expr *cur, uint64_t key, const char *value,
 707                 const char *format)
 708 {
 709         char     errbuf[BUFSIZ];
 710         char    *cp;
 711         int      irc;
 712
 713         mandoc_asprintf(&cp, format, value);
 714         cur->next = mandoc_calloc(1, sizeof(struct expr));
 715         cur = cur->next;
 716         cur->and = 1;
 717         cur->bits = key;
 718         if (0 != (irc = regcomp(&cur->regexp, cp,
 719             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 720                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 721                 fprintf(stderr, "regcomp: %s\n", errbuf);
 722                 cur->substr = value;
 723         }
 724         free(cp);
 725         return(cur);
 726 }
 727
 728 static struct expr *
 729 exprterm(const struct mansearch *search, char *buf, int cs)
 730 {
 731         char             errbuf[BUFSIZ];
 732         struct expr     *e;
 733         char            *key, *val;
 734         uint64_t         iterbit;
 735         int              i, irc;
 736
 737         if ('\0' == *buf)
 738                 return(NULL);
 739
 740         e = mandoc_calloc(1, sizeof(struct expr));
 741
 742         if (MANSEARCH_MAN & search->flags) {
 743                 e->bits = search->deftype;
 744                 e->substr = buf;
 745                 e->equal = 1;
 746                 return(e);
 747         }
 748
 749         /*
 750          * Look for an '=' or '~' operator,
 751          * unless forced to some fixed macro keys.
 752          */
 753
 754         if (MANSEARCH_WHATIS & search->flags)
 755                 val = NULL;
 756         else
 757                 val = strpbrk(buf, "=~");
 758
 759         if (NULL == val) {
 760                 e->bits = search->deftype;
 761                 e->substr = buf;
 762
 763         /*
 764          * Found an operator.
 765          * Regexp search is requested by !e->substr.
 766          */
 767
 768         } else {
 769                 if (val == buf)
 770                         e->bits = search->deftype;
 771                 if ('=' == *val)
 772                         e->substr = val + 1;
 773                 *val++ = '\0';
 774                 if (NULL != strstr(buf, "arch"))
 775                         cs = 0;
 776         }
 777
 778         /* Compile regular expressions. */
 779
 780         if (MANSEARCH_WHATIS & search->flags) {
 781                 e->substr = NULL;
 782                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 783         }
 784
 785         if (NULL == e->substr) {
 786                 irc = regcomp(&e->regexp, val,
 787                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 788                 if (MANSEARCH_WHATIS & search->flags)
 789                         free(val);
 790                 if (irc) {
 791                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 792                         fprintf(stderr, "regcomp: %s\n", errbuf);
 793                         free(e);
 794                         return(NULL);
 795                 }
 796         }
 797
 798         if (e->bits)
 799                 return(e);
 800
 801         /*
 802          * Parse out all possible fields.
 803          * If the field doesn't resolve, bail.
 804          */
 805
 806         while (NULL != (key = strsep(&buf, ","))) {
 807                 if ('\0' == *key)
 808                         continue;
 809                 for (i = 0, iterbit = 1;
 810                      i < mansearch_keymax;
 811                      i++, iterbit <<= 1) {
 812                         if (0 == strcasecmp(key,
 813                             mansearch_keynames[i])) {
 814                                 e->bits |= iterbit;
 815                                 break;
 816                         }
 817                 }
 818                 if (i == mansearch_keymax) {
 819                         if (strcasecmp(key, "any")) {
 820                                 free(e);
 821                                 return(NULL);
 822                         }
 823                         e->bits |= ~0ULL;
 824                 }
 825         }
 826
 827         return(e);
 828 }
 829
 830 static void
 831 exprfree(struct expr *p)
 832 {
 833         struct expr     *pp;
 834
 835         while (NULL != p) {
 836                 pp = p->next;
 837                 free(p);
 838                 p = pp;
 839         }
 840 }
 841
 842 static void *
 843 hash_calloc(size_t nmemb, size_t sz, void *arg)
 844 {
 845
 846         return(mandoc_calloc(nmemb, sz));
 847 }
 848
 849 static void *
 850 hash_alloc(size_t sz, void *arg)
 851 {
 852
 853         return(mandoc_malloc(sz));
 854 }
 855
 856 static void
 857 hash_free(void *p, void *arg)
 858 {
 859
 860         free(p);
 861 }