mansearch.c

   1 /*      $Id: mansearch.c,v 1.50 2014/11/18 01:15:21 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 #include "config.h"
  19
  20 #include <sys/mman.h>
  21 #include <sys/types.h>
  22
  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 #if 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         uint64_t         bits; /* name type mask */
  83         char            *desc; /* manual page description */
  84         int              form; /* bit field: formatted, zipped? */
  85 };
  86
  87 static  void             buildnames(struct manpage *, sqlite3 *,
  88                                 sqlite3_stmt *, uint64_t,
  89                                 const char *, int form);
  90 static  char            *buildoutput(sqlite3 *, sqlite3_stmt *,
  91                                  uint64_t, uint64_t);
  92 static  void            *hash_alloc(size_t, void *);
  93 static  void             hash_free(void *, void *);
  94 static  void            *hash_calloc(size_t, size_t, void *);
  95 static  struct expr     *exprcomp(const struct mansearch *,
  96                                 int, char *[]);
  97 static  void             exprfree(struct expr *);
  98 static  struct expr     *exprspec(struct expr *, uint64_t,
  99                                  const char *, const char *);
 100 static  struct expr     *exprterm(const struct mansearch *, char *, int);
 101 static  int              manpage_compare(const void *, const void *);
 102 static  void             sql_append(char **sql, size_t *sz,
 103                                 const char *newstr, int count);
 104 static  void             sql_match(sqlite3_context *context,
 105                                 int argc, sqlite3_value **argv);
 106 static  void             sql_regexp(sqlite3_context *context,
 107                                 int argc, sqlite3_value **argv);
 108 static  char            *sql_statement(const struct expr *);
 109
 110
 111 int
 112 mansearch_setup(int start)
 113 {
 114         static void     *pagecache;
 115         int              c;
 116
 117 #define PC_PAGESIZE     1280
 118 #define PC_NUMPAGES     256
 119
 120         if (start) {
 121                 if (NULL != pagecache) {
 122                         fprintf(stderr, "pagecache already enabled\n");
 123                         return((int)MANDOCLEVEL_BADARG);
 124                 }
 125
 126                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 127                     PROT_READ | PROT_WRITE,
 128                     MAP_SHARED | MAP_ANON, -1, 0);
 129
 130                 if (MAP_FAILED == pagecache) {
 131                         perror("mmap");
 132                         pagecache = NULL;
 133                         return((int)MANDOCLEVEL_SYSERR);
 134                 }
 135
 136                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 137                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 138
 139                 if (SQLITE_OK == c)
 140                         return((int)MANDOCLEVEL_OK);
 141
 142                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 143
 144         } else if (NULL == pagecache) {
 145                 fprintf(stderr, "pagecache missing\n");
 146                 return((int)MANDOCLEVEL_BADARG);
 147         }
 148
 149         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 150                 perror("munmap");
 151                 pagecache = NULL;
 152                 return((int)MANDOCLEVEL_SYSERR);
 153         }
 154
 155         pagecache = NULL;
 156         return((int)MANDOCLEVEL_OK);
 157 }
 158
 159 int
 160 mansearch(const struct mansearch *search,
 161                 const struct manpaths *paths,
 162                 int argc, char *argv[],
 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 != search->outkey) {
 199                 for (indexbit = 0, iterbit = 1;
 200                      indexbit < mansearch_keymax;
 201                      indexbit++, iterbit <<= 1) {
 202                         if (0 == strcasecmp(search->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                         mp->bits = sqlite3_column_int64(s, 3);
 306                         if (TYPE_Nd == outbit)
 307                                 mp->desc = mandoc_strdup((const char *)
 308                                     sqlite3_column_text(s, 0));
 309                         ohash_insert(&htab, idx, mp);
 310                 }
 311
 312                 if (SQLITE_DONE != c)
 313                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 314
 315                 sqlite3_finalize(s);
 316
 317                 c = sqlite3_prepare_v2(db,
 318                     "SELECT sec, arch, name, pageid FROM mlinks "
 319                     "WHERE pageid=? ORDER BY sec, arch, name",
 320                     -1, &s, NULL);
 321                 if (SQLITE_OK != c)
 322                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 323
 324                 c = sqlite3_prepare_v2(db,
 325                     "SELECT bits, key, pageid FROM keys "
 326                     "WHERE pageid=? AND bits & ?",
 327                     -1, &s2, NULL);
 328                 if (SQLITE_OK != c)
 329                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 330
 331                 for (mp = ohash_first(&htab, &idx);
 332                                 NULL != mp;
 333                                 mp = ohash_next(&htab, &idx)) {
 334                         if (cur + 1 > maxres) {
 335                                 maxres += 1024;
 336                                 *res = mandoc_reallocarray(*res,
 337                                     maxres, sizeof(struct manpage));
 338                         }
 339                         mpage = *res + cur;
 340                         mpage->ipath = i;
 341                         mpage->bits = mp->bits;
 342                         mpage->sec = 10;
 343                         mpage->form = mp->form;
 344                         buildnames(mpage, db, s, mp->pageid,
 345                             paths->paths[i], mp->form);
 346                         mpage->output = TYPE_Nd & outbit ?
 347                             mp->desc : outbit ?
 348                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 349
 350                         free(mp);
 351                         cur++;
 352                 }
 353
 354                 sqlite3_finalize(s);
 355                 sqlite3_finalize(s2);
 356                 sqlite3_close(db);
 357                 ohash_delete(&htab);
 358
 359                 /*
 360                  * In man(1) mode, prefer matches in earlier trees
 361                  * over matches in later trees.
 362                  */
 363
 364                 if (cur && search->firstmatch)
 365                         break;
 366         }
 367         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 368         rc = 1;
 369 out:
 370         if (-1 != fd) {
 371                 if (-1 == fchdir(fd))
 372                         perror(buf);
 373                 close(fd);
 374         }
 375         exprfree(e);
 376         free(sql);
 377         *sz = cur;
 378         return(rc);
 379 }
 380
 381 void
 382 mansearch_free(struct manpage *res, size_t sz)
 383 {
 384         size_t   i;
 385
 386         for (i = 0; i < sz; i++) {
 387                 free(res[i].file);
 388                 free(res[i].names);
 389                 free(res[i].output);
 390         }
 391         free(res);
 392 }
 393
 394 static int
 395 manpage_compare(const void *vp1, const void *vp2)
 396 {
 397         const struct manpage    *mp1, *mp2;
 398         int                      diff;
 399
 400         mp1 = vp1;
 401         mp2 = vp2;
 402         return( (diff = mp2->bits - mp1->bits) ? diff :
 403                 (diff = mp1->sec - mp2->sec) ? diff :
 404                 strcasecmp(mp1->names, mp2->names));
 405 }
 406
 407 static void
 408 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 409                 uint64_t pageid, const char *path, int form)
 410 {
 411         char            *newnames, *prevsec, *prevarch;
 412         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 413         const char      *gzip;
 414         size_t           i;
 415         int              c;
 416
 417         mpage->file = NULL;
 418         mpage->names = NULL;
 419         prevsec = prevarch = NULL;
 420         i = 1;
 421         SQL_BIND_INT64(db, s, i, pageid);
 422         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 423
 424                 /* Decide whether we already have some names. */
 425
 426                 if (NULL == mpage->names) {
 427                         oldnames = "";
 428                         sep1 = "";
 429                 } else {
 430                         oldnames = mpage->names;
 431                         sep1 = ", ";
 432                 }
 433
 434                 /* Fetch the next name. */
 435
 436                 sec = (const char *)sqlite3_column_text(s, 0);
 437                 arch = (const char *)sqlite3_column_text(s, 1);
 438                 name = (const char *)sqlite3_column_text(s, 2);
 439
 440                 /* Remember the first section found. */
 441
 442                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 443                         mpage->sec = (*sec - '1') + 1;
 444
 445                 /* If the section changed, append the old one. */
 446
 447                 if (NULL != prevsec &&
 448                     (strcmp(sec, prevsec) ||
 449                      strcmp(arch, prevarch))) {
 450                         sep2 = '\0' == *prevarch ? "" : "/";
 451                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 452                             oldnames, prevsec, sep2, prevarch);
 453                         free(mpage->names);
 454                         oldnames = mpage->names = newnames;
 455                         free(prevsec);
 456                         free(prevarch);
 457                         prevsec = prevarch = NULL;
 458                 }
 459
 460                 /* Save the new section, to append it later. */
 461
 462                 if (NULL == prevsec) {
 463                         prevsec = mandoc_strdup(sec);
 464                         prevarch = mandoc_strdup(arch);
 465                 }
 466
 467                 /* Append the new name. */
 468
 469                 mandoc_asprintf(&newnames, "%s%s%s",
 470                     oldnames, sep1, name);
 471                 free(mpage->names);
 472                 mpage->names = newnames;
 473
 474                 /* Also save the first file name encountered. */
 475
 476                 if (NULL != mpage->file)
 477                         continue;
 478
 479                 if (form & FORM_SRC) {
 480                         sep1 = "man";
 481                         fsec = sec;
 482                 } else {
 483                         sep1 = "cat";
 484                         fsec = "0";
 485                 }
 486                 if (form & FORM_GZ)
 487                         gzip = ".gz";
 488                 else
 489                         gzip = "";
 490                 sep2 = '\0' == *arch ? "" : "/";
 491                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s%s",
 492                     path, sep1, sec, sep2, arch, name, fsec, gzip);
 493         }
 494         if (SQLITE_DONE != c)
 495                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 496         sqlite3_reset(s);
 497
 498         /* Append one final section to the names. */
 499
 500         if (NULL != prevsec) {
 501                 sep2 = '\0' == *prevarch ? "" : "/";
 502                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 503                     mpage->names, prevsec, sep2, prevarch);
 504                 free(mpage->names);
 505                 mpage->names = newnames;
 506                 free(prevsec);
 507                 free(prevarch);
 508         }
 509 }
 510
 511 static char *
 512 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 513 {
 514         char            *output, *newoutput;
 515         const char      *oldoutput, *sep1, *data;
 516         size_t           i;
 517         int              c;
 518
 519         output = NULL;
 520         i = 1;
 521         SQL_BIND_INT64(db, s, i, pageid);
 522         SQL_BIND_INT64(db, s, i, outbit);
 523         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 524                 if (NULL == output) {
 525                         oldoutput = "";
 526                         sep1 = "";
 527                 } else {
 528                         oldoutput = output;
 529                         sep1 = " # ";
 530                 }
 531                 data = (const char *)sqlite3_column_text(s, 1);
 532                 mandoc_asprintf(&newoutput, "%s%s%s",
 533                     oldoutput, sep1, data);
 534                 free(output);
 535                 output = newoutput;
 536         }
 537         if (SQLITE_DONE != c)
 538                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 539         sqlite3_reset(s);
 540         return(output);
 541 }
 542
 543 /*
 544  * Implement substring match as an application-defined SQL function.
 545  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 546  * because that would require escaping metacharacters in the string
 547  * being searched for.
 548  */
 549 static void
 550 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 551 {
 552
 553         assert(2 == argc);
 554         sqlite3_result_int(context, NULL != strcasestr(
 555             (const char *)sqlite3_value_text(argv[1]),
 556             (const char *)sqlite3_value_text(argv[0])));
 557 }
 558
 559 /*
 560  * Implement regular expression match
 561  * as an application-defined SQL function.
 562  */
 563 static void
 564 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 565 {
 566
 567         assert(2 == argc);
 568         sqlite3_result_int(context, !regexec(
 569             (regex_t *)sqlite3_value_blob(argv[0]),
 570             (const char *)sqlite3_value_text(argv[1]),
 571             0, NULL, 0));
 572 }
 573
 574 static void
 575 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 576 {
 577         size_t           newsz;
 578
 579         newsz = 1 < count ? (size_t)count : strlen(newstr);
 580         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 581         if (1 < count)
 582                 memset(*sql + *sz, *newstr, (size_t)count);
 583         else
 584                 memcpy(*sql + *sz, newstr, newsz);
 585         *sz += newsz;
 586         (*sql)[*sz] = '\0';
 587 }
 588
 589 /*
 590  * Prepare the search SQL statement.
 591  */
 592 static char *
 593 sql_statement(const struct expr *e)
 594 {
 595         char            *sql;
 596         size_t           sz;
 597         int              needop;
 598
 599         sql = mandoc_strdup(e->equal ?
 600             "SELECT desc, form, pageid, bits "
 601                 "FROM mpages NATURAL JOIN names WHERE " :
 602             "SELECT desc, form, pageid, 0 FROM mpages WHERE ");
 603         sz = strlen(sql);
 604
 605         for (needop = 0; NULL != e; e = e->next) {
 606                 if (e->and)
 607                         sql_append(&sql, &sz, " AND ", 1);
 608                 else if (needop)
 609                         sql_append(&sql, &sz, " OR ", 1);
 610                 if (e->open)
 611                         sql_append(&sql, &sz, "(", e->open);
 612                 sql_append(&sql, &sz,
 613                     TYPE_Nd & e->bits
 614                     ? (NULL == e->substr
 615                         ? "desc REGEXP ?"
 616                         : "desc MATCH ?")
 617                     : TYPE_Nm == e->bits
 618                     ? (NULL == e->substr
 619                         ? "pageid IN (SELECT pageid FROM names "
 620                           "WHERE name REGEXP ?)"
 621                         : e->equal
 622                         ? "name = ? "
 623                         : "pageid IN (SELECT pageid FROM names "
 624                           "WHERE name MATCH ?)")
 625                     : (NULL == e->substr
 626                         ? "pageid IN (SELECT pageid FROM keys "
 627                           "WHERE key REGEXP ? AND bits & ?)"
 628                         : "pageid IN (SELECT pageid FROM keys "
 629                           "WHERE key MATCH ? AND bits & ?)"), 1);
 630                 if (e->close)
 631                         sql_append(&sql, &sz, ")", e->close);
 632                 needop = 1;
 633         }
 634
 635         return(sql);
 636 }
 637
 638 /*
 639  * Compile a set of string tokens into an expression.
 640  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 641  * "(", "foo=bar", etc.).
 642  */
 643 static struct expr *
 644 exprcomp(const struct mansearch *search, int argc, char *argv[])
 645 {
 646         uint64_t         mask;
 647         int              i, toopen, logic, igncase, toclose;
 648         struct expr     *first, *prev, *cur, *next;
 649
 650         first = cur = NULL;
 651         logic = igncase = toclose = 0;
 652         toopen = NULL != search->sec || NULL != search->arch;
 653
 654         for (i = 0; i < argc; i++) {
 655                 if (0 == strcmp("(", argv[i])) {
 656                         if (igncase)
 657                                 goto fail;
 658                         toopen++;
 659                         toclose++;
 660                         continue;
 661                 } else if (0 == strcmp(")", argv[i])) {
 662                         if (toopen || logic || igncase || NULL == cur)
 663                                 goto fail;
 664                         cur->close++;
 665                         if (0 > --toclose)
 666                                 goto fail;
 667                         continue;
 668                 } else if (0 == strcmp("-a", argv[i])) {
 669                         if (toopen || logic || igncase || NULL == cur)
 670                                 goto fail;
 671                         logic = 1;
 672                         continue;
 673                 } else if (0 == strcmp("-o", argv[i])) {
 674                         if (toopen || logic || igncase || NULL == cur)
 675                                 goto fail;
 676                         logic = 2;
 677                         continue;
 678                 } else if (0 == strcmp("-i", argv[i])) {
 679                         if (igncase)
 680                                 goto fail;
 681                         igncase = 1;
 682                         continue;
 683                 }
 684                 next = exprterm(search, argv[i], !igncase);
 685                 if (NULL == next)
 686                         goto fail;
 687                 if (NULL == first)
 688                         first = next;
 689                 else
 690                         cur->next = next;
 691                 prev = cur = next;
 692
 693                 /*
 694                  * Searching for descriptions must be split out
 695                  * because they are stored in the mpages table,
 696                  * not in the keys table.
 697                  */
 698
 699                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 700                         if (mask & cur->bits && ~mask & cur->bits) {
 701                                 next = mandoc_calloc(1,
 702                                     sizeof(struct expr));
 703                                 memcpy(next, cur, sizeof(struct expr));
 704                                 prev->open = 1;
 705                                 cur->bits = mask;
 706                                 cur->next = next;
 707                                 cur = next;
 708                                 cur->bits &= ~mask;
 709                         }
 710                 }
 711                 prev->and = (1 == logic);
 712                 prev->open += toopen;
 713                 if (cur != prev)
 714                         cur->close = 1;
 715
 716                 toopen = logic = igncase = 0;
 717         }
 718         if (toopen || logic || igncase || toclose)
 719                 goto fail;
 720
 721         if (NULL != search->sec || NULL != search->arch)
 722                 cur->close++;
 723         if (NULL != search->arch)
 724                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 725         if (NULL != search->sec)
 726                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 727
 728         return(first);
 729
 730 fail:
 731         if (NULL != first)
 732                 exprfree(first);
 733         return(NULL);
 734 }
 735
 736 static struct expr *
 737 exprspec(struct expr *cur, uint64_t key, const char *value,
 738                 const char *format)
 739 {
 740         char     errbuf[BUFSIZ];
 741         char    *cp;
 742         int      irc;
 743
 744         mandoc_asprintf(&cp, format, value);
 745         cur->next = mandoc_calloc(1, sizeof(struct expr));
 746         cur = cur->next;
 747         cur->and = 1;
 748         cur->bits = key;
 749         if (0 != (irc = regcomp(&cur->regexp, cp,
 750             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 751                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 752                 fprintf(stderr, "regcomp: %s\n", errbuf);
 753                 cur->substr = value;
 754         }
 755         free(cp);
 756         return(cur);
 757 }
 758
 759 static struct expr *
 760 exprterm(const struct mansearch *search, char *buf, int cs)
 761 {
 762         char             errbuf[BUFSIZ];
 763         struct expr     *e;
 764         char            *key, *val;
 765         uint64_t         iterbit;
 766         int              i, irc;
 767
 768         if ('\0' == *buf)
 769                 return(NULL);
 770
 771         e = mandoc_calloc(1, sizeof(struct expr));
 772
 773         if (search->argmode == ARG_NAME) {
 774                 e->bits = TYPE_Nm;
 775                 e->substr = buf;
 776                 e->equal = 1;
 777                 return(e);
 778         }
 779
 780         /*
 781          * Separate macro keys from search string.
 782          * If needed, request regular expression handling
 783          * by setting e->substr to NULL.
 784          */
 785
 786         if (search->argmode == ARG_WORD) {
 787                 e->bits = TYPE_Nm;
 788                 e->substr = NULL;
 789                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 790                 cs = 0;
 791         } else if ((val = strpbrk(buf, "=~")) == NULL) {
 792                 e->bits = TYPE_Nm | TYPE_Nd;
 793                 e->substr = buf;
 794         } else {
 795                 if (val == buf)
 796                         e->bits = TYPE_Nm | TYPE_Nd;
 797                 if ('=' == *val)
 798                         e->substr = val + 1;
 799                 *val++ = '\0';
 800                 if (NULL != strstr(buf, "arch"))
 801                         cs = 0;
 802         }
 803
 804         /* Compile regular expressions. */
 805
 806         if (NULL == e->substr) {
 807                 irc = regcomp(&e->regexp, val,
 808                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 809                 if (search->argmode == ARG_WORD)
 810                         free(val);
 811                 if (irc) {
 812                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 813                         fprintf(stderr, "regcomp: %s\n", errbuf);
 814                         free(e);
 815                         return(NULL);
 816                 }
 817         }
 818
 819         if (e->bits)
 820                 return(e);
 821
 822         /*
 823          * Parse out all possible fields.
 824          * If the field doesn't resolve, bail.
 825          */
 826
 827         while (NULL != (key = strsep(&buf, ","))) {
 828                 if ('\0' == *key)
 829                         continue;
 830                 for (i = 0, iterbit = 1;
 831                      i < mansearch_keymax;
 832                      i++, iterbit <<= 1) {
 833                         if (0 == strcasecmp(key,
 834                             mansearch_keynames[i])) {
 835                                 e->bits |= iterbit;
 836                                 break;
 837                         }
 838                 }
 839                 if (i == mansearch_keymax) {
 840                         if (strcasecmp(key, "any")) {
 841                                 free(e);
 842                                 return(NULL);
 843                         }
 844                         e->bits |= ~0ULL;
 845                 }
 846         }
 847
 848         return(e);
 849 }
 850
 851 static void
 852 exprfree(struct expr *p)
 853 {
 854         struct expr     *pp;
 855
 856         while (NULL != p) {
 857                 pp = p->next;
 858                 free(p);
 859                 p = pp;
 860         }
 861 }
 862
 863 static void *
 864 hash_calloc(size_t nmemb, size_t sz, void *arg)
 865 {
 866
 867         return(mandoc_calloc(nmemb, sz));
 868 }
 869
 870 static void *
 871 hash_alloc(size_t sz, void *arg)
 872 {
 873
 874         return(mandoc_malloc(sz));
 875 }
 876
 877 static void
 878 hash_free(void *p, void *arg)
 879 {
 880
 881         free(p);
 882 }