mansearch.c

   1 /*      $Id: mansearch.c,v 1.52 2014/12/06 01:23:24 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 <errno.h>
  25 #include <fcntl.h>
  26 #include <getopt.h>
  27 #include <limits.h>
  28 #include <regex.h>
  29 #include <stdio.h>
  30 #include <stdint.h>
  31 #include <stddef.h>
  32 #include <stdlib.h>
  33 #include <string.h>
  34 #include <unistd.h>
  35
  36 #if HAVE_OHASH
  37 #include <ohash.h>
  38 #else
  39 #include "compat_ohash.h"
  40 #endif
  41 #include <sqlite3.h>
  42 #ifndef SQLITE_DETERMINISTIC
  43 #define SQLITE_DETERMINISTIC 0
  44 #endif
  45
  46 #include "mandoc.h"
  47 #include "mandoc_aux.h"
  48 #include "manpath.h"
  49 #include "mansearch.h"
  50
  51 extern int mansearch_keymax;
  52 extern const char *const mansearch_keynames[];
  53
  54 #define SQL_BIND_TEXT(_db, _s, _i, _v) \
  55         do { if (SQLITE_OK != sqlite3_bind_text \
  56                 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
  57                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  58         } while (0)
  59 #define SQL_BIND_INT64(_db, _s, _i, _v) \
  60         do { if (SQLITE_OK != sqlite3_bind_int64 \
  61                 ((_s), (_i)++, (_v))) \
  62                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  63         } while (0)
  64 #define SQL_BIND_BLOB(_db, _s, _i, _v) \
  65         do { if (SQLITE_OK != sqlite3_bind_blob \
  66                 ((_s), (_i)++, (&_v), sizeof(_v), SQLITE_STATIC)) \
  67                 fprintf(stderr, "%s\n", sqlite3_errmsg((_db))); \
  68         } while (0)
  69
  70 struct  expr {
  71         regex_t          regexp;  /* compiled regexp, if applicable */
  72         const char      *substr;  /* to search for, if applicable */
  73         struct expr     *next;    /* next in sequence */
  74         uint64_t         bits;    /* type-mask */
  75         int              equal;   /* equality, not subsring match */
  76         int              open;    /* opening parentheses before */
  77         int              and;     /* logical AND before */
  78         int              close;   /* closing parentheses after */
  79 };
  80
  81 struct  match {
  82         uint64_t         pageid; /* identifier in database */
  83         uint64_t         bits; /* name type mask */
  84         char            *desc; /* manual page description */
  85         int              form; /* bit field: formatted, zipped? */
  86 };
  87
  88 static  void             buildnames(struct manpage *, sqlite3 *,
  89                                 sqlite3_stmt *, uint64_t,
  90                                 const char *, int form);
  91 static  char            *buildoutput(sqlite3 *, sqlite3_stmt *,
  92                                  uint64_t, uint64_t);
  93 static  void            *hash_alloc(size_t, void *);
  94 static  void             hash_free(void *, void *);
  95 static  void            *hash_calloc(size_t, size_t, void *);
  96 static  struct expr     *exprcomp(const struct mansearch *,
  97                                 int, char *[]);
  98 static  void             exprfree(struct expr *);
  99 static  struct expr     *exprspec(struct expr *, uint64_t,
 100                                  const char *, const char *);
 101 static  struct expr     *exprterm(const struct mansearch *, char *, int);
 102 static  int              manpage_compare(const void *, const void *);
 103 static  void             sql_append(char **sql, size_t *sz,
 104                                 const char *newstr, int count);
 105 static  void             sql_match(sqlite3_context *context,
 106                                 int argc, sqlite3_value **argv);
 107 static  void             sql_regexp(sqlite3_context *context,
 108                                 int argc, sqlite3_value **argv);
 109 static  char            *sql_statement(const struct expr *);
 110
 111
 112 int
 113 mansearch_setup(int start)
 114 {
 115         static void     *pagecache;
 116         int              c;
 117
 118 #define PC_PAGESIZE     1280
 119 #define PC_NUMPAGES     256
 120
 121         if (start) {
 122                 if (NULL != pagecache) {
 123                         fprintf(stderr, "pagecache already enabled\n");
 124                         return((int)MANDOCLEVEL_BADARG);
 125                 }
 126
 127                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 128                     PROT_READ | PROT_WRITE,
 129                     MAP_SHARED | MAP_ANON, -1, 0);
 130
 131                 if (MAP_FAILED == pagecache) {
 132                         perror("mmap");
 133                         pagecache = NULL;
 134                         return((int)MANDOCLEVEL_SYSERR);
 135                 }
 136
 137                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 138                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 139
 140                 if (SQLITE_OK == c)
 141                         return((int)MANDOCLEVEL_OK);
 142
 143                 fprintf(stderr, "pagecache: %s\n", sqlite3_errstr(c));
 144
 145         } else if (NULL == pagecache) {
 146                 fprintf(stderr, "pagecache missing\n");
 147                 return((int)MANDOCLEVEL_BADARG);
 148         }
 149
 150         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 151                 perror("munmap");
 152                 pagecache = NULL;
 153                 return((int)MANDOCLEVEL_SYSERR);
 154         }
 155
 156         pagecache = NULL;
 157         return((int)MANDOCLEVEL_OK);
 158 }
 159
 160 int
 161 mansearch(const struct mansearch *search,
 162                 const struct manpaths *paths,
 163                 int argc, char *argv[],
 164                 struct manpage **res, size_t *sz)
 165 {
 166         int              fd, rc, c, indexbit;
 167         int64_t          pageid;
 168         uint64_t         outbit, iterbit;
 169         char             buf[PATH_MAX];
 170         char            *sql;
 171         struct manpage  *mpage;
 172         struct expr     *e, *ep;
 173         sqlite3         *db;
 174         sqlite3_stmt    *s, *s2;
 175         struct match    *mp;
 176         struct ohash_info info;
 177         struct ohash     htab;
 178         unsigned int     idx;
 179         size_t           i, j, cur, maxres;
 180
 181         info.calloc = hash_calloc;
 182         info.alloc = hash_alloc;
 183         info.free = hash_free;
 184         info.key_offset = offsetof(struct match, pageid);
 185
 186         *sz = cur = maxres = 0;
 187         sql = NULL;
 188         *res = NULL;
 189         fd = -1;
 190         e = NULL;
 191         rc = 0;
 192
 193         if (0 == argc)
 194                 goto out;
 195         if (NULL == (e = exprcomp(search, argc, argv)))
 196                 goto out;
 197
 198         outbit = 0;
 199         if (NULL != search->outkey) {
 200                 for (indexbit = 0, iterbit = 1;
 201                      indexbit < mansearch_keymax;
 202                      indexbit++, iterbit <<= 1) {
 203                         if (0 == strcasecmp(search->outkey,
 204                             mansearch_keynames[indexbit])) {
 205                                 outbit = iterbit;
 206                                 break;
 207                         }
 208                 }
 209         }
 210
 211         /*
 212          * Save a descriptor to the current working directory.
 213          * Since pathnames in the "paths" variable might be relative,
 214          * and we'll be chdir()ing into them, we need to keep a handle
 215          * on our current directory from which to start the chdir().
 216          */
 217
 218         if (NULL == getcwd(buf, PATH_MAX)) {
 219                 perror("getcwd");
 220                 goto out;
 221         } else if (-1 == (fd = open(buf, O_RDONLY, 0))) {
 222                 perror(buf);
 223                 goto out;
 224         }
 225
 226         sql = sql_statement(e);
 227
 228         /*
 229          * Loop over the directories (containing databases) for us to
 230          * search.
 231          * Don't let missing/bad databases/directories phase us.
 232          * In each, try to open the resident database and, if it opens,
 233          * scan it for our match expression.
 234          */
 235
 236         for (i = 0; i < paths->sz; i++) {
 237                 if (-1 == fchdir(fd)) {
 238                         perror(buf);
 239                         free(*res);
 240                         break;
 241                 } else if (-1 == chdir(paths->paths[i])) {
 242                         perror(paths->paths[i]);
 243                         continue;
 244                 }
 245
 246                 c = sqlite3_open_v2(MANDOC_DB, &db,
 247                     SQLITE_OPEN_READONLY, NULL);
 248
 249                 if (SQLITE_OK != c) {
 250                         fprintf(stderr, "%s/%s: %s\n",
 251                             paths->paths[i], MANDOC_DB, strerror(errno));
 252                         sqlite3_close(db);
 253                         continue;
 254                 }
 255
 256                 /*
 257                  * Define the SQL functions for substring
 258                  * and regular expression matching.
 259                  */
 260
 261                 c = sqlite3_create_function(db, "match", 2,
 262                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 263                     NULL, sql_match, NULL, NULL);
 264                 assert(SQLITE_OK == c);
 265                 c = sqlite3_create_function(db, "regexp", 2,
 266                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 267                     NULL, sql_regexp, NULL, NULL);
 268                 assert(SQLITE_OK == c);
 269
 270                 j = 1;
 271                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 272                 if (SQLITE_OK != c)
 273                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 274
 275                 for (ep = e; NULL != ep; ep = ep->next) {
 276                         if (NULL == ep->substr) {
 277                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 278                         } else
 279                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 280                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 281                                 SQL_BIND_INT64(db, s, j, ep->bits);
 282                 }
 283
 284                 memset(&htab, 0, sizeof(struct ohash));
 285                 ohash_init(&htab, 4, &info);
 286
 287                 /*
 288                  * Hash each entry on its [unique] document identifier.
 289                  * This is a uint64_t.
 290                  * Instead of using a hash function, simply convert the
 291                  * uint64_t to a uint32_t, the hash value's type.
 292                  * This gives good performance and preserves the
 293                  * distribution of buckets in the table.
 294                  */
 295                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 296                         pageid = sqlite3_column_int64(s, 2);
 297                         idx = ohash_lookup_memory(&htab,
 298                             (char *)&pageid, sizeof(uint64_t),
 299                             (uint32_t)pageid);
 300
 301                         if (NULL != ohash_find(&htab, idx))
 302                                 continue;
 303
 304                         mp = mandoc_calloc(1, sizeof(struct match));
 305                         mp->pageid = pageid;
 306                         mp->form = sqlite3_column_int(s, 1);
 307                         mp->bits = sqlite3_column_int64(s, 3);
 308                         if (TYPE_Nd == outbit)
 309                                 mp->desc = mandoc_strdup((const char *)
 310                                     sqlite3_column_text(s, 0));
 311                         ohash_insert(&htab, idx, mp);
 312                 }
 313
 314                 if (SQLITE_DONE != c)
 315                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 316
 317                 sqlite3_finalize(s);
 318
 319                 c = sqlite3_prepare_v2(db,
 320                     "SELECT sec, arch, name, pageid FROM mlinks "
 321                     "WHERE pageid=? ORDER BY sec, arch, name",
 322                     -1, &s, NULL);
 323                 if (SQLITE_OK != c)
 324                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 325
 326                 c = sqlite3_prepare_v2(db,
 327                     "SELECT bits, key, pageid FROM keys "
 328                     "WHERE pageid=? AND bits & ?",
 329                     -1, &s2, NULL);
 330                 if (SQLITE_OK != c)
 331                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 332
 333                 for (mp = ohash_first(&htab, &idx);
 334                                 NULL != mp;
 335                                 mp = ohash_next(&htab, &idx)) {
 336                         if (cur + 1 > maxres) {
 337                                 maxres += 1024;
 338                                 *res = mandoc_reallocarray(*res,
 339                                     maxres, sizeof(struct manpage));
 340                         }
 341                         mpage = *res + cur;
 342                         mpage->ipath = i;
 343                         mpage->bits = mp->bits;
 344                         mpage->sec = 10;
 345                         mpage->form = mp->form;
 346                         buildnames(mpage, db, s, mp->pageid,
 347                             paths->paths[i], mp->form);
 348                         mpage->output = TYPE_Nd & outbit ?
 349                             mp->desc : outbit ?
 350                             buildoutput(db, s2, mp->pageid, outbit) : NULL;
 351
 352                         free(mp);
 353                         cur++;
 354                 }
 355
 356                 sqlite3_finalize(s);
 357                 sqlite3_finalize(s2);
 358                 sqlite3_close(db);
 359                 ohash_delete(&htab);
 360
 361                 /*
 362                  * In man(1) mode, prefer matches in earlier trees
 363                  * over matches in later trees.
 364                  */
 365
 366                 if (cur && search->firstmatch)
 367                         break;
 368         }
 369         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 370         rc = 1;
 371 out:
 372         if (-1 != fd) {
 373                 if (-1 == fchdir(fd))
 374                         perror(buf);
 375                 close(fd);
 376         }
 377         exprfree(e);
 378         free(sql);
 379         *sz = cur;
 380         return(rc);
 381 }
 382
 383 void
 384 mansearch_free(struct manpage *res, size_t sz)
 385 {
 386         size_t   i;
 387
 388         for (i = 0; i < sz; i++) {
 389                 free(res[i].file);
 390                 free(res[i].names);
 391                 free(res[i].output);
 392         }
 393         free(res);
 394 }
 395
 396 static int
 397 manpage_compare(const void *vp1, const void *vp2)
 398 {
 399         const struct manpage    *mp1, *mp2;
 400         int                      diff;
 401
 402         mp1 = vp1;
 403         mp2 = vp2;
 404         return( (diff = mp2->bits - mp1->bits) ? diff :
 405                 (diff = mp1->sec - mp2->sec) ? diff :
 406                 strcasecmp(mp1->names, mp2->names));
 407 }
 408
 409 static void
 410 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 411                 uint64_t pageid, const char *path, int form)
 412 {
 413         char            *newnames, *prevsec, *prevarch;
 414         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 415         size_t           i;
 416         int              c;
 417
 418         mpage->file = NULL;
 419         mpage->names = NULL;
 420         prevsec = prevarch = NULL;
 421         i = 1;
 422         SQL_BIND_INT64(db, s, i, pageid);
 423         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 424
 425                 /* Decide whether we already have some names. */
 426
 427                 if (NULL == mpage->names) {
 428                         oldnames = "";
 429                         sep1 = "";
 430                 } else {
 431                         oldnames = mpage->names;
 432                         sep1 = ", ";
 433                 }
 434
 435                 /* Fetch the next name. */
 436
 437                 sec = (const char *)sqlite3_column_text(s, 0);
 438                 arch = (const char *)sqlite3_column_text(s, 1);
 439                 name = (const char *)sqlite3_column_text(s, 2);
 440
 441                 /* Remember the first section found. */
 442
 443                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 444                         mpage->sec = (*sec - '1') + 1;
 445
 446                 /* If the section changed, append the old one. */
 447
 448                 if (NULL != prevsec &&
 449                     (strcmp(sec, prevsec) ||
 450                      strcmp(arch, prevarch))) {
 451                         sep2 = '\0' == *prevarch ? "" : "/";
 452                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 453                             oldnames, prevsec, sep2, prevarch);
 454                         free(mpage->names);
 455                         oldnames = mpage->names = newnames;
 456                         free(prevsec);
 457                         free(prevarch);
 458                         prevsec = prevarch = NULL;
 459                 }
 460
 461                 /* Save the new section, to append it later. */
 462
 463                 if (NULL == prevsec) {
 464                         prevsec = mandoc_strdup(sec);
 465                         prevarch = mandoc_strdup(arch);
 466                 }
 467
 468                 /* Append the new name. */
 469
 470                 mandoc_asprintf(&newnames, "%s%s%s",
 471                     oldnames, sep1, name);
 472                 free(mpage->names);
 473                 mpage->names = newnames;
 474
 475                 /* Also save the first file name encountered. */
 476
 477                 if (mpage->file != NULL)
 478                         continue;
 479
 480                 if (form & FORM_SRC) {
 481                         sep1 = "man";
 482                         fsec = sec;
 483                 } else {
 484                         sep1 = "cat";
 485                         fsec = "0";
 486                 }
 487                 sep2 = *arch == '\0' ? "" : "/";
 488                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 489                     path, sep1, sec, sep2, arch, name, fsec);
 490         }
 491         if (c != SQLITE_DONE)
 492                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 493         sqlite3_reset(s);
 494
 495         /* Append one final section to the names. */
 496
 497         if (prevsec != NULL) {
 498                 sep2 = *prevarch == '\0' ? "" : "/";
 499                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 500                     mpage->names, prevsec, sep2, prevarch);
 501                 free(mpage->names);
 502                 mpage->names = newnames;
 503                 free(prevsec);
 504                 free(prevarch);
 505         }
 506 }
 507
 508 static char *
 509 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 510 {
 511         char            *output, *newoutput;
 512         const char      *oldoutput, *sep1, *data;
 513         size_t           i;
 514         int              c;
 515
 516         output = NULL;
 517         i = 1;
 518         SQL_BIND_INT64(db, s, i, pageid);
 519         SQL_BIND_INT64(db, s, i, outbit);
 520         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 521                 if (NULL == output) {
 522                         oldoutput = "";
 523                         sep1 = "";
 524                 } else {
 525                         oldoutput = output;
 526                         sep1 = " # ";
 527                 }
 528                 data = (const char *)sqlite3_column_text(s, 1);
 529                 mandoc_asprintf(&newoutput, "%s%s%s",
 530                     oldoutput, sep1, data);
 531                 free(output);
 532                 output = newoutput;
 533         }
 534         if (SQLITE_DONE != c)
 535                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 536         sqlite3_reset(s);
 537         return(output);
 538 }
 539
 540 /*
 541  * Implement substring match as an application-defined SQL function.
 542  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 543  * because that would require escaping metacharacters in the string
 544  * being searched for.
 545  */
 546 static void
 547 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 548 {
 549
 550         assert(2 == argc);
 551         sqlite3_result_int(context, NULL != strcasestr(
 552             (const char *)sqlite3_value_text(argv[1]),
 553             (const char *)sqlite3_value_text(argv[0])));
 554 }
 555
 556 /*
 557  * Implement regular expression match
 558  * as an application-defined SQL function.
 559  */
 560 static void
 561 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 562 {
 563
 564         assert(2 == argc);
 565         sqlite3_result_int(context, !regexec(
 566             (regex_t *)sqlite3_value_blob(argv[0]),
 567             (const char *)sqlite3_value_text(argv[1]),
 568             0, NULL, 0));
 569 }
 570
 571 static void
 572 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 573 {
 574         size_t           newsz;
 575
 576         newsz = 1 < count ? (size_t)count : strlen(newstr);
 577         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 578         if (1 < count)
 579                 memset(*sql + *sz, *newstr, (size_t)count);
 580         else
 581                 memcpy(*sql + *sz, newstr, newsz);
 582         *sz += newsz;
 583         (*sql)[*sz] = '\0';
 584 }
 585
 586 /*
 587  * Prepare the search SQL statement.
 588  */
 589 static char *
 590 sql_statement(const struct expr *e)
 591 {
 592         char            *sql;
 593         size_t           sz;
 594         int              needop;
 595
 596         sql = mandoc_strdup(e->equal ?
 597             "SELECT desc, form, pageid, bits "
 598                 "FROM mpages NATURAL JOIN names WHERE " :
 599             "SELECT desc, form, pageid, 0 FROM mpages WHERE ");
 600         sz = strlen(sql);
 601
 602         for (needop = 0; NULL != e; e = e->next) {
 603                 if (e->and)
 604                         sql_append(&sql, &sz, " AND ", 1);
 605                 else if (needop)
 606                         sql_append(&sql, &sz, " OR ", 1);
 607                 if (e->open)
 608                         sql_append(&sql, &sz, "(", e->open);
 609                 sql_append(&sql, &sz,
 610                     TYPE_Nd & e->bits
 611                     ? (NULL == e->substr
 612                         ? "desc REGEXP ?"
 613                         : "desc MATCH ?")
 614                     : TYPE_Nm == e->bits
 615                     ? (NULL == e->substr
 616                         ? "pageid IN (SELECT pageid FROM names "
 617                           "WHERE name REGEXP ?)"
 618                         : e->equal
 619                         ? "name = ? "
 620                         : "pageid IN (SELECT pageid FROM names "
 621                           "WHERE name MATCH ?)")
 622                     : (NULL == e->substr
 623                         ? "pageid IN (SELECT pageid FROM keys "
 624                           "WHERE key REGEXP ? AND bits & ?)"
 625                         : "pageid IN (SELECT pageid FROM keys "
 626                           "WHERE key MATCH ? AND bits & ?)"), 1);
 627                 if (e->close)
 628                         sql_append(&sql, &sz, ")", e->close);
 629                 needop = 1;
 630         }
 631
 632         return(sql);
 633 }
 634
 635 /*
 636  * Compile a set of string tokens into an expression.
 637  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 638  * "(", "foo=bar", etc.).
 639  */
 640 static struct expr *
 641 exprcomp(const struct mansearch *search, int argc, char *argv[])
 642 {
 643         uint64_t         mask;
 644         int              i, toopen, logic, igncase, toclose;
 645         struct expr     *first, *prev, *cur, *next;
 646
 647         first = cur = NULL;
 648         logic = igncase = toclose = 0;
 649         toopen = NULL != search->sec || NULL != search->arch;
 650
 651         for (i = 0; i < argc; i++) {
 652                 if (0 == strcmp("(", argv[i])) {
 653                         if (igncase)
 654                                 goto fail;
 655                         toopen++;
 656                         toclose++;
 657                         continue;
 658                 } else if (0 == strcmp(")", argv[i])) {
 659                         if (toopen || logic || igncase || NULL == cur)
 660                                 goto fail;
 661                         cur->close++;
 662                         if (0 > --toclose)
 663                                 goto fail;
 664                         continue;
 665                 } else if (0 == strcmp("-a", argv[i])) {
 666                         if (toopen || logic || igncase || NULL == cur)
 667                                 goto fail;
 668                         logic = 1;
 669                         continue;
 670                 } else if (0 == strcmp("-o", argv[i])) {
 671                         if (toopen || logic || igncase || NULL == cur)
 672                                 goto fail;
 673                         logic = 2;
 674                         continue;
 675                 } else if (0 == strcmp("-i", argv[i])) {
 676                         if (igncase)
 677                                 goto fail;
 678                         igncase = 1;
 679                         continue;
 680                 }
 681                 next = exprterm(search, argv[i], !igncase);
 682                 if (NULL == next)
 683                         goto fail;
 684                 if (NULL == first)
 685                         first = next;
 686                 else
 687                         cur->next = next;
 688                 prev = cur = next;
 689
 690                 /*
 691                  * Searching for descriptions must be split out
 692                  * because they are stored in the mpages table,
 693                  * not in the keys table.
 694                  */
 695
 696                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 697                         if (mask & cur->bits && ~mask & cur->bits) {
 698                                 next = mandoc_calloc(1,
 699                                     sizeof(struct expr));
 700                                 memcpy(next, cur, sizeof(struct expr));
 701                                 prev->open = 1;
 702                                 cur->bits = mask;
 703                                 cur->next = next;
 704                                 cur = next;
 705                                 cur->bits &= ~mask;
 706                         }
 707                 }
 708                 prev->and = (1 == logic);
 709                 prev->open += toopen;
 710                 if (cur != prev)
 711                         cur->close = 1;
 712
 713                 toopen = logic = igncase = 0;
 714         }
 715         if (toopen || logic || igncase || toclose)
 716                 goto fail;
 717
 718         if (NULL != search->sec || NULL != search->arch)
 719                 cur->close++;
 720         if (NULL != search->arch)
 721                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 722         if (NULL != search->sec)
 723                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 724
 725         return(first);
 726
 727 fail:
 728         if (NULL != first)
 729                 exprfree(first);
 730         return(NULL);
 731 }
 732
 733 static struct expr *
 734 exprspec(struct expr *cur, uint64_t key, const char *value,
 735                 const char *format)
 736 {
 737         char     errbuf[BUFSIZ];
 738         char    *cp;
 739         int      irc;
 740
 741         mandoc_asprintf(&cp, format, value);
 742         cur->next = mandoc_calloc(1, sizeof(struct expr));
 743         cur = cur->next;
 744         cur->and = 1;
 745         cur->bits = key;
 746         if (0 != (irc = regcomp(&cur->regexp, cp,
 747             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 748                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 749                 fprintf(stderr, "regcomp: %s\n", errbuf);
 750                 cur->substr = value;
 751         }
 752         free(cp);
 753         return(cur);
 754 }
 755
 756 static struct expr *
 757 exprterm(const struct mansearch *search, char *buf, int cs)
 758 {
 759         char             errbuf[BUFSIZ];
 760         struct expr     *e;
 761         char            *key, *val;
 762         uint64_t         iterbit;
 763         int              i, irc;
 764
 765         if ('\0' == *buf)
 766                 return(NULL);
 767
 768         e = mandoc_calloc(1, sizeof(struct expr));
 769
 770         if (search->argmode == ARG_NAME) {
 771                 e->bits = TYPE_Nm;
 772                 e->substr = buf;
 773                 e->equal = 1;
 774                 return(e);
 775         }
 776
 777         /*
 778          * Separate macro keys from search string.
 779          * If needed, request regular expression handling
 780          * by setting e->substr to NULL.
 781          */
 782
 783         if (search->argmode == ARG_WORD) {
 784                 e->bits = TYPE_Nm;
 785                 e->substr = NULL;
 786                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 787                 cs = 0;
 788         } else if ((val = strpbrk(buf, "=~")) == NULL) {
 789                 e->bits = TYPE_Nm | TYPE_Nd;
 790                 e->substr = buf;
 791         } else {
 792                 if (val == buf)
 793                         e->bits = TYPE_Nm | TYPE_Nd;
 794                 if ('=' == *val)
 795                         e->substr = val + 1;
 796                 *val++ = '\0';
 797                 if (NULL != strstr(buf, "arch"))
 798                         cs = 0;
 799         }
 800
 801         /* Compile regular expressions. */
 802
 803         if (NULL == e->substr) {
 804                 irc = regcomp(&e->regexp, val,
 805                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 806                 if (search->argmode == ARG_WORD)
 807                         free(val);
 808                 if (irc) {
 809                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 810                         fprintf(stderr, "regcomp: %s\n", errbuf);
 811                         free(e);
 812                         return(NULL);
 813                 }
 814         }
 815
 816         if (e->bits)
 817                 return(e);
 818
 819         /*
 820          * Parse out all possible fields.
 821          * If the field doesn't resolve, bail.
 822          */
 823
 824         while (NULL != (key = strsep(&buf, ","))) {
 825                 if ('\0' == *key)
 826                         continue;
 827                 for (i = 0, iterbit = 1;
 828                      i < mansearch_keymax;
 829                      i++, iterbit <<= 1) {
 830                         if (0 == strcasecmp(key,
 831                             mansearch_keynames[i])) {
 832                                 e->bits |= iterbit;
 833                                 break;
 834                         }
 835                 }
 836                 if (i == mansearch_keymax) {
 837                         if (strcasecmp(key, "any")) {
 838                                 free(e);
 839                                 return(NULL);
 840                         }
 841                         e->bits |= ~0ULL;
 842                 }
 843         }
 844
 845         return(e);
 846 }
 847
 848 static void
 849 exprfree(struct expr *p)
 850 {
 851         struct expr     *pp;
 852
 853         while (NULL != p) {
 854                 pp = p->next;
 855                 free(p);
 856                 p = pp;
 857         }
 858 }
 859
 860 static void *
 861 hash_calloc(size_t nmemb, size_t sz, void *arg)
 862 {
 863
 864         return(mandoc_calloc(nmemb, sz));
 865 }
 866
 867 static void *
 868 hash_alloc(size_t sz, void *arg)
 869 {
 870
 871         return(mandoc_malloc(sz));
 872 }
 873
 874 static void
 875 hash_free(void *p, void *arg)
 876 {
 877
 878         free(p);
 879 }