mansearch.c

   1 /*      $Id: mansearch.c,v 1.48 2014/09/03 18:09:14 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         char            *desc; /* manual page description */
  83         int              form; /* bit field: formatted, zipped? */
  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                 struct manpage **res, size_t *sz)
 163 {
 164         int              fd, rc, c, indexbit;
 165         int64_t          pageid;
 166         uint64_t         outbit, iterbit;
 167         char             buf[PATH_MAX];
 168         char            *sql;
 169         struct manpage  *mpage;
 170         struct expr     *e, *ep;
 171         sqlite3         *db;
 172         sqlite3_stmt    *s, *s2;
 173         struct match    *mp;
 174         struct ohash_info info;
 175         struct ohash     htab;
 176         unsigned int     idx;
 177         size_t           i, j, cur, maxres;
 178
 179         info.calloc = hash_calloc;
 180         info.alloc = hash_alloc;
 181         info.free = hash_free;
 182         info.key_offset = offsetof(struct match, pageid);
 183
 184         *sz = cur = maxres = 0;
 185         sql = NULL;
 186         *res = NULL;
 187         fd = -1;
 188         e = NULL;
 189         rc = 0;
 190
 191         if (0 == argc)
 192                 goto out;
 193         if (NULL == (e = exprcomp(search, argc, argv)))
 194                 goto out;
 195
 196         outbit = 0;
 197         if (NULL != search->outkey) {
 198                 for (indexbit = 0, iterbit = 1;
 199                      indexbit < mansearch_keymax;
 200                      indexbit++, iterbit <<= 1) {
 201                         if (0 == strcasecmp(search->outkey,
 202                             mansearch_keynames[indexbit])) {
 203                                 outbit = iterbit;
 204                                 break;
 205                         }
 206                 }
 207         }
 208
 209         /*
 210          * Save a descriptor to the current working directory.
 211          * Since pathnames in the "paths" variable might be relative,
 212          * and we'll be chdir()ing into them, we need to keep a handle
 213          * on our current directory from which to start the chdir().
 214          */
 215
 216         if (NULL == getcwd(buf, PATH_MAX)) {
 217                 perror("getcwd");
 218                 goto out;
 219         } else if (-1 == (fd = open(buf, O_RDONLY, 0))) {
 220                 perror(buf);
 221                 goto out;
 222         }
 223
 224         sql = sql_statement(e);
 225
 226         /*
 227          * Loop over the directories (containing databases) for us to
 228          * search.
 229          * Don't let missing/bad databases/directories phase us.
 230          * In each, try to open the resident database and, if it opens,
 231          * scan it for our match expression.
 232          */
 233
 234         for (i = 0; i < paths->sz; i++) {
 235                 if (-1 == fchdir(fd)) {
 236                         perror(buf);
 237                         free(*res);
 238                         break;
 239                 } else if (-1 == chdir(paths->paths[i])) {
 240                         perror(paths->paths[i]);
 241                         continue;
 242                 }
 243
 244                 c = sqlite3_open_v2(MANDOC_DB, &db,
 245                     SQLITE_OPEN_READONLY, NULL);
 246
 247                 if (SQLITE_OK != c) {
 248                         perror(MANDOC_DB);
 249                         sqlite3_close(db);
 250                         continue;
 251                 }
 252
 253                 /*
 254                  * Define the SQL functions for substring
 255                  * and regular expression matching.
 256                  */
 257
 258                 c = sqlite3_create_function(db, "match", 2,
 259                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 260                     NULL, sql_match, NULL, NULL);
 261                 assert(SQLITE_OK == c);
 262                 c = sqlite3_create_function(db, "regexp", 2,
 263                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 264                     NULL, sql_regexp, NULL, NULL);
 265                 assert(SQLITE_OK == c);
 266
 267                 j = 1;
 268                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 269                 if (SQLITE_OK != c)
 270                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 271
 272                 for (ep = e; NULL != ep; ep = ep->next) {
 273                         if (NULL == ep->substr) {
 274                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 275                         } else
 276                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 277                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 278                                 SQL_BIND_INT64(db, s, j, ep->bits);
 279                 }
 280
 281                 memset(&htab, 0, sizeof(struct ohash));
 282                 ohash_init(&htab, 4, &info);
 283
 284                 /*
 285                  * Hash each entry on its [unique] document identifier.
 286                  * This is a uint64_t.
 287                  * Instead of using a hash function, simply convert the
 288                  * uint64_t to a uint32_t, the hash value's type.
 289                  * This gives good performance and preserves the
 290                  * distribution of buckets in the table.
 291                  */
 292                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 293                         pageid = sqlite3_column_int64(s, 2);
 294                         idx = ohash_lookup_memory(&htab,
 295                             (char *)&pageid, sizeof(uint64_t),
 296                             (uint32_t)pageid);
 297
 298                         if (NULL != ohash_find(&htab, idx))
 299                                 continue;
 300
 301                         mp = mandoc_calloc(1, sizeof(struct match));
 302                         mp->pageid = pageid;
 303                         mp->form = sqlite3_column_int(s, 1);
 304                         if (TYPE_Nd == outbit)
 305                                 mp->desc = mandoc_strdup((const char *)
 306                                     sqlite3_column_text(s, 0));
 307                         ohash_insert(&htab, idx, mp);
 308                 }
 309
 310                 if (SQLITE_DONE != c)
 311                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 312
 313                 sqlite3_finalize(s);
 314
 315                 c = sqlite3_prepare_v2(db,
 316                     "SELECT sec, arch, name, pageid FROM mlinks "
 317                     "WHERE pageid=? ORDER BY sec, arch, name",
 318                     -1, &s, NULL);
 319                 if (SQLITE_OK != c)
 320                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 321
 322                 c = sqlite3_prepare_v2(db,
 323                     "SELECT bits, key, pageid FROM keys "
 324                     "WHERE pageid=? AND bits & ?",
 325                     -1, &s2, NULL);
 326                 if (SQLITE_OK != c)
 327                         fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 328
 329                 for (mp = ohash_first(&htab, &idx);
 330                                 NULL != mp;
 331                                 mp = ohash_next(&htab, &idx)) {
 332                         if (cur + 1 > maxres) {
 333                                 maxres += 1024;
 334                                 *res = mandoc_reallocarray(*res,
 335                                     maxres, sizeof(struct manpage));
 336                         }
 337                         mpage = *res + cur;
 338                         mpage->ipath = i;
 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 void
 371 mansearch_free(struct manpage *res, size_t sz)
 372 {
 373         size_t   i;
 374
 375         for (i = 0; i < sz; i++) {
 376                 free(res[i].file);
 377                 free(res[i].names);
 378                 free(res[i].output);
 379         }
 380         free(res);
 381 }
 382
 383 static int
 384 manpage_compare(const void *vp1, const void *vp2)
 385 {
 386         const struct manpage    *mp1, *mp2;
 387         int                      diff;
 388
 389         mp1 = vp1;
 390         mp2 = vp2;
 391         diff = mp1->sec - mp2->sec;
 392         return(diff ? diff : strcasecmp(mp1->names, mp2->names));
 393 }
 394
 395 static void
 396 buildnames(struct manpage *mpage, sqlite3 *db, sqlite3_stmt *s,
 397                 uint64_t pageid, const char *path, int form)
 398 {
 399         char            *newnames, *prevsec, *prevarch;
 400         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 401         const char      *gzip;
 402         size_t           i;
 403         int              c;
 404
 405         mpage->file = NULL;
 406         mpage->names = NULL;
 407         prevsec = prevarch = NULL;
 408         i = 1;
 409         SQL_BIND_INT64(db, s, i, pageid);
 410         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 411
 412                 /* Decide whether we already have some names. */
 413
 414                 if (NULL == mpage->names) {
 415                         oldnames = "";
 416                         sep1 = "";
 417                 } else {
 418                         oldnames = mpage->names;
 419                         sep1 = ", ";
 420                 }
 421
 422                 /* Fetch the next name. */
 423
 424                 sec = (const char *)sqlite3_column_text(s, 0);
 425                 arch = (const char *)sqlite3_column_text(s, 1);
 426                 name = (const char *)sqlite3_column_text(s, 2);
 427
 428                 /* Remember the first section found. */
 429
 430                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 431                         mpage->sec = (*sec - '1') + 1;
 432
 433                 /* If the section changed, append the old one. */
 434
 435                 if (NULL != prevsec &&
 436                     (strcmp(sec, prevsec) ||
 437                      strcmp(arch, prevarch))) {
 438                         sep2 = '\0' == *prevarch ? "" : "/";
 439                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 440                             oldnames, prevsec, sep2, prevarch);
 441                         free(mpage->names);
 442                         oldnames = mpage->names = newnames;
 443                         free(prevsec);
 444                         free(prevarch);
 445                         prevsec = prevarch = NULL;
 446                 }
 447
 448                 /* Save the new section, to append it later. */
 449
 450                 if (NULL == prevsec) {
 451                         prevsec = mandoc_strdup(sec);
 452                         prevarch = mandoc_strdup(arch);
 453                 }
 454
 455                 /* Append the new name. */
 456
 457                 mandoc_asprintf(&newnames, "%s%s%s",
 458                     oldnames, sep1, name);
 459                 free(mpage->names);
 460                 mpage->names = newnames;
 461
 462                 /* Also save the first file name encountered. */
 463
 464                 if (NULL != mpage->file)
 465                         continue;
 466
 467                 if (form & FORM_SRC) {
 468                         sep1 = "man";
 469                         fsec = sec;
 470                 } else {
 471                         sep1 = "cat";
 472                         fsec = "0";
 473                 }
 474                 if (form & FORM_GZ)
 475                         gzip = ".gz";
 476                 else
 477                         gzip = "";
 478                 sep2 = '\0' == *arch ? "" : "/";
 479                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s%s",
 480                     path, sep1, sec, sep2, arch, name, fsec, gzip);
 481         }
 482         if (SQLITE_DONE != c)
 483                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 484         sqlite3_reset(s);
 485
 486         /* Append one final section to the names. */
 487
 488         if (NULL != prevsec) {
 489                 sep2 = '\0' == *prevarch ? "" : "/";
 490                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 491                     mpage->names, prevsec, sep2, prevarch);
 492                 free(mpage->names);
 493                 mpage->names = newnames;
 494                 free(prevsec);
 495                 free(prevarch);
 496         }
 497 }
 498
 499 static char *
 500 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 501 {
 502         char            *output, *newoutput;
 503         const char      *oldoutput, *sep1, *data;
 504         size_t           i;
 505         int              c;
 506
 507         output = NULL;
 508         i = 1;
 509         SQL_BIND_INT64(db, s, i, pageid);
 510         SQL_BIND_INT64(db, s, i, outbit);
 511         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 512                 if (NULL == output) {
 513                         oldoutput = "";
 514                         sep1 = "";
 515                 } else {
 516                         oldoutput = output;
 517                         sep1 = " # ";
 518                 }
 519                 data = (const char *)sqlite3_column_text(s, 1);
 520                 mandoc_asprintf(&newoutput, "%s%s%s",
 521                     oldoutput, sep1, data);
 522                 free(output);
 523                 output = newoutput;
 524         }
 525         if (SQLITE_DONE != c)
 526                 fprintf(stderr, "%s\n", sqlite3_errmsg(db));
 527         sqlite3_reset(s);
 528         return(output);
 529 }
 530
 531 /*
 532  * Implement substring match as an application-defined SQL function.
 533  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 534  * because that would require escaping metacharacters in the string
 535  * being searched for.
 536  */
 537 static void
 538 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 539 {
 540
 541         assert(2 == argc);
 542         sqlite3_result_int(context, NULL != strcasestr(
 543             (const char *)sqlite3_value_text(argv[1]),
 544             (const char *)sqlite3_value_text(argv[0])));
 545 }
 546
 547 /*
 548  * Implement regular expression match
 549  * as an application-defined SQL function.
 550  */
 551 static void
 552 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 553 {
 554
 555         assert(2 == argc);
 556         sqlite3_result_int(context, !regexec(
 557             (regex_t *)sqlite3_value_blob(argv[0]),
 558             (const char *)sqlite3_value_text(argv[1]),
 559             0, NULL, 0));
 560 }
 561
 562 static void
 563 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 564 {
 565         size_t           newsz;
 566
 567         newsz = 1 < count ? (size_t)count : strlen(newstr);
 568         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 569         if (1 < count)
 570                 memset(*sql + *sz, *newstr, (size_t)count);
 571         else
 572                 memcpy(*sql + *sz, newstr, newsz);
 573         *sz += newsz;
 574         (*sql)[*sz] = '\0';
 575 }
 576
 577 /*
 578  * Prepare the search SQL statement.
 579  */
 580 static char *
 581 sql_statement(const struct expr *e)
 582 {
 583         char            *sql;
 584         size_t           sz;
 585         int              needop;
 586
 587         sql = mandoc_strdup(
 588             "SELECT desc, form, pageid FROM mpages WHERE ");
 589         sz = strlen(sql);
 590
 591         for (needop = 0; NULL != e; e = e->next) {
 592                 if (e->and)
 593                         sql_append(&sql, &sz, " AND ", 1);
 594                 else if (needop)
 595                         sql_append(&sql, &sz, " OR ", 1);
 596                 if (e->open)
 597                         sql_append(&sql, &sz, "(", e->open);
 598                 sql_append(&sql, &sz,
 599                     TYPE_Nd & e->bits
 600                     ? (NULL == e->substr
 601                         ? "desc REGEXP ?"
 602                         : "desc MATCH ?")
 603                     : TYPE_Nm == e->bits
 604                     ? (NULL == e->substr
 605                         ? "pageid IN (SELECT pageid FROM names "
 606                           "WHERE name REGEXP ?)"
 607                         : e->equal
 608                         ? "pageid IN (SELECT pageid FROM names "
 609                           "WHERE name = ?)"
 610                         : "pageid IN (SELECT pageid FROM names "
 611                           "WHERE name MATCH ?)")
 612                     : (NULL == e->substr
 613                         ? "pageid IN (SELECT pageid FROM keys "
 614                           "WHERE key REGEXP ? AND bits & ?)"
 615                         : "pageid IN (SELECT pageid FROM keys "
 616                           "WHERE key MATCH ? AND bits & ?)"), 1);
 617                 if (e->close)
 618                         sql_append(&sql, &sz, ")", e->close);
 619                 needop = 1;
 620         }
 621
 622         return(sql);
 623 }
 624
 625 /*
 626  * Compile a set of string tokens into an expression.
 627  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 628  * "(", "foo=bar", etc.).
 629  */
 630 static struct expr *
 631 exprcomp(const struct mansearch *search, int argc, char *argv[])
 632 {
 633         uint64_t         mask;
 634         int              i, toopen, logic, igncase, toclose;
 635         struct expr     *first, *prev, *cur, *next;
 636
 637         first = cur = NULL;
 638         logic = igncase = toclose = 0;
 639         toopen = NULL != search->sec || NULL != search->arch;
 640
 641         for (i = 0; i < argc; i++) {
 642                 if (0 == strcmp("(", argv[i])) {
 643                         if (igncase)
 644                                 goto fail;
 645                         toopen++;
 646                         toclose++;
 647                         continue;
 648                 } else if (0 == strcmp(")", argv[i])) {
 649                         if (toopen || logic || igncase || NULL == cur)
 650                                 goto fail;
 651                         cur->close++;
 652                         if (0 > --toclose)
 653                                 goto fail;
 654                         continue;
 655                 } else if (0 == strcmp("-a", argv[i])) {
 656                         if (toopen || logic || igncase || NULL == cur)
 657                                 goto fail;
 658                         logic = 1;
 659                         continue;
 660                 } else if (0 == strcmp("-o", argv[i])) {
 661                         if (toopen || logic || igncase || NULL == cur)
 662                                 goto fail;
 663                         logic = 2;
 664                         continue;
 665                 } else if (0 == strcmp("-i", argv[i])) {
 666                         if (igncase)
 667                                 goto fail;
 668                         igncase = 1;
 669                         continue;
 670                 }
 671                 next = exprterm(search, argv[i], !igncase);
 672                 if (NULL == next)
 673                         goto fail;
 674                 if (NULL == first)
 675                         first = next;
 676                 else
 677                         cur->next = next;
 678                 prev = cur = next;
 679
 680                 /*
 681                  * Searching for descriptions must be split out
 682                  * because they are stored in the mpages table,
 683                  * not in the keys table.
 684                  */
 685
 686                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 687                         if (mask & cur->bits && ~mask & cur->bits) {
 688                                 next = mandoc_calloc(1,
 689                                     sizeof(struct expr));
 690                                 memcpy(next, cur, sizeof(struct expr));
 691                                 prev->open = 1;
 692                                 cur->bits = mask;
 693                                 cur->next = next;
 694                                 cur = next;
 695                                 cur->bits &= ~mask;
 696                         }
 697                 }
 698                 prev->and = (1 == logic);
 699                 prev->open += toopen;
 700                 if (cur != prev)
 701                         cur->close = 1;
 702
 703                 toopen = logic = igncase = 0;
 704         }
 705         if (toopen || logic || igncase || toclose)
 706                 goto fail;
 707
 708         if (NULL != search->sec || NULL != search->arch)
 709                 cur->close++;
 710         if (NULL != search->arch)
 711                 cur = exprspec(cur, TYPE_arch, search->arch, "^(%s|any)$");
 712         if (NULL != search->sec)
 713                 exprspec(cur, TYPE_sec, search->sec, "^%s$");
 714
 715         return(first);
 716
 717 fail:
 718         if (NULL != first)
 719                 exprfree(first);
 720         return(NULL);
 721 }
 722
 723 static struct expr *
 724 exprspec(struct expr *cur, uint64_t key, const char *value,
 725                 const char *format)
 726 {
 727         char     errbuf[BUFSIZ];
 728         char    *cp;
 729         int      irc;
 730
 731         mandoc_asprintf(&cp, format, value);
 732         cur->next = mandoc_calloc(1, sizeof(struct expr));
 733         cur = cur->next;
 734         cur->and = 1;
 735         cur->bits = key;
 736         if (0 != (irc = regcomp(&cur->regexp, cp,
 737             REG_EXTENDED | REG_NOSUB | REG_ICASE))) {
 738                 regerror(irc, &cur->regexp, errbuf, sizeof(errbuf));
 739                 fprintf(stderr, "regcomp: %s\n", errbuf);
 740                 cur->substr = value;
 741         }
 742         free(cp);
 743         return(cur);
 744 }
 745
 746 static struct expr *
 747 exprterm(const struct mansearch *search, char *buf, int cs)
 748 {
 749         char             errbuf[BUFSIZ];
 750         struct expr     *e;
 751         char            *key, *val;
 752         uint64_t         iterbit;
 753         int              i, irc;
 754
 755         if ('\0' == *buf)
 756                 return(NULL);
 757
 758         e = mandoc_calloc(1, sizeof(struct expr));
 759
 760         if (search->argmode == ARG_NAME) {
 761                 e->bits = TYPE_Nm;
 762                 e->substr = buf;
 763                 e->equal = 1;
 764                 return(e);
 765         }
 766
 767         /*
 768          * Separate macro keys from search string.
 769          * If needed, request regular expression handling
 770          * by setting e->substr to NULL.
 771          */
 772
 773         if (search->argmode == ARG_WORD) {
 774                 e->bits = TYPE_Nm;
 775                 e->substr = NULL;
 776                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 777                 cs = 0;
 778         } else if ((val = strpbrk(buf, "=~")) == NULL) {
 779                 e->bits = TYPE_Nm | TYPE_Nd;
 780                 e->substr = buf;
 781         } else {
 782                 if (val == buf)
 783                         e->bits = TYPE_Nm | TYPE_Nd;
 784                 if ('=' == *val)
 785                         e->substr = val + 1;
 786                 *val++ = '\0';
 787                 if (NULL != strstr(buf, "arch"))
 788                         cs = 0;
 789         }
 790
 791         /* Compile regular expressions. */
 792
 793         if (NULL == e->substr) {
 794                 irc = regcomp(&e->regexp, val,
 795                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 796                 if (search->argmode == ARG_WORD)
 797                         free(val);
 798                 if (irc) {
 799                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 800                         fprintf(stderr, "regcomp: %s\n", errbuf);
 801                         free(e);
 802                         return(NULL);
 803                 }
 804         }
 805
 806         if (e->bits)
 807                 return(e);
 808
 809         /*
 810          * Parse out all possible fields.
 811          * If the field doesn't resolve, bail.
 812          */
 813
 814         while (NULL != (key = strsep(&buf, ","))) {
 815                 if ('\0' == *key)
 816                         continue;
 817                 for (i = 0, iterbit = 1;
 818                      i < mansearch_keymax;
 819                      i++, iterbit <<= 1) {
 820                         if (0 == strcasecmp(key,
 821                             mansearch_keynames[i])) {
 822                                 e->bits |= iterbit;
 823                                 break;
 824                         }
 825                 }
 826                 if (i == mansearch_keymax) {
 827                         if (strcasecmp(key, "any")) {
 828                                 free(e);
 829                                 return(NULL);
 830                         }
 831                         e->bits |= ~0ULL;
 832                 }
 833         }
 834
 835         return(e);
 836 }
 837
 838 static void
 839 exprfree(struct expr *p)
 840 {
 841         struct expr     *pp;
 842
 843         while (NULL != p) {
 844                 pp = p->next;
 845                 free(p);
 846                 p = pp;
 847         }
 848 }
 849
 850 static void *
 851 hash_calloc(size_t nmemb, size_t sz, void *arg)
 852 {
 853
 854         return(mandoc_calloc(nmemb, sz));
 855 }
 856
 857 static void *
 858 hash_alloc(size_t sz, void *arg)
 859 {
 860
 861         return(mandoc_malloc(sz));
 862 }
 863
 864 static void
 865 hash_free(void *p, void *arg)
 866 {
 867
 868         free(p);
 869 }