mansearch.c

   1 /*      $Id: mansearch.c,v 1.65 2016/07/09 15:24:19 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2012 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014, 2015 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 AUTHORS DISCLAIM ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 #if HAVE_ERR
  25 #include <err.h>
  26 #endif
  27 #include <errno.h>
  28 #include <fcntl.h>
  29 #include <glob.h>
  30 #include <limits.h>
  31 #include <regex.h>
  32 #include <stdio.h>
  33 #include <stdint.h>
  34 #include <stddef.h>
  35 #include <stdlib.h>
  36 #include <string.h>
  37 #include <unistd.h>
  38
  39 #include <sqlite3.h>
  40 #ifndef SQLITE_DETERMINISTIC
  41 #define SQLITE_DETERMINISTIC 0
  42 #endif
  43
  44 #include "mandoc.h"
  45 #include "mandoc_aux.h"
  46 #include "mandoc_ohash.h"
  47 #include "manconf.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                 errx((int)MANDOCLEVEL_SYSERR, "%s", 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                 errx((int)MANDOCLEVEL_SYSERR, "%s", 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                 errx((int)MANDOCLEVEL_SYSERR, "%s", 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(const struct mansearch *,
  88                                 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  struct expr     *exprcomp(const struct mansearch *,
  94                                 int, char *[]);
  95 static  void             exprfree(struct expr *);
  96 static  struct expr     *exprterm(const struct mansearch *, char *, int);
  97 static  int              manpage_compare(const void *, const void *);
  98 static  void             sql_append(char **sql, size_t *sz,
  99                                 const char *newstr, int count);
 100 static  void             sql_match(sqlite3_context *context,
 101                                 int argc, sqlite3_value **argv);
 102 static  void             sql_regexp(sqlite3_context *context,
 103                                 int argc, sqlite3_value **argv);
 104 static  char            *sql_statement(const struct expr *);
 105
 106
 107 int
 108 mansearch_setup(int start)
 109 {
 110         static void     *pagecache;
 111         int              c;
 112
 113 #define PC_PAGESIZE     1280
 114 #define PC_NUMPAGES     256
 115
 116         if (start) {
 117                 if (NULL != pagecache) {
 118                         warnx("pagecache already enabled");
 119                         return (int)MANDOCLEVEL_BADARG;
 120                 }
 121
 122                 pagecache = mmap(NULL, PC_PAGESIZE * PC_NUMPAGES,
 123                     PROT_READ | PROT_WRITE,
 124                     MAP_SHARED | MAP_ANON, -1, 0);
 125
 126                 if (MAP_FAILED == pagecache) {
 127                         warn("mmap");
 128                         pagecache = NULL;
 129                         return (int)MANDOCLEVEL_SYSERR;
 130                 }
 131
 132                 c = sqlite3_config(SQLITE_CONFIG_PAGECACHE,
 133                     pagecache, PC_PAGESIZE, PC_NUMPAGES);
 134
 135                 if (SQLITE_OK == c)
 136                         return (int)MANDOCLEVEL_OK;
 137
 138                 warnx("pagecache: %s", sqlite3_errstr(c));
 139
 140         } else if (NULL == pagecache) {
 141                 warnx("pagecache missing");
 142                 return (int)MANDOCLEVEL_BADARG;
 143         }
 144
 145         if (-1 == munmap(pagecache, PC_PAGESIZE * PC_NUMPAGES)) {
 146                 warn("munmap");
 147                 pagecache = NULL;
 148                 return (int)MANDOCLEVEL_SYSERR;
 149         }
 150
 151         pagecache = NULL;
 152         return (int)MANDOCLEVEL_OK;
 153 }
 154
 155 int
 156 mansearch(const struct mansearch *search,
 157                 const struct manpaths *paths,
 158                 int argc, char *argv[],
 159                 struct manpage **res, size_t *sz)
 160 {
 161         int64_t          pageid;
 162         uint64_t         outbit, iterbit;
 163         char             buf[PATH_MAX];
 164         char            *sql;
 165         struct manpage  *mpage;
 166         struct expr     *e, *ep;
 167         sqlite3         *db;
 168         sqlite3_stmt    *s, *s2;
 169         struct match    *mp;
 170         struct ohash     htab;
 171         unsigned int     idx;
 172         size_t           i, j, cur, maxres;
 173         int              c, chdir_status, getcwd_status, indexbit;
 174
 175         if (argc == 0 || (e = exprcomp(search, argc, argv)) == NULL) {
 176                 *sz = 0;
 177                 return 0;
 178         }
 179
 180         cur = maxres = 0;
 181         *res = NULL;
 182
 183         if (NULL != search->outkey) {
 184                 outbit = TYPE_Nd;
 185                 for (indexbit = 0, iterbit = 1;
 186                      indexbit < mansearch_keymax;
 187                      indexbit++, iterbit <<= 1) {
 188                         if (0 == strcasecmp(search->outkey,
 189                             mansearch_keynames[indexbit])) {
 190                                 outbit = iterbit;
 191                                 break;
 192                         }
 193                 }
 194         } else
 195                 outbit = 0;
 196
 197         /*
 198          * Remember the original working directory, if possible.
 199          * This will be needed if the second or a later directory
 200          * is given as a relative path.
 201          * Do not error out if the current directory is not
 202          * searchable: Maybe it won't be needed after all.
 203          */
 204
 205         if (getcwd(buf, PATH_MAX) == NULL) {
 206                 getcwd_status = 0;
 207                 (void)strlcpy(buf, strerror(errno), sizeof(buf));
 208         } else
 209                 getcwd_status = 1;
 210
 211         sql = sql_statement(e);
 212
 213         /*
 214          * Loop over the directories (containing databases) for us to
 215          * search.
 216          * Don't let missing/bad databases/directories phase us.
 217          * In each, try to open the resident database and, if it opens,
 218          * scan it for our match expression.
 219          */
 220
 221         chdir_status = 0;
 222         for (i = 0; i < paths->sz; i++) {
 223                 if (chdir_status && paths->paths[i][0] != '/') {
 224                         if ( ! getcwd_status) {
 225                                 warnx("%s: getcwd: %s", paths->paths[i], buf);
 226                                 continue;
 227                         } else if (chdir(buf) == -1) {
 228                                 warn("%s", buf);
 229                                 continue;
 230                         }
 231                 }
 232                 if (chdir(paths->paths[i]) == -1) {
 233                         warn("%s", paths->paths[i]);
 234                         continue;
 235                 }
 236                 chdir_status = 1;
 237
 238                 c = sqlite3_open_v2(MANDOC_DB, &db,
 239                     SQLITE_OPEN_READONLY, NULL);
 240
 241                 if (SQLITE_OK != c) {
 242                         warn("%s/%s", paths->paths[i], MANDOC_DB);
 243                         sqlite3_close(db);
 244                         continue;
 245                 }
 246
 247                 /*
 248                  * Define the SQL functions for substring
 249                  * and regular expression matching.
 250                  */
 251
 252                 c = sqlite3_create_function(db, "match", 2,
 253                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 254                     NULL, sql_match, NULL, NULL);
 255                 assert(SQLITE_OK == c);
 256                 c = sqlite3_create_function(db, "regexp", 2,
 257                     SQLITE_UTF8 | SQLITE_DETERMINISTIC,
 258                     NULL, sql_regexp, NULL, NULL);
 259                 assert(SQLITE_OK == c);
 260
 261                 j = 1;
 262                 c = sqlite3_prepare_v2(db, sql, -1, &s, NULL);
 263                 if (SQLITE_OK != c)
 264                         errx((int)MANDOCLEVEL_SYSERR,
 265                             "%s", sqlite3_errmsg(db));
 266
 267                 for (ep = e; NULL != ep; ep = ep->next) {
 268                         if (NULL == ep->substr) {
 269                                 SQL_BIND_BLOB(db, s, j, ep->regexp);
 270                         } else
 271                                 SQL_BIND_TEXT(db, s, j, ep->substr);
 272                         if (0 == ((TYPE_Nd | TYPE_Nm) & ep->bits))
 273                                 SQL_BIND_INT64(db, s, j, ep->bits);
 274                 }
 275
 276                 mandoc_ohash_init(&htab, 4, offsetof(struct match, pageid));
 277
 278                 /*
 279                  * Hash each entry on its [unique] document identifier.
 280                  * This is a uint64_t.
 281                  * Instead of using a hash function, simply convert the
 282                  * uint64_t to a uint32_t, the hash value's type.
 283                  * This gives good performance and preserves the
 284                  * distribution of buckets in the table.
 285                  */
 286                 while (SQLITE_ROW == (c = sqlite3_step(s))) {
 287                         pageid = sqlite3_column_int64(s, 2);
 288                         idx = ohash_lookup_memory(&htab,
 289                             (char *)&pageid, sizeof(uint64_t),
 290                             (uint32_t)pageid);
 291
 292                         if (NULL != ohash_find(&htab, idx))
 293                                 continue;
 294
 295                         mp = mandoc_calloc(1, sizeof(struct match));
 296                         mp->pageid = pageid;
 297                         mp->form = sqlite3_column_int(s, 1);
 298                         mp->bits = sqlite3_column_int64(s, 3);
 299                         if (TYPE_Nd == outbit)
 300                                 mp->desc = mandoc_strdup((const char *)
 301                                     sqlite3_column_text(s, 0));
 302                         ohash_insert(&htab, idx, mp);
 303                 }
 304
 305                 if (SQLITE_DONE != c)
 306                         warnx("%s", sqlite3_errmsg(db));
 307
 308                 sqlite3_finalize(s);
 309
 310                 c = sqlite3_prepare_v2(db,
 311                     "SELECT sec, arch, name, pageid FROM mlinks "
 312                     "WHERE pageid=? ORDER BY sec, arch, name",
 313                     -1, &s, NULL);
 314                 if (SQLITE_OK != c)
 315                         errx((int)MANDOCLEVEL_SYSERR,
 316                             "%s", sqlite3_errmsg(db));
 317
 318                 c = sqlite3_prepare_v2(db,
 319                     "SELECT bits, key, pageid FROM keys "
 320                     "WHERE pageid=? AND bits & ?",
 321                     -1, &s2, NULL);
 322                 if (SQLITE_OK != c)
 323                         errx((int)MANDOCLEVEL_SYSERR,
 324                             "%s", sqlite3_errmsg(db));
 325
 326                 for (mp = ohash_first(&htab, &idx);
 327                                 NULL != mp;
 328                                 mp = ohash_next(&htab, &idx)) {
 329                         if (cur + 1 > maxres) {
 330                                 maxres += 1024;
 331                                 *res = mandoc_reallocarray(*res,
 332                                     maxres, sizeof(struct manpage));
 333                         }
 334                         mpage = *res + cur;
 335                         mpage->ipath = i;
 336                         mpage->bits = mp->bits;
 337                         mpage->sec = 10;
 338                         mpage->form = mp->form;
 339                         buildnames(search, mpage, db, s, mp->pageid,
 340                             paths->paths[i], mp->form);
 341                         if (mpage->names != NULL) {
 342                                 mpage->output = TYPE_Nd & outbit ?
 343                                     mp->desc : outbit ?
 344                                     buildoutput(db, s2, mp->pageid, outbit) :
 345                                     NULL;
 346                                 cur++;
 347                         }
 348                         free(mp);
 349                 }
 350
 351                 sqlite3_finalize(s);
 352                 sqlite3_finalize(s2);
 353                 sqlite3_close(db);
 354                 ohash_delete(&htab);
 355
 356                 /*
 357                  * In man(1) mode, prefer matches in earlier trees
 358                  * over matches in later trees.
 359                  */
 360
 361                 if (cur && search->firstmatch)
 362                         break;
 363         }
 364         qsort(*res, cur, sizeof(struct manpage), manpage_compare);
 365         if (chdir_status && getcwd_status && chdir(buf) == -1)
 366                 warn("%s", buf);
 367         exprfree(e);
 368         free(sql);
 369         *sz = cur;
 370         return 1;
 371 }
 372
 373 void
 374 mansearch_free(struct manpage *res, size_t sz)
 375 {
 376         size_t   i;
 377
 378         for (i = 0; i < sz; i++) {
 379                 free(res[i].file);
 380                 free(res[i].names);
 381                 free(res[i].output);
 382         }
 383         free(res);
 384 }
 385
 386 static int
 387 manpage_compare(const void *vp1, const void *vp2)
 388 {
 389         const struct manpage    *mp1, *mp2;
 390         int                      diff;
 391
 392         mp1 = vp1;
 393         mp2 = vp2;
 394         return (diff = mp2->bits - mp1->bits) ? diff :
 395             (diff = mp1->sec - mp2->sec) ? diff :
 396             strcasecmp(mp1->names, mp2->names);
 397 }
 398
 399 static void
 400 buildnames(const struct mansearch *search, struct manpage *mpage,
 401                 sqlite3 *db, sqlite3_stmt *s,
 402                 uint64_t pageid, const char *path, int form)
 403 {
 404         glob_t           globinfo;
 405         char            *firstname, *newnames, *prevsec, *prevarch;
 406         const char      *oldnames, *sep1, *name, *sec, *sep2, *arch, *fsec;
 407         size_t           i;
 408         int              c, globres;
 409
 410         mpage->file = NULL;
 411         mpage->names = NULL;
 412         firstname = prevsec = prevarch = NULL;
 413         i = 1;
 414         SQL_BIND_INT64(db, s, i, pageid);
 415         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 416
 417                 /* Decide whether we already have some names. */
 418
 419                 if (NULL == mpage->names) {
 420                         oldnames = "";
 421                         sep1 = "";
 422                 } else {
 423                         oldnames = mpage->names;
 424                         sep1 = ", ";
 425                 }
 426
 427                 /* Fetch the next name, rejecting sec/arch mismatches. */
 428
 429                 sec = (const char *)sqlite3_column_text(s, 0);
 430                 if (search->sec != NULL && strcasecmp(sec, search->sec))
 431                         continue;
 432                 arch = (const char *)sqlite3_column_text(s, 1);
 433                 if (search->arch != NULL && *arch != '\0' &&
 434                     strcasecmp(arch, search->arch))
 435                         continue;
 436                 name = (const char *)sqlite3_column_text(s, 2);
 437
 438                 /* Remember the first section found. */
 439
 440                 if (9 < mpage->sec && '1' <= *sec && '9' >= *sec)
 441                         mpage->sec = (*sec - '1') + 1;
 442
 443                 /* If the section changed, append the old one. */
 444
 445                 if (NULL != prevsec &&
 446                     (strcmp(sec, prevsec) ||
 447                      strcmp(arch, prevarch))) {
 448                         sep2 = '\0' == *prevarch ? "" : "/";
 449                         mandoc_asprintf(&newnames, "%s(%s%s%s)",
 450                             oldnames, prevsec, sep2, prevarch);
 451                         free(mpage->names);
 452                         oldnames = mpage->names = newnames;
 453                         free(prevsec);
 454                         free(prevarch);
 455                         prevsec = prevarch = NULL;
 456                 }
 457
 458                 /* Save the new section, to append it later. */
 459
 460                 if (NULL == prevsec) {
 461                         prevsec = mandoc_strdup(sec);
 462                         prevarch = mandoc_strdup(arch);
 463                 }
 464
 465                 /* Append the new name. */
 466
 467                 mandoc_asprintf(&newnames, "%s%s%s",
 468                     oldnames, sep1, name);
 469                 free(mpage->names);
 470                 mpage->names = newnames;
 471
 472                 /* Also save the first file name encountered. */
 473
 474                 if (mpage->file != NULL)
 475                         continue;
 476
 477                 if (form & FORM_SRC) {
 478                         sep1 = "man";
 479                         fsec = sec;
 480                 } else {
 481                         sep1 = "cat";
 482                         fsec = "0";
 483                 }
 484                 sep2 = *arch == '\0' ? "" : "/";
 485                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.%s",
 486                     path, sep1, sec, sep2, arch, name, fsec);
 487                 if (access(mpage->file, R_OK) != -1)
 488                         continue;
 489
 490                 /* Handle unusual file name extensions. */
 491
 492                 if (firstname == NULL)
 493                         firstname = mpage->file;
 494                 else
 495                         free(mpage->file);
 496                 mandoc_asprintf(&mpage->file, "%s/%s%s%s%s/%s.*",
 497                     path, sep1, sec, sep2, arch, name);
 498                 globres = glob(mpage->file, 0, NULL, &globinfo);
 499                 free(mpage->file);
 500                 mpage->file = globres ? NULL :
 501                     mandoc_strdup(*globinfo.gl_pathv);
 502                 globfree(&globinfo);
 503         }
 504         if (c != SQLITE_DONE)
 505                 warnx("%s", sqlite3_errmsg(db));
 506         sqlite3_reset(s);
 507
 508         /* If none of the files is usable, use the first name. */
 509
 510         if (mpage->file == NULL)
 511                 mpage->file = firstname;
 512         else if (mpage->file != firstname)
 513                 free(firstname);
 514
 515         /* Append one final section to the names. */
 516
 517         if (prevsec != NULL) {
 518                 sep2 = *prevarch == '\0' ? "" : "/";
 519                 mandoc_asprintf(&newnames, "%s(%s%s%s)",
 520                     mpage->names, prevsec, sep2, prevarch);
 521                 free(mpage->names);
 522                 mpage->names = newnames;
 523                 free(prevsec);
 524                 free(prevarch);
 525         }
 526 }
 527
 528 static char *
 529 buildoutput(sqlite3 *db, sqlite3_stmt *s, uint64_t pageid, uint64_t outbit)
 530 {
 531         char            *output, *newoutput;
 532         const char      *oldoutput, *sep1, *data;
 533         size_t           i;
 534         int              c;
 535
 536         output = NULL;
 537         i = 1;
 538         SQL_BIND_INT64(db, s, i, pageid);
 539         SQL_BIND_INT64(db, s, i, outbit);
 540         while (SQLITE_ROW == (c = sqlite3_step(s))) {
 541                 if (NULL == output) {
 542                         oldoutput = "";
 543                         sep1 = "";
 544                 } else {
 545                         oldoutput = output;
 546                         sep1 = " # ";
 547                 }
 548                 data = (const char *)sqlite3_column_text(s, 1);
 549                 mandoc_asprintf(&newoutput, "%s%s%s",
 550                     oldoutput, sep1, data);
 551                 free(output);
 552                 output = newoutput;
 553         }
 554         if (SQLITE_DONE != c)
 555                 warnx("%s", sqlite3_errmsg(db));
 556         sqlite3_reset(s);
 557         return output;
 558 }
 559
 560 /*
 561  * Implement substring match as an application-defined SQL function.
 562  * Using the SQL LIKE or GLOB operators instead would be a bad idea
 563  * because that would require escaping metacharacters in the string
 564  * being searched for.
 565  */
 566 static void
 567 sql_match(sqlite3_context *context, int argc, sqlite3_value **argv)
 568 {
 569
 570         assert(2 == argc);
 571         sqlite3_result_int(context, NULL != strcasestr(
 572             (const char *)sqlite3_value_text(argv[1]),
 573             (const char *)sqlite3_value_text(argv[0])));
 574 }
 575
 576 /*
 577  * Implement regular expression match
 578  * as an application-defined SQL function.
 579  */
 580 static void
 581 sql_regexp(sqlite3_context *context, int argc, sqlite3_value **argv)
 582 {
 583
 584         assert(2 == argc);
 585         sqlite3_result_int(context, !regexec(
 586             (regex_t *)sqlite3_value_blob(argv[0]),
 587             (const char *)sqlite3_value_text(argv[1]),
 588             0, NULL, 0));
 589 }
 590
 591 static void
 592 sql_append(char **sql, size_t *sz, const char *newstr, int count)
 593 {
 594         size_t           newsz;
 595
 596         newsz = 1 < count ? (size_t)count : strlen(newstr);
 597         *sql = mandoc_realloc(*sql, *sz + newsz + 1);
 598         if (1 < count)
 599                 memset(*sql + *sz, *newstr, (size_t)count);
 600         else
 601                 memcpy(*sql + *sz, newstr, newsz);
 602         *sz += newsz;
 603         (*sql)[*sz] = '\0';
 604 }
 605
 606 /*
 607  * Prepare the search SQL statement.
 608  */
 609 static char *
 610 sql_statement(const struct expr *e)
 611 {
 612         char            *sql;
 613         size_t           sz;
 614         int              needop;
 615
 616         sql = mandoc_strdup(e->equal ?
 617             "SELECT desc, form, pageid, bits "
 618                 "FROM mpages NATURAL JOIN names WHERE " :
 619             "SELECT desc, form, pageid, 0 FROM mpages WHERE ");
 620         sz = strlen(sql);
 621
 622         for (needop = 0; NULL != e; e = e->next) {
 623                 if (e->and)
 624                         sql_append(&sql, &sz, " AND ", 1);
 625                 else if (needop)
 626                         sql_append(&sql, &sz, " OR ", 1);
 627                 if (e->open)
 628                         sql_append(&sql, &sz, "(", e->open);
 629                 sql_append(&sql, &sz,
 630                     TYPE_Nd & e->bits
 631                     ? (NULL == e->substr
 632                         ? "desc REGEXP ?"
 633                         : "desc MATCH ?")
 634                     : TYPE_Nm == e->bits
 635                     ? (NULL == e->substr
 636                         ? "pageid IN (SELECT pageid FROM names "
 637                           "WHERE name REGEXP ?)"
 638                         : e->equal
 639                         ? "name = ? "
 640                         : "pageid IN (SELECT pageid FROM names "
 641                           "WHERE name MATCH ?)")
 642                     : (NULL == e->substr
 643                         ? "pageid IN (SELECT pageid FROM keys "
 644                           "WHERE key REGEXP ? AND bits & ?)"
 645                         : "pageid IN (SELECT pageid FROM keys "
 646                           "WHERE key MATCH ? AND bits & ?)"), 1);
 647                 if (e->close)
 648                         sql_append(&sql, &sz, ")", e->close);
 649                 needop = 1;
 650         }
 651
 652         return sql;
 653 }
 654
 655 /*
 656  * Compile a set of string tokens into an expression.
 657  * Tokens in "argv" are assumed to be individual expression atoms (e.g.,
 658  * "(", "foo=bar", etc.).
 659  */
 660 static struct expr *
 661 exprcomp(const struct mansearch *search, int argc, char *argv[])
 662 {
 663         uint64_t         mask;
 664         int              i, toopen, logic, igncase, toclose;
 665         struct expr     *first, *prev, *cur, *next;
 666
 667         first = cur = NULL;
 668         logic = igncase = toopen = toclose = 0;
 669
 670         for (i = 0; i < argc; i++) {
 671                 if (0 == strcmp("(", argv[i])) {
 672                         if (igncase)
 673                                 goto fail;
 674                         toopen++;
 675                         toclose++;
 676                         continue;
 677                 } else if (0 == strcmp(")", argv[i])) {
 678                         if (toopen || logic || igncase || NULL == cur)
 679                                 goto fail;
 680                         cur->close++;
 681                         if (0 > --toclose)
 682                                 goto fail;
 683                         continue;
 684                 } else if (0 == strcmp("-a", argv[i])) {
 685                         if (toopen || logic || igncase || NULL == cur)
 686                                 goto fail;
 687                         logic = 1;
 688                         continue;
 689                 } else if (0 == strcmp("-o", argv[i])) {
 690                         if (toopen || logic || igncase || NULL == cur)
 691                                 goto fail;
 692                         logic = 2;
 693                         continue;
 694                 } else if (0 == strcmp("-i", argv[i])) {
 695                         if (igncase)
 696                                 goto fail;
 697                         igncase = 1;
 698                         continue;
 699                 }
 700                 next = exprterm(search, argv[i], !igncase);
 701                 if (NULL == next)
 702                         goto fail;
 703                 if (NULL == first)
 704                         first = next;
 705                 else
 706                         cur->next = next;
 707                 prev = cur = next;
 708
 709                 /*
 710                  * Searching for descriptions must be split out
 711                  * because they are stored in the mpages table,
 712                  * not in the keys table.
 713                  */
 714
 715                 for (mask = TYPE_Nm; mask <= TYPE_Nd; mask <<= 1) {
 716                         if (mask & cur->bits && ~mask & cur->bits) {
 717                                 next = mandoc_calloc(1,
 718                                     sizeof(struct expr));
 719                                 memcpy(next, cur, sizeof(struct expr));
 720                                 prev->open = 1;
 721                                 cur->bits = mask;
 722                                 cur->next = next;
 723                                 cur = next;
 724                                 cur->bits &= ~mask;
 725                         }
 726                 }
 727                 prev->and = (1 == logic);
 728                 prev->open += toopen;
 729                 if (cur != prev)
 730                         cur->close = 1;
 731
 732                 toopen = logic = igncase = 0;
 733         }
 734         if ( ! (toopen || logic || igncase || toclose))
 735                 return first;
 736
 737 fail:
 738         if (NULL != first)
 739                 exprfree(first);
 740         return NULL;
 741 }
 742
 743 static struct expr *
 744 exprterm(const struct mansearch *search, char *buf, int cs)
 745 {
 746         char             errbuf[BUFSIZ];
 747         struct expr     *e;
 748         char            *key, *val;
 749         uint64_t         iterbit;
 750         int              i, irc;
 751
 752         if ('\0' == *buf)
 753                 return NULL;
 754
 755         e = mandoc_calloc(1, sizeof(struct expr));
 756
 757         if (search->argmode == ARG_NAME) {
 758                 e->bits = TYPE_Nm;
 759                 e->substr = buf;
 760                 e->equal = 1;
 761                 return e;
 762         }
 763
 764         /*
 765          * Separate macro keys from search string.
 766          * If needed, request regular expression handling
 767          * by setting e->substr to NULL.
 768          */
 769
 770         if (search->argmode == ARG_WORD) {
 771                 e->bits = TYPE_Nm;
 772                 e->substr = NULL;
 773 #if HAVE_REWB_BSD
 774                 mandoc_asprintf(&val, "[[:<:]]%s[[:>:]]", buf);
 775 #elif HAVE_REWB_SYSV
 776                 mandoc_asprintf(&val, "\\<%s\\>", buf);
 777 #else
 778                 mandoc_asprintf(&val,
 779                     "(^|[^a-zA-Z01-9_])%s([^a-zA-Z01-9_]|$)", buf);
 780 #endif
 781                 cs = 0;
 782         } else if ((val = strpbrk(buf, "=~")) == NULL) {
 783                 e->bits = TYPE_Nm | TYPE_Nd;
 784                 e->substr = buf;
 785         } else {
 786                 if (val == buf)
 787                         e->bits = TYPE_Nm | TYPE_Nd;
 788                 if ('=' == *val)
 789                         e->substr = val + 1;
 790                 *val++ = '\0';
 791                 if (NULL != strstr(buf, "arch"))
 792                         cs = 0;
 793         }
 794
 795         /* Compile regular expressions. */
 796
 797         if (NULL == e->substr) {
 798                 irc = regcomp(&e->regexp, val,
 799                     REG_EXTENDED | REG_NOSUB | (cs ? 0 : REG_ICASE));
 800                 if (search->argmode == ARG_WORD)
 801                         free(val);
 802                 if (irc) {
 803                         regerror(irc, &e->regexp, errbuf, sizeof(errbuf));
 804                         warnx("regcomp: %s", errbuf);
 805                         free(e);
 806                         return NULL;
 807                 }
 808         }
 809
 810         if (e->bits)
 811                 return e;
 812
 813         /*
 814          * Parse out all possible fields.
 815          * If the field doesn't resolve, bail.
 816          */
 817
 818         while (NULL != (key = strsep(&buf, ","))) {
 819                 if ('\0' == *key)
 820                         continue;
 821                 for (i = 0, iterbit = 1;
 822                      i < mansearch_keymax;
 823                      i++, iterbit <<= 1) {
 824                         if (0 == strcasecmp(key,
 825                             mansearch_keynames[i])) {
 826                                 e->bits |= iterbit;
 827                                 break;
 828                         }
 829                 }
 830                 if (i == mansearch_keymax) {
 831                         if (strcasecmp(key, "any")) {
 832                                 free(e);
 833                                 return NULL;
 834                         }
 835                         e->bits |= ~0ULL;
 836                 }
 837         }
 838
 839         return e;
 840 }
 841
 842 static void
 843 exprfree(struct expr *p)
 844 {
 845         struct expr     *pp;
 846
 847         while (NULL != p) {
 848                 pp = p->next;
 849                 free(p);
 850                 p = pp;
 851         }
 852 }