makemandb.c revision 1.32
1 /* $NetBSD: makemandb.c,v 1.32 2016/03/24 17:28:03 christos Exp $ */ 2 /* 3 * Copyright (c) 2011 Abhinav Upadhyay <er.abhinav.upadhyay (at) gmail.com> 4 * Copyright (c) 2011 Kristaps Dzonsons <kristaps (at) bsd.lv> 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 19 #include <sys/cdefs.h> 20 __RCSID("$NetBSD: makemandb.c,v 1.32 2016/03/24 17:28:03 christos Exp $"); 21 22 #include <sys/stat.h> 23 #include <sys/types.h> 24 25 #include <assert.h> 26 #include <dirent.h> 27 #include <err.h> 28 #include <archive.h> 29 #include <libgen.h> 30 #include <md5.h> 31 #include <stdio.h> 32 #include <stdlib.h> 33 #include <string.h> 34 #include <unistd.h> 35 #include <util.h> 36 37 #include "apropos-utils.h" 38 #include "dist/man.h" 39 #include "dist/mandoc.h" 40 #include "dist/mdoc.h" 41 42 #define BUFLEN 1024 43 #define MDOC 0 //If the page is of mdoc(7) type 44 #define MAN 1 //If the page is of man(7) type 45 46 /* 47 * A data structure for holding section specific data. 48 */ 49 typedef struct secbuff { 50 char *data; 51 size_t buflen; //Total length of buffer allocated initially 52 size_t offset; // Current offset in the buffer. 53 } secbuff; 54 55 typedef struct makemandb_flags { 56 int optimize; 57 int limit; // limit the indexing to only NAME section 58 int recreate; // Database was created from scratch 59 int verbosity; // 0: quiet, 1: default, 2: verbose 60 } makemandb_flags; 61 62 typedef struct mandb_rec { 63 /* Fields for mandb table */ 64 char *name; // for storing the name of the man page 65 char *name_desc; // for storing the one line description (.Nd) 66 secbuff desc; // for storing the DESCRIPTION section 67 secbuff lib; // for the LIBRARY section 68 secbuff return_vals; // RETURN VALUES 69 secbuff env; // ENVIRONMENT 70 secbuff files; // FILES 71 secbuff exit_status; // EXIT STATUS 72 secbuff diagnostics; // DIAGNOSTICS 73 secbuff errors; // ERRORS 74 char section[2]; 75 76 int xr_found; // To track whether a .Xr was seen when parsing a section 77 78 /* Fields for mandb_meta table */ 79 char *md5_hash; 80 dev_t device; 81 ino_t inode; 82 time_t mtime; 83 84 /* Fields for mandb_links table */ 85 char *machine; 86 char *links; //all the links to a page in a space separated form 87 char *file_path; 88 89 /* Non-db fields */ 90 int page_type; //Indicates the type of page: mdoc or man 91 } mandb_rec; 92 93 static void append(secbuff *sbuff, const char *src); 94 static void init_secbuffs(mandb_rec *); 95 static void free_secbuffs(mandb_rec *); 96 static int check_md5(const char *, sqlite3 *, const char *, char **, void *, size_t); 97 static void cleanup(mandb_rec *); 98 static void set_section(const struct mdoc *, const struct man *, mandb_rec *); 99 static void set_machine(const struct mdoc *, mandb_rec *); 100 static int insert_into_db(sqlite3 *, mandb_rec *); 101 static void begin_parse(const char *, struct mparse *, mandb_rec *, 102 const void *, size_t len); 103 static void pmdoc_node(const struct mdoc_node *, mandb_rec *); 104 static void pmdoc_Nm(const struct mdoc_node *, mandb_rec *); 105 static void pmdoc_Nd(const struct mdoc_node *, mandb_rec *); 106 static void pmdoc_Sh(const struct mdoc_node *, mandb_rec *); 107 static void pmdoc_Xr(const struct mdoc_node *, mandb_rec *); 108 static void pmdoc_Pp(const struct mdoc_node *, mandb_rec *); 109 static void pmdoc_macro_handler(const struct mdoc_node *, mandb_rec *, 110 enum mdoct); 111 static void pman_node(const struct man_node *n, mandb_rec *); 112 static void pman_parse_node(const struct man_node *, secbuff *); 113 static void pman_parse_name(const struct man_node *, mandb_rec *); 114 static void pman_sh(const struct man_node *, mandb_rec *); 115 static void pman_block(const struct man_node *, mandb_rec *); 116 static void traversedir(const char *, const char *, sqlite3 *, struct mparse *); 117 static void mdoc_parse_section(enum mdoc_sec, const char *, mandb_rec *); 118 static void man_parse_section(enum man_sec, const struct man_node *, mandb_rec *); 119 static void build_file_cache(sqlite3 *, const char *, const char *, 120 struct stat *); 121 static void update_db(sqlite3 *, struct mparse *, mandb_rec *); 122 __dead static void usage(void); 123 static void optimize(sqlite3 *); 124 static char *parse_escape(const char *); 125 static void replace_hyph(char *); 126 static makemandb_flags mflags = { .verbosity = 1 }; 127 128 typedef void (*pman_nf)(const struct man_node *n, mandb_rec *); 129 typedef void (*pmdoc_nf)(const struct mdoc_node *n, mandb_rec *); 130 static const pmdoc_nf mdocs[MDOC_MAX + 1] = { 131 NULL, /* Ap */ 132 NULL, /* Dd */ 133 NULL, /* Dt */ 134 NULL, /* Os */ 135 136 pmdoc_Sh, /* Sh */ 137 NULL, /* Ss */ 138 pmdoc_Pp, /* Pp */ 139 NULL, /* D1 */ 140 141 NULL, /* Dl */ 142 NULL, /* Bd */ 143 NULL, /* Ed */ 144 NULL, /* Bl */ 145 146 NULL, /* El */ 147 NULL, /* It */ 148 NULL, /* Ad */ 149 NULL, /* An */ 150 151 NULL, /* Ar */ 152 NULL, /* Cd */ 153 NULL, /* Cm */ 154 NULL, /* Dv */ 155 156 NULL, /* Er */ 157 NULL, /* Ev */ 158 NULL, /* Ex */ 159 NULL, /* Fa */ 160 161 NULL, /* Fd */ 162 NULL, /* Fl */ 163 NULL, /* Fn */ 164 NULL, /* Ft */ 165 166 NULL, /* Ic */ 167 NULL, /* In */ 168 NULL, /* Li */ 169 pmdoc_Nd, /* Nd */ 170 171 pmdoc_Nm, /* Nm */ 172 NULL, /* Op */ 173 NULL, /* Ot */ 174 NULL, /* Pa */ 175 176 NULL, /* Rv */ 177 NULL, /* St */ 178 NULL, /* Va */ 179 NULL, /* Vt */ 180 181 pmdoc_Xr, /* Xr */ 182 NULL, /* %A */ 183 NULL, /* %B */ 184 NULL, /* %D */ 185 186 NULL, /* %I */ 187 NULL, /* %J */ 188 NULL, /* %N */ 189 NULL, /* %O */ 190 191 NULL, /* %P */ 192 NULL, /* %R */ 193 NULL, /* %T */ 194 NULL, /* %V */ 195 196 NULL, /* Ac */ 197 NULL, /* Ao */ 198 NULL, /* Aq */ 199 NULL, /* At */ 200 201 NULL, /* Bc */ 202 NULL, /* Bf */ 203 NULL, /* Bo */ 204 NULL, /* Bq */ 205 206 NULL, /* Bsx */ 207 NULL, /* Bx */ 208 NULL, /* Db */ 209 NULL, /* Dc */ 210 211 NULL, /* Do */ 212 NULL, /* Dq */ 213 NULL, /* Ec */ 214 NULL, /* Ef */ 215 216 NULL, /* Em */ 217 NULL, /* Eo */ 218 NULL, /* Fx */ 219 NULL, /* Ms */ 220 221 NULL, /* No */ 222 NULL, /* Ns */ 223 NULL, /* Nx */ 224 NULL, /* Ox */ 225 226 NULL, /* Pc */ 227 NULL, /* Pf */ 228 NULL, /* Po */ 229 NULL, /* Pq */ 230 231 NULL, /* Qc */ 232 NULL, /* Ql */ 233 NULL, /* Qo */ 234 NULL, /* Qq */ 235 236 NULL, /* Re */ 237 NULL, /* Rs */ 238 NULL, /* Sc */ 239 NULL, /* So */ 240 241 NULL, /* Sq */ 242 NULL, /* Sm */ 243 NULL, /* Sx */ 244 NULL, /* Sy */ 245 246 NULL, /* Tn */ 247 NULL, /* Ux */ 248 NULL, /* Xc */ 249 NULL, /* Xo */ 250 251 NULL, /* Fo */ 252 NULL, /* Fc */ 253 NULL, /* Oo */ 254 NULL, /* Oc */ 255 256 NULL, /* Bk */ 257 NULL, /* Ek */ 258 NULL, /* Bt */ 259 NULL, /* Hf */ 260 261 NULL, /* Fr */ 262 NULL, /* Ud */ 263 NULL, /* Lb */ 264 NULL, /* Lp */ 265 266 NULL, /* Lk */ 267 NULL, /* Mt */ 268 NULL, /* Brq */ 269 NULL, /* Bro */ 270 271 NULL, /* Brc */ 272 NULL, /* %C */ 273 NULL, /* Es */ 274 NULL, /* En */ 275 276 NULL, /* Dx */ 277 NULL, /* %Q */ 278 NULL, /* br */ 279 NULL, /* sp */ 280 281 NULL, /* %U */ 282 NULL, /* Ta */ 283 NULL, /* ll */ 284 NULL, /* text */ 285 }; 286 287 static const pman_nf mans[MAN_MAX] = { 288 NULL, //br 289 NULL, //TH 290 pman_sh, //SH 291 NULL, //SS 292 NULL, //TP 293 NULL, //LP 294 NULL, //PP 295 NULL, //P 296 NULL, //IP 297 NULL, //HP 298 NULL, //SM 299 NULL, //SB 300 NULL, //BI 301 NULL, //IB 302 NULL, //BR 303 NULL, //RB 304 NULL, //R 305 pman_block, //B 306 NULL, //I 307 NULL, //IR 308 NULL, //RI 309 NULL, //sp 310 NULL, //nf 311 NULL, //fi 312 NULL, //RE 313 NULL, //RS 314 NULL, //DT 315 NULL, //UC 316 NULL, //PD 317 NULL, //AT 318 NULL, //in 319 NULL, //ft 320 NULL, //OP 321 NULL, //EX 322 NULL, //EE 323 NULL, //UR 324 NULL, //UE 325 NULL, //ll 326 }; 327 328 329 int 330 main(int argc, char *argv[]) 331 { 332 FILE *file; 333 struct mchars *mchars; 334 const char *sqlstr, *manconf = NULL; 335 char *line, *command, *parent; 336 char *errmsg; 337 int ch; 338 struct mparse *mp; 339 sqlite3 *db; 340 ssize_t len; 341 size_t linesize; 342 struct mandb_rec rec; 343 344 while ((ch = getopt(argc, argv, "C:floQqv")) != -1) { 345 switch (ch) { 346 case 'C': 347 manconf = optarg; 348 break; 349 case 'f': 350 mflags.recreate = 1; 351 break; 352 case 'l': 353 mflags.limit = 1; 354 break; 355 case 'o': 356 mflags.optimize = 1; 357 break; 358 case 'Q': 359 mflags.verbosity = 0; 360 break; 361 case 'q': 362 mflags.verbosity = 1; 363 break; 364 case 'v': 365 mflags.verbosity = 2; 366 break; 367 default: 368 usage(); 369 } 370 } 371 372 memset(&rec, 0, sizeof(rec)); 373 374 init_secbuffs(&rec); 375 mchars = mchars_alloc(); 376 if (mchars == NULL) 377 errx(EXIT_FAILURE, "Can't allocate mchars"); 378 mp = mparse_alloc(0, MANDOCLEVEL_BADARG, NULL, mchars, NULL); 379 380 if (manconf) { 381 char *arg; 382 size_t command_len = shquote(manconf, NULL, 0) + 1; 383 arg = emalloc(command_len); 384 shquote(manconf, arg, command_len); 385 easprintf(&command, "man -p -C %s", arg); 386 free(arg); 387 } else { 388 command = estrdup("man -p"); 389 manconf = MANCONF; 390 } 391 392 if (mflags.recreate) { 393 char *dbp = get_dbpath(manconf); 394 /* No error here, it will fail in init_db in the same call */ 395 if (dbp != NULL) 396 remove(dbp); 397 } 398 399 if ((db = init_db(MANDB_CREATE, manconf)) == NULL) 400 exit(EXIT_FAILURE); 401 402 sqlite3_exec(db, "PRAGMA synchronous = 0", NULL, NULL, &errmsg); 403 if (errmsg != NULL) { 404 warnx("%s", errmsg); 405 free(errmsg); 406 close_db(db); 407 exit(EXIT_FAILURE); 408 } 409 410 sqlite3_exec(db, "ATTACH DATABASE \':memory:\' AS metadb", NULL, NULL, 411 &errmsg); 412 if (errmsg != NULL) { 413 warnx("%s", errmsg); 414 free(errmsg); 415 close_db(db); 416 exit(EXIT_FAILURE); 417 } 418 419 420 /* Call man -p to get the list of man page dirs */ 421 if ((file = popen(command, "r")) == NULL) { 422 close_db(db); 423 err(EXIT_FAILURE, "fopen failed"); 424 } 425 free(command); 426 427 /* Begin the transaction for indexing the pages */ 428 sqlite3_exec(db, "BEGIN", NULL, NULL, &errmsg); 429 if (errmsg != NULL) { 430 warnx("%s", errmsg); 431 free(errmsg); 432 exit(EXIT_FAILURE); 433 } 434 435 sqlstr = "CREATE TABLE metadb.file_cache(device, inode, mtime, parent," 436 " file PRIMARY KEY);" 437 "CREATE UNIQUE INDEX metadb.index_file_cache_dev" 438 " ON file_cache (device, inode)"; 439 440 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 441 if (errmsg != NULL) { 442 warnx("%s", errmsg); 443 free(errmsg); 444 close_db(db); 445 exit(EXIT_FAILURE); 446 } 447 448 if (mflags.verbosity) 449 printf("Building temporary file cache\n"); 450 line = NULL; 451 linesize = 0; 452 while ((len = getline(&line, &linesize, file)) != -1) { 453 /* Replace the new line character at the end of string with '\0' */ 454 line[len - 1] = '\0'; 455 parent = estrdup(line); 456 char *pdir = estrdup(dirname(parent)); 457 free(parent); 458 /* Traverse the man page directories and parse the pages */ 459 traversedir(pdir, line, db, mp); 460 free(pdir); 461 } 462 free(line); 463 464 if (pclose(file) == -1) { 465 close_db(db); 466 cleanup(&rec); 467 free_secbuffs(&rec); 468 err(EXIT_FAILURE, "pclose error"); 469 } 470 471 if (mflags.verbosity) 472 printf("Performing index update\n"); 473 update_db(db, mp, &rec); 474 mparse_free(mp); 475 mchars_free(mchars); 476 free_secbuffs(&rec); 477 478 /* Commit the transaction */ 479 sqlite3_exec(db, "COMMIT", NULL, NULL, &errmsg); 480 if (errmsg != NULL) { 481 warnx("%s", errmsg); 482 free(errmsg); 483 exit(EXIT_FAILURE); 484 } 485 486 if (mflags.optimize) 487 optimize(db); 488 489 close_db(db); 490 return 0; 491 } 492 493 /* 494 * traversedir -- 495 * Traverses the given directory recursively and passes all the man page files 496 * in the way to build_file_cache() 497 */ 498 static void 499 traversedir(const char *parent, const char *file, sqlite3 *db, 500 struct mparse *mp) 501 { 502 struct stat sb; 503 struct dirent *dirp; 504 DIR *dp; 505 char *buf; 506 507 if (stat(file, &sb) < 0) { 508 if (mflags.verbosity) 509 warn("stat failed: %s", file); 510 return; 511 } 512 513 /* If it is a directory, traverse it recursively */ 514 if (S_ISDIR(sb.st_mode)) { 515 if ((dp = opendir(file)) == NULL) { 516 if (mflags.verbosity) 517 warn("opendir error: %s", file); 518 return; 519 } 520 521 while ((dirp = readdir(dp)) != NULL) { 522 /* Avoid . and .. entries in a directory */ 523 if (strncmp(dirp->d_name, ".", 1)) { 524 easprintf(&buf, "%s/%s", file, dirp->d_name); 525 traversedir(parent, buf, db, mp); 526 free(buf); 527 } 528 } 529 closedir(dp); 530 } 531 532 if (!S_ISREG(sb.st_mode) && !S_ISLNK(sb.st_mode)) 533 return; 534 535 if (sb.st_size == 0) { 536 if (mflags.verbosity) 537 warnx("Empty file: %s", file); 538 return; 539 } 540 build_file_cache(db, parent, file, &sb); 541 } 542 543 /* build_file_cache -- 544 * This function generates an md5 hash of the file passed as its 2nd parameter 545 * and stores it in a temporary table file_cache along with the full file path. 546 * This is done to support incremental updation of the database. 547 * The temporary table file_cache is dropped thereafter in the function 548 * update_db(), once the database has been updated. 549 */ 550 static void 551 build_file_cache(sqlite3 *db, const char *parent, const char *file, 552 struct stat *sb) 553 { 554 const char *sqlstr; 555 sqlite3_stmt *stmt = NULL; 556 int rc, idx; 557 assert(file != NULL); 558 dev_t device_cache = sb->st_dev; 559 ino_t inode_cache = sb->st_ino; 560 time_t mtime_cache = sb->st_mtime; 561 562 sqlstr = "INSERT INTO metadb.file_cache VALUES (:device, :inode," 563 " :mtime, :parent, :file)"; 564 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 565 if (rc != SQLITE_OK) { 566 if (mflags.verbosity) 567 warnx("%s", sqlite3_errmsg(db)); 568 return; 569 } 570 571 idx = sqlite3_bind_parameter_index(stmt, ":device"); 572 rc = sqlite3_bind_int64(stmt, idx, device_cache); 573 if (rc != SQLITE_OK) { 574 if (mflags.verbosity) 575 warnx("%s", sqlite3_errmsg(db)); 576 sqlite3_finalize(stmt); 577 return; 578 } 579 580 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 581 rc = sqlite3_bind_int64(stmt, idx, inode_cache); 582 if (rc != SQLITE_OK) { 583 if (mflags.verbosity) 584 warnx("%s", sqlite3_errmsg(db)); 585 sqlite3_finalize(stmt); 586 return; 587 } 588 589 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 590 rc = sqlite3_bind_int64(stmt, idx, mtime_cache); 591 if (rc != SQLITE_OK) { 592 if (mflags.verbosity) 593 warnx("%s", sqlite3_errmsg(db)); 594 sqlite3_finalize(stmt); 595 return; 596 } 597 598 idx = sqlite3_bind_parameter_index(stmt, ":parent"); 599 rc = sqlite3_bind_text(stmt, idx, parent, -1, NULL); 600 if (rc != SQLITE_OK) { 601 if (mflags.verbosity) 602 warnx("%s", sqlite3_errmsg(db)); 603 sqlite3_finalize(stmt); 604 return; 605 } 606 607 idx = sqlite3_bind_parameter_index(stmt, ":file"); 608 rc = sqlite3_bind_text(stmt, idx, file, -1, NULL); 609 if (rc != SQLITE_OK) { 610 if (mflags.verbosity) 611 warnx("%s", sqlite3_errmsg(db)); 612 sqlite3_finalize(stmt); 613 return; 614 } 615 616 sqlite3_step(stmt); 617 sqlite3_finalize(stmt); 618 } 619 620 static void 621 update_existing_entry(sqlite3 *db, const char *file, const char *hash, 622 mandb_rec *rec, int *new_count, int *link_count, int *err_count) 623 { 624 int update_count, rc, idx; 625 const char *inner_sqlstr; 626 sqlite3_stmt *inner_stmt; 627 628 update_count = sqlite3_total_changes(db); 629 inner_sqlstr = "UPDATE mandb_meta SET device = :device," 630 " inode = :inode, mtime = :mtime WHERE" 631 " md5_hash = :md5 AND file = :file AND" 632 " (device <> :device2 OR inode <> " 633 " :inode2 OR mtime <> :mtime2)"; 634 rc = sqlite3_prepare_v2(db, inner_sqlstr, -1, &inner_stmt, NULL); 635 if (rc != SQLITE_OK) { 636 if (mflags.verbosity) 637 warnx("%s", sqlite3_errmsg(db)); 638 return; 639 } 640 idx = sqlite3_bind_parameter_index(inner_stmt, ":device"); 641 sqlite3_bind_int64(inner_stmt, idx, rec->device); 642 idx = sqlite3_bind_parameter_index(inner_stmt, ":inode"); 643 sqlite3_bind_int64(inner_stmt, idx, rec->inode); 644 idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime"); 645 sqlite3_bind_int64(inner_stmt, idx, rec->mtime); 646 idx = sqlite3_bind_parameter_index(inner_stmt, ":md5"); 647 sqlite3_bind_text(inner_stmt, idx, hash, -1, NULL); 648 idx = sqlite3_bind_parameter_index(inner_stmt, ":file"); 649 sqlite3_bind_text(inner_stmt, idx, file, -1, NULL); 650 idx = sqlite3_bind_parameter_index(inner_stmt, ":device2"); 651 sqlite3_bind_int64(inner_stmt, idx, rec->device); 652 idx = sqlite3_bind_parameter_index(inner_stmt, ":inode2"); 653 sqlite3_bind_int64(inner_stmt, idx, rec->inode); 654 idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime2"); 655 sqlite3_bind_int64(inner_stmt, idx, rec->mtime); 656 657 rc = sqlite3_step(inner_stmt); 658 if (rc == SQLITE_DONE) { 659 /* Check if an update has been performed. */ 660 if (update_count != sqlite3_total_changes(db)) { 661 if (mflags.verbosity == 2) 662 printf("Updated %s\n", file); 663 (*new_count)++; 664 } else { 665 /* Otherwise it was a hardlink. */ 666 (*link_count)++; 667 } 668 } else { 669 if (mflags.verbosity == 2) 670 warnx("Could not update the meta data for %s", file); 671 (*err_count)++; 672 } 673 sqlite3_finalize(inner_stmt); 674 } 675 676 /* read_and_decompress -- 677 * Reads the given file into memory. If it is compressed, decompress 678 * it before returning to the caller. 679 */ 680 static int 681 read_and_decompress(const char *file, void **bufp, size_t *len) 682 { 683 size_t off; 684 ssize_t r; 685 struct archive *a; 686 struct archive_entry *ae; 687 char *buf; 688 689 if ((a = archive_read_new()) == NULL) 690 errx(EXIT_FAILURE, "memory allocation failed"); 691 692 *bufp = NULL; 693 if (archive_read_support_compression_all(a) != ARCHIVE_OK || 694 archive_read_support_format_raw(a) != ARCHIVE_OK || 695 archive_read_open_filename(a, file, 65536) != ARCHIVE_OK || 696 archive_read_next_header(a, &ae) != ARCHIVE_OK) 697 goto archive_error; 698 *len = 65536; 699 buf = emalloc(*len); 700 off = 0; 701 for (;;) { 702 r = archive_read_data(a, buf + off, *len - off); 703 if (r == ARCHIVE_OK) { 704 archive_read_close(a); 705 *bufp = buf; 706 *len = off; 707 return 0; 708 } 709 if (r <= 0) { 710 free(buf); 711 break; 712 } 713 off += r; 714 if (off == *len) { 715 *len *= 2; 716 if (*len < off) { 717 if (mflags.verbosity) 718 warnx("File too large: %s", file); 719 free(buf); 720 archive_read_close(a); 721 return -1; 722 } 723 buf = erealloc(buf, *len); 724 } 725 } 726 727 archive_error: 728 warnx("Error while reading `%s': %s", file, archive_error_string(a)); 729 archive_read_close(a); 730 return -1; 731 } 732 733 /* update_db -- 734 * Does an incremental updation of the database by checking the file_cache. 735 * It parses and adds the pages which are present in file_cache, 736 * but not in the database. 737 * It also removes the pages which are present in the databse, 738 * but not in the file_cache. 739 */ 740 static void 741 update_db(sqlite3 *db, struct mparse *mp, mandb_rec *rec) 742 { 743 const char *sqlstr; 744 sqlite3_stmt *stmt = NULL; 745 char *file; 746 char *parent; 747 char *errmsg = NULL; 748 char *md5sum; 749 void *buf; 750 size_t buflen; 751 struct sql_row { 752 struct sql_row *next; 753 dev_t device; 754 ino_t inode; 755 time_t mtime; 756 char *parent; 757 char *file; 758 } *rows, *row; 759 int new_count = 0; /* Counter for newly indexed/updated pages */ 760 int total_count = 0; /* Counter for total number of pages */ 761 int err_count = 0; /* Counter for number of failed pages */ 762 int link_count = 0; /* Counter for number of hard/sym links */ 763 int md5_status; 764 int rc; 765 766 sqlstr = "SELECT device, inode, mtime, parent, file" 767 " FROM metadb.file_cache fc" 768 " WHERE NOT EXISTS(SELECT 1 FROM mandb_meta WHERE" 769 " device = fc.device AND inode = fc.inode AND " 770 " mtime = fc.mtime AND file = fc.file)"; 771 772 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 773 if (rc != SQLITE_OK) { 774 if (mflags.verbosity) 775 warnx("%s", sqlite3_errmsg(db)); 776 close_db(db); 777 errx(EXIT_FAILURE, "Could not query file cache"); 778 } 779 780 buf = NULL; 781 rows = NULL; 782 while (sqlite3_step(stmt) == SQLITE_ROW) { 783 row = emalloc(sizeof(struct sql_row)); 784 row->device = sqlite3_column_int64(stmt, 0); 785 row->inode = sqlite3_column_int64(stmt, 1); 786 row->mtime = sqlite3_column_int64(stmt, 2); 787 row->parent = estrdup((const char *) sqlite3_column_text(stmt, 3)); 788 row->file = estrdup((const char *) sqlite3_column_text(stmt, 4)); 789 row->next = rows; 790 rows = row; 791 total_count++; 792 } 793 sqlite3_finalize(stmt); 794 795 for ( ; rows != NULL; free(parent), free(file), free(buf)) { 796 row = rows; 797 rows = rows->next; 798 799 rec->device = row->device; 800 rec->inode = row->inode; 801 rec->mtime = row->mtime; 802 parent = row->parent; 803 file = row->file; 804 free(row); 805 806 if (read_and_decompress(file, &buf, &buflen)) { 807 err_count++; 808 continue; 809 } 810 md5_status = check_md5(file, db, "mandb_meta", &md5sum, buf, buflen); 811 assert(md5sum != NULL); 812 if (md5_status == -1) { 813 if (mflags.verbosity) 814 warnx("An error occurred in checking md5 value" 815 " for file %s", file); 816 err_count++; 817 continue; 818 } 819 820 if (md5_status == 0) { 821 /* 822 * The MD5 hash is already present in the database, 823 * so simply update the metadata, ignoring symlinks. 824 */ 825 struct stat sb; 826 stat(file, &sb); 827 if (S_ISLNK(sb.st_mode)) { 828 free(md5sum); 829 link_count++; 830 continue; 831 } 832 update_existing_entry(db, file, md5sum, rec, 833 &new_count, &link_count, &err_count); 834 free(md5sum); 835 continue; 836 } 837 838 if (md5_status == 1) { 839 /* 840 * The MD5 hash was not present in the database. 841 * This means is either a new file or an updated file. 842 * We should go ahead with parsing. 843 */ 844 if (mflags.verbosity == 2) 845 printf("Parsing: %s\n", file); 846 rec->md5_hash = md5sum; 847 rec->file_path = estrdup(file); 848 // file_path is freed by insert_into_db itself. 849 chdir(parent); 850 begin_parse(file, mp, rec, buf, buflen); 851 if (insert_into_db(db, rec) < 0) { 852 if (mflags.verbosity) 853 warnx("Error in indexing %s", file); 854 err_count++; 855 } else { 856 new_count++; 857 } 858 } 859 } 860 861 if (mflags.verbosity == 2) { 862 printf("Total Number of new or updated pages encountered = %d\n" 863 "Total number of (hard or symbolic) links found = %d\n" 864 "Total number of pages that were successfully" 865 " indexed/updated = %d\n" 866 "Total number of pages that could not be indexed" 867 " due to errors = %d\n", 868 total_count - link_count, link_count, new_count, err_count); 869 } 870 871 if (mflags.recreate) 872 return; 873 874 if (mflags.verbosity == 2) 875 printf("Deleting stale index entries\n"); 876 877 sqlstr = "DELETE FROM mandb_meta WHERE file NOT IN" 878 " (SELECT file FROM metadb.file_cache);" 879 "DELETE FROM mandb_links WHERE md5_hash NOT IN" 880 " (SELECT md5_hash from mandb_meta);" 881 "DROP TABLE metadb.file_cache;" 882 "DELETE FROM mandb WHERE rowid NOT IN" 883 " (SELECT id FROM mandb_meta);"; 884 885 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 886 if (errmsg != NULL) { 887 warnx("Removing old entries failed: %s", errmsg); 888 warnx("Please rebuild database from scratch with -f."); 889 free(errmsg); 890 return; 891 } 892 } 893 894 /* 895 * begin_parse -- 896 * parses the man page using libmandoc 897 */ 898 static void 899 begin_parse(const char *file, struct mparse *mp, mandb_rec *rec, 900 const void *buf, size_t len) 901 { 902 struct mdoc *mdoc; 903 struct man *man; 904 mparse_reset(mp); 905 906 rec->xr_found = 0; 907 908 if (mparse_readmem(mp, buf, len, file) >= MANDOCLEVEL_BADARG) { 909 /* Printing this warning at verbosity level 2 910 * because some packages from pkgsrc might trigger several 911 * of such warnings. 912 */ 913 if (mflags.verbosity == 2) 914 warnx("%s: Parse failure", file); 915 return; 916 } 917 918 mparse_result(mp, &mdoc, &man, NULL); 919 if (mdoc == NULL && man == NULL) { 920 if (mflags.verbosity == 2) 921 warnx("Not a man(7) or mdoc(7) page"); 922 return; 923 } 924 925 set_machine(mdoc, rec); 926 set_section(mdoc, man, rec); 927 if (mdoc) { 928 rec->page_type = MDOC; 929 pmdoc_node(mdoc_node(mdoc), rec); 930 } else { 931 rec->page_type = MAN; 932 pman_node(man_node(man), rec); 933 } 934 } 935 936 /* 937 * set_section -- 938 * Extracts the section number and normalizes it to only the numeric part 939 * (Which should be the first character of the string). 940 */ 941 static void 942 set_section(const struct mdoc *md, const struct man *m, mandb_rec *rec) 943 { 944 if (md) { 945 const struct mdoc_meta *md_meta = mdoc_meta(md); 946 if (md_meta->msec == NULL) { 947 rec->section[0] = '?'; 948 } else 949 rec->section[0] = md_meta->msec[0]; 950 } else if (m) { 951 const struct man_meta *m_meta = man_meta(m); 952 if (m_meta->msec == NULL) 953 rec->section[0] = '?'; 954 else 955 rec->section[0] = m_meta->msec[0]; 956 } else 957 return; 958 959 if (rec->section[0] == '?' && mflags.verbosity == 2) 960 warnx("%s: Missing section number", rec->file_path); 961 } 962 963 /* 964 * get_machine -- 965 * Extracts the machine architecture information if available. 966 */ 967 static void 968 set_machine(const struct mdoc *md, mandb_rec *rec) 969 { 970 if (md == NULL) 971 return; 972 const struct mdoc_meta *md_meta = mdoc_meta(md); 973 if (md_meta->arch) 974 rec->machine = estrdup(md_meta->arch); 975 } 976 977 static void 978 pmdoc_node(const struct mdoc_node *n, mandb_rec *rec) 979 { 980 981 if (n == NULL) 982 return; 983 984 switch (n->type) { 985 case (MDOC_BODY): 986 /* FALLTHROUGH */ 987 case (MDOC_TAIL): 988 /* FALLTHROUGH */ 989 case (MDOC_ELEM): 990 if (mdocs[n->tok] == NULL) 991 break; 992 (*mdocs[n->tok])(n, rec); 993 break; 994 default: 995 break; 996 } 997 998 pmdoc_node(n->child, rec); 999 pmdoc_node(n->next, rec); 1000 } 1001 1002 /* 1003 * pmdoc_Nm -- 1004 * Extracts the Name of the manual page from the .Nm macro 1005 */ 1006 static void 1007 pmdoc_Nm(const struct mdoc_node *n, mandb_rec *rec) 1008 { 1009 if (n->sec != SEC_NAME) 1010 return; 1011 1012 for (n = n->child; n; n = n->next) { 1013 if (n->type == MDOC_TEXT) { 1014 char *escaped_name = parse_escape(n->string); 1015 concat(&rec->name, escaped_name); 1016 free(escaped_name); 1017 } 1018 } 1019 } 1020 1021 /* 1022 * pmdoc_Nd -- 1023 * Extracts the one line description of the man page from the .Nd macro 1024 */ 1025 static void 1026 pmdoc_Nd(const struct mdoc_node *n, mandb_rec *rec) 1027 { 1028 char *buf = NULL; 1029 char *name; 1030 char *nd_text; 1031 1032 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1033 return; 1034 1035 if (n->type == MDOC_TEXT) { 1036 if (rec->xr_found && n->next) { 1037 /* 1038 * An Xr macro was seen previously, so parse this 1039 * and the next node, as "Name(Section)". 1040 */ 1041 name = n->string; 1042 n = n->next; 1043 assert(n->type == MDOC_TEXT); 1044 easprintf(&buf, "%s(%s)", name, n->string); 1045 concat(&rec->name_desc, buf); 1046 free(buf); 1047 } else { 1048 nd_text = parse_escape(n->string); 1049 concat(&rec->name_desc, nd_text); 1050 free(nd_text); 1051 } 1052 rec->xr_found = 0; 1053 } else if (mdocs[n->tok] == pmdoc_Xr) { 1054 /* Remember that we have encountered an Xr macro */ 1055 rec->xr_found = 1; 1056 } 1057 1058 if (n->child) 1059 pmdoc_Nd(n->child, rec); 1060 1061 if(n->next) 1062 pmdoc_Nd(n->next, rec); 1063 } 1064 1065 /* 1066 * pmdoc_macro_handler-- 1067 * This function is a single point of handling all the special macros that we 1068 * want to handle especially. For example the .Xr macro for properly parsing 1069 * the referenced page name along with the section number, or the .Pp macro 1070 * for adding a new line whenever we encounter it. 1071 */ 1072 static void 1073 pmdoc_macro_handler(const struct mdoc_node *n, mandb_rec *rec, enum mdoct doct) 1074 { 1075 const struct mdoc_node *sn; 1076 assert(n); 1077 1078 switch (doct) { 1079 /* Parse the man page references. 1080 * Basically the .Xr macros are used like: 1081 * .Xr ls 1 1082 * and formatted like this: 1083 * ls(1) 1084 * Prepare a buffer to format the data like the above example and call 1085 * pmdoc_parse_section to append it. 1086 */ 1087 case MDOC_Xr: 1088 n = n->child; 1089 while (n->type != MDOC_TEXT && n->next) 1090 n = n->next; 1091 1092 if (n && n->type != MDOC_TEXT) 1093 return; 1094 sn = n; 1095 if (n->next) 1096 n = n->next; 1097 1098 while (n->type != MDOC_TEXT && n->next) 1099 n = n->next; 1100 1101 if (n && n->type == MDOC_TEXT) { 1102 char *buf; 1103 easprintf(&buf, "%s(%s)", sn->string, n->string); 1104 mdoc_parse_section(n->sec, buf, rec); 1105 free(buf); 1106 } 1107 1108 break; 1109 1110 /* Parse the .Pp macro to add a new line */ 1111 case MDOC_Pp: 1112 if (n->type == MDOC_TEXT) 1113 mdoc_parse_section(n->sec, "\n", rec); 1114 break; 1115 default: 1116 break; 1117 } 1118 1119 } 1120 1121 /* 1122 * pmdoc_Xr, pmdoc_Pp-- 1123 * Empty stubs. 1124 * The parser calls these functions each time it encounters 1125 * a .Xr or .Pp macro. We are parsing all the data from 1126 * the pmdoc_Sh function, so don't do anything here. 1127 * (See if else blocks in pmdoc_Sh.) 1128 */ 1129 static void 1130 pmdoc_Xr(const struct mdoc_node *n, mandb_rec *rec) 1131 { 1132 } 1133 1134 static void 1135 pmdoc_Pp(const struct mdoc_node *n, mandb_rec *rec) 1136 { 1137 } 1138 1139 /* 1140 * pmdoc_Sh -- 1141 * Called when a .Sh macro is encountered and loops through its body, calling 1142 * mdoc_parse_section to append the data to the section specific buffer. 1143 * Two special macros which may occur inside the body of Sh are .Nm and .Xr and 1144 * they need special handling, thus the separate if branches for them. 1145 */ 1146 static void 1147 pmdoc_Sh(const struct mdoc_node *n, mandb_rec *rec) 1148 { 1149 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1150 return; 1151 int xr_found = 0; 1152 1153 if (n->type == MDOC_TEXT) { 1154 mdoc_parse_section(n->sec, n->string, rec); 1155 } else if (mdocs[n->tok] == pmdoc_Nm && rec->name != NULL) { 1156 /* 1157 * When encountering a .Nm macro, substitute it 1158 * with its previously cached value of the argument. 1159 */ 1160 mdoc_parse_section(n->sec, rec->name, rec); 1161 } else if (mdocs[n->tok] == pmdoc_Xr) { 1162 /* 1163 * When encountering other inline macros, 1164 * call pmdoc_macro_handler. 1165 */ 1166 pmdoc_macro_handler(n, rec, MDOC_Xr); 1167 xr_found = 1; 1168 } else if (mdocs[n->tok] == pmdoc_Pp) { 1169 pmdoc_macro_handler(n, rec, MDOC_Pp); 1170 } 1171 1172 /* 1173 * If an Xr macro was encountered then the child node has 1174 * already been explored by pmdoc_macro_handler. 1175 */ 1176 if (xr_found == 0) 1177 pmdoc_Sh(n->child, rec); 1178 pmdoc_Sh(n->next, rec); 1179 } 1180 1181 /* 1182 * mdoc_parse_section-- 1183 * Utility function for parsing sections of the mdoc type pages. 1184 * Takes two params: 1185 * 1. sec is an enum which indicates the section in which we are present 1186 * 2. string is the string which we need to append to the secbuff for this 1187 * particular section. 1188 * The function appends string to the global section buffer and returns. 1189 */ 1190 static void 1191 mdoc_parse_section(enum mdoc_sec sec, const char *string, mandb_rec *rec) 1192 { 1193 /* 1194 * If the user specified the 'l' flag, then parse and store only the 1195 * NAME section. Ignore the rest. 1196 */ 1197 if (mflags.limit) 1198 return; 1199 1200 switch (sec) { 1201 case SEC_LIBRARY: 1202 append(&rec->lib, string); 1203 break; 1204 case SEC_RETURN_VALUES: 1205 append(&rec->return_vals, string); 1206 break; 1207 case SEC_ENVIRONMENT: 1208 append(&rec->env, string); 1209 break; 1210 case SEC_FILES: 1211 append(&rec->files, string); 1212 break; 1213 case SEC_EXIT_STATUS: 1214 append(&rec->exit_status, string); 1215 break; 1216 case SEC_DIAGNOSTICS: 1217 append(&rec->diagnostics, string); 1218 break; 1219 case SEC_ERRORS: 1220 append(&rec->errors, string); 1221 break; 1222 case SEC_NAME: 1223 case SEC_SYNOPSIS: 1224 case SEC_EXAMPLES: 1225 case SEC_STANDARDS: 1226 case SEC_HISTORY: 1227 case SEC_AUTHORS: 1228 case SEC_BUGS: 1229 break; 1230 default: 1231 append(&rec->desc, string); 1232 break; 1233 } 1234 } 1235 1236 static void 1237 pman_node(const struct man_node *n, mandb_rec *rec) 1238 { 1239 if (n == NULL) 1240 return; 1241 1242 switch (n->type) { 1243 case (MAN_BODY): 1244 /* FALLTHROUGH */ 1245 case (MAN_BLOCK): 1246 /* FALLTHROUGH */ 1247 case (MAN_ELEM): 1248 if (mans[n->tok] != NULL) 1249 (*mans[n->tok])(n, rec); 1250 break; 1251 default: 1252 break; 1253 } 1254 1255 pman_node(n->child, rec); 1256 pman_node(n->next, rec); 1257 } 1258 1259 /* 1260 * pman_parse_name -- 1261 * Parses the NAME section and puts the complete content in the name_desc 1262 * variable. 1263 */ 1264 static void 1265 pman_parse_name(const struct man_node *n, mandb_rec *rec) 1266 { 1267 if (n == NULL) 1268 return; 1269 1270 if (n->type == MAN_TEXT) { 1271 char *tmp = parse_escape(n->string); 1272 concat(&rec->name_desc, tmp); 1273 free(tmp); 1274 } 1275 1276 if (n->child) 1277 pman_parse_name(n->child, rec); 1278 1279 if(n->next) 1280 pman_parse_name(n->next, rec); 1281 } 1282 1283 /* 1284 * A stub function to be able to parse the macros like .B embedded inside 1285 * a section. 1286 */ 1287 static void 1288 pman_block(const struct man_node *n, mandb_rec *rec) 1289 { 1290 } 1291 1292 /* 1293 * pman_sh -- 1294 * This function does one of the two things: 1295 * 1. If the present section is NAME, then it will: 1296 * (a) Extract the name of the page (in case of multiple comma separated 1297 * names, it will pick up the first one). 1298 * (b) Build a space spearated list of all the symlinks/hardlinks to 1299 * this page and store in the buffer 'links'. These are extracted from 1300 * the comma separated list of names in the NAME section as well. 1301 * (c) Move on to the one line description section, which is after the list 1302 * of names in the NAME section. 1303 * 2. Otherwise, it will check the section name and call the man_parse_section 1304 * function, passing the enum corresponding that section. 1305 */ 1306 static void 1307 pman_sh(const struct man_node *n, mandb_rec *rec) 1308 { 1309 static const struct { 1310 enum man_sec section; 1311 const char *header; 1312 } mapping[] = { 1313 { MANSEC_DESCRIPTION, "DESCRIPTION" }, 1314 { MANSEC_SYNOPSIS, "SYNOPSIS" }, 1315 { MANSEC_LIBRARY, "LIBRARY" }, 1316 { MANSEC_ERRORS, "ERRORS" }, 1317 { MANSEC_FILES, "FILES" }, 1318 { MANSEC_RETURN_VALUES, "RETURN VALUE" }, 1319 { MANSEC_RETURN_VALUES, "RETURN VALUES" }, 1320 { MANSEC_EXIT_STATUS, "EXIT STATUS" }, 1321 { MANSEC_EXAMPLES, "EXAMPLES" }, 1322 { MANSEC_EXAMPLES, "EXAMPLE" }, 1323 { MANSEC_STANDARDS, "STANDARDS" }, 1324 { MANSEC_HISTORY, "HISTORY" }, 1325 { MANSEC_BUGS, "BUGS" }, 1326 { MANSEC_AUTHORS, "AUTHORS" }, 1327 { MANSEC_COPYRIGHT, "COPYRIGHT" }, 1328 }; 1329 const struct man_node *head; 1330 char *name_desc; 1331 size_t sz; 1332 size_t i; 1333 1334 if ((head = n->parent->head) == NULL || (head = head->child) == NULL || 1335 head->type != MAN_TEXT) 1336 return; 1337 1338 /* 1339 * Check if this section should be extracted and 1340 * where it should be stored. Handled the trival cases first. 1341 */ 1342 for (i = 0; i < sizeof(mapping) / sizeof(mapping[0]); ++i) { 1343 if (strcmp(head->string, mapping[i].header) == 0) { 1344 man_parse_section(mapping[i].section, n, rec); 1345 return; 1346 } 1347 } 1348 1349 if (strcmp(head->string, "NAME") == 0) { 1350 /* 1351 * We are in the NAME section. 1352 * pman_parse_name will put the complete content in name_desc. 1353 */ 1354 pman_parse_name(n, rec); 1355 1356 name_desc = rec->name_desc; 1357 if (name_desc == NULL) 1358 return; 1359 1360 /* Remove any leading spaces. */ 1361 while (name_desc[0] == ' ') 1362 name_desc++; 1363 1364 /* If the line begins with a "\&", avoid those */ 1365 if (name_desc[0] == '\\' && name_desc[1] == '&') 1366 name_desc += 2; 1367 1368 /* Now name_desc should be left with a comma-space 1369 * separated list of names and the one line description 1370 * of the page: 1371 * "a, b, c \- sample description" 1372 * Take out the first name, before the first comma 1373 * (or space) and store it in rec->name. 1374 * If the page has aliases then they should be 1375 * in the form of a comma separated list. 1376 * Keep looping while there is a comma in name_desc, 1377 * extract the alias name and store in rec->links. 1378 * When there are no more commas left, break out. 1379 */ 1380 int has_alias = 0; // Any more aliases left? 1381 while (*name_desc) { 1382 /* Remove any leading spaces or hyphens. */ 1383 if (name_desc[0] == ' ' || name_desc[0] =='-') { 1384 name_desc++; 1385 continue; 1386 } 1387 sz = strcspn(name_desc, ", "); 1388 1389 /* Extract the first term and store it in rec->name. */ 1390 if (rec->name == NULL) { 1391 if (name_desc[sz] == ',') 1392 has_alias = 1; 1393 name_desc[sz] = 0; 1394 rec->name = emalloc(sz + 1); 1395 memcpy(rec->name, name_desc, sz + 1); 1396 name_desc += sz + 1; 1397 continue; 1398 } 1399 1400 /* 1401 * Once rec->name is set, rest of the names 1402 * are to be treated as links or aliases. 1403 */ 1404 if (rec->name && has_alias) { 1405 if (name_desc[sz] != ',') { 1406 /* No more commas left --> 1407 * no more aliases to take out 1408 */ 1409 has_alias = 0; 1410 } 1411 name_desc[sz] = 0; 1412 concat2(&rec->links, name_desc, sz); 1413 name_desc += sz + 1; 1414 continue; 1415 } 1416 break; 1417 } 1418 1419 /* Parse any escape sequences that might be there */ 1420 char *temp = parse_escape(name_desc); 1421 free(rec->name_desc); 1422 rec->name_desc = temp; 1423 temp = parse_escape(rec->name); 1424 free(rec->name); 1425 rec->name = temp; 1426 return; 1427 } 1428 1429 /* The RETURN VALUE section might be specified in multiple ways */ 1430 if (strcmp(head->string, "RETURN") == 0 && 1431 head->next != NULL && head->next->type == MAN_TEXT && 1432 (strcmp(head->next->string, "VALUE") == 0 || 1433 strcmp(head->next->string, "VALUES") == 0)) { 1434 man_parse_section(MANSEC_RETURN_VALUES, n, rec); 1435 return; 1436 } 1437 1438 /* 1439 * EXIT STATUS section can also be specified all on one line or on two 1440 * separate lines. 1441 */ 1442 if (strcmp(head->string, "EXIT") == 0 && 1443 head->next != NULL && head->next->type == MAN_TEXT && 1444 strcmp(head->next->string, "STATUS") == 0) { 1445 man_parse_section(MANSEC_EXIT_STATUS, n, rec); 1446 return; 1447 } 1448 1449 /* Store the rest of the content in desc. */ 1450 man_parse_section(MANSEC_NONE, n, rec); 1451 } 1452 1453 /* 1454 * pman_parse_node -- 1455 * Generic function to iterate through a node. Usually called from 1456 * man_parse_section to parse a particular section of the man page. 1457 */ 1458 static void 1459 pman_parse_node(const struct man_node *n, secbuff *s) 1460 { 1461 if (n == NULL) 1462 return; 1463 1464 if (n->type == MAN_TEXT) 1465 append(s, n->string); 1466 1467 pman_parse_node(n->child, s); 1468 pman_parse_node(n->next, s); 1469 } 1470 1471 /* 1472 * man_parse_section -- 1473 * Takes two parameters: 1474 * sec: Tells which section we are present in 1475 * n: Is the present node of the AST. 1476 * Depending on the section, we call pman_parse_node to parse that section and 1477 * concatenate the content from that section into the buffer for that section. 1478 */ 1479 static void 1480 man_parse_section(enum man_sec sec, const struct man_node *n, mandb_rec *rec) 1481 { 1482 /* 1483 * If the user sepecified the 'l' flag then just parse 1484 * the NAME section, ignore the rest. 1485 */ 1486 if (mflags.limit) 1487 return; 1488 1489 switch (sec) { 1490 case MANSEC_LIBRARY: 1491 pman_parse_node(n, &rec->lib); 1492 break; 1493 case MANSEC_RETURN_VALUES: 1494 pman_parse_node(n, &rec->return_vals); 1495 break; 1496 case MANSEC_ENVIRONMENT: 1497 pman_parse_node(n, &rec->env); 1498 break; 1499 case MANSEC_FILES: 1500 pman_parse_node(n, &rec->files); 1501 break; 1502 case MANSEC_EXIT_STATUS: 1503 pman_parse_node(n, &rec->exit_status); 1504 break; 1505 case MANSEC_DIAGNOSTICS: 1506 pman_parse_node(n, &rec->diagnostics); 1507 break; 1508 case MANSEC_ERRORS: 1509 pman_parse_node(n, &rec->errors); 1510 break; 1511 case MANSEC_NAME: 1512 case MANSEC_SYNOPSIS: 1513 case MANSEC_EXAMPLES: 1514 case MANSEC_STANDARDS: 1515 case MANSEC_HISTORY: 1516 case MANSEC_BUGS: 1517 case MANSEC_AUTHORS: 1518 case MANSEC_COPYRIGHT: 1519 break; 1520 default: 1521 pman_parse_node(n, &rec->desc); 1522 break; 1523 } 1524 1525 } 1526 1527 /* 1528 * insert_into_db -- 1529 * Inserts the parsed data of the man page in the Sqlite databse. 1530 * If any of the values is NULL, then we cleanup and return -1 indicating 1531 * an error. 1532 * Otherwise, store the data in the database and return 0. 1533 */ 1534 static int 1535 insert_into_db(sqlite3 *db, mandb_rec *rec) 1536 { 1537 int rc = 0; 1538 int idx = -1; 1539 const char *sqlstr = NULL; 1540 sqlite3_stmt *stmt = NULL; 1541 char *ln = NULL; 1542 char *errmsg = NULL; 1543 long int mandb_rowid; 1544 1545 /* 1546 * At the very minimum we want to make sure that we store 1547 * the following data: 1548 * Name, one line description, and the MD5 hash 1549 */ 1550 if (rec->name == NULL || rec->name_desc == NULL || 1551 rec->md5_hash == NULL) { 1552 cleanup(rec); 1553 return -1; 1554 } 1555 1556 /* Write null byte at the end of all the sec_buffs */ 1557 rec->desc.data[rec->desc.offset] = 0; 1558 rec->lib.data[rec->lib.offset] = 0; 1559 rec->env.data[rec->env.offset] = 0; 1560 rec->return_vals.data[rec->return_vals.offset] = 0; 1561 rec->exit_status.data[rec->exit_status.offset] = 0; 1562 rec->files.data[rec->files.offset] = 0; 1563 rec->diagnostics.data[rec->diagnostics.offset] = 0; 1564 rec->errors.data[rec->errors.offset] = 0; 1565 1566 /* 1567 * In case of a mdoc page: (sorry, no better place to put this code) 1568 * parse the comma separated list of names of man pages, 1569 * the first name will be stored in the mandb table, rest will be 1570 * treated as links and put in the mandb_links table. 1571 */ 1572 if (rec->page_type == MDOC) { 1573 char *tmp; 1574 rec->links = estrdup(rec->name); 1575 free(rec->name); 1576 int sz = strcspn(rec->links, " \0"); 1577 rec->name = emalloc(sz + 1); 1578 memcpy(rec->name, rec->links, sz); 1579 if(rec->name[sz - 1] == ',') 1580 rec->name[sz - 1] = 0; 1581 else 1582 rec->name[sz] = 0; 1583 while (rec->links[sz] == ' ') 1584 ++sz; 1585 tmp = estrdup(rec->links + sz); 1586 free(rec->links); 1587 rec->links = tmp; 1588 } 1589 1590 /*------------------------ Populate the mandb table---------------------------*/ 1591 sqlstr = "INSERT INTO mandb VALUES (:section, :name, :name_desc, :desc," 1592 " :lib, :return_vals, :env, :files, :exit_status," 1593 " :diagnostics, :errors, :md5_hash, :machine)"; 1594 1595 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1596 if (rc != SQLITE_OK) 1597 goto Out; 1598 1599 idx = sqlite3_bind_parameter_index(stmt, ":name"); 1600 rc = sqlite3_bind_text(stmt, idx, rec->name, -1, NULL); 1601 if (rc != SQLITE_OK) { 1602 sqlite3_finalize(stmt); 1603 goto Out; 1604 } 1605 1606 idx = sqlite3_bind_parameter_index(stmt, ":section"); 1607 rc = sqlite3_bind_text(stmt, idx, rec->section, -1, NULL); 1608 if (rc != SQLITE_OK) { 1609 sqlite3_finalize(stmt); 1610 goto Out; 1611 } 1612 1613 idx = sqlite3_bind_parameter_index(stmt, ":name_desc"); 1614 rc = sqlite3_bind_text(stmt, idx, rec->name_desc, -1, NULL); 1615 if (rc != SQLITE_OK) { 1616 sqlite3_finalize(stmt); 1617 goto Out; 1618 } 1619 1620 idx = sqlite3_bind_parameter_index(stmt, ":desc"); 1621 rc = sqlite3_bind_text(stmt, idx, rec->desc.data, 1622 rec->desc.offset + 1, NULL); 1623 if (rc != SQLITE_OK) { 1624 sqlite3_finalize(stmt); 1625 goto Out; 1626 } 1627 1628 idx = sqlite3_bind_parameter_index(stmt, ":lib"); 1629 rc = sqlite3_bind_text(stmt, idx, rec->lib.data, rec->lib.offset + 1, NULL); 1630 if (rc != SQLITE_OK) { 1631 sqlite3_finalize(stmt); 1632 goto Out; 1633 } 1634 1635 idx = sqlite3_bind_parameter_index(stmt, ":return_vals"); 1636 rc = sqlite3_bind_text(stmt, idx, rec->return_vals.data, 1637 rec->return_vals.offset + 1, NULL); 1638 if (rc != SQLITE_OK) { 1639 sqlite3_finalize(stmt); 1640 goto Out; 1641 } 1642 1643 idx = sqlite3_bind_parameter_index(stmt, ":env"); 1644 rc = sqlite3_bind_text(stmt, idx, rec->env.data, rec->env.offset + 1, NULL); 1645 if (rc != SQLITE_OK) { 1646 sqlite3_finalize(stmt); 1647 goto Out; 1648 } 1649 1650 idx = sqlite3_bind_parameter_index(stmt, ":files"); 1651 rc = sqlite3_bind_text(stmt, idx, rec->files.data, 1652 rec->files.offset + 1, NULL); 1653 if (rc != SQLITE_OK) { 1654 sqlite3_finalize(stmt); 1655 goto Out; 1656 } 1657 1658 idx = sqlite3_bind_parameter_index(stmt, ":exit_status"); 1659 rc = sqlite3_bind_text(stmt, idx, rec->exit_status.data, 1660 rec->exit_status.offset + 1, NULL); 1661 if (rc != SQLITE_OK) { 1662 sqlite3_finalize(stmt); 1663 goto Out; 1664 } 1665 1666 idx = sqlite3_bind_parameter_index(stmt, ":diagnostics"); 1667 rc = sqlite3_bind_text(stmt, idx, rec->diagnostics.data, 1668 rec->diagnostics.offset + 1, NULL); 1669 if (rc != SQLITE_OK) { 1670 sqlite3_finalize(stmt); 1671 goto Out; 1672 } 1673 1674 idx = sqlite3_bind_parameter_index(stmt, ":errors"); 1675 rc = sqlite3_bind_text(stmt, idx, rec->errors.data, 1676 rec->errors.offset + 1, NULL); 1677 if (rc != SQLITE_OK) { 1678 sqlite3_finalize(stmt); 1679 goto Out; 1680 } 1681 1682 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1683 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1684 if (rc != SQLITE_OK) { 1685 sqlite3_finalize(stmt); 1686 goto Out; 1687 } 1688 1689 idx = sqlite3_bind_parameter_index(stmt, ":machine"); 1690 if (rec->machine) 1691 rc = sqlite3_bind_text(stmt, idx, rec->machine, -1, NULL); 1692 else 1693 rc = sqlite3_bind_null(stmt, idx); 1694 if (rc != SQLITE_OK) { 1695 sqlite3_finalize(stmt); 1696 goto Out; 1697 } 1698 1699 rc = sqlite3_step(stmt); 1700 if (rc != SQLITE_DONE) { 1701 sqlite3_finalize(stmt); 1702 goto Out; 1703 } 1704 1705 sqlite3_finalize(stmt); 1706 1707 /* Get the row id of the last inserted row */ 1708 mandb_rowid = sqlite3_last_insert_rowid(db); 1709 1710 /*------------------------Populate the mandb_meta table-----------------------*/ 1711 sqlstr = "INSERT INTO mandb_meta VALUES (:device, :inode, :mtime," 1712 " :file, :md5_hash, :id)"; 1713 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1714 if (rc != SQLITE_OK) 1715 goto Out; 1716 1717 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1718 rc = sqlite3_bind_int64(stmt, idx, rec->device); 1719 if (rc != SQLITE_OK) { 1720 sqlite3_finalize(stmt); 1721 goto Out; 1722 } 1723 1724 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1725 rc = sqlite3_bind_int64(stmt, idx, rec->inode); 1726 if (rc != SQLITE_OK) { 1727 sqlite3_finalize(stmt); 1728 goto Out; 1729 } 1730 1731 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1732 rc = sqlite3_bind_int64(stmt, idx, rec->mtime); 1733 if (rc != SQLITE_OK) { 1734 sqlite3_finalize(stmt); 1735 goto Out; 1736 } 1737 1738 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1739 rc = sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1740 if (rc != SQLITE_OK) { 1741 sqlite3_finalize(stmt); 1742 goto Out; 1743 } 1744 1745 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1746 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1747 if (rc != SQLITE_OK) { 1748 sqlite3_finalize(stmt); 1749 goto Out; 1750 } 1751 1752 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1753 rc = sqlite3_bind_int64(stmt, idx, mandb_rowid); 1754 if (rc != SQLITE_OK) { 1755 sqlite3_finalize(stmt); 1756 goto Out; 1757 } 1758 1759 rc = sqlite3_step(stmt); 1760 sqlite3_finalize(stmt); 1761 if (rc == SQLITE_CONSTRAINT_UNIQUE) { 1762 /* The *most* probable reason for reaching here is that 1763 * the UNIQUE contraint on the file column of the mandb_meta 1764 * table was violated. 1765 * This can happen when a file was updated/modified. 1766 * To fix this we need to do two things: 1767 * 1. Delete the row for the older version of this file 1768 * from mandb table. 1769 * 2. Run an UPDATE query to update the row for this file 1770 * in the mandb_meta table. 1771 */ 1772 warnx("Trying to update index for %s", rec->file_path); 1773 char *sql = sqlite3_mprintf("DELETE FROM mandb " 1774 "WHERE rowid = (SELECT id" 1775 " FROM mandb_meta" 1776 " WHERE file = %Q)", 1777 rec->file_path); 1778 sqlite3_exec(db, sql, NULL, NULL, &errmsg); 1779 sqlite3_free(sql); 1780 if (errmsg != NULL) { 1781 if (mflags.verbosity) 1782 warnx("%s", errmsg); 1783 free(errmsg); 1784 } 1785 sqlstr = "UPDATE mandb_meta SET device = :device," 1786 " inode = :inode, mtime = :mtime, id = :id," 1787 " md5_hash = :md5 WHERE file = :file"; 1788 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1789 if (rc != SQLITE_OK) { 1790 if (mflags.verbosity) 1791 warnx("Update failed with error: %s", 1792 sqlite3_errmsg(db)); 1793 close_db(db); 1794 cleanup(rec); 1795 errx(EXIT_FAILURE, 1796 "Consider running makemandb with -f option"); 1797 } 1798 1799 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1800 sqlite3_bind_int64(stmt, idx, rec->device); 1801 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1802 sqlite3_bind_int64(stmt, idx, rec->inode); 1803 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1804 sqlite3_bind_int64(stmt, idx, rec->mtime); 1805 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1806 sqlite3_bind_int64(stmt, idx, mandb_rowid); 1807 idx = sqlite3_bind_parameter_index(stmt, ":md5"); 1808 sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1809 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1810 sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1811 rc = sqlite3_step(stmt); 1812 sqlite3_finalize(stmt); 1813 1814 if (rc != SQLITE_DONE) { 1815 if (mflags.verbosity) 1816 warnx("%s", sqlite3_errmsg(db)); 1817 close_db(db); 1818 cleanup(rec); 1819 errx(EXIT_FAILURE, 1820 "Consider running makemandb with -f option"); 1821 } 1822 } else if (rc != SQLITE_DONE) { 1823 /* Otherwise make this error fatal */ 1824 warnx("Failed at %s\n%s", rec->file_path, sqlite3_errmsg(db)); 1825 cleanup(rec); 1826 close_db(db); 1827 exit(EXIT_FAILURE); 1828 } 1829 1830 /*------------------------ Populate the mandb_links table---------------------*/ 1831 char *str = NULL; 1832 char *links; 1833 if (rec->links && strlen(rec->links)) { 1834 links = rec->links; 1835 for(ln = strtok(links, " "); ln; ln = strtok(NULL, " ")) { 1836 if (ln[0] == ',') 1837 ln++; 1838 if(ln[strlen(ln) - 1] == ',') 1839 ln[strlen(ln) - 1] = 0; 1840 1841 str = sqlite3_mprintf("INSERT INTO mandb_links" 1842 " VALUES (%Q, %Q, %Q, %Q, %Q)", 1843 ln, rec->name, rec->section, 1844 rec->machine, rec->md5_hash); 1845 sqlite3_exec(db, str, NULL, NULL, &errmsg); 1846 sqlite3_free(str); 1847 if (errmsg != NULL) { 1848 warnx("%s", errmsg); 1849 cleanup(rec); 1850 free(errmsg); 1851 return -1; 1852 } 1853 } 1854 } 1855 1856 cleanup(rec); 1857 return 0; 1858 1859 Out: 1860 if (mflags.verbosity) 1861 warnx("%s", sqlite3_errmsg(db)); 1862 cleanup(rec); 1863 return -1; 1864 } 1865 1866 /* 1867 * check_md5-- 1868 * Generates the md5 hash of the file and checks if it already doesn't exist 1869 * in the table (passed as the 3rd parameter). 1870 * This function is being used to avoid hardlinks. 1871 * On successful completion it will also set the value of the fourth parameter 1872 * to the md5 hash of the file (computed previously). It is the responsibility 1873 * of the caller to free this buffer. 1874 * Return values: 1875 * -1: If an error occurs somewhere and sets the md5 return buffer to NULL. 1876 * 0: If the md5 hash does not exist in the table. 1877 * 1: If the hash exists in the database. 1878 */ 1879 static int 1880 check_md5(const char *file, sqlite3 *db, const char *table, char **md5sum, 1881 void *buf, size_t buflen) 1882 { 1883 int rc = 0; 1884 int idx = -1; 1885 char *sqlstr = NULL; 1886 sqlite3_stmt *stmt = NULL; 1887 1888 assert(file != NULL); 1889 *md5sum = MD5Data(buf, buflen, NULL); 1890 if (*md5sum == NULL) { 1891 if (mflags.verbosity) 1892 warn("md5 failed: %s", file); 1893 return -1; 1894 } 1895 1896 easprintf(&sqlstr, "SELECT * FROM %s WHERE md5_hash = :md5_hash", 1897 table); 1898 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1899 if (rc != SQLITE_OK) { 1900 free(sqlstr); 1901 free(*md5sum); 1902 *md5sum = NULL; 1903 return -1; 1904 } 1905 1906 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1907 rc = sqlite3_bind_text(stmt, idx, *md5sum, -1, NULL); 1908 if (rc != SQLITE_OK) { 1909 if (mflags.verbosity) 1910 warnx("%s", sqlite3_errmsg(db)); 1911 sqlite3_finalize(stmt); 1912 free(sqlstr); 1913 free(*md5sum); 1914 *md5sum = NULL; 1915 return -1; 1916 } 1917 1918 if (sqlite3_step(stmt) == SQLITE_ROW) { 1919 sqlite3_finalize(stmt); 1920 free(sqlstr); 1921 return 0; 1922 } 1923 1924 sqlite3_finalize(stmt); 1925 free(sqlstr); 1926 return 1; 1927 } 1928 1929 /* Optimize the index for faster search */ 1930 static void 1931 optimize(sqlite3 *db) 1932 { 1933 const char *sqlstr; 1934 char *errmsg = NULL; 1935 1936 if (mflags.verbosity == 2) 1937 printf("Optimizing the database index\n"); 1938 sqlstr = "INSERT INTO mandb(mandb) VALUES (\'optimize\');" 1939 "VACUUM"; 1940 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 1941 if (errmsg != NULL) { 1942 if (mflags.verbosity) 1943 warnx("%s", errmsg); 1944 free(errmsg); 1945 return; 1946 } 1947 } 1948 1949 /* 1950 * cleanup -- 1951 * cleans up the global buffers 1952 */ 1953 static void 1954 cleanup(mandb_rec *rec) 1955 { 1956 rec->desc.offset = 0; 1957 rec->lib.offset = 0; 1958 rec->return_vals.offset = 0; 1959 rec->env.offset = 0; 1960 rec->exit_status.offset = 0; 1961 rec->diagnostics.offset = 0; 1962 rec->errors.offset = 0; 1963 rec->files.offset = 0; 1964 1965 free(rec->machine); 1966 rec->machine = NULL; 1967 1968 free(rec->links); 1969 rec->links = NULL; 1970 1971 free(rec->file_path); 1972 rec->file_path = NULL; 1973 1974 free(rec->name); 1975 rec->name = NULL; 1976 1977 free(rec->name_desc); 1978 rec->name_desc = NULL; 1979 1980 free(rec->md5_hash); 1981 rec->md5_hash = NULL; 1982 } 1983 1984 /* 1985 * init_secbuffs-- 1986 * Sets the value of buflen for all the sec_buff field of rec. And then 1987 * allocate memory to each sec_buff member of rec. 1988 */ 1989 static void 1990 init_secbuffs(mandb_rec *rec) 1991 { 1992 /* 1993 * Some sec_buff might need more memory, for example desc, 1994 * which stores the data of the DESCRIPTION section, 1995 * while some might need very small amount of memory. 1996 * Therefore explicitly setting the value of buflen field for 1997 * each sec_buff. 1998 */ 1999 rec->desc.buflen = 10 * BUFLEN; 2000 rec->desc.data = emalloc(rec->desc.buflen); 2001 rec->desc.offset = 0; 2002 2003 rec->lib.buflen = BUFLEN / 2; 2004 rec->lib.data = emalloc(rec->lib.buflen); 2005 rec->lib.offset = 0; 2006 2007 rec->return_vals.buflen = BUFLEN; 2008 rec->return_vals.data = emalloc(rec->return_vals.buflen); 2009 rec->return_vals.offset = 0; 2010 2011 rec->exit_status.buflen = BUFLEN; 2012 rec->exit_status.data = emalloc(rec->exit_status.buflen); 2013 rec->exit_status.offset = 0; 2014 2015 rec->env.buflen = BUFLEN; 2016 rec->env.data = emalloc(rec->env.buflen); 2017 rec->env.offset = 0; 2018 2019 rec->files.buflen = BUFLEN; 2020 rec->files.data = emalloc(rec->files.buflen); 2021 rec->files.offset = 0; 2022 2023 rec->diagnostics.buflen = BUFLEN; 2024 rec->diagnostics.data = emalloc(rec->diagnostics.buflen); 2025 rec->diagnostics.offset = 0; 2026 2027 rec->errors.buflen = BUFLEN; 2028 rec->errors.data = emalloc(rec->errors.buflen); 2029 rec->errors.offset = 0; 2030 } 2031 2032 /* 2033 * free_secbuffs-- 2034 * This function should be called at the end, when all the pages have been 2035 * parsed. 2036 * It frees the memory allocated to sec_buffs by init_secbuffs in the starting. 2037 */ 2038 static void 2039 free_secbuffs(mandb_rec *rec) 2040 { 2041 free(rec->desc.data); 2042 free(rec->lib.data); 2043 free(rec->return_vals.data); 2044 free(rec->exit_status.data); 2045 free(rec->env.data); 2046 free(rec->files.data); 2047 free(rec->diagnostics.data); 2048 free(rec->errors.data); 2049 } 2050 2051 static void 2052 replace_hyph(char *str) 2053 { 2054 char *iter = str; 2055 while ((iter = strchr(iter, ASCII_HYPH)) != NULL) 2056 *iter = '-'; 2057 2058 iter = str; 2059 while ((iter = strchr(iter, ASCII_NBRSP)) != NULL) 2060 *iter = '-'; 2061 } 2062 2063 static char * 2064 parse_escape(const char *str) 2065 { 2066 const char *backslash, *last_backslash; 2067 char *result, *iter; 2068 size_t len; 2069 2070 assert(str); 2071 2072 last_backslash = str; 2073 backslash = strchr(str, '\\'); 2074 if (backslash == NULL) { 2075 result = estrdup(str); 2076 replace_hyph(result); 2077 return result; 2078 } 2079 2080 result = emalloc(strlen(str) + 1); 2081 iter = result; 2082 2083 do { 2084 len = backslash - last_backslash; 2085 memcpy(iter, last_backslash, len); 2086 iter += len; 2087 if (backslash[1] == '-' || backslash[1] == ' ') { 2088 *iter++ = backslash[1]; 2089 last_backslash = backslash + 2; 2090 backslash = strchr(backslash + 2, '\\'); 2091 } else { 2092 ++backslash; 2093 mandoc_escape(&backslash, NULL, NULL); 2094 last_backslash = backslash; 2095 if (backslash == NULL) 2096 break; 2097 backslash = strchr(last_backslash, '\\'); 2098 } 2099 } while (backslash != NULL); 2100 if (last_backslash != NULL) 2101 strcpy(iter, last_backslash); 2102 2103 replace_hyph(result); 2104 return result; 2105 } 2106 2107 /* 2108 * append-- 2109 * Concatenates a space and src at the end of sbuff->data (much like concat in 2110 * apropos-utils.c). 2111 * Rather than reallocating space for writing data, it uses the value of the 2112 * offset field of sec_buff to write new data at the free space left in the 2113 * buffer. 2114 * In case the size of the data to be appended exceeds the number of bytes left 2115 * in the buffer, it reallocates buflen number of bytes and then continues. 2116 * Value of offset field should be adjusted as new data is written. 2117 * 2118 * NOTE: This function does not write the null byte at the end of the buffers, 2119 * write a null byte at the position pointed to by offset before inserting data 2120 * in the db. 2121 */ 2122 static void 2123 append(secbuff *sbuff, const char *src) 2124 { 2125 short flag = 0; 2126 size_t srclen, newlen; 2127 char *temp; 2128 2129 assert(src != NULL); 2130 temp = parse_escape(src); 2131 srclen = strlen(temp); 2132 2133 if (sbuff->data == NULL) { 2134 sbuff->data = emalloc(sbuff->buflen); 2135 sbuff->offset = 0; 2136 } 2137 2138 newlen = sbuff->offset + srclen + 2; 2139 if (newlen >= sbuff->buflen) { 2140 while (sbuff->buflen < newlen) 2141 sbuff->buflen += sbuff->buflen; 2142 sbuff->data = erealloc(sbuff->data, sbuff->buflen); 2143 flag = 1; 2144 } 2145 2146 /* Append a space at the end of the buffer. */ 2147 if (sbuff->offset || flag) 2148 sbuff->data[sbuff->offset++] = ' '; 2149 /* Now, copy src at the end of the buffer. */ 2150 memcpy(sbuff->data + sbuff->offset, temp, srclen); 2151 sbuff->offset += srclen; 2152 free(temp); 2153 } 2154 2155 static void 2156 usage(void) 2157 { 2158 fprintf(stderr, "Usage: %s [-floQqv] [-C path]\n", getprogname()); 2159 exit(1); 2160 } 2161
Indexes created Mon Sep 29 21:09:56 GMT 2025