makemandb.c revision 1.36
1 /* $NetBSD: makemandb.c,v 1.36 2016/04/13 01:41:18 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 1.36 2016/04/13 01:41:18 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 close_db(db); 433 exit(EXIT_FAILURE); 434 } 435 436 sqlstr = "CREATE TABLE metadb.file_cache(device, inode, mtime, parent," 437 " file PRIMARY KEY);" 438 "CREATE UNIQUE INDEX metadb.index_file_cache_dev" 439 " ON file_cache (device, inode)"; 440 441 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 442 if (errmsg != NULL) { 443 warnx("%s", errmsg); 444 free(errmsg); 445 close_db(db); 446 exit(EXIT_FAILURE); 447 } 448 449 if (mflags.verbosity) 450 printf("Building temporary file cache\n"); 451 line = NULL; 452 linesize = 0; 453 while ((len = getline(&line, &linesize, file)) != -1) { 454 /* Replace the new line character at the end of string with '\0' */ 455 line[len - 1] = '\0'; 456 parent = estrdup(line); 457 char *pdir = estrdup(dirname(parent)); 458 free(parent); 459 /* Traverse the man page directories and parse the pages */ 460 traversedir(pdir, line, db, mp); 461 free(pdir); 462 } 463 free(line); 464 465 if (pclose(file) == -1) { 466 close_db(db); 467 cleanup(&rec); 468 free_secbuffs(&rec); 469 err(EXIT_FAILURE, "pclose error"); 470 } 471 472 if (mflags.verbosity) 473 printf("Performing index update\n"); 474 update_db(db, mp, &rec); 475 mparse_free(mp); 476 mchars_free(mchars); 477 free_secbuffs(&rec); 478 479 /* Commit the transaction */ 480 sqlite3_exec(db, "COMMIT", NULL, NULL, &errmsg); 481 if (errmsg != NULL) { 482 warnx("%s", errmsg); 483 free(errmsg); 484 close_db(db); 485 exit(EXIT_FAILURE); 486 } 487 488 if (mflags.optimize) 489 optimize(db); 490 491 close_db(db); 492 return 0; 493 } 494 495 /* 496 * traversedir -- 497 * Traverses the given directory recursively and passes all the man page files 498 * in the way to build_file_cache() 499 */ 500 static void 501 traversedir(const char *parent, const char *file, sqlite3 *db, 502 struct mparse *mp) 503 { 504 struct stat sb; 505 struct dirent *dirp; 506 DIR *dp; 507 char *buf; 508 509 if (stat(file, &sb) < 0) { 510 if (mflags.verbosity) 511 warn("stat failed: %s", file); 512 return; 513 } 514 515 /* If it is a directory, traverse it recursively */ 516 if (S_ISDIR(sb.st_mode)) { 517 if ((dp = opendir(file)) == NULL) { 518 if (mflags.verbosity) 519 warn("opendir error: %s", file); 520 return; 521 } 522 523 while ((dirp = readdir(dp)) != NULL) { 524 /* Avoid . and .. entries in a directory */ 525 if (strncmp(dirp->d_name, ".", 1)) { 526 easprintf(&buf, "%s/%s", file, dirp->d_name); 527 traversedir(parent, buf, db, mp); 528 free(buf); 529 } 530 } 531 closedir(dp); 532 } 533 534 if (!S_ISREG(sb.st_mode) && !S_ISLNK(sb.st_mode)) 535 return; 536 537 if (sb.st_size == 0) { 538 if (mflags.verbosity) 539 warnx("Empty file: %s", file); 540 return; 541 } 542 build_file_cache(db, parent, file, &sb); 543 } 544 545 /* build_file_cache -- 546 * This function generates an md5 hash of the file passed as its 2nd parameter 547 * and stores it in a temporary table file_cache along with the full file path. 548 * This is done to support incremental updation of the database. 549 * The temporary table file_cache is dropped thereafter in the function 550 * update_db(), once the database has been updated. 551 */ 552 static void 553 build_file_cache(sqlite3 *db, const char *parent, const char *file, 554 struct stat *sb) 555 { 556 const char *sqlstr; 557 sqlite3_stmt *stmt = NULL; 558 int rc, idx; 559 assert(file != NULL); 560 dev_t device_cache = sb->st_dev; 561 ino_t inode_cache = sb->st_ino; 562 time_t mtime_cache = sb->st_mtime; 563 564 sqlstr = "INSERT INTO metadb.file_cache VALUES (:device, :inode," 565 " :mtime, :parent, :file)"; 566 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 567 if (rc != SQLITE_OK) { 568 if (mflags.verbosity) 569 warnx("%s", sqlite3_errmsg(db)); 570 return; 571 } 572 573 idx = sqlite3_bind_parameter_index(stmt, ":device"); 574 rc = sqlite3_bind_int64(stmt, idx, device_cache); 575 if (rc != SQLITE_OK) { 576 if (mflags.verbosity) 577 warnx("%s", sqlite3_errmsg(db)); 578 sqlite3_finalize(stmt); 579 return; 580 } 581 582 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 583 rc = sqlite3_bind_int64(stmt, idx, inode_cache); 584 if (rc != SQLITE_OK) { 585 if (mflags.verbosity) 586 warnx("%s", sqlite3_errmsg(db)); 587 sqlite3_finalize(stmt); 588 return; 589 } 590 591 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 592 rc = sqlite3_bind_int64(stmt, idx, mtime_cache); 593 if (rc != SQLITE_OK) { 594 if (mflags.verbosity) 595 warnx("%s", sqlite3_errmsg(db)); 596 sqlite3_finalize(stmt); 597 return; 598 } 599 600 idx = sqlite3_bind_parameter_index(stmt, ":parent"); 601 rc = sqlite3_bind_text(stmt, idx, parent, -1, NULL); 602 if (rc != SQLITE_OK) { 603 if (mflags.verbosity) 604 warnx("%s", sqlite3_errmsg(db)); 605 sqlite3_finalize(stmt); 606 return; 607 } 608 609 idx = sqlite3_bind_parameter_index(stmt, ":file"); 610 rc = sqlite3_bind_text(stmt, idx, file, -1, NULL); 611 if (rc != SQLITE_OK) { 612 if (mflags.verbosity) 613 warnx("%s", sqlite3_errmsg(db)); 614 sqlite3_finalize(stmt); 615 return; 616 } 617 618 sqlite3_step(stmt); 619 sqlite3_finalize(stmt); 620 } 621 622 static void 623 update_existing_entry(sqlite3 *db, const char *file, const char *hash, 624 mandb_rec *rec, int *new_count, int *link_count, int *err_count) 625 { 626 int update_count, rc, idx; 627 const char *inner_sqlstr; 628 sqlite3_stmt *inner_stmt; 629 630 update_count = sqlite3_total_changes(db); 631 inner_sqlstr = "UPDATE mandb_meta SET device = :device," 632 " inode = :inode, mtime = :mtime WHERE" 633 " md5_hash = :md5 AND file = :file AND" 634 " (device <> :device2 OR inode <> " 635 " :inode2 OR mtime <> :mtime2)"; 636 rc = sqlite3_prepare_v2(db, inner_sqlstr, -1, &inner_stmt, NULL); 637 if (rc != SQLITE_OK) { 638 if (mflags.verbosity) 639 warnx("%s", sqlite3_errmsg(db)); 640 return; 641 } 642 idx = sqlite3_bind_parameter_index(inner_stmt, ":device"); 643 sqlite3_bind_int64(inner_stmt, idx, rec->device); 644 idx = sqlite3_bind_parameter_index(inner_stmt, ":inode"); 645 sqlite3_bind_int64(inner_stmt, idx, rec->inode); 646 idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime"); 647 sqlite3_bind_int64(inner_stmt, idx, rec->mtime); 648 idx = sqlite3_bind_parameter_index(inner_stmt, ":md5"); 649 sqlite3_bind_text(inner_stmt, idx, hash, -1, NULL); 650 idx = sqlite3_bind_parameter_index(inner_stmt, ":file"); 651 sqlite3_bind_text(inner_stmt, idx, file, -1, NULL); 652 idx = sqlite3_bind_parameter_index(inner_stmt, ":device2"); 653 sqlite3_bind_int64(inner_stmt, idx, rec->device); 654 idx = sqlite3_bind_parameter_index(inner_stmt, ":inode2"); 655 sqlite3_bind_int64(inner_stmt, idx, rec->inode); 656 idx = sqlite3_bind_parameter_index(inner_stmt, ":mtime2"); 657 sqlite3_bind_int64(inner_stmt, idx, rec->mtime); 658 659 rc = sqlite3_step(inner_stmt); 660 if (rc == SQLITE_DONE) { 661 /* Check if an update has been performed. */ 662 if (update_count != sqlite3_total_changes(db)) { 663 if (mflags.verbosity == 2) 664 printf("Updated %s\n", file); 665 (*new_count)++; 666 } else { 667 /* Otherwise it was a hardlink. */ 668 (*link_count)++; 669 } 670 } else { 671 if (mflags.verbosity == 2) 672 warnx("Could not update the meta data for %s", file); 673 (*err_count)++; 674 } 675 sqlite3_finalize(inner_stmt); 676 } 677 678 /* read_and_decompress -- 679 * Reads the given file into memory. If it is compressed, decompress 680 * it before returning to the caller. 681 */ 682 static int 683 read_and_decompress(const char *file, void **bufp, size_t *len) 684 { 685 size_t off; 686 ssize_t r; 687 struct archive *a; 688 struct archive_entry *ae; 689 char *buf; 690 691 if ((a = archive_read_new()) == NULL) 692 errx(EXIT_FAILURE, "memory allocation failed"); 693 694 *bufp = NULL; 695 if (archive_read_support_compression_all(a) != ARCHIVE_OK || 696 archive_read_support_format_raw(a) != ARCHIVE_OK || 697 archive_read_open_filename(a, file, 65536) != ARCHIVE_OK || 698 archive_read_next_header(a, &ae) != ARCHIVE_OK) 699 goto archive_error; 700 *len = 65536; 701 buf = emalloc(*len); 702 off = 0; 703 for (;;) { 704 r = archive_read_data(a, buf + off, *len - off); 705 if (r == ARCHIVE_OK) { 706 archive_read_finish(a); 707 *bufp = buf; 708 *len = off; 709 return 0; 710 } 711 if (r <= 0) { 712 free(buf); 713 break; 714 } 715 off += r; 716 if (off == *len) { 717 *len *= 2; 718 if (*len < off) { 719 if (mflags.verbosity) 720 warnx("File too large: %s", file); 721 free(buf); 722 archive_read_finish(a); 723 return -1; 724 } 725 buf = erealloc(buf, *len); 726 } 727 } 728 729 archive_error: 730 warnx("Error while reading `%s': %s", file, archive_error_string(a)); 731 archive_read_finish(a); 732 return -1; 733 } 734 735 /* update_db -- 736 * Does an incremental updation of the database by checking the file_cache. 737 * It parses and adds the pages which are present in file_cache, 738 * but not in the database. 739 * It also removes the pages which are present in the databse, 740 * but not in the file_cache. 741 */ 742 static void 743 update_db(sqlite3 *db, struct mparse *mp, mandb_rec *rec) 744 { 745 const char *sqlstr; 746 sqlite3_stmt *stmt = NULL; 747 char *file; 748 char *parent; 749 char *errmsg = NULL; 750 char *md5sum; 751 void *buf; 752 size_t buflen; 753 struct sql_row { 754 struct sql_row *next; 755 dev_t device; 756 ino_t inode; 757 time_t mtime; 758 char *parent; 759 char *file; 760 } *rows, *row; 761 int new_count = 0; /* Counter for newly indexed/updated pages */ 762 int total_count = 0; /* Counter for total number of pages */ 763 int err_count = 0; /* Counter for number of failed pages */ 764 int link_count = 0; /* Counter for number of hard/sym links */ 765 int md5_status; 766 int rc; 767 768 sqlstr = "SELECT device, inode, mtime, parent, file" 769 " FROM metadb.file_cache fc" 770 " WHERE NOT EXISTS(SELECT 1 FROM mandb_meta WHERE" 771 " device = fc.device AND inode = fc.inode AND " 772 " mtime = fc.mtime AND file = fc.file)"; 773 774 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 775 if (rc != SQLITE_OK) { 776 if (mflags.verbosity) 777 warnx("%s", sqlite3_errmsg(db)); 778 close_db(db); 779 errx(EXIT_FAILURE, "Could not query file cache"); 780 } 781 782 buf = NULL; 783 rows = NULL; 784 while (sqlite3_step(stmt) == SQLITE_ROW) { 785 row = emalloc(sizeof(struct sql_row)); 786 row->device = sqlite3_column_int64(stmt, 0); 787 row->inode = sqlite3_column_int64(stmt, 1); 788 row->mtime = sqlite3_column_int64(stmt, 2); 789 row->parent = estrdup((const char *) sqlite3_column_text(stmt, 3)); 790 row->file = estrdup((const char *) sqlite3_column_text(stmt, 4)); 791 row->next = rows; 792 rows = row; 793 total_count++; 794 } 795 sqlite3_finalize(stmt); 796 797 for ( ; rows != NULL; free(parent), free(file), free(buf)) { 798 row = rows; 799 rows = rows->next; 800 801 rec->device = row->device; 802 rec->inode = row->inode; 803 rec->mtime = row->mtime; 804 parent = row->parent; 805 file = row->file; 806 free(row); 807 808 if (read_and_decompress(file, &buf, &buflen)) { 809 err_count++; 810 continue; 811 } 812 md5_status = check_md5(file, db, "mandb_meta", &md5sum, buf, buflen); 813 assert(md5sum != NULL); 814 if (md5_status == -1) { 815 if (mflags.verbosity) 816 warnx("An error occurred in checking md5 value" 817 " for file %s", file); 818 err_count++; 819 continue; 820 } 821 822 if (md5_status == 0) { 823 /* 824 * The MD5 hash is already present in the database, 825 * so simply update the metadata, ignoring symlinks. 826 */ 827 struct stat sb; 828 stat(file, &sb); 829 if (S_ISLNK(sb.st_mode)) { 830 free(md5sum); 831 link_count++; 832 continue; 833 } 834 update_existing_entry(db, file, md5sum, rec, 835 &new_count, &link_count, &err_count); 836 free(md5sum); 837 continue; 838 } 839 840 if (md5_status == 1) { 841 /* 842 * The MD5 hash was not present in the database. 843 * This means is either a new file or an updated file. 844 * We should go ahead with parsing. 845 */ 846 if (chdir(parent) == -1) { 847 if (mflags.verbosity) 848 warn("chdir failed for `%s', could " 849 "not index `%s'", parent, file); 850 err_count++; 851 free(md5sum); 852 continue; 853 } 854 855 if (mflags.verbosity == 2) 856 printf("Parsing: %s\n", file); 857 rec->md5_hash = md5sum; 858 rec->file_path = estrdup(file); 859 // file_path is freed by insert_into_db itself. 860 begin_parse(file, mp, rec, buf, buflen); 861 if (insert_into_db(db, rec) < 0) { 862 if (mflags.verbosity) 863 warnx("Error in indexing `%s'", file); 864 err_count++; 865 } else { 866 new_count++; 867 } 868 } 869 } 870 871 if (mflags.verbosity == 2) { 872 printf("Total Number of new or updated pages encountered = %d\n" 873 "Total number of (hard or symbolic) links found = %d\n" 874 "Total number of pages that were successfully" 875 " indexed/updated = %d\n" 876 "Total number of pages that could not be indexed" 877 " due to errors = %d\n", 878 total_count - link_count, link_count, new_count, err_count); 879 } 880 881 if (mflags.recreate) 882 return; 883 884 if (mflags.verbosity == 2) 885 printf("Deleting stale index entries\n"); 886 887 sqlstr = "DELETE FROM mandb_meta WHERE file NOT IN" 888 " (SELECT file FROM metadb.file_cache);" 889 "DELETE FROM mandb_links WHERE md5_hash NOT IN" 890 " (SELECT md5_hash from mandb_meta);" 891 "DROP TABLE metadb.file_cache;" 892 "DELETE FROM mandb WHERE rowid NOT IN" 893 " (SELECT id FROM mandb_meta);"; 894 895 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 896 if (errmsg != NULL) { 897 warnx("Removing old entries failed: %s", errmsg); 898 warnx("Please rebuild database from scratch with -f."); 899 free(errmsg); 900 return; 901 } 902 } 903 904 /* 905 * begin_parse -- 906 * parses the man page using libmandoc 907 */ 908 static void 909 begin_parse(const char *file, struct mparse *mp, mandb_rec *rec, 910 const void *buf, size_t len) 911 { 912 struct mdoc *mdoc; 913 struct man *man; 914 mparse_reset(mp); 915 916 rec->xr_found = 0; 917 918 if (mparse_readmem(mp, buf, len, file) >= MANDOCLEVEL_BADARG) { 919 /* Printing this warning at verbosity level 2 920 * because some packages from pkgsrc might trigger several 921 * of such warnings. 922 */ 923 if (mflags.verbosity == 2) 924 warnx("%s: Parse failure", file); 925 return; 926 } 927 928 mparse_result(mp, &mdoc, &man, NULL); 929 if (mdoc == NULL && man == NULL) { 930 if (mflags.verbosity == 2) 931 warnx("Not a man(7) or mdoc(7) page"); 932 return; 933 } 934 935 set_machine(mdoc, rec); 936 set_section(mdoc, man, rec); 937 if (mdoc) { 938 rec->page_type = MDOC; 939 pmdoc_node(mdoc_node(mdoc), rec); 940 } else { 941 rec->page_type = MAN; 942 pman_node(man_node(man), rec); 943 } 944 } 945 946 /* 947 * set_section -- 948 * Extracts the section number and normalizes it to only the numeric part 949 * (Which should be the first character of the string). 950 */ 951 static void 952 set_section(const struct mdoc *md, const struct man *m, mandb_rec *rec) 953 { 954 if (md) { 955 const struct mdoc_meta *md_meta = mdoc_meta(md); 956 if (md_meta->msec == NULL) { 957 rec->section[0] = '?'; 958 } else 959 rec->section[0] = md_meta->msec[0]; 960 } else if (m) { 961 const struct man_meta *m_meta = man_meta(m); 962 if (m_meta->msec == NULL) 963 rec->section[0] = '?'; 964 else 965 rec->section[0] = m_meta->msec[0]; 966 } else 967 return; 968 969 if (rec->section[0] == '?' && mflags.verbosity == 2) 970 warnx("%s: Missing section number", rec->file_path); 971 } 972 973 /* 974 * get_machine -- 975 * Extracts the machine architecture information if available. 976 */ 977 static void 978 set_machine(const struct mdoc *md, mandb_rec *rec) 979 { 980 if (md == NULL) 981 return; 982 const struct mdoc_meta *md_meta = mdoc_meta(md); 983 if (md_meta->arch) 984 rec->machine = estrdup(md_meta->arch); 985 } 986 987 static void 988 pmdoc_node(const struct mdoc_node *n, mandb_rec *rec) 989 { 990 991 if (n == NULL) 992 return; 993 994 switch (n->type) { 995 case (MDOC_BODY): 996 /* FALLTHROUGH */ 997 case (MDOC_TAIL): 998 /* FALLTHROUGH */ 999 case (MDOC_ELEM): 1000 if (mdocs[n->tok] == NULL) 1001 break; 1002 (*mdocs[n->tok])(n, rec); 1003 break; 1004 default: 1005 break; 1006 } 1007 1008 pmdoc_node(n->child, rec); 1009 pmdoc_node(n->next, rec); 1010 } 1011 1012 /* 1013 * pmdoc_Nm -- 1014 * Extracts the Name of the manual page from the .Nm macro 1015 */ 1016 static void 1017 pmdoc_Nm(const struct mdoc_node *n, mandb_rec *rec) 1018 { 1019 if (n->sec != SEC_NAME) 1020 return; 1021 1022 for (n = n->child; n; n = n->next) { 1023 if (n->type == MDOC_TEXT) { 1024 char *escaped_name = parse_escape(n->string); 1025 concat(&rec->name, escaped_name); 1026 free(escaped_name); 1027 } 1028 } 1029 } 1030 1031 /* 1032 * pmdoc_Nd -- 1033 * Extracts the one line description of the man page from the .Nd macro 1034 */ 1035 static void 1036 pmdoc_Nd(const struct mdoc_node *n, mandb_rec *rec) 1037 { 1038 char *buf = NULL; 1039 char *name; 1040 char *nd_text; 1041 1042 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1043 return; 1044 1045 if (n->type == MDOC_TEXT) { 1046 if (rec->xr_found && n->next) { 1047 /* 1048 * An Xr macro was seen previously, so parse this 1049 * and the next node, as "Name(Section)". 1050 */ 1051 name = n->string; 1052 n = n->next; 1053 assert(n->type == MDOC_TEXT); 1054 easprintf(&buf, "%s(%s)", name, n->string); 1055 concat(&rec->name_desc, buf); 1056 free(buf); 1057 } else { 1058 nd_text = parse_escape(n->string); 1059 concat(&rec->name_desc, nd_text); 1060 free(nd_text); 1061 } 1062 rec->xr_found = 0; 1063 } else if (mdocs[n->tok] == pmdoc_Xr) { 1064 /* Remember that we have encountered an Xr macro */ 1065 rec->xr_found = 1; 1066 } 1067 1068 if (n->child) 1069 pmdoc_Nd(n->child, rec); 1070 1071 if(n->next) 1072 pmdoc_Nd(n->next, rec); 1073 } 1074 1075 /* 1076 * pmdoc_macro_handler-- 1077 * This function is a single point of handling all the special macros that we 1078 * want to handle especially. For example the .Xr macro for properly parsing 1079 * the referenced page name along with the section number, or the .Pp macro 1080 * for adding a new line whenever we encounter it. 1081 */ 1082 static void 1083 pmdoc_macro_handler(const struct mdoc_node *n, mandb_rec *rec, enum mdoct doct) 1084 { 1085 const struct mdoc_node *sn; 1086 assert(n); 1087 1088 switch (doct) { 1089 /* Parse the man page references. 1090 * Basically the .Xr macros are used like: 1091 * .Xr ls 1 1092 * and formatted like this: 1093 * ls(1) 1094 * Prepare a buffer to format the data like the above example and call 1095 * pmdoc_parse_section to append it. 1096 */ 1097 case MDOC_Xr: 1098 n = n->child; 1099 while (n->type != MDOC_TEXT && n->next) 1100 n = n->next; 1101 1102 if (n && n->type != MDOC_TEXT) 1103 return; 1104 sn = n; 1105 if (n->next) 1106 n = n->next; 1107 1108 while (n->type != MDOC_TEXT && n->next) 1109 n = n->next; 1110 1111 if (n && n->type == MDOC_TEXT) { 1112 char *buf; 1113 easprintf(&buf, "%s(%s)", sn->string, n->string); 1114 mdoc_parse_section(n->sec, buf, rec); 1115 free(buf); 1116 } 1117 1118 break; 1119 1120 /* Parse the .Pp macro to add a new line */ 1121 case MDOC_Pp: 1122 if (n->type == MDOC_TEXT) 1123 mdoc_parse_section(n->sec, "\n", rec); 1124 break; 1125 default: 1126 break; 1127 } 1128 1129 } 1130 1131 /* 1132 * pmdoc_Xr, pmdoc_Pp-- 1133 * Empty stubs. 1134 * The parser calls these functions each time it encounters 1135 * a .Xr or .Pp macro. We are parsing all the data from 1136 * the pmdoc_Sh function, so don't do anything here. 1137 * (See if else blocks in pmdoc_Sh.) 1138 */ 1139 static void 1140 pmdoc_Xr(const struct mdoc_node *n, mandb_rec *rec) 1141 { 1142 } 1143 1144 static void 1145 pmdoc_Pp(const struct mdoc_node *n, mandb_rec *rec) 1146 { 1147 } 1148 1149 /* 1150 * pmdoc_Sh -- 1151 * Called when a .Sh macro is encountered and loops through its body, calling 1152 * mdoc_parse_section to append the data to the section specific buffer. 1153 * Two special macros which may occur inside the body of Sh are .Nm and .Xr and 1154 * they need special handling, thus the separate if branches for them. 1155 */ 1156 static void 1157 pmdoc_Sh(const struct mdoc_node *n, mandb_rec *rec) 1158 { 1159 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1160 return; 1161 int xr_found = 0; 1162 1163 if (n->type == MDOC_TEXT) { 1164 mdoc_parse_section(n->sec, n->string, rec); 1165 } else if (mdocs[n->tok] == pmdoc_Nm && rec->name != NULL) { 1166 /* 1167 * When encountering a .Nm macro, substitute it 1168 * with its previously cached value of the argument. 1169 */ 1170 mdoc_parse_section(n->sec, rec->name, rec); 1171 } else if (mdocs[n->tok] == pmdoc_Xr) { 1172 /* 1173 * When encountering other inline macros, 1174 * call pmdoc_macro_handler. 1175 */ 1176 pmdoc_macro_handler(n, rec, MDOC_Xr); 1177 xr_found = 1; 1178 } else if (mdocs[n->tok] == pmdoc_Pp) { 1179 pmdoc_macro_handler(n, rec, MDOC_Pp); 1180 } 1181 1182 /* 1183 * If an Xr macro was encountered then the child node has 1184 * already been explored by pmdoc_macro_handler. 1185 */ 1186 if (xr_found == 0) 1187 pmdoc_Sh(n->child, rec); 1188 pmdoc_Sh(n->next, rec); 1189 } 1190 1191 /* 1192 * mdoc_parse_section-- 1193 * Utility function for parsing sections of the mdoc type pages. 1194 * Takes two params: 1195 * 1. sec is an enum which indicates the section in which we are present 1196 * 2. string is the string which we need to append to the secbuff for this 1197 * particular section. 1198 * The function appends string to the global section buffer and returns. 1199 */ 1200 static void 1201 mdoc_parse_section(enum mdoc_sec sec, const char *string, mandb_rec *rec) 1202 { 1203 /* 1204 * If the user specified the 'l' flag, then parse and store only the 1205 * NAME section. Ignore the rest. 1206 */ 1207 if (mflags.limit) 1208 return; 1209 1210 switch (sec) { 1211 case SEC_LIBRARY: 1212 append(&rec->lib, string); 1213 break; 1214 case SEC_RETURN_VALUES: 1215 append(&rec->return_vals, string); 1216 break; 1217 case SEC_ENVIRONMENT: 1218 append(&rec->env, string); 1219 break; 1220 case SEC_FILES: 1221 append(&rec->files, string); 1222 break; 1223 case SEC_EXIT_STATUS: 1224 append(&rec->exit_status, string); 1225 break; 1226 case SEC_DIAGNOSTICS: 1227 append(&rec->diagnostics, string); 1228 break; 1229 case SEC_ERRORS: 1230 append(&rec->errors, string); 1231 break; 1232 case SEC_NAME: 1233 case SEC_SYNOPSIS: 1234 case SEC_EXAMPLES: 1235 case SEC_STANDARDS: 1236 case SEC_HISTORY: 1237 case SEC_AUTHORS: 1238 case SEC_BUGS: 1239 break; 1240 default: 1241 append(&rec->desc, string); 1242 break; 1243 } 1244 } 1245 1246 static void 1247 pman_node(const struct man_node *n, mandb_rec *rec) 1248 { 1249 if (n == NULL) 1250 return; 1251 1252 switch (n->type) { 1253 case (MAN_BODY): 1254 /* FALLTHROUGH */ 1255 case (MAN_BLOCK): 1256 /* FALLTHROUGH */ 1257 case (MAN_ELEM): 1258 if (mans[n->tok] != NULL) 1259 (*mans[n->tok])(n, rec); 1260 break; 1261 default: 1262 break; 1263 } 1264 1265 pman_node(n->child, rec); 1266 pman_node(n->next, rec); 1267 } 1268 1269 /* 1270 * pman_parse_name -- 1271 * Parses the NAME section and puts the complete content in the name_desc 1272 * variable. 1273 */ 1274 static void 1275 pman_parse_name(const struct man_node *n, mandb_rec *rec) 1276 { 1277 if (n == NULL) 1278 return; 1279 1280 if (n->type == MAN_TEXT) { 1281 char *tmp = parse_escape(n->string); 1282 concat(&rec->name_desc, tmp); 1283 free(tmp); 1284 } 1285 1286 if (n->child) 1287 pman_parse_name(n->child, rec); 1288 1289 if(n->next) 1290 pman_parse_name(n->next, rec); 1291 } 1292 1293 /* 1294 * A stub function to be able to parse the macros like .B embedded inside 1295 * a section. 1296 */ 1297 static void 1298 pman_block(const struct man_node *n, mandb_rec *rec) 1299 { 1300 } 1301 1302 /* 1303 * pman_sh -- 1304 * This function does one of the two things: 1305 * 1. If the present section is NAME, then it will: 1306 * (a) Extract the name of the page (in case of multiple comma separated 1307 * names, it will pick up the first one). 1308 * (b) Build a space spearated list of all the symlinks/hardlinks to 1309 * this page and store in the buffer 'links'. These are extracted from 1310 * the comma separated list of names in the NAME section as well. 1311 * (c) Move on to the one line description section, which is after the list 1312 * of names in the NAME section. 1313 * 2. Otherwise, it will check the section name and call the man_parse_section 1314 * function, passing the enum corresponding that section. 1315 */ 1316 static void 1317 pman_sh(const struct man_node *n, mandb_rec *rec) 1318 { 1319 static const struct { 1320 enum man_sec section; 1321 const char *header; 1322 } mapping[] = { 1323 { MANSEC_DESCRIPTION, "DESCRIPTION" }, 1324 { MANSEC_SYNOPSIS, "SYNOPSIS" }, 1325 { MANSEC_LIBRARY, "LIBRARY" }, 1326 { MANSEC_ERRORS, "ERRORS" }, 1327 { MANSEC_FILES, "FILES" }, 1328 { MANSEC_RETURN_VALUES, "RETURN VALUE" }, 1329 { MANSEC_RETURN_VALUES, "RETURN VALUES" }, 1330 { MANSEC_EXIT_STATUS, "EXIT STATUS" }, 1331 { MANSEC_EXAMPLES, "EXAMPLES" }, 1332 { MANSEC_EXAMPLES, "EXAMPLE" }, 1333 { MANSEC_STANDARDS, "STANDARDS" }, 1334 { MANSEC_HISTORY, "HISTORY" }, 1335 { MANSEC_BUGS, "BUGS" }, 1336 { MANSEC_AUTHORS, "AUTHORS" }, 1337 { MANSEC_COPYRIGHT, "COPYRIGHT" }, 1338 }; 1339 const struct man_node *head; 1340 char *name_desc; 1341 size_t sz; 1342 size_t i; 1343 1344 if ((head = n->parent->head) == NULL || (head = head->child) == NULL || 1345 head->type != MAN_TEXT) 1346 return; 1347 1348 /* 1349 * Check if this section should be extracted and 1350 * where it should be stored. Handled the trival cases first. 1351 */ 1352 for (i = 0; i < sizeof(mapping) / sizeof(mapping[0]); ++i) { 1353 if (strcmp(head->string, mapping[i].header) == 0) { 1354 man_parse_section(mapping[i].section, n, rec); 1355 return; 1356 } 1357 } 1358 1359 if (strcmp(head->string, "NAME") == 0) { 1360 /* 1361 * We are in the NAME section. 1362 * pman_parse_name will put the complete content in name_desc. 1363 */ 1364 pman_parse_name(n, rec); 1365 1366 name_desc = rec->name_desc; 1367 if (name_desc == NULL) 1368 return; 1369 1370 /* Remove any leading spaces. */ 1371 while (name_desc[0] == ' ') 1372 name_desc++; 1373 1374 /* If the line begins with a "\&", avoid those */ 1375 if (name_desc[0] == '\\' && name_desc[1] == '&') 1376 name_desc += 2; 1377 1378 /* Now name_desc should be left with a comma-space 1379 * separated list of names and the one line description 1380 * of the page: 1381 * "a, b, c \- sample description" 1382 * Take out the first name, before the first comma 1383 * (or space) and store it in rec->name. 1384 * If the page has aliases then they should be 1385 * in the form of a comma separated list. 1386 * Keep looping while there is a comma in name_desc, 1387 * extract the alias name and store in rec->links. 1388 * When there are no more commas left, break out. 1389 */ 1390 int has_alias = 0; // Any more aliases left? 1391 while (*name_desc) { 1392 /* Remove any leading spaces or hyphens. */ 1393 if (name_desc[0] == ' ' || name_desc[0] =='-') { 1394 name_desc++; 1395 continue; 1396 } 1397 sz = strcspn(name_desc, ", "); 1398 1399 /* Extract the first term and store it in rec->name. */ 1400 if (rec->name == NULL) { 1401 if (name_desc[sz] == ',') 1402 has_alias = 1; 1403 name_desc[sz] = 0; 1404 rec->name = emalloc(sz + 1); 1405 memcpy(rec->name, name_desc, sz + 1); 1406 name_desc += sz + 1; 1407 continue; 1408 } 1409 1410 /* 1411 * Once rec->name is set, rest of the names 1412 * are to be treated as links or aliases. 1413 */ 1414 if (rec->name && has_alias) { 1415 if (name_desc[sz] != ',') { 1416 /* No more commas left --> 1417 * no more aliases to take out 1418 */ 1419 has_alias = 0; 1420 } 1421 name_desc[sz] = 0; 1422 concat2(&rec->links, name_desc, sz); 1423 name_desc += sz + 1; 1424 continue; 1425 } 1426 break; 1427 } 1428 1429 /* Parse any escape sequences that might be there */ 1430 char *temp = parse_escape(name_desc); 1431 free(rec->name_desc); 1432 rec->name_desc = temp; 1433 temp = parse_escape(rec->name); 1434 free(rec->name); 1435 rec->name = temp; 1436 return; 1437 } 1438 1439 /* The RETURN VALUE section might be specified in multiple ways */ 1440 if (strcmp(head->string, "RETURN") == 0 && 1441 head->next != NULL && head->next->type == MAN_TEXT && 1442 (strcmp(head->next->string, "VALUE") == 0 || 1443 strcmp(head->next->string, "VALUES") == 0)) { 1444 man_parse_section(MANSEC_RETURN_VALUES, n, rec); 1445 return; 1446 } 1447 1448 /* 1449 * EXIT STATUS section can also be specified all on one line or on two 1450 * separate lines. 1451 */ 1452 if (strcmp(head->string, "EXIT") == 0 && 1453 head->next != NULL && head->next->type == MAN_TEXT && 1454 strcmp(head->next->string, "STATUS") == 0) { 1455 man_parse_section(MANSEC_EXIT_STATUS, n, rec); 1456 return; 1457 } 1458 1459 /* Store the rest of the content in desc. */ 1460 man_parse_section(MANSEC_NONE, n, rec); 1461 } 1462 1463 /* 1464 * pman_parse_node -- 1465 * Generic function to iterate through a node. Usually called from 1466 * man_parse_section to parse a particular section of the man page. 1467 */ 1468 static void 1469 pman_parse_node(const struct man_node *n, secbuff *s) 1470 { 1471 if (n == NULL) 1472 return; 1473 1474 if (n->type == MAN_TEXT) 1475 append(s, n->string); 1476 1477 pman_parse_node(n->child, s); 1478 pman_parse_node(n->next, s); 1479 } 1480 1481 /* 1482 * man_parse_section -- 1483 * Takes two parameters: 1484 * sec: Tells which section we are present in 1485 * n: Is the present node of the AST. 1486 * Depending on the section, we call pman_parse_node to parse that section and 1487 * concatenate the content from that section into the buffer for that section. 1488 */ 1489 static void 1490 man_parse_section(enum man_sec sec, const struct man_node *n, mandb_rec *rec) 1491 { 1492 /* 1493 * If the user sepecified the 'l' flag then just parse 1494 * the NAME section, ignore the rest. 1495 */ 1496 if (mflags.limit) 1497 return; 1498 1499 switch (sec) { 1500 case MANSEC_LIBRARY: 1501 pman_parse_node(n, &rec->lib); 1502 break; 1503 case MANSEC_RETURN_VALUES: 1504 pman_parse_node(n, &rec->return_vals); 1505 break; 1506 case MANSEC_ENVIRONMENT: 1507 pman_parse_node(n, &rec->env); 1508 break; 1509 case MANSEC_FILES: 1510 pman_parse_node(n, &rec->files); 1511 break; 1512 case MANSEC_EXIT_STATUS: 1513 pman_parse_node(n, &rec->exit_status); 1514 break; 1515 case MANSEC_DIAGNOSTICS: 1516 pman_parse_node(n, &rec->diagnostics); 1517 break; 1518 case MANSEC_ERRORS: 1519 pman_parse_node(n, &rec->errors); 1520 break; 1521 case MANSEC_NAME: 1522 case MANSEC_SYNOPSIS: 1523 case MANSEC_EXAMPLES: 1524 case MANSEC_STANDARDS: 1525 case MANSEC_HISTORY: 1526 case MANSEC_BUGS: 1527 case MANSEC_AUTHORS: 1528 case MANSEC_COPYRIGHT: 1529 break; 1530 default: 1531 pman_parse_node(n, &rec->desc); 1532 break; 1533 } 1534 1535 } 1536 1537 /* 1538 * insert_into_db -- 1539 * Inserts the parsed data of the man page in the Sqlite databse. 1540 * If any of the values is NULL, then we cleanup and return -1 indicating 1541 * an error. 1542 * Otherwise, store the data in the database and return 0. 1543 */ 1544 static int 1545 insert_into_db(sqlite3 *db, mandb_rec *rec) 1546 { 1547 int rc = 0; 1548 int idx = -1; 1549 const char *sqlstr = NULL; 1550 sqlite3_stmt *stmt = NULL; 1551 char *ln = NULL; 1552 char *errmsg = NULL; 1553 long int mandb_rowid; 1554 1555 /* 1556 * At the very minimum we want to make sure that we store 1557 * the following data: 1558 * Name, one line description, and the MD5 hash 1559 */ 1560 if (rec->name == NULL || rec->name_desc == NULL || 1561 rec->md5_hash == NULL) { 1562 cleanup(rec); 1563 return -1; 1564 } 1565 1566 /* Write null byte at the end of all the sec_buffs */ 1567 rec->desc.data[rec->desc.offset] = 0; 1568 rec->lib.data[rec->lib.offset] = 0; 1569 rec->env.data[rec->env.offset] = 0; 1570 rec->return_vals.data[rec->return_vals.offset] = 0; 1571 rec->exit_status.data[rec->exit_status.offset] = 0; 1572 rec->files.data[rec->files.offset] = 0; 1573 rec->diagnostics.data[rec->diagnostics.offset] = 0; 1574 rec->errors.data[rec->errors.offset] = 0; 1575 1576 /* 1577 * In case of a mdoc page: (sorry, no better place to put this code) 1578 * parse the comma separated list of names of man pages, 1579 * the first name will be stored in the mandb table, rest will be 1580 * treated as links and put in the mandb_links table. 1581 */ 1582 if (rec->page_type == MDOC) { 1583 char *tmp; 1584 rec->links = estrdup(rec->name); 1585 free(rec->name); 1586 int sz = strcspn(rec->links, " \0"); 1587 rec->name = emalloc(sz + 1); 1588 memcpy(rec->name, rec->links, sz); 1589 if(rec->name[sz - 1] == ',') 1590 rec->name[sz - 1] = 0; 1591 else 1592 rec->name[sz] = 0; 1593 while (rec->links[sz] == ' ') 1594 ++sz; 1595 tmp = estrdup(rec->links + sz); 1596 free(rec->links); 1597 rec->links = tmp; 1598 } 1599 1600 /*------------------------ Populate the mandb table---------------------------*/ 1601 sqlstr = "INSERT INTO mandb VALUES (:section, :name, :name_desc, :desc," 1602 " :lib, :return_vals, :env, :files, :exit_status," 1603 " :diagnostics, :errors, :md5_hash, :machine)"; 1604 1605 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1606 if (rc != SQLITE_OK) 1607 goto Out; 1608 1609 idx = sqlite3_bind_parameter_index(stmt, ":name"); 1610 rc = sqlite3_bind_text(stmt, idx, rec->name, -1, NULL); 1611 if (rc != SQLITE_OK) { 1612 sqlite3_finalize(stmt); 1613 goto Out; 1614 } 1615 1616 idx = sqlite3_bind_parameter_index(stmt, ":section"); 1617 rc = sqlite3_bind_text(stmt, idx, rec->section, -1, NULL); 1618 if (rc != SQLITE_OK) { 1619 sqlite3_finalize(stmt); 1620 goto Out; 1621 } 1622 1623 idx = sqlite3_bind_parameter_index(stmt, ":name_desc"); 1624 rc = sqlite3_bind_text(stmt, idx, rec->name_desc, -1, NULL); 1625 if (rc != SQLITE_OK) { 1626 sqlite3_finalize(stmt); 1627 goto Out; 1628 } 1629 1630 idx = sqlite3_bind_parameter_index(stmt, ":desc"); 1631 rc = sqlite3_bind_text(stmt, idx, rec->desc.data, 1632 rec->desc.offset + 1, NULL); 1633 if (rc != SQLITE_OK) { 1634 sqlite3_finalize(stmt); 1635 goto Out; 1636 } 1637 1638 idx = sqlite3_bind_parameter_index(stmt, ":lib"); 1639 rc = sqlite3_bind_text(stmt, idx, rec->lib.data, rec->lib.offset + 1, NULL); 1640 if (rc != SQLITE_OK) { 1641 sqlite3_finalize(stmt); 1642 goto Out; 1643 } 1644 1645 idx = sqlite3_bind_parameter_index(stmt, ":return_vals"); 1646 rc = sqlite3_bind_text(stmt, idx, rec->return_vals.data, 1647 rec->return_vals.offset + 1, NULL); 1648 if (rc != SQLITE_OK) { 1649 sqlite3_finalize(stmt); 1650 goto Out; 1651 } 1652 1653 idx = sqlite3_bind_parameter_index(stmt, ":env"); 1654 rc = sqlite3_bind_text(stmt, idx, rec->env.data, rec->env.offset + 1, NULL); 1655 if (rc != SQLITE_OK) { 1656 sqlite3_finalize(stmt); 1657 goto Out; 1658 } 1659 1660 idx = sqlite3_bind_parameter_index(stmt, ":files"); 1661 rc = sqlite3_bind_text(stmt, idx, rec->files.data, 1662 rec->files.offset + 1, NULL); 1663 if (rc != SQLITE_OK) { 1664 sqlite3_finalize(stmt); 1665 goto Out; 1666 } 1667 1668 idx = sqlite3_bind_parameter_index(stmt, ":exit_status"); 1669 rc = sqlite3_bind_text(stmt, idx, rec->exit_status.data, 1670 rec->exit_status.offset + 1, NULL); 1671 if (rc != SQLITE_OK) { 1672 sqlite3_finalize(stmt); 1673 goto Out; 1674 } 1675 1676 idx = sqlite3_bind_parameter_index(stmt, ":diagnostics"); 1677 rc = sqlite3_bind_text(stmt, idx, rec->diagnostics.data, 1678 rec->diagnostics.offset + 1, NULL); 1679 if (rc != SQLITE_OK) { 1680 sqlite3_finalize(stmt); 1681 goto Out; 1682 } 1683 1684 idx = sqlite3_bind_parameter_index(stmt, ":errors"); 1685 rc = sqlite3_bind_text(stmt, idx, rec->errors.data, 1686 rec->errors.offset + 1, NULL); 1687 if (rc != SQLITE_OK) { 1688 sqlite3_finalize(stmt); 1689 goto Out; 1690 } 1691 1692 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1693 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1694 if (rc != SQLITE_OK) { 1695 sqlite3_finalize(stmt); 1696 goto Out; 1697 } 1698 1699 idx = sqlite3_bind_parameter_index(stmt, ":machine"); 1700 if (rec->machine) 1701 rc = sqlite3_bind_text(stmt, idx, rec->machine, -1, NULL); 1702 else 1703 rc = sqlite3_bind_null(stmt, idx); 1704 if (rc != SQLITE_OK) { 1705 sqlite3_finalize(stmt); 1706 goto Out; 1707 } 1708 1709 rc = sqlite3_step(stmt); 1710 if (rc != SQLITE_DONE) { 1711 sqlite3_finalize(stmt); 1712 goto Out; 1713 } 1714 1715 sqlite3_finalize(stmt); 1716 1717 /* Get the row id of the last inserted row */ 1718 mandb_rowid = sqlite3_last_insert_rowid(db); 1719 1720 /*------------------------Populate the mandb_meta table-----------------------*/ 1721 sqlstr = "INSERT INTO mandb_meta VALUES (:device, :inode, :mtime," 1722 " :file, :md5_hash, :id)"; 1723 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1724 if (rc != SQLITE_OK) 1725 goto Out; 1726 1727 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1728 rc = sqlite3_bind_int64(stmt, idx, rec->device); 1729 if (rc != SQLITE_OK) { 1730 sqlite3_finalize(stmt); 1731 goto Out; 1732 } 1733 1734 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1735 rc = sqlite3_bind_int64(stmt, idx, rec->inode); 1736 if (rc != SQLITE_OK) { 1737 sqlite3_finalize(stmt); 1738 goto Out; 1739 } 1740 1741 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1742 rc = sqlite3_bind_int64(stmt, idx, rec->mtime); 1743 if (rc != SQLITE_OK) { 1744 sqlite3_finalize(stmt); 1745 goto Out; 1746 } 1747 1748 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1749 rc = sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1750 if (rc != SQLITE_OK) { 1751 sqlite3_finalize(stmt); 1752 goto Out; 1753 } 1754 1755 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1756 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1757 if (rc != SQLITE_OK) { 1758 sqlite3_finalize(stmt); 1759 goto Out; 1760 } 1761 1762 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1763 rc = sqlite3_bind_int64(stmt, idx, mandb_rowid); 1764 if (rc != SQLITE_OK) { 1765 sqlite3_finalize(stmt); 1766 goto Out; 1767 } 1768 1769 rc = sqlite3_step(stmt); 1770 sqlite3_finalize(stmt); 1771 if (rc == SQLITE_CONSTRAINT_UNIQUE) { 1772 /* The *most* probable reason for reaching here is that 1773 * the UNIQUE contraint on the file column of the mandb_meta 1774 * table was violated. 1775 * This can happen when a file was updated/modified. 1776 * To fix this we need to do two things: 1777 * 1. Delete the row for the older version of this file 1778 * from mandb table. 1779 * 2. Run an UPDATE query to update the row for this file 1780 * in the mandb_meta table. 1781 */ 1782 warnx("Trying to update index for %s", rec->file_path); 1783 char *sql = sqlite3_mprintf("DELETE FROM mandb " 1784 "WHERE rowid = (SELECT id" 1785 " FROM mandb_meta" 1786 " WHERE file = %Q)", 1787 rec->file_path); 1788 sqlite3_exec(db, sql, NULL, NULL, &errmsg); 1789 sqlite3_free(sql); 1790 if (errmsg != NULL) { 1791 if (mflags.verbosity) 1792 warnx("%s", errmsg); 1793 free(errmsg); 1794 } 1795 sqlstr = "UPDATE mandb_meta SET device = :device," 1796 " inode = :inode, mtime = :mtime, id = :id," 1797 " md5_hash = :md5 WHERE file = :file"; 1798 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1799 if (rc != SQLITE_OK) { 1800 if (mflags.verbosity) 1801 warnx("Update failed with error: %s", 1802 sqlite3_errmsg(db)); 1803 close_db(db); 1804 cleanup(rec); 1805 errx(EXIT_FAILURE, 1806 "Consider running makemandb with -f option"); 1807 } 1808 1809 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1810 sqlite3_bind_int64(stmt, idx, rec->device); 1811 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1812 sqlite3_bind_int64(stmt, idx, rec->inode); 1813 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1814 sqlite3_bind_int64(stmt, idx, rec->mtime); 1815 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1816 sqlite3_bind_int64(stmt, idx, mandb_rowid); 1817 idx = sqlite3_bind_parameter_index(stmt, ":md5"); 1818 sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1819 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1820 sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1821 rc = sqlite3_step(stmt); 1822 sqlite3_finalize(stmt); 1823 1824 if (rc != SQLITE_DONE) { 1825 if (mflags.verbosity) 1826 warnx("%s", sqlite3_errmsg(db)); 1827 close_db(db); 1828 cleanup(rec); 1829 errx(EXIT_FAILURE, 1830 "Consider running makemandb with -f option"); 1831 } 1832 } else if (rc != SQLITE_DONE) { 1833 /* Otherwise make this error fatal */ 1834 warnx("Failed at %s\n%s", rec->file_path, sqlite3_errmsg(db)); 1835 cleanup(rec); 1836 close_db(db); 1837 exit(EXIT_FAILURE); 1838 } 1839 1840 /*------------------------ Populate the mandb_links table---------------------*/ 1841 char *str = NULL; 1842 char *links; 1843 if (rec->links && strlen(rec->links)) { 1844 links = rec->links; 1845 for(ln = strtok(links, " "); ln; ln = strtok(NULL, " ")) { 1846 if (ln[0] == ',') 1847 ln++; 1848 if(ln[strlen(ln) - 1] == ',') 1849 ln[strlen(ln) - 1] = 0; 1850 1851 str = sqlite3_mprintf("INSERT INTO mandb_links" 1852 " VALUES (%Q, %Q, %Q, %Q, %Q)", 1853 ln, rec->name, rec->section, 1854 rec->machine, rec->md5_hash); 1855 sqlite3_exec(db, str, NULL, NULL, &errmsg); 1856 sqlite3_free(str); 1857 if (errmsg != NULL) { 1858 warnx("%s", errmsg); 1859 cleanup(rec); 1860 free(errmsg); 1861 return -1; 1862 } 1863 } 1864 } 1865 1866 cleanup(rec); 1867 return 0; 1868 1869 Out: 1870 if (mflags.verbosity) 1871 warnx("%s", sqlite3_errmsg(db)); 1872 cleanup(rec); 1873 return -1; 1874 } 1875 1876 /* 1877 * check_md5-- 1878 * Generates the md5 hash of the file and checks if it already doesn't exist 1879 * in the table (passed as the 3rd parameter). 1880 * This function is being used to avoid hardlinks. 1881 * On successful completion it will also set the value of the fourth parameter 1882 * to the md5 hash of the file (computed previously). It is the responsibility 1883 * of the caller to free this buffer. 1884 * Return values: 1885 * -1: If an error occurs somewhere and sets the md5 return buffer to NULL. 1886 * 0: If the md5 hash does not exist in the table. 1887 * 1: If the hash exists in the database. 1888 */ 1889 static int 1890 check_md5(const char *file, sqlite3 *db, const char *table, char **md5sum, 1891 void *buf, size_t buflen) 1892 { 1893 int rc = 0; 1894 int idx = -1; 1895 char *sqlstr = NULL; 1896 sqlite3_stmt *stmt = NULL; 1897 1898 assert(file != NULL); 1899 *md5sum = MD5Data(buf, buflen, NULL); 1900 if (*md5sum == NULL) { 1901 if (mflags.verbosity) 1902 warn("md5 failed: %s", file); 1903 return -1; 1904 } 1905 1906 easprintf(&sqlstr, "SELECT * FROM %s WHERE md5_hash = :md5_hash", 1907 table); 1908 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1909 if (rc != SQLITE_OK) { 1910 free(sqlstr); 1911 free(*md5sum); 1912 *md5sum = NULL; 1913 return -1; 1914 } 1915 1916 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1917 rc = sqlite3_bind_text(stmt, idx, *md5sum, -1, NULL); 1918 if (rc != SQLITE_OK) { 1919 if (mflags.verbosity) 1920 warnx("%s", sqlite3_errmsg(db)); 1921 sqlite3_finalize(stmt); 1922 free(sqlstr); 1923 free(*md5sum); 1924 *md5sum = NULL; 1925 return -1; 1926 } 1927 1928 if (sqlite3_step(stmt) == SQLITE_ROW) { 1929 sqlite3_finalize(stmt); 1930 free(sqlstr); 1931 return 0; 1932 } 1933 1934 sqlite3_finalize(stmt); 1935 free(sqlstr); 1936 return 1; 1937 } 1938 1939 /* Optimize the index for faster search */ 1940 static void 1941 optimize(sqlite3 *db) 1942 { 1943 const char *sqlstr; 1944 char *errmsg = NULL; 1945 1946 if (mflags.verbosity == 2) 1947 printf("Optimizing the database index\n"); 1948 sqlstr = "INSERT INTO mandb(mandb) VALUES (\'optimize\');" 1949 "VACUUM"; 1950 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 1951 if (errmsg != NULL) { 1952 if (mflags.verbosity) 1953 warnx("%s", errmsg); 1954 free(errmsg); 1955 return; 1956 } 1957 } 1958 1959 /* 1960 * cleanup -- 1961 * cleans up the global buffers 1962 */ 1963 static void 1964 cleanup(mandb_rec *rec) 1965 { 1966 rec->desc.offset = 0; 1967 rec->lib.offset = 0; 1968 rec->return_vals.offset = 0; 1969 rec->env.offset = 0; 1970 rec->exit_status.offset = 0; 1971 rec->diagnostics.offset = 0; 1972 rec->errors.offset = 0; 1973 rec->files.offset = 0; 1974 1975 free(rec->machine); 1976 rec->machine = NULL; 1977 1978 free(rec->links); 1979 rec->links = NULL; 1980 1981 free(rec->file_path); 1982 rec->file_path = NULL; 1983 1984 free(rec->name); 1985 rec->name = NULL; 1986 1987 free(rec->name_desc); 1988 rec->name_desc = NULL; 1989 1990 free(rec->md5_hash); 1991 rec->md5_hash = NULL; 1992 } 1993 1994 /* 1995 * init_secbuffs-- 1996 * Sets the value of buflen for all the sec_buff field of rec. And then 1997 * allocate memory to each sec_buff member of rec. 1998 */ 1999 static void 2000 init_secbuffs(mandb_rec *rec) 2001 { 2002 /* 2003 * Some sec_buff might need more memory, for example desc, 2004 * which stores the data of the DESCRIPTION section, 2005 * while some might need very small amount of memory. 2006 * Therefore explicitly setting the value of buflen field for 2007 * each sec_buff. 2008 */ 2009 rec->desc.buflen = 10 * BUFLEN; 2010 rec->desc.data = emalloc(rec->desc.buflen); 2011 rec->desc.offset = 0; 2012 2013 rec->lib.buflen = BUFLEN / 2; 2014 rec->lib.data = emalloc(rec->lib.buflen); 2015 rec->lib.offset = 0; 2016 2017 rec->return_vals.buflen = BUFLEN; 2018 rec->return_vals.data = emalloc(rec->return_vals.buflen); 2019 rec->return_vals.offset = 0; 2020 2021 rec->exit_status.buflen = BUFLEN; 2022 rec->exit_status.data = emalloc(rec->exit_status.buflen); 2023 rec->exit_status.offset = 0; 2024 2025 rec->env.buflen = BUFLEN; 2026 rec->env.data = emalloc(rec->env.buflen); 2027 rec->env.offset = 0; 2028 2029 rec->files.buflen = BUFLEN; 2030 rec->files.data = emalloc(rec->files.buflen); 2031 rec->files.offset = 0; 2032 2033 rec->diagnostics.buflen = BUFLEN; 2034 rec->diagnostics.data = emalloc(rec->diagnostics.buflen); 2035 rec->diagnostics.offset = 0; 2036 2037 rec->errors.buflen = BUFLEN; 2038 rec->errors.data = emalloc(rec->errors.buflen); 2039 rec->errors.offset = 0; 2040 } 2041 2042 /* 2043 * free_secbuffs-- 2044 * This function should be called at the end, when all the pages have been 2045 * parsed. 2046 * It frees the memory allocated to sec_buffs by init_secbuffs in the starting. 2047 */ 2048 static void 2049 free_secbuffs(mandb_rec *rec) 2050 { 2051 free(rec->desc.data); 2052 free(rec->lib.data); 2053 free(rec->return_vals.data); 2054 free(rec->exit_status.data); 2055 free(rec->env.data); 2056 free(rec->files.data); 2057 free(rec->diagnostics.data); 2058 free(rec->errors.data); 2059 } 2060 2061 static void 2062 replace_hyph(char *str) 2063 { 2064 char *iter = str; 2065 while ((iter = strchr(iter, ASCII_HYPH)) != NULL) 2066 *iter = '-'; 2067 2068 iter = str; 2069 while ((iter = strchr(iter, ASCII_NBRSP)) != NULL) 2070 *iter = '-'; 2071 } 2072 2073 static char * 2074 parse_escape(const char *str) 2075 { 2076 const char *backslash, *last_backslash; 2077 char *result, *iter; 2078 size_t len; 2079 2080 assert(str); 2081 2082 last_backslash = str; 2083 backslash = strchr(str, '\\'); 2084 if (backslash == NULL) { 2085 result = estrdup(str); 2086 replace_hyph(result); 2087 return result; 2088 } 2089 2090 result = emalloc(strlen(str) + 1); 2091 iter = result; 2092 2093 do { 2094 len = backslash - last_backslash; 2095 memcpy(iter, last_backslash, len); 2096 iter += len; 2097 if (backslash[1] == '-' || backslash[1] == ' ') { 2098 *iter++ = backslash[1]; 2099 last_backslash = backslash + 2; 2100 backslash = strchr(backslash + 2, '\\'); 2101 } else { 2102 ++backslash; 2103 mandoc_escape(&backslash, NULL, NULL); 2104 last_backslash = backslash; 2105 if (backslash == NULL) 2106 break; 2107 backslash = strchr(last_backslash, '\\'); 2108 } 2109 } while (backslash != NULL); 2110 if (last_backslash != NULL) 2111 strcpy(iter, last_backslash); 2112 2113 replace_hyph(result); 2114 return result; 2115 } 2116 2117 /* 2118 * append-- 2119 * Concatenates a space and src at the end of sbuff->data (much like concat in 2120 * apropos-utils.c). 2121 * Rather than reallocating space for writing data, it uses the value of the 2122 * offset field of sec_buff to write new data at the free space left in the 2123 * buffer. 2124 * In case the size of the data to be appended exceeds the number of bytes left 2125 * in the buffer, it reallocates buflen number of bytes and then continues. 2126 * Value of offset field should be adjusted as new data is written. 2127 * 2128 * NOTE: This function does not write the null byte at the end of the buffers, 2129 * write a null byte at the position pointed to by offset before inserting data 2130 * in the db. 2131 */ 2132 static void 2133 append(secbuff *sbuff, const char *src) 2134 { 2135 short flag = 0; 2136 size_t srclen, newlen; 2137 char *temp; 2138 2139 assert(src != NULL); 2140 temp = parse_escape(src); 2141 srclen = strlen(temp); 2142 2143 if (sbuff->data == NULL) { 2144 sbuff->data = emalloc(sbuff->buflen); 2145 sbuff->offset = 0; 2146 } 2147 2148 newlen = sbuff->offset + srclen + 2; 2149 if (newlen >= sbuff->buflen) { 2150 while (sbuff->buflen < newlen) 2151 sbuff->buflen += sbuff->buflen; 2152 sbuff->data = erealloc(sbuff->data, sbuff->buflen); 2153 flag = 1; 2154 } 2155 2156 /* Append a space at the end of the buffer. */ 2157 if (sbuff->offset || flag) 2158 sbuff->data[sbuff->offset++] = ' '; 2159 /* Now, copy src at the end of the buffer. */ 2160 memcpy(sbuff->data + sbuff->offset, temp, srclen); 2161 sbuff->offset += srclen; 2162 free(temp); 2163 } 2164 2165 static void 2166 usage(void) 2167 { 2168 fprintf(stderr, "Usage: %s [-floQqv] [-C path]\n", getprogname()); 2169 exit(1); 2170 } 2171
Indexes created Mon Sep 29 13:09:55 GMT 2025