makemandb.c revision 1.34
1 /* $NetBSD: makemandb.c,v 1.34 2016/04/13 01:32:00 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 1.34 2016/04/13 01:32:00 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_close(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_close(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_close(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 (mflags.verbosity == 2) 847 printf("Parsing: %s\n", file); 848 rec->md5_hash = md5sum; 849 rec->file_path = estrdup(file); 850 // file_path is freed by insert_into_db itself. 851 chdir(parent); 852 begin_parse(file, mp, rec, buf, buflen); 853 if (insert_into_db(db, rec) < 0) { 854 if (mflags.verbosity) 855 warnx("Error in indexing %s", file); 856 err_count++; 857 } else { 858 new_count++; 859 } 860 } 861 } 862 863 if (mflags.verbosity == 2) { 864 printf("Total Number of new or updated pages encountered = %d\n" 865 "Total number of (hard or symbolic) links found = %d\n" 866 "Total number of pages that were successfully" 867 " indexed/updated = %d\n" 868 "Total number of pages that could not be indexed" 869 " due to errors = %d\n", 870 total_count - link_count, link_count, new_count, err_count); 871 } 872 873 if (mflags.recreate) 874 return; 875 876 if (mflags.verbosity == 2) 877 printf("Deleting stale index entries\n"); 878 879 sqlstr = "DELETE FROM mandb_meta WHERE file NOT IN" 880 " (SELECT file FROM metadb.file_cache);" 881 "DELETE FROM mandb_links WHERE md5_hash NOT IN" 882 " (SELECT md5_hash from mandb_meta);" 883 "DROP TABLE metadb.file_cache;" 884 "DELETE FROM mandb WHERE rowid NOT IN" 885 " (SELECT id FROM mandb_meta);"; 886 887 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 888 if (errmsg != NULL) { 889 warnx("Removing old entries failed: %s", errmsg); 890 warnx("Please rebuild database from scratch with -f."); 891 free(errmsg); 892 return; 893 } 894 } 895 896 /* 897 * begin_parse -- 898 * parses the man page using libmandoc 899 */ 900 static void 901 begin_parse(const char *file, struct mparse *mp, mandb_rec *rec, 902 const void *buf, size_t len) 903 { 904 struct mdoc *mdoc; 905 struct man *man; 906 mparse_reset(mp); 907 908 rec->xr_found = 0; 909 910 if (mparse_readmem(mp, buf, len, file) >= MANDOCLEVEL_BADARG) { 911 /* Printing this warning at verbosity level 2 912 * because some packages from pkgsrc might trigger several 913 * of such warnings. 914 */ 915 if (mflags.verbosity == 2) 916 warnx("%s: Parse failure", file); 917 return; 918 } 919 920 mparse_result(mp, &mdoc, &man, NULL); 921 if (mdoc == NULL && man == NULL) { 922 if (mflags.verbosity == 2) 923 warnx("Not a man(7) or mdoc(7) page"); 924 return; 925 } 926 927 set_machine(mdoc, rec); 928 set_section(mdoc, man, rec); 929 if (mdoc) { 930 rec->page_type = MDOC; 931 pmdoc_node(mdoc_node(mdoc), rec); 932 } else { 933 rec->page_type = MAN; 934 pman_node(man_node(man), rec); 935 } 936 } 937 938 /* 939 * set_section -- 940 * Extracts the section number and normalizes it to only the numeric part 941 * (Which should be the first character of the string). 942 */ 943 static void 944 set_section(const struct mdoc *md, const struct man *m, mandb_rec *rec) 945 { 946 if (md) { 947 const struct mdoc_meta *md_meta = mdoc_meta(md); 948 if (md_meta->msec == NULL) { 949 rec->section[0] = '?'; 950 } else 951 rec->section[0] = md_meta->msec[0]; 952 } else if (m) { 953 const struct man_meta *m_meta = man_meta(m); 954 if (m_meta->msec == NULL) 955 rec->section[0] = '?'; 956 else 957 rec->section[0] = m_meta->msec[0]; 958 } else 959 return; 960 961 if (rec->section[0] == '?' && mflags.verbosity == 2) 962 warnx("%s: Missing section number", rec->file_path); 963 } 964 965 /* 966 * get_machine -- 967 * Extracts the machine architecture information if available. 968 */ 969 static void 970 set_machine(const struct mdoc *md, mandb_rec *rec) 971 { 972 if (md == NULL) 973 return; 974 const struct mdoc_meta *md_meta = mdoc_meta(md); 975 if (md_meta->arch) 976 rec->machine = estrdup(md_meta->arch); 977 } 978 979 static void 980 pmdoc_node(const struct mdoc_node *n, mandb_rec *rec) 981 { 982 983 if (n == NULL) 984 return; 985 986 switch (n->type) { 987 case (MDOC_BODY): 988 /* FALLTHROUGH */ 989 case (MDOC_TAIL): 990 /* FALLTHROUGH */ 991 case (MDOC_ELEM): 992 if (mdocs[n->tok] == NULL) 993 break; 994 (*mdocs[n->tok])(n, rec); 995 break; 996 default: 997 break; 998 } 999 1000 pmdoc_node(n->child, rec); 1001 pmdoc_node(n->next, rec); 1002 } 1003 1004 /* 1005 * pmdoc_Nm -- 1006 * Extracts the Name of the manual page from the .Nm macro 1007 */ 1008 static void 1009 pmdoc_Nm(const struct mdoc_node *n, mandb_rec *rec) 1010 { 1011 if (n->sec != SEC_NAME) 1012 return; 1013 1014 for (n = n->child; n; n = n->next) { 1015 if (n->type == MDOC_TEXT) { 1016 char *escaped_name = parse_escape(n->string); 1017 concat(&rec->name, escaped_name); 1018 free(escaped_name); 1019 } 1020 } 1021 } 1022 1023 /* 1024 * pmdoc_Nd -- 1025 * Extracts the one line description of the man page from the .Nd macro 1026 */ 1027 static void 1028 pmdoc_Nd(const struct mdoc_node *n, mandb_rec *rec) 1029 { 1030 char *buf = NULL; 1031 char *name; 1032 char *nd_text; 1033 1034 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1035 return; 1036 1037 if (n->type == MDOC_TEXT) { 1038 if (rec->xr_found && n->next) { 1039 /* 1040 * An Xr macro was seen previously, so parse this 1041 * and the next node, as "Name(Section)". 1042 */ 1043 name = n->string; 1044 n = n->next; 1045 assert(n->type == MDOC_TEXT); 1046 easprintf(&buf, "%s(%s)", name, n->string); 1047 concat(&rec->name_desc, buf); 1048 free(buf); 1049 } else { 1050 nd_text = parse_escape(n->string); 1051 concat(&rec->name_desc, nd_text); 1052 free(nd_text); 1053 } 1054 rec->xr_found = 0; 1055 } else if (mdocs[n->tok] == pmdoc_Xr) { 1056 /* Remember that we have encountered an Xr macro */ 1057 rec->xr_found = 1; 1058 } 1059 1060 if (n->child) 1061 pmdoc_Nd(n->child, rec); 1062 1063 if(n->next) 1064 pmdoc_Nd(n->next, rec); 1065 } 1066 1067 /* 1068 * pmdoc_macro_handler-- 1069 * This function is a single point of handling all the special macros that we 1070 * want to handle especially. For example the .Xr macro for properly parsing 1071 * the referenced page name along with the section number, or the .Pp macro 1072 * for adding a new line whenever we encounter it. 1073 */ 1074 static void 1075 pmdoc_macro_handler(const struct mdoc_node *n, mandb_rec *rec, enum mdoct doct) 1076 { 1077 const struct mdoc_node *sn; 1078 assert(n); 1079 1080 switch (doct) { 1081 /* Parse the man page references. 1082 * Basically the .Xr macros are used like: 1083 * .Xr ls 1 1084 * and formatted like this: 1085 * ls(1) 1086 * Prepare a buffer to format the data like the above example and call 1087 * pmdoc_parse_section to append it. 1088 */ 1089 case MDOC_Xr: 1090 n = n->child; 1091 while (n->type != MDOC_TEXT && n->next) 1092 n = n->next; 1093 1094 if (n && n->type != MDOC_TEXT) 1095 return; 1096 sn = n; 1097 if (n->next) 1098 n = n->next; 1099 1100 while (n->type != MDOC_TEXT && n->next) 1101 n = n->next; 1102 1103 if (n && n->type == MDOC_TEXT) { 1104 char *buf; 1105 easprintf(&buf, "%s(%s)", sn->string, n->string); 1106 mdoc_parse_section(n->sec, buf, rec); 1107 free(buf); 1108 } 1109 1110 break; 1111 1112 /* Parse the .Pp macro to add a new line */ 1113 case MDOC_Pp: 1114 if (n->type == MDOC_TEXT) 1115 mdoc_parse_section(n->sec, "\n", rec); 1116 break; 1117 default: 1118 break; 1119 } 1120 1121 } 1122 1123 /* 1124 * pmdoc_Xr, pmdoc_Pp-- 1125 * Empty stubs. 1126 * The parser calls these functions each time it encounters 1127 * a .Xr or .Pp macro. We are parsing all the data from 1128 * the pmdoc_Sh function, so don't do anything here. 1129 * (See if else blocks in pmdoc_Sh.) 1130 */ 1131 static void 1132 pmdoc_Xr(const struct mdoc_node *n, mandb_rec *rec) 1133 { 1134 } 1135 1136 static void 1137 pmdoc_Pp(const struct mdoc_node *n, mandb_rec *rec) 1138 { 1139 } 1140 1141 /* 1142 * pmdoc_Sh -- 1143 * Called when a .Sh macro is encountered and loops through its body, calling 1144 * mdoc_parse_section to append the data to the section specific buffer. 1145 * Two special macros which may occur inside the body of Sh are .Nm and .Xr and 1146 * they need special handling, thus the separate if branches for them. 1147 */ 1148 static void 1149 pmdoc_Sh(const struct mdoc_node *n, mandb_rec *rec) 1150 { 1151 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1152 return; 1153 int xr_found = 0; 1154 1155 if (n->type == MDOC_TEXT) { 1156 mdoc_parse_section(n->sec, n->string, rec); 1157 } else if (mdocs[n->tok] == pmdoc_Nm && rec->name != NULL) { 1158 /* 1159 * When encountering a .Nm macro, substitute it 1160 * with its previously cached value of the argument. 1161 */ 1162 mdoc_parse_section(n->sec, rec->name, rec); 1163 } else if (mdocs[n->tok] == pmdoc_Xr) { 1164 /* 1165 * When encountering other inline macros, 1166 * call pmdoc_macro_handler. 1167 */ 1168 pmdoc_macro_handler(n, rec, MDOC_Xr); 1169 xr_found = 1; 1170 } else if (mdocs[n->tok] == pmdoc_Pp) { 1171 pmdoc_macro_handler(n, rec, MDOC_Pp); 1172 } 1173 1174 /* 1175 * If an Xr macro was encountered then the child node has 1176 * already been explored by pmdoc_macro_handler. 1177 */ 1178 if (xr_found == 0) 1179 pmdoc_Sh(n->child, rec); 1180 pmdoc_Sh(n->next, rec); 1181 } 1182 1183 /* 1184 * mdoc_parse_section-- 1185 * Utility function for parsing sections of the mdoc type pages. 1186 * Takes two params: 1187 * 1. sec is an enum which indicates the section in which we are present 1188 * 2. string is the string which we need to append to the secbuff for this 1189 * particular section. 1190 * The function appends string to the global section buffer and returns. 1191 */ 1192 static void 1193 mdoc_parse_section(enum mdoc_sec sec, const char *string, mandb_rec *rec) 1194 { 1195 /* 1196 * If the user specified the 'l' flag, then parse and store only the 1197 * NAME section. Ignore the rest. 1198 */ 1199 if (mflags.limit) 1200 return; 1201 1202 switch (sec) { 1203 case SEC_LIBRARY: 1204 append(&rec->lib, string); 1205 break; 1206 case SEC_RETURN_VALUES: 1207 append(&rec->return_vals, string); 1208 break; 1209 case SEC_ENVIRONMENT: 1210 append(&rec->env, string); 1211 break; 1212 case SEC_FILES: 1213 append(&rec->files, string); 1214 break; 1215 case SEC_EXIT_STATUS: 1216 append(&rec->exit_status, string); 1217 break; 1218 case SEC_DIAGNOSTICS: 1219 append(&rec->diagnostics, string); 1220 break; 1221 case SEC_ERRORS: 1222 append(&rec->errors, string); 1223 break; 1224 case SEC_NAME: 1225 case SEC_SYNOPSIS: 1226 case SEC_EXAMPLES: 1227 case SEC_STANDARDS: 1228 case SEC_HISTORY: 1229 case SEC_AUTHORS: 1230 case SEC_BUGS: 1231 break; 1232 default: 1233 append(&rec->desc, string); 1234 break; 1235 } 1236 } 1237 1238 static void 1239 pman_node(const struct man_node *n, mandb_rec *rec) 1240 { 1241 if (n == NULL) 1242 return; 1243 1244 switch (n->type) { 1245 case (MAN_BODY): 1246 /* FALLTHROUGH */ 1247 case (MAN_BLOCK): 1248 /* FALLTHROUGH */ 1249 case (MAN_ELEM): 1250 if (mans[n->tok] != NULL) 1251 (*mans[n->tok])(n, rec); 1252 break; 1253 default: 1254 break; 1255 } 1256 1257 pman_node(n->child, rec); 1258 pman_node(n->next, rec); 1259 } 1260 1261 /* 1262 * pman_parse_name -- 1263 * Parses the NAME section and puts the complete content in the name_desc 1264 * variable. 1265 */ 1266 static void 1267 pman_parse_name(const struct man_node *n, mandb_rec *rec) 1268 { 1269 if (n == NULL) 1270 return; 1271 1272 if (n->type == MAN_TEXT) { 1273 char *tmp = parse_escape(n->string); 1274 concat(&rec->name_desc, tmp); 1275 free(tmp); 1276 } 1277 1278 if (n->child) 1279 pman_parse_name(n->child, rec); 1280 1281 if(n->next) 1282 pman_parse_name(n->next, rec); 1283 } 1284 1285 /* 1286 * A stub function to be able to parse the macros like .B embedded inside 1287 * a section. 1288 */ 1289 static void 1290 pman_block(const struct man_node *n, mandb_rec *rec) 1291 { 1292 } 1293 1294 /* 1295 * pman_sh -- 1296 * This function does one of the two things: 1297 * 1. If the present section is NAME, then it will: 1298 * (a) Extract the name of the page (in case of multiple comma separated 1299 * names, it will pick up the first one). 1300 * (b) Build a space spearated list of all the symlinks/hardlinks to 1301 * this page and store in the buffer 'links'. These are extracted from 1302 * the comma separated list of names in the NAME section as well. 1303 * (c) Move on to the one line description section, which is after the list 1304 * of names in the NAME section. 1305 * 2. Otherwise, it will check the section name and call the man_parse_section 1306 * function, passing the enum corresponding that section. 1307 */ 1308 static void 1309 pman_sh(const struct man_node *n, mandb_rec *rec) 1310 { 1311 static const struct { 1312 enum man_sec section; 1313 const char *header; 1314 } mapping[] = { 1315 { MANSEC_DESCRIPTION, "DESCRIPTION" }, 1316 { MANSEC_SYNOPSIS, "SYNOPSIS" }, 1317 { MANSEC_LIBRARY, "LIBRARY" }, 1318 { MANSEC_ERRORS, "ERRORS" }, 1319 { MANSEC_FILES, "FILES" }, 1320 { MANSEC_RETURN_VALUES, "RETURN VALUE" }, 1321 { MANSEC_RETURN_VALUES, "RETURN VALUES" }, 1322 { MANSEC_EXIT_STATUS, "EXIT STATUS" }, 1323 { MANSEC_EXAMPLES, "EXAMPLES" }, 1324 { MANSEC_EXAMPLES, "EXAMPLE" }, 1325 { MANSEC_STANDARDS, "STANDARDS" }, 1326 { MANSEC_HISTORY, "HISTORY" }, 1327 { MANSEC_BUGS, "BUGS" }, 1328 { MANSEC_AUTHORS, "AUTHORS" }, 1329 { MANSEC_COPYRIGHT, "COPYRIGHT" }, 1330 }; 1331 const struct man_node *head; 1332 char *name_desc; 1333 size_t sz; 1334 size_t i; 1335 1336 if ((head = n->parent->head) == NULL || (head = head->child) == NULL || 1337 head->type != MAN_TEXT) 1338 return; 1339 1340 /* 1341 * Check if this section should be extracted and 1342 * where it should be stored. Handled the trival cases first. 1343 */ 1344 for (i = 0; i < sizeof(mapping) / sizeof(mapping[0]); ++i) { 1345 if (strcmp(head->string, mapping[i].header) == 0) { 1346 man_parse_section(mapping[i].section, n, rec); 1347 return; 1348 } 1349 } 1350 1351 if (strcmp(head->string, "NAME") == 0) { 1352 /* 1353 * We are in the NAME section. 1354 * pman_parse_name will put the complete content in name_desc. 1355 */ 1356 pman_parse_name(n, rec); 1357 1358 name_desc = rec->name_desc; 1359 if (name_desc == NULL) 1360 return; 1361 1362 /* Remove any leading spaces. */ 1363 while (name_desc[0] == ' ') 1364 name_desc++; 1365 1366 /* If the line begins with a "\&", avoid those */ 1367 if (name_desc[0] == '\\' && name_desc[1] == '&') 1368 name_desc += 2; 1369 1370 /* Now name_desc should be left with a comma-space 1371 * separated list of names and the one line description 1372 * of the page: 1373 * "a, b, c \- sample description" 1374 * Take out the first name, before the first comma 1375 * (or space) and store it in rec->name. 1376 * If the page has aliases then they should be 1377 * in the form of a comma separated list. 1378 * Keep looping while there is a comma in name_desc, 1379 * extract the alias name and store in rec->links. 1380 * When there are no more commas left, break out. 1381 */ 1382 int has_alias = 0; // Any more aliases left? 1383 while (*name_desc) { 1384 /* Remove any leading spaces or hyphens. */ 1385 if (name_desc[0] == ' ' || name_desc[0] =='-') { 1386 name_desc++; 1387 continue; 1388 } 1389 sz = strcspn(name_desc, ", "); 1390 1391 /* Extract the first term and store it in rec->name. */ 1392 if (rec->name == NULL) { 1393 if (name_desc[sz] == ',') 1394 has_alias = 1; 1395 name_desc[sz] = 0; 1396 rec->name = emalloc(sz + 1); 1397 memcpy(rec->name, name_desc, sz + 1); 1398 name_desc += sz + 1; 1399 continue; 1400 } 1401 1402 /* 1403 * Once rec->name is set, rest of the names 1404 * are to be treated as links or aliases. 1405 */ 1406 if (rec->name && has_alias) { 1407 if (name_desc[sz] != ',') { 1408 /* No more commas left --> 1409 * no more aliases to take out 1410 */ 1411 has_alias = 0; 1412 } 1413 name_desc[sz] = 0; 1414 concat2(&rec->links, name_desc, sz); 1415 name_desc += sz + 1; 1416 continue; 1417 } 1418 break; 1419 } 1420 1421 /* Parse any escape sequences that might be there */ 1422 char *temp = parse_escape(name_desc); 1423 free(rec->name_desc); 1424 rec->name_desc = temp; 1425 temp = parse_escape(rec->name); 1426 free(rec->name); 1427 rec->name = temp; 1428 return; 1429 } 1430 1431 /* The RETURN VALUE section might be specified in multiple ways */ 1432 if (strcmp(head->string, "RETURN") == 0 && 1433 head->next != NULL && head->next->type == MAN_TEXT && 1434 (strcmp(head->next->string, "VALUE") == 0 || 1435 strcmp(head->next->string, "VALUES") == 0)) { 1436 man_parse_section(MANSEC_RETURN_VALUES, n, rec); 1437 return; 1438 } 1439 1440 /* 1441 * EXIT STATUS section can also be specified all on one line or on two 1442 * separate lines. 1443 */ 1444 if (strcmp(head->string, "EXIT") == 0 && 1445 head->next != NULL && head->next->type == MAN_TEXT && 1446 strcmp(head->next->string, "STATUS") == 0) { 1447 man_parse_section(MANSEC_EXIT_STATUS, n, rec); 1448 return; 1449 } 1450 1451 /* Store the rest of the content in desc. */ 1452 man_parse_section(MANSEC_NONE, n, rec); 1453 } 1454 1455 /* 1456 * pman_parse_node -- 1457 * Generic function to iterate through a node. Usually called from 1458 * man_parse_section to parse a particular section of the man page. 1459 */ 1460 static void 1461 pman_parse_node(const struct man_node *n, secbuff *s) 1462 { 1463 if (n == NULL) 1464 return; 1465 1466 if (n->type == MAN_TEXT) 1467 append(s, n->string); 1468 1469 pman_parse_node(n->child, s); 1470 pman_parse_node(n->next, s); 1471 } 1472 1473 /* 1474 * man_parse_section -- 1475 * Takes two parameters: 1476 * sec: Tells which section we are present in 1477 * n: Is the present node of the AST. 1478 * Depending on the section, we call pman_parse_node to parse that section and 1479 * concatenate the content from that section into the buffer for that section. 1480 */ 1481 static void 1482 man_parse_section(enum man_sec sec, const struct man_node *n, mandb_rec *rec) 1483 { 1484 /* 1485 * If the user sepecified the 'l' flag then just parse 1486 * the NAME section, ignore the rest. 1487 */ 1488 if (mflags.limit) 1489 return; 1490 1491 switch (sec) { 1492 case MANSEC_LIBRARY: 1493 pman_parse_node(n, &rec->lib); 1494 break; 1495 case MANSEC_RETURN_VALUES: 1496 pman_parse_node(n, &rec->return_vals); 1497 break; 1498 case MANSEC_ENVIRONMENT: 1499 pman_parse_node(n, &rec->env); 1500 break; 1501 case MANSEC_FILES: 1502 pman_parse_node(n, &rec->files); 1503 break; 1504 case MANSEC_EXIT_STATUS: 1505 pman_parse_node(n, &rec->exit_status); 1506 break; 1507 case MANSEC_DIAGNOSTICS: 1508 pman_parse_node(n, &rec->diagnostics); 1509 break; 1510 case MANSEC_ERRORS: 1511 pman_parse_node(n, &rec->errors); 1512 break; 1513 case MANSEC_NAME: 1514 case MANSEC_SYNOPSIS: 1515 case MANSEC_EXAMPLES: 1516 case MANSEC_STANDARDS: 1517 case MANSEC_HISTORY: 1518 case MANSEC_BUGS: 1519 case MANSEC_AUTHORS: 1520 case MANSEC_COPYRIGHT: 1521 break; 1522 default: 1523 pman_parse_node(n, &rec->desc); 1524 break; 1525 } 1526 1527 } 1528 1529 /* 1530 * insert_into_db -- 1531 * Inserts the parsed data of the man page in the Sqlite databse. 1532 * If any of the values is NULL, then we cleanup and return -1 indicating 1533 * an error. 1534 * Otherwise, store the data in the database and return 0. 1535 */ 1536 static int 1537 insert_into_db(sqlite3 *db, mandb_rec *rec) 1538 { 1539 int rc = 0; 1540 int idx = -1; 1541 const char *sqlstr = NULL; 1542 sqlite3_stmt *stmt = NULL; 1543 char *ln = NULL; 1544 char *errmsg = NULL; 1545 long int mandb_rowid; 1546 1547 /* 1548 * At the very minimum we want to make sure that we store 1549 * the following data: 1550 * Name, one line description, and the MD5 hash 1551 */ 1552 if (rec->name == NULL || rec->name_desc == NULL || 1553 rec->md5_hash == NULL) { 1554 cleanup(rec); 1555 return -1; 1556 } 1557 1558 /* Write null byte at the end of all the sec_buffs */ 1559 rec->desc.data[rec->desc.offset] = 0; 1560 rec->lib.data[rec->lib.offset] = 0; 1561 rec->env.data[rec->env.offset] = 0; 1562 rec->return_vals.data[rec->return_vals.offset] = 0; 1563 rec->exit_status.data[rec->exit_status.offset] = 0; 1564 rec->files.data[rec->files.offset] = 0; 1565 rec->diagnostics.data[rec->diagnostics.offset] = 0; 1566 rec->errors.data[rec->errors.offset] = 0; 1567 1568 /* 1569 * In case of a mdoc page: (sorry, no better place to put this code) 1570 * parse the comma separated list of names of man pages, 1571 * the first name will be stored in the mandb table, rest will be 1572 * treated as links and put in the mandb_links table. 1573 */ 1574 if (rec->page_type == MDOC) { 1575 char *tmp; 1576 rec->links = estrdup(rec->name); 1577 free(rec->name); 1578 int sz = strcspn(rec->links, " \0"); 1579 rec->name = emalloc(sz + 1); 1580 memcpy(rec->name, rec->links, sz); 1581 if(rec->name[sz - 1] == ',') 1582 rec->name[sz - 1] = 0; 1583 else 1584 rec->name[sz] = 0; 1585 while (rec->links[sz] == ' ') 1586 ++sz; 1587 tmp = estrdup(rec->links + sz); 1588 free(rec->links); 1589 rec->links = tmp; 1590 } 1591 1592 /*------------------------ Populate the mandb table---------------------------*/ 1593 sqlstr = "INSERT INTO mandb VALUES (:section, :name, :name_desc, :desc," 1594 " :lib, :return_vals, :env, :files, :exit_status," 1595 " :diagnostics, :errors, :md5_hash, :machine)"; 1596 1597 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1598 if (rc != SQLITE_OK) 1599 goto Out; 1600 1601 idx = sqlite3_bind_parameter_index(stmt, ":name"); 1602 rc = sqlite3_bind_text(stmt, idx, rec->name, -1, NULL); 1603 if (rc != SQLITE_OK) { 1604 sqlite3_finalize(stmt); 1605 goto Out; 1606 } 1607 1608 idx = sqlite3_bind_parameter_index(stmt, ":section"); 1609 rc = sqlite3_bind_text(stmt, idx, rec->section, -1, NULL); 1610 if (rc != SQLITE_OK) { 1611 sqlite3_finalize(stmt); 1612 goto Out; 1613 } 1614 1615 idx = sqlite3_bind_parameter_index(stmt, ":name_desc"); 1616 rc = sqlite3_bind_text(stmt, idx, rec->name_desc, -1, NULL); 1617 if (rc != SQLITE_OK) { 1618 sqlite3_finalize(stmt); 1619 goto Out; 1620 } 1621 1622 idx = sqlite3_bind_parameter_index(stmt, ":desc"); 1623 rc = sqlite3_bind_text(stmt, idx, rec->desc.data, 1624 rec->desc.offset + 1, NULL); 1625 if (rc != SQLITE_OK) { 1626 sqlite3_finalize(stmt); 1627 goto Out; 1628 } 1629 1630 idx = sqlite3_bind_parameter_index(stmt, ":lib"); 1631 rc = sqlite3_bind_text(stmt, idx, rec->lib.data, rec->lib.offset + 1, NULL); 1632 if (rc != SQLITE_OK) { 1633 sqlite3_finalize(stmt); 1634 goto Out; 1635 } 1636 1637 idx = sqlite3_bind_parameter_index(stmt, ":return_vals"); 1638 rc = sqlite3_bind_text(stmt, idx, rec->return_vals.data, 1639 rec->return_vals.offset + 1, NULL); 1640 if (rc != SQLITE_OK) { 1641 sqlite3_finalize(stmt); 1642 goto Out; 1643 } 1644 1645 idx = sqlite3_bind_parameter_index(stmt, ":env"); 1646 rc = sqlite3_bind_text(stmt, idx, rec->env.data, rec->env.offset + 1, NULL); 1647 if (rc != SQLITE_OK) { 1648 sqlite3_finalize(stmt); 1649 goto Out; 1650 } 1651 1652 idx = sqlite3_bind_parameter_index(stmt, ":files"); 1653 rc = sqlite3_bind_text(stmt, idx, rec->files.data, 1654 rec->files.offset + 1, NULL); 1655 if (rc != SQLITE_OK) { 1656 sqlite3_finalize(stmt); 1657 goto Out; 1658 } 1659 1660 idx = sqlite3_bind_parameter_index(stmt, ":exit_status"); 1661 rc = sqlite3_bind_text(stmt, idx, rec->exit_status.data, 1662 rec->exit_status.offset + 1, NULL); 1663 if (rc != SQLITE_OK) { 1664 sqlite3_finalize(stmt); 1665 goto Out; 1666 } 1667 1668 idx = sqlite3_bind_parameter_index(stmt, ":diagnostics"); 1669 rc = sqlite3_bind_text(stmt, idx, rec->diagnostics.data, 1670 rec->diagnostics.offset + 1, NULL); 1671 if (rc != SQLITE_OK) { 1672 sqlite3_finalize(stmt); 1673 goto Out; 1674 } 1675 1676 idx = sqlite3_bind_parameter_index(stmt, ":errors"); 1677 rc = sqlite3_bind_text(stmt, idx, rec->errors.data, 1678 rec->errors.offset + 1, NULL); 1679 if (rc != SQLITE_OK) { 1680 sqlite3_finalize(stmt); 1681 goto Out; 1682 } 1683 1684 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1685 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1686 if (rc != SQLITE_OK) { 1687 sqlite3_finalize(stmt); 1688 goto Out; 1689 } 1690 1691 idx = sqlite3_bind_parameter_index(stmt, ":machine"); 1692 if (rec->machine) 1693 rc = sqlite3_bind_text(stmt, idx, rec->machine, -1, NULL); 1694 else 1695 rc = sqlite3_bind_null(stmt, idx); 1696 if (rc != SQLITE_OK) { 1697 sqlite3_finalize(stmt); 1698 goto Out; 1699 } 1700 1701 rc = sqlite3_step(stmt); 1702 if (rc != SQLITE_DONE) { 1703 sqlite3_finalize(stmt); 1704 goto Out; 1705 } 1706 1707 sqlite3_finalize(stmt); 1708 1709 /* Get the row id of the last inserted row */ 1710 mandb_rowid = sqlite3_last_insert_rowid(db); 1711 1712 /*------------------------Populate the mandb_meta table-----------------------*/ 1713 sqlstr = "INSERT INTO mandb_meta VALUES (:device, :inode, :mtime," 1714 " :file, :md5_hash, :id)"; 1715 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1716 if (rc != SQLITE_OK) 1717 goto Out; 1718 1719 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1720 rc = sqlite3_bind_int64(stmt, idx, rec->device); 1721 if (rc != SQLITE_OK) { 1722 sqlite3_finalize(stmt); 1723 goto Out; 1724 } 1725 1726 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1727 rc = sqlite3_bind_int64(stmt, idx, rec->inode); 1728 if (rc != SQLITE_OK) { 1729 sqlite3_finalize(stmt); 1730 goto Out; 1731 } 1732 1733 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1734 rc = sqlite3_bind_int64(stmt, idx, rec->mtime); 1735 if (rc != SQLITE_OK) { 1736 sqlite3_finalize(stmt); 1737 goto Out; 1738 } 1739 1740 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1741 rc = sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1742 if (rc != SQLITE_OK) { 1743 sqlite3_finalize(stmt); 1744 goto Out; 1745 } 1746 1747 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1748 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1749 if (rc != SQLITE_OK) { 1750 sqlite3_finalize(stmt); 1751 goto Out; 1752 } 1753 1754 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1755 rc = sqlite3_bind_int64(stmt, idx, mandb_rowid); 1756 if (rc != SQLITE_OK) { 1757 sqlite3_finalize(stmt); 1758 goto Out; 1759 } 1760 1761 rc = sqlite3_step(stmt); 1762 sqlite3_finalize(stmt); 1763 if (rc == SQLITE_CONSTRAINT_UNIQUE) { 1764 /* The *most* probable reason for reaching here is that 1765 * the UNIQUE contraint on the file column of the mandb_meta 1766 * table was violated. 1767 * This can happen when a file was updated/modified. 1768 * To fix this we need to do two things: 1769 * 1. Delete the row for the older version of this file 1770 * from mandb table. 1771 * 2. Run an UPDATE query to update the row for this file 1772 * in the mandb_meta table. 1773 */ 1774 warnx("Trying to update index for %s", rec->file_path); 1775 char *sql = sqlite3_mprintf("DELETE FROM mandb " 1776 "WHERE rowid = (SELECT id" 1777 " FROM mandb_meta" 1778 " WHERE file = %Q)", 1779 rec->file_path); 1780 sqlite3_exec(db, sql, NULL, NULL, &errmsg); 1781 sqlite3_free(sql); 1782 if (errmsg != NULL) { 1783 if (mflags.verbosity) 1784 warnx("%s", errmsg); 1785 free(errmsg); 1786 } 1787 sqlstr = "UPDATE mandb_meta SET device = :device," 1788 " inode = :inode, mtime = :mtime, id = :id," 1789 " md5_hash = :md5 WHERE file = :file"; 1790 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1791 if (rc != SQLITE_OK) { 1792 if (mflags.verbosity) 1793 warnx("Update failed with error: %s", 1794 sqlite3_errmsg(db)); 1795 close_db(db); 1796 cleanup(rec); 1797 errx(EXIT_FAILURE, 1798 "Consider running makemandb with -f option"); 1799 } 1800 1801 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1802 sqlite3_bind_int64(stmt, idx, rec->device); 1803 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1804 sqlite3_bind_int64(stmt, idx, rec->inode); 1805 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1806 sqlite3_bind_int64(stmt, idx, rec->mtime); 1807 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1808 sqlite3_bind_int64(stmt, idx, mandb_rowid); 1809 idx = sqlite3_bind_parameter_index(stmt, ":md5"); 1810 sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1811 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1812 sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1813 rc = sqlite3_step(stmt); 1814 sqlite3_finalize(stmt); 1815 1816 if (rc != SQLITE_DONE) { 1817 if (mflags.verbosity) 1818 warnx("%s", sqlite3_errmsg(db)); 1819 close_db(db); 1820 cleanup(rec); 1821 errx(EXIT_FAILURE, 1822 "Consider running makemandb with -f option"); 1823 } 1824 } else if (rc != SQLITE_DONE) { 1825 /* Otherwise make this error fatal */ 1826 warnx("Failed at %s\n%s", rec->file_path, sqlite3_errmsg(db)); 1827 cleanup(rec); 1828 close_db(db); 1829 exit(EXIT_FAILURE); 1830 } 1831 1832 /*------------------------ Populate the mandb_links table---------------------*/ 1833 char *str = NULL; 1834 char *links; 1835 if (rec->links && strlen(rec->links)) { 1836 links = rec->links; 1837 for(ln = strtok(links, " "); ln; ln = strtok(NULL, " ")) { 1838 if (ln[0] == ',') 1839 ln++; 1840 if(ln[strlen(ln) - 1] == ',') 1841 ln[strlen(ln) - 1] = 0; 1842 1843 str = sqlite3_mprintf("INSERT INTO mandb_links" 1844 " VALUES (%Q, %Q, %Q, %Q, %Q)", 1845 ln, rec->name, rec->section, 1846 rec->machine, rec->md5_hash); 1847 sqlite3_exec(db, str, NULL, NULL, &errmsg); 1848 sqlite3_free(str); 1849 if (errmsg != NULL) { 1850 warnx("%s", errmsg); 1851 cleanup(rec); 1852 free(errmsg); 1853 return -1; 1854 } 1855 } 1856 } 1857 1858 cleanup(rec); 1859 return 0; 1860 1861 Out: 1862 if (mflags.verbosity) 1863 warnx("%s", sqlite3_errmsg(db)); 1864 cleanup(rec); 1865 return -1; 1866 } 1867 1868 /* 1869 * check_md5-- 1870 * Generates the md5 hash of the file and checks if it already doesn't exist 1871 * in the table (passed as the 3rd parameter). 1872 * This function is being used to avoid hardlinks. 1873 * On successful completion it will also set the value of the fourth parameter 1874 * to the md5 hash of the file (computed previously). It is the responsibility 1875 * of the caller to free this buffer. 1876 * Return values: 1877 * -1: If an error occurs somewhere and sets the md5 return buffer to NULL. 1878 * 0: If the md5 hash does not exist in the table. 1879 * 1: If the hash exists in the database. 1880 */ 1881 static int 1882 check_md5(const char *file, sqlite3 *db, const char *table, char **md5sum, 1883 void *buf, size_t buflen) 1884 { 1885 int rc = 0; 1886 int idx = -1; 1887 char *sqlstr = NULL; 1888 sqlite3_stmt *stmt = NULL; 1889 1890 assert(file != NULL); 1891 *md5sum = MD5Data(buf, buflen, NULL); 1892 if (*md5sum == NULL) { 1893 if (mflags.verbosity) 1894 warn("md5 failed: %s", file); 1895 return -1; 1896 } 1897 1898 easprintf(&sqlstr, "SELECT * FROM %s WHERE md5_hash = :md5_hash", 1899 table); 1900 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1901 if (rc != SQLITE_OK) { 1902 free(sqlstr); 1903 free(*md5sum); 1904 *md5sum = NULL; 1905 return -1; 1906 } 1907 1908 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1909 rc = sqlite3_bind_text(stmt, idx, *md5sum, -1, NULL); 1910 if (rc != SQLITE_OK) { 1911 if (mflags.verbosity) 1912 warnx("%s", sqlite3_errmsg(db)); 1913 sqlite3_finalize(stmt); 1914 free(sqlstr); 1915 free(*md5sum); 1916 *md5sum = NULL; 1917 return -1; 1918 } 1919 1920 if (sqlite3_step(stmt) == SQLITE_ROW) { 1921 sqlite3_finalize(stmt); 1922 free(sqlstr); 1923 return 0; 1924 } 1925 1926 sqlite3_finalize(stmt); 1927 free(sqlstr); 1928 return 1; 1929 } 1930 1931 /* Optimize the index for faster search */ 1932 static void 1933 optimize(sqlite3 *db) 1934 { 1935 const char *sqlstr; 1936 char *errmsg = NULL; 1937 1938 if (mflags.verbosity == 2) 1939 printf("Optimizing the database index\n"); 1940 sqlstr = "INSERT INTO mandb(mandb) VALUES (\'optimize\');" 1941 "VACUUM"; 1942 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 1943 if (errmsg != NULL) { 1944 if (mflags.verbosity) 1945 warnx("%s", errmsg); 1946 free(errmsg); 1947 return; 1948 } 1949 } 1950 1951 /* 1952 * cleanup -- 1953 * cleans up the global buffers 1954 */ 1955 static void 1956 cleanup(mandb_rec *rec) 1957 { 1958 rec->desc.offset = 0; 1959 rec->lib.offset = 0; 1960 rec->return_vals.offset = 0; 1961 rec->env.offset = 0; 1962 rec->exit_status.offset = 0; 1963 rec->diagnostics.offset = 0; 1964 rec->errors.offset = 0; 1965 rec->files.offset = 0; 1966 1967 free(rec->machine); 1968 rec->machine = NULL; 1969 1970 free(rec->links); 1971 rec->links = NULL; 1972 1973 free(rec->file_path); 1974 rec->file_path = NULL; 1975 1976 free(rec->name); 1977 rec->name = NULL; 1978 1979 free(rec->name_desc); 1980 rec->name_desc = NULL; 1981 1982 free(rec->md5_hash); 1983 rec->md5_hash = NULL; 1984 } 1985 1986 /* 1987 * init_secbuffs-- 1988 * Sets the value of buflen for all the sec_buff field of rec. And then 1989 * allocate memory to each sec_buff member of rec. 1990 */ 1991 static void 1992 init_secbuffs(mandb_rec *rec) 1993 { 1994 /* 1995 * Some sec_buff might need more memory, for example desc, 1996 * which stores the data of the DESCRIPTION section, 1997 * while some might need very small amount of memory. 1998 * Therefore explicitly setting the value of buflen field for 1999 * each sec_buff. 2000 */ 2001 rec->desc.buflen = 10 * BUFLEN; 2002 rec->desc.data = emalloc(rec->desc.buflen); 2003 rec->desc.offset = 0; 2004 2005 rec->lib.buflen = BUFLEN / 2; 2006 rec->lib.data = emalloc(rec->lib.buflen); 2007 rec->lib.offset = 0; 2008 2009 rec->return_vals.buflen = BUFLEN; 2010 rec->return_vals.data = emalloc(rec->return_vals.buflen); 2011 rec->return_vals.offset = 0; 2012 2013 rec->exit_status.buflen = BUFLEN; 2014 rec->exit_status.data = emalloc(rec->exit_status.buflen); 2015 rec->exit_status.offset = 0; 2016 2017 rec->env.buflen = BUFLEN; 2018 rec->env.data = emalloc(rec->env.buflen); 2019 rec->env.offset = 0; 2020 2021 rec->files.buflen = BUFLEN; 2022 rec->files.data = emalloc(rec->files.buflen); 2023 rec->files.offset = 0; 2024 2025 rec->diagnostics.buflen = BUFLEN; 2026 rec->diagnostics.data = emalloc(rec->diagnostics.buflen); 2027 rec->diagnostics.offset = 0; 2028 2029 rec->errors.buflen = BUFLEN; 2030 rec->errors.data = emalloc(rec->errors.buflen); 2031 rec->errors.offset = 0; 2032 } 2033 2034 /* 2035 * free_secbuffs-- 2036 * This function should be called at the end, when all the pages have been 2037 * parsed. 2038 * It frees the memory allocated to sec_buffs by init_secbuffs in the starting. 2039 */ 2040 static void 2041 free_secbuffs(mandb_rec *rec) 2042 { 2043 free(rec->desc.data); 2044 free(rec->lib.data); 2045 free(rec->return_vals.data); 2046 free(rec->exit_status.data); 2047 free(rec->env.data); 2048 free(rec->files.data); 2049 free(rec->diagnostics.data); 2050 free(rec->errors.data); 2051 } 2052 2053 static void 2054 replace_hyph(char *str) 2055 { 2056 char *iter = str; 2057 while ((iter = strchr(iter, ASCII_HYPH)) != NULL) 2058 *iter = '-'; 2059 2060 iter = str; 2061 while ((iter = strchr(iter, ASCII_NBRSP)) != NULL) 2062 *iter = '-'; 2063 } 2064 2065 static char * 2066 parse_escape(const char *str) 2067 { 2068 const char *backslash, *last_backslash; 2069 char *result, *iter; 2070 size_t len; 2071 2072 assert(str); 2073 2074 last_backslash = str; 2075 backslash = strchr(str, '\\'); 2076 if (backslash == NULL) { 2077 result = estrdup(str); 2078 replace_hyph(result); 2079 return result; 2080 } 2081 2082 result = emalloc(strlen(str) + 1); 2083 iter = result; 2084 2085 do { 2086 len = backslash - last_backslash; 2087 memcpy(iter, last_backslash, len); 2088 iter += len; 2089 if (backslash[1] == '-' || backslash[1] == ' ') { 2090 *iter++ = backslash[1]; 2091 last_backslash = backslash + 2; 2092 backslash = strchr(backslash + 2, '\\'); 2093 } else { 2094 ++backslash; 2095 mandoc_escape(&backslash, NULL, NULL); 2096 last_backslash = backslash; 2097 if (backslash == NULL) 2098 break; 2099 backslash = strchr(last_backslash, '\\'); 2100 } 2101 } while (backslash != NULL); 2102 if (last_backslash != NULL) 2103 strcpy(iter, last_backslash); 2104 2105 replace_hyph(result); 2106 return result; 2107 } 2108 2109 /* 2110 * append-- 2111 * Concatenates a space and src at the end of sbuff->data (much like concat in 2112 * apropos-utils.c). 2113 * Rather than reallocating space for writing data, it uses the value of the 2114 * offset field of sec_buff to write new data at the free space left in the 2115 * buffer. 2116 * In case the size of the data to be appended exceeds the number of bytes left 2117 * in the buffer, it reallocates buflen number of bytes and then continues. 2118 * Value of offset field should be adjusted as new data is written. 2119 * 2120 * NOTE: This function does not write the null byte at the end of the buffers, 2121 * write a null byte at the position pointed to by offset before inserting data 2122 * in the db. 2123 */ 2124 static void 2125 append(secbuff *sbuff, const char *src) 2126 { 2127 short flag = 0; 2128 size_t srclen, newlen; 2129 char *temp; 2130 2131 assert(src != NULL); 2132 temp = parse_escape(src); 2133 srclen = strlen(temp); 2134 2135 if (sbuff->data == NULL) { 2136 sbuff->data = emalloc(sbuff->buflen); 2137 sbuff->offset = 0; 2138 } 2139 2140 newlen = sbuff->offset + srclen + 2; 2141 if (newlen >= sbuff->buflen) { 2142 while (sbuff->buflen < newlen) 2143 sbuff->buflen += sbuff->buflen; 2144 sbuff->data = erealloc(sbuff->data, sbuff->buflen); 2145 flag = 1; 2146 } 2147 2148 /* Append a space at the end of the buffer. */ 2149 if (sbuff->offset || flag) 2150 sbuff->data[sbuff->offset++] = ' '; 2151 /* Now, copy src at the end of the buffer. */ 2152 memcpy(sbuff->data + sbuff->offset, temp, srclen); 2153 sbuff->offset += srclen; 2154 free(temp); 2155 } 2156 2157 static void 2158 usage(void) 2159 { 2160 fprintf(stderr, "Usage: %s [-floQqv] [-C path]\n", getprogname()); 2161 exit(1); 2162 } 2163
Indexes created Mon Sep 29 21:09:56 GMT 2025