makemandb.c revision 1.38
1/* $NetBSD: makemandb.c,v 1.38 2016/07/05 16:24:18 abhinav Exp $ */ 2/* 3 * Copyright (c) 2011 Abhinav Upadhyay <er.abhinav.upadhyay@gmail.com> 4 * Copyright (c) 2011 Kristaps Dzonsons <kristaps@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 1.38 2016/07/05 16:24:18 abhinav 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 */ 49typedef 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 55typedef 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 62typedef 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; 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 93static void append(secbuff *sbuff, const char *src); 94static void init_secbuffs(mandb_rec *); 95static void free_secbuffs(mandb_rec *); 96static int check_md5(const char *, sqlite3 *, const char *, char **, void *, size_t); 97static void cleanup(mandb_rec *); 98static void set_section(const struct mdoc *, const struct man *, mandb_rec *); 99static void set_machine(const struct mdoc *, mandb_rec *); 100static int insert_into_db(sqlite3 *, mandb_rec *); 101static void begin_parse(const char *, struct mparse *, mandb_rec *, 102 const void *, size_t len); 103static void pmdoc_node(const struct mdoc_node *, mandb_rec *); 104static void pmdoc_Nm(const struct mdoc_node *, mandb_rec *); 105static void pmdoc_Nd(const struct mdoc_node *, mandb_rec *); 106static void pmdoc_Sh(const struct mdoc_node *, mandb_rec *); 107static void pmdoc_Xr(const struct mdoc_node *, mandb_rec *); 108static void pmdoc_Pp(const struct mdoc_node *, mandb_rec *); 109static void pmdoc_macro_handler(const struct mdoc_node *, mandb_rec *, 110 enum mdoct); 111static void pman_node(const struct man_node *n, mandb_rec *); 112static void pman_parse_node(const struct man_node *, secbuff *); 113static void pman_parse_name(const struct man_node *, mandb_rec *); 114static void pman_sh(const struct man_node *, mandb_rec *); 115static void pman_block(const struct man_node *, mandb_rec *); 116static void traversedir(const char *, const char *, sqlite3 *, struct mparse *); 117static void mdoc_parse_section(enum mdoc_sec, const char *, mandb_rec *); 118static void man_parse_section(enum man_sec, const struct man_node *, mandb_rec *); 119static void build_file_cache(sqlite3 *, const char *, const char *, 120 struct stat *); 121static void update_db(sqlite3 *, struct mparse *, mandb_rec *); 122__dead static void usage(void); 123static void optimize(sqlite3 *); 124static char *parse_escape(const char *); 125static void replace_hyph(char *); 126static makemandb_flags mflags = { .verbosity = 1 }; 127 128typedef void (*pman_nf)(const struct man_node *n, mandb_rec *); 129typedef void (*pmdoc_nf)(const struct mdoc_node *n, mandb_rec *); 130static 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 287static 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 329int 330main(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 */ 500static void 501traversedir(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 */ 552static void 553build_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 622static void 623update_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 */ 682static int 683read_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 729archive_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 */ 742static void 743update_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, 813 buflen); 814 assert(md5sum != NULL); 815 if (md5_status == -1) { 816 if (mflags.verbosity) 817 warnx("An error occurred in checking md5 value" 818 " for file %s", file); 819 err_count++; 820 continue; 821 } 822 823 if (md5_status == 0) { 824 /* 825 * The MD5 hash is already present in the database, 826 * so simply update the metadata, ignoring symlinks. 827 */ 828 struct stat sb; 829 stat(file, &sb); 830 if (S_ISLNK(sb.st_mode)) { 831 free(md5sum); 832 link_count++; 833 continue; 834 } 835 update_existing_entry(db, file, md5sum, rec, 836 &new_count, &link_count, &err_count); 837 free(md5sum); 838 continue; 839 } 840 841 if (md5_status == 1) { 842 /* 843 * The MD5 hash was not present in the database. 844 * This means is either a new file or an updated file. 845 * We should go ahead with parsing. 846 */ 847 if (chdir(parent) == -1) { 848 if (mflags.verbosity) 849 warn("chdir failed for `%s', could " 850 "not index `%s'", parent, file); 851 err_count++; 852 free(md5sum); 853 continue; 854 } 855 856 if (mflags.verbosity == 2) 857 printf("Parsing: %s\n", file); 858 rec->md5_hash = md5sum; 859 rec->file_path = estrdup(file); 860 // file_path is freed by insert_into_db itself. 861 begin_parse(file, mp, rec, buf, buflen); 862 if (insert_into_db(db, rec) < 0) { 863 if (mflags.verbosity) 864 warnx("Error in indexing `%s'", file); 865 err_count++; 866 } else { 867 new_count++; 868 } 869 } 870 } 871 872 if (mflags.verbosity == 2) { 873 printf("Total Number of new or updated pages encountered = %d\n" 874 "Total number of (hard or symbolic) links found = %d\n" 875 "Total number of pages that were successfully" 876 " indexed/updated = %d\n" 877 "Total number of pages that could not be indexed" 878 " due to errors = %d\n", 879 total_count - link_count, link_count, new_count, err_count); 880 } 881 882 if (mflags.recreate) 883 return; 884 885 if (mflags.verbosity == 2) 886 printf("Deleting stale index entries\n"); 887 888 sqlstr = "DELETE FROM mandb_meta WHERE file NOT IN" 889 " (SELECT file FROM metadb.file_cache);" 890 "DELETE FROM mandb_links WHERE md5_hash NOT IN" 891 " (SELECT md5_hash from mandb_meta);" 892 "DROP TABLE metadb.file_cache;" 893 "DELETE FROM mandb WHERE rowid NOT IN" 894 " (SELECT id FROM mandb_meta);"; 895 896 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 897 if (errmsg != NULL) { 898 warnx("Removing old entries failed: %s", errmsg); 899 warnx("Please rebuild database from scratch with -f."); 900 free(errmsg); 901 return; 902 } 903} 904 905/* 906 * begin_parse -- 907 * parses the man page using libmandoc 908 */ 909static void 910begin_parse(const char *file, struct mparse *mp, mandb_rec *rec, 911 const void *buf, size_t len) 912{ 913 struct mdoc *mdoc; 914 struct man *man; 915 mparse_reset(mp); 916 917 rec->xr_found = 0; 918 919 if (mparse_readmem(mp, buf, len, file) >= MANDOCLEVEL_BADARG) { 920 /* Printing this warning at verbosity level 2 921 * because some packages from pkgsrc might trigger several 922 * of such warnings. 923 */ 924 if (mflags.verbosity == 2) 925 warnx("%s: Parse failure", file); 926 return; 927 } 928 929 mparse_result(mp, &mdoc, &man, NULL); 930 if (mdoc == NULL && man == NULL) { 931 if (mflags.verbosity == 2) 932 warnx("Not a man(7) or mdoc(7) page"); 933 return; 934 } 935 936 set_machine(mdoc, rec); 937 set_section(mdoc, man, rec); 938 if (mdoc) { 939 rec->page_type = MDOC; 940 pmdoc_node(mdoc_node(mdoc), rec); 941 } else { 942 rec->page_type = MAN; 943 pman_node(man_node(man), rec); 944 } 945} 946 947/* 948 * set_section -- 949 * Extracts the section number and normalizes it to only the numeric part 950 * (Which should be the first character of the string). 951 */ 952static void 953set_section(const struct mdoc *md, const struct man *m, mandb_rec *rec) 954{ 955 if (md) { 956 const struct mdoc_meta *md_meta = mdoc_meta(md); 957 if (md_meta->msec == NULL) { 958 easprintf(&rec->section, "%s", "?"); 959 } else 960 rec->section = estrdup(md_meta->msec); 961 } else if (m) { 962 const struct man_meta *m_meta = man_meta(m); 963 if (m_meta->msec == NULL) 964 easprintf(&rec->section, "%s", "?"); 965 else 966 rec->section = estrdup(m_meta->msec); 967 } else 968 return; 969 970 if (rec->section[0] == '?' && mflags.verbosity == 2) 971 warnx("%s: Missing section number", rec->file_path); 972} 973 974/* 975 * get_machine -- 976 * Extracts the machine architecture information if available. 977 */ 978static void 979set_machine(const struct mdoc *md, mandb_rec *rec) 980{ 981 if (md == NULL) 982 return; 983 const struct mdoc_meta *md_meta = mdoc_meta(md); 984 if (md_meta->arch) 985 rec->machine = estrdup(md_meta->arch); 986} 987 988static void 989pmdoc_node(const struct mdoc_node *n, mandb_rec *rec) 990{ 991 992 if (n == NULL) 993 return; 994 995 switch (n->type) { 996 case (MDOC_BODY): 997 /* FALLTHROUGH */ 998 case (MDOC_TAIL): 999 /* FALLTHROUGH */ 1000 case (MDOC_ELEM): 1001 if (mdocs[n->tok] == NULL) 1002 break; 1003 (*mdocs[n->tok])(n, rec); 1004 break; 1005 default: 1006 break; 1007 } 1008 1009 pmdoc_node(n->child, rec); 1010 pmdoc_node(n->next, rec); 1011} 1012 1013/* 1014 * pmdoc_Nm -- 1015 * Extracts the Name of the manual page from the .Nm macro 1016 */ 1017static void 1018pmdoc_Nm(const struct mdoc_node *n, mandb_rec *rec) 1019{ 1020 if (n->sec != SEC_NAME) 1021 return; 1022 1023 for (n = n->child; n; n = n->next) { 1024 if (n->type == MDOC_TEXT) { 1025 char *escaped_name = parse_escape(n->string); 1026 concat(&rec->name, escaped_name); 1027 free(escaped_name); 1028 } 1029 } 1030} 1031 1032/* 1033 * pmdoc_Nd -- 1034 * Extracts the one line description of the man page from the .Nd macro 1035 */ 1036static void 1037pmdoc_Nd(const struct mdoc_node *n, mandb_rec *rec) 1038{ 1039 char *buf = NULL; 1040 char *name; 1041 char *nd_text; 1042 1043 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1044 return; 1045 1046 if (n->type == MDOC_TEXT) { 1047 if (rec->xr_found && n->next) { 1048 /* 1049 * An Xr macro was seen previously, so parse this 1050 * and the next node, as "Name(Section)". 1051 */ 1052 name = n->string; 1053 n = n->next; 1054 assert(n->type == MDOC_TEXT); 1055 easprintf(&buf, "%s(%s)", name, n->string); 1056 concat(&rec->name_desc, buf); 1057 free(buf); 1058 } else { 1059 nd_text = parse_escape(n->string); 1060 concat(&rec->name_desc, nd_text); 1061 free(nd_text); 1062 } 1063 rec->xr_found = 0; 1064 } else if (mdocs[n->tok] == pmdoc_Xr) { 1065 /* Remember that we have encountered an Xr macro */ 1066 rec->xr_found = 1; 1067 } 1068 1069 if (n->child) 1070 pmdoc_Nd(n->child, rec); 1071 1072 if(n->next) 1073 pmdoc_Nd(n->next, rec); 1074} 1075 1076/* 1077 * pmdoc_macro_handler-- 1078 * This function is a single point of handling all the special macros that we 1079 * want to handle especially. For example the .Xr macro for properly parsing 1080 * the referenced page name along with the section number, or the .Pp macro 1081 * for adding a new line whenever we encounter it. 1082 */ 1083static void 1084pmdoc_macro_handler(const struct mdoc_node *n, mandb_rec *rec, enum mdoct doct) 1085{ 1086 const struct mdoc_node *sn; 1087 assert(n); 1088 1089 switch (doct) { 1090 /* Parse the man page references. 1091 * Basically the .Xr macros are used like: 1092 * .Xr ls 1 1093 * and formatted like this: 1094 * ls(1) 1095 * Prepare a buffer to format the data like the above example and call 1096 * pmdoc_parse_section to append it. 1097 */ 1098 case MDOC_Xr: 1099 n = n->child; 1100 while (n->type != MDOC_TEXT && n->next) 1101 n = n->next; 1102 1103 if (n && n->type != MDOC_TEXT) 1104 return; 1105 sn = n; 1106 if (n->next) 1107 n = n->next; 1108 1109 while (n->type != MDOC_TEXT && n->next) 1110 n = n->next; 1111 1112 if (n && n->type == MDOC_TEXT) { 1113 char *buf; 1114 easprintf(&buf, "%s(%s)", sn->string, n->string); 1115 mdoc_parse_section(n->sec, buf, rec); 1116 free(buf); 1117 } 1118 1119 break; 1120 1121 /* Parse the .Pp macro to add a new line */ 1122 case MDOC_Pp: 1123 if (n->type == MDOC_TEXT) 1124 mdoc_parse_section(n->sec, "\n", rec); 1125 break; 1126 default: 1127 break; 1128 } 1129 1130} 1131 1132/* 1133 * pmdoc_Xr, pmdoc_Pp-- 1134 * Empty stubs. 1135 * The parser calls these functions each time it encounters 1136 * a .Xr or .Pp macro. We are parsing all the data from 1137 * the pmdoc_Sh function, so don't do anything here. 1138 * (See if else blocks in pmdoc_Sh.) 1139 */ 1140static void 1141pmdoc_Xr(const struct mdoc_node *n, mandb_rec *rec) 1142{ 1143} 1144 1145static void 1146pmdoc_Pp(const struct mdoc_node *n, mandb_rec *rec) 1147{ 1148} 1149 1150/* 1151 * pmdoc_Sh -- 1152 * Called when a .Sh macro is encountered and loops through its body, calling 1153 * mdoc_parse_section to append the data to the section specific buffer. 1154 * Two special macros which may occur inside the body of Sh are .Nm and .Xr and 1155 * they need special handling, thus the separate if branches for them. 1156 */ 1157static void 1158pmdoc_Sh(const struct mdoc_node *n, mandb_rec *rec) 1159{ 1160 if (n == NULL || (n->type != MDOC_TEXT && n->tok == MDOC_MAX)) 1161 return; 1162 int xr_found = 0; 1163 1164 if (n->type == MDOC_TEXT) { 1165 mdoc_parse_section(n->sec, n->string, rec); 1166 } else if (mdocs[n->tok] == pmdoc_Nm && rec->name != NULL) { 1167 /* 1168 * When encountering a .Nm macro, substitute it 1169 * with its previously cached value of the argument. 1170 */ 1171 mdoc_parse_section(n->sec, rec->name, rec); 1172 } else if (mdocs[n->tok] == pmdoc_Xr) { 1173 /* 1174 * When encountering other inline macros, 1175 * call pmdoc_macro_handler. 1176 */ 1177 pmdoc_macro_handler(n, rec, MDOC_Xr); 1178 xr_found = 1; 1179 } else if (mdocs[n->tok] == pmdoc_Pp) { 1180 pmdoc_macro_handler(n, rec, MDOC_Pp); 1181 } 1182 1183 /* 1184 * If an Xr macro was encountered then the child node has 1185 * already been explored by pmdoc_macro_handler. 1186 */ 1187 if (xr_found == 0) 1188 pmdoc_Sh(n->child, rec); 1189 pmdoc_Sh(n->next, rec); 1190} 1191 1192/* 1193 * mdoc_parse_section-- 1194 * Utility function for parsing sections of the mdoc type pages. 1195 * Takes two params: 1196 * 1. sec is an enum which indicates the section in which we are present 1197 * 2. string is the string which we need to append to the secbuff for this 1198 * particular section. 1199 * The function appends string to the global section buffer and returns. 1200 */ 1201static void 1202mdoc_parse_section(enum mdoc_sec sec, const char *string, mandb_rec *rec) 1203{ 1204 /* 1205 * If the user specified the 'l' flag, then parse and store only the 1206 * NAME section. Ignore the rest. 1207 */ 1208 if (mflags.limit) 1209 return; 1210 1211 switch (sec) { 1212 case SEC_LIBRARY: 1213 append(&rec->lib, string); 1214 break; 1215 case SEC_RETURN_VALUES: 1216 append(&rec->return_vals, string); 1217 break; 1218 case SEC_ENVIRONMENT: 1219 append(&rec->env, string); 1220 break; 1221 case SEC_FILES: 1222 append(&rec->files, string); 1223 break; 1224 case SEC_EXIT_STATUS: 1225 append(&rec->exit_status, string); 1226 break; 1227 case SEC_DIAGNOSTICS: 1228 append(&rec->diagnostics, string); 1229 break; 1230 case SEC_ERRORS: 1231 append(&rec->errors, string); 1232 break; 1233 case SEC_NAME: 1234 case SEC_SYNOPSIS: 1235 case SEC_EXAMPLES: 1236 case SEC_STANDARDS: 1237 case SEC_HISTORY: 1238 case SEC_AUTHORS: 1239 case SEC_BUGS: 1240 break; 1241 default: 1242 append(&rec->desc, string); 1243 break; 1244 } 1245} 1246 1247static void 1248pman_node(const struct man_node *n, mandb_rec *rec) 1249{ 1250 if (n == NULL) 1251 return; 1252 1253 switch (n->type) { 1254 case (MAN_BODY): 1255 /* FALLTHROUGH */ 1256 case (MAN_BLOCK): 1257 /* FALLTHROUGH */ 1258 case (MAN_ELEM): 1259 if (mans[n->tok] != NULL) 1260 (*mans[n->tok])(n, rec); 1261 break; 1262 default: 1263 break; 1264 } 1265 1266 pman_node(n->child, rec); 1267 pman_node(n->next, rec); 1268} 1269 1270/* 1271 * pman_parse_name -- 1272 * Parses the NAME section and puts the complete content in the name_desc 1273 * variable. 1274 */ 1275static void 1276pman_parse_name(const struct man_node *n, mandb_rec *rec) 1277{ 1278 if (n == NULL) 1279 return; 1280 1281 if (n->type == MAN_TEXT) { 1282 char *tmp = parse_escape(n->string); 1283 concat(&rec->name_desc, tmp); 1284 free(tmp); 1285 } 1286 1287 if (n->child) 1288 pman_parse_name(n->child, rec); 1289 1290 if(n->next) 1291 pman_parse_name(n->next, rec); 1292} 1293 1294/* 1295 * A stub function to be able to parse the macros like .B embedded inside 1296 * a section. 1297 */ 1298static void 1299pman_block(const struct man_node *n, mandb_rec *rec) 1300{ 1301} 1302 1303/* 1304 * pman_sh -- 1305 * This function does one of the two things: 1306 * 1. If the present section is NAME, then it will: 1307 * (a) Extract the name of the page (in case of multiple comma separated 1308 * names, it will pick up the first one). 1309 * (b) Build a space spearated list of all the symlinks/hardlinks to 1310 * this page and store in the buffer 'links'. These are extracted from 1311 * the comma separated list of names in the NAME section as well. 1312 * (c) Move on to the one line description section, which is after the list 1313 * of names in the NAME section. 1314 * 2. Otherwise, it will check the section name and call the man_parse_section 1315 * function, passing the enum corresponding that section. 1316 */ 1317static void 1318pman_sh(const struct man_node *n, mandb_rec *rec) 1319{ 1320 static const struct { 1321 enum man_sec section; 1322 const char *header; 1323 } mapping[] = { 1324 { MANSEC_DESCRIPTION, "DESCRIPTION" }, 1325 { MANSEC_SYNOPSIS, "SYNOPSIS" }, 1326 { MANSEC_LIBRARY, "LIBRARY" }, 1327 { MANSEC_ERRORS, "ERRORS" }, 1328 { MANSEC_FILES, "FILES" }, 1329 { MANSEC_RETURN_VALUES, "RETURN VALUE" }, 1330 { MANSEC_RETURN_VALUES, "RETURN VALUES" }, 1331 { MANSEC_EXIT_STATUS, "EXIT STATUS" }, 1332 { MANSEC_EXAMPLES, "EXAMPLES" }, 1333 { MANSEC_EXAMPLES, "EXAMPLE" }, 1334 { MANSEC_STANDARDS, "STANDARDS" }, 1335 { MANSEC_HISTORY, "HISTORY" }, 1336 { MANSEC_BUGS, "BUGS" }, 1337 { MANSEC_AUTHORS, "AUTHORS" }, 1338 { MANSEC_COPYRIGHT, "COPYRIGHT" }, 1339 }; 1340 const struct man_node *head; 1341 char *name_desc; 1342 size_t sz; 1343 size_t i; 1344 1345 if ((head = n->parent->head) == NULL || (head = head->child) == NULL || 1346 head->type != MAN_TEXT) 1347 return; 1348 1349 /* 1350 * Check if this section should be extracted and 1351 * where it should be stored. Handled the trival cases first. 1352 */ 1353 for (i = 0; i < sizeof(mapping) / sizeof(mapping[0]); ++i) { 1354 if (strcmp(head->string, mapping[i].header) == 0) { 1355 man_parse_section(mapping[i].section, n, rec); 1356 return; 1357 } 1358 } 1359 1360 if (strcmp(head->string, "NAME") == 0) { 1361 /* 1362 * We are in the NAME section. 1363 * pman_parse_name will put the complete content in name_desc. 1364 */ 1365 pman_parse_name(n, rec); 1366 1367 name_desc = rec->name_desc; 1368 if (name_desc == NULL) 1369 return; 1370 1371 /* Remove any leading spaces. */ 1372 while (name_desc[0] == ' ') 1373 name_desc++; 1374 1375 /* If the line begins with a "\&", avoid those */ 1376 if (name_desc[0] == '\\' && name_desc[1] == '&') 1377 name_desc += 2; 1378 1379 /* Now name_desc should be left with a comma-space 1380 * separated list of names and the one line description 1381 * of the page: 1382 * "a, b, c \- sample description" 1383 * Take out the first name, before the first comma 1384 * (or space) and store it in rec->name. 1385 * If the page has aliases then they should be 1386 * in the form of a comma separated list. 1387 * Keep looping while there is a comma in name_desc, 1388 * extract the alias name and store in rec->links. 1389 * When there are no more commas left, break out. 1390 */ 1391 int has_alias = 0; // Any more aliases left? 1392 while (*name_desc) { 1393 /* Remove any leading spaces or hyphens. */ 1394 if (name_desc[0] == ' ' || name_desc[0] =='-') { 1395 name_desc++; 1396 continue; 1397 } 1398 sz = strcspn(name_desc, ", "); 1399 1400 /* Extract the first term and store it in rec->name. */ 1401 if (rec->name == NULL) { 1402 if (name_desc[sz] == ',') 1403 has_alias = 1; 1404 name_desc[sz] = 0; 1405 rec->name = emalloc(sz + 1); 1406 memcpy(rec->name, name_desc, sz + 1); 1407 name_desc += sz + 1; 1408 continue; 1409 } 1410 1411 /* 1412 * Once rec->name is set, rest of the names 1413 * are to be treated as links or aliases. 1414 */ 1415 if (rec->name && has_alias) { 1416 if (name_desc[sz] != ',') { 1417 /* No more commas left --> 1418 * no more aliases to take out 1419 */ 1420 has_alias = 0; 1421 } 1422 name_desc[sz] = 0; 1423 concat2(&rec->links, name_desc, sz); 1424 name_desc += sz + 1; 1425 continue; 1426 } 1427 break; 1428 } 1429 1430 /* Parse any escape sequences that might be there */ 1431 char *temp = parse_escape(name_desc); 1432 free(rec->name_desc); 1433 rec->name_desc = temp; 1434 temp = parse_escape(rec->name); 1435 free(rec->name); 1436 rec->name = temp; 1437 return; 1438 } 1439 1440 /* The RETURN VALUE section might be specified in multiple ways */ 1441 if (strcmp(head->string, "RETURN") == 0 && 1442 head->next != NULL && head->next->type == MAN_TEXT && 1443 (strcmp(head->next->string, "VALUE") == 0 || 1444 strcmp(head->next->string, "VALUES") == 0)) { 1445 man_parse_section(MANSEC_RETURN_VALUES, n, rec); 1446 return; 1447 } 1448 1449 /* 1450 * EXIT STATUS section can also be specified all on one line or on two 1451 * separate lines. 1452 */ 1453 if (strcmp(head->string, "EXIT") == 0 && 1454 head->next != NULL && head->next->type == MAN_TEXT && 1455 strcmp(head->next->string, "STATUS") == 0) { 1456 man_parse_section(MANSEC_EXIT_STATUS, n, rec); 1457 return; 1458 } 1459 1460 /* Store the rest of the content in desc. */ 1461 man_parse_section(MANSEC_NONE, n, rec); 1462} 1463 1464/* 1465 * pman_parse_node -- 1466 * Generic function to iterate through a node. Usually called from 1467 * man_parse_section to parse a particular section of the man page. 1468 */ 1469static void 1470pman_parse_node(const struct man_node *n, secbuff *s) 1471{ 1472 if (n == NULL) 1473 return; 1474 1475 if (n->type == MAN_TEXT) 1476 append(s, n->string); 1477 1478 pman_parse_node(n->child, s); 1479 pman_parse_node(n->next, s); 1480} 1481 1482/* 1483 * man_parse_section -- 1484 * Takes two parameters: 1485 * sec: Tells which section we are present in 1486 * n: Is the present node of the AST. 1487 * Depending on the section, we call pman_parse_node to parse that section and 1488 * concatenate the content from that section into the buffer for that section. 1489 */ 1490static void 1491man_parse_section(enum man_sec sec, const struct man_node *n, mandb_rec *rec) 1492{ 1493 /* 1494 * If the user sepecified the 'l' flag then just parse 1495 * the NAME section, ignore the rest. 1496 */ 1497 if (mflags.limit) 1498 return; 1499 1500 switch (sec) { 1501 case MANSEC_LIBRARY: 1502 pman_parse_node(n, &rec->lib); 1503 break; 1504 case MANSEC_RETURN_VALUES: 1505 pman_parse_node(n, &rec->return_vals); 1506 break; 1507 case MANSEC_ENVIRONMENT: 1508 pman_parse_node(n, &rec->env); 1509 break; 1510 case MANSEC_FILES: 1511 pman_parse_node(n, &rec->files); 1512 break; 1513 case MANSEC_EXIT_STATUS: 1514 pman_parse_node(n, &rec->exit_status); 1515 break; 1516 case MANSEC_DIAGNOSTICS: 1517 pman_parse_node(n, &rec->diagnostics); 1518 break; 1519 case MANSEC_ERRORS: 1520 pman_parse_node(n, &rec->errors); 1521 break; 1522 case MANSEC_NAME: 1523 case MANSEC_SYNOPSIS: 1524 case MANSEC_EXAMPLES: 1525 case MANSEC_STANDARDS: 1526 case MANSEC_HISTORY: 1527 case MANSEC_BUGS: 1528 case MANSEC_AUTHORS: 1529 case MANSEC_COPYRIGHT: 1530 break; 1531 default: 1532 pman_parse_node(n, &rec->desc); 1533 break; 1534 } 1535 1536} 1537 1538/* 1539 * insert_into_db -- 1540 * Inserts the parsed data of the man page in the Sqlite databse. 1541 * If any of the values is NULL, then we cleanup and return -1 indicating 1542 * an error. 1543 * Otherwise, store the data in the database and return 0. 1544 */ 1545static int 1546insert_into_db(sqlite3 *db, mandb_rec *rec) 1547{ 1548 int rc = 0; 1549 int idx = -1; 1550 const char *sqlstr = NULL; 1551 sqlite3_stmt *stmt = NULL; 1552 char *ln = NULL; 1553 char *errmsg = NULL; 1554 long int mandb_rowid; 1555 1556 /* 1557 * At the very minimum we want to make sure that we store 1558 * the following data: 1559 * Name, one line description, and the MD5 hash 1560 */ 1561 if (rec->name == NULL || rec->name_desc == NULL || 1562 rec->md5_hash == NULL) { 1563 cleanup(rec); 1564 return -1; 1565 } 1566 1567 /* Write null byte at the end of all the sec_buffs */ 1568 rec->desc.data[rec->desc.offset] = 0; 1569 rec->lib.data[rec->lib.offset] = 0; 1570 rec->env.data[rec->env.offset] = 0; 1571 rec->return_vals.data[rec->return_vals.offset] = 0; 1572 rec->exit_status.data[rec->exit_status.offset] = 0; 1573 rec->files.data[rec->files.offset] = 0; 1574 rec->diagnostics.data[rec->diagnostics.offset] = 0; 1575 rec->errors.data[rec->errors.offset] = 0; 1576 1577 /* 1578 * In case of a mdoc page: (sorry, no better place to put this code) 1579 * parse the comma separated list of names of man pages, 1580 * the first name will be stored in the mandb table, rest will be 1581 * treated as links and put in the mandb_links table. 1582 */ 1583 if (rec->page_type == MDOC) { 1584 char *tmp; 1585 rec->links = estrdup(rec->name); 1586 free(rec->name); 1587 size_t sz = strcspn(rec->links, " \0"); 1588 rec->name = emalloc(sz + 1); 1589 memcpy(rec->name, rec->links, sz); 1590 if(rec->name[sz - 1] == ',') 1591 rec->name[sz - 1] = 0; 1592 else 1593 rec->name[sz] = 0; 1594 while (rec->links[sz] == ' ') 1595 ++sz; 1596 tmp = estrdup(rec->links + sz); 1597 free(rec->links); 1598 rec->links = tmp; 1599 } 1600 1601/*------------------------ Populate the mandb table---------------------------*/ 1602 sqlstr = "INSERT INTO mandb VALUES (:section, :name, :name_desc, :desc," 1603 " :lib, :return_vals, :env, :files, :exit_status," 1604 " :diagnostics, :errors, :md5_hash, :machine)"; 1605 1606 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1607 if (rc != SQLITE_OK) 1608 goto Out; 1609 1610 idx = sqlite3_bind_parameter_index(stmt, ":name"); 1611 rc = sqlite3_bind_text(stmt, idx, rec->name, -1, NULL); 1612 if (rc != SQLITE_OK) { 1613 sqlite3_finalize(stmt); 1614 goto Out; 1615 } 1616 1617 idx = sqlite3_bind_parameter_index(stmt, ":section"); 1618 rc = sqlite3_bind_text(stmt, idx, rec->section, -1, NULL); 1619 if (rc != SQLITE_OK) { 1620 sqlite3_finalize(stmt); 1621 goto Out; 1622 } 1623 1624 idx = sqlite3_bind_parameter_index(stmt, ":name_desc"); 1625 rc = sqlite3_bind_text(stmt, idx, rec->name_desc, -1, NULL); 1626 if (rc != SQLITE_OK) { 1627 sqlite3_finalize(stmt); 1628 goto Out; 1629 } 1630 1631 idx = sqlite3_bind_parameter_index(stmt, ":desc"); 1632 rc = sqlite3_bind_text(stmt, idx, rec->desc.data, 1633 rec->desc.offset + 1, NULL); 1634 if (rc != SQLITE_OK) { 1635 sqlite3_finalize(stmt); 1636 goto Out; 1637 } 1638 1639 idx = sqlite3_bind_parameter_index(stmt, ":lib"); 1640 rc = sqlite3_bind_text(stmt, idx, rec->lib.data, 1641 rec->lib.offset + 1, NULL); 1642 if (rc != SQLITE_OK) { 1643 sqlite3_finalize(stmt); 1644 goto Out; 1645 } 1646 1647 idx = sqlite3_bind_parameter_index(stmt, ":return_vals"); 1648 rc = sqlite3_bind_text(stmt, idx, rec->return_vals.data, 1649 rec->return_vals.offset + 1, NULL); 1650 if (rc != SQLITE_OK) { 1651 sqlite3_finalize(stmt); 1652 goto Out; 1653 } 1654 1655 idx = sqlite3_bind_parameter_index(stmt, ":env"); 1656 rc = sqlite3_bind_text(stmt, idx, rec->env.data, 1657 rec->env.offset + 1, NULL); 1658 if (rc != SQLITE_OK) { 1659 sqlite3_finalize(stmt); 1660 goto Out; 1661 } 1662 1663 idx = sqlite3_bind_parameter_index(stmt, ":files"); 1664 rc = sqlite3_bind_text(stmt, idx, rec->files.data, 1665 rec->files.offset + 1, NULL); 1666 if (rc != SQLITE_OK) { 1667 sqlite3_finalize(stmt); 1668 goto Out; 1669 } 1670 1671 idx = sqlite3_bind_parameter_index(stmt, ":exit_status"); 1672 rc = sqlite3_bind_text(stmt, idx, rec->exit_status.data, 1673 rec->exit_status.offset + 1, NULL); 1674 if (rc != SQLITE_OK) { 1675 sqlite3_finalize(stmt); 1676 goto Out; 1677 } 1678 1679 idx = sqlite3_bind_parameter_index(stmt, ":diagnostics"); 1680 rc = sqlite3_bind_text(stmt, idx, rec->diagnostics.data, 1681 rec->diagnostics.offset + 1, NULL); 1682 if (rc != SQLITE_OK) { 1683 sqlite3_finalize(stmt); 1684 goto Out; 1685 } 1686 1687 idx = sqlite3_bind_parameter_index(stmt, ":errors"); 1688 rc = sqlite3_bind_text(stmt, idx, rec->errors.data, 1689 rec->errors.offset + 1, NULL); 1690 if (rc != SQLITE_OK) { 1691 sqlite3_finalize(stmt); 1692 goto Out; 1693 } 1694 1695 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1696 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1697 if (rc != SQLITE_OK) { 1698 sqlite3_finalize(stmt); 1699 goto Out; 1700 } 1701 1702 idx = sqlite3_bind_parameter_index(stmt, ":machine"); 1703 if (rec->machine) 1704 rc = sqlite3_bind_text(stmt, idx, rec->machine, -1, NULL); 1705 else 1706 rc = sqlite3_bind_null(stmt, idx); 1707 if (rc != SQLITE_OK) { 1708 sqlite3_finalize(stmt); 1709 goto Out; 1710 } 1711 1712 rc = sqlite3_step(stmt); 1713 if (rc != SQLITE_DONE) { 1714 sqlite3_finalize(stmt); 1715 goto Out; 1716 } 1717 1718 sqlite3_finalize(stmt); 1719 1720 /* Get the row id of the last inserted row */ 1721 mandb_rowid = sqlite3_last_insert_rowid(db); 1722 1723/*------------------------Populate the mandb_meta table-----------------------*/ 1724 sqlstr = "INSERT INTO mandb_meta VALUES (:device, :inode, :mtime," 1725 " :file, :md5_hash, :id)"; 1726 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1727 if (rc != SQLITE_OK) 1728 goto Out; 1729 1730 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1731 rc = sqlite3_bind_int64(stmt, idx, rec->device); 1732 if (rc != SQLITE_OK) { 1733 sqlite3_finalize(stmt); 1734 goto Out; 1735 } 1736 1737 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1738 rc = sqlite3_bind_int64(stmt, idx, rec->inode); 1739 if (rc != SQLITE_OK) { 1740 sqlite3_finalize(stmt); 1741 goto Out; 1742 } 1743 1744 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1745 rc = sqlite3_bind_int64(stmt, idx, rec->mtime); 1746 if (rc != SQLITE_OK) { 1747 sqlite3_finalize(stmt); 1748 goto Out; 1749 } 1750 1751 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1752 rc = sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1753 if (rc != SQLITE_OK) { 1754 sqlite3_finalize(stmt); 1755 goto Out; 1756 } 1757 1758 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1759 rc = sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1760 if (rc != SQLITE_OK) { 1761 sqlite3_finalize(stmt); 1762 goto Out; 1763 } 1764 1765 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1766 rc = sqlite3_bind_int64(stmt, idx, mandb_rowid); 1767 if (rc != SQLITE_OK) { 1768 sqlite3_finalize(stmt); 1769 goto Out; 1770 } 1771 1772 rc = sqlite3_step(stmt); 1773 sqlite3_finalize(stmt); 1774 if (rc == SQLITE_CONSTRAINT_UNIQUE) { 1775 /* The *most* probable reason for reaching here is that 1776 * the UNIQUE contraint on the file column of the mandb_meta 1777 * table was violated. 1778 * This can happen when a file was updated/modified. 1779 * To fix this we need to do two things: 1780 * 1. Delete the row for the older version of this file 1781 * from mandb table. 1782 * 2. Run an UPDATE query to update the row for this file 1783 * in the mandb_meta table. 1784 */ 1785 warnx("Trying to update index for %s", rec->file_path); 1786 char *sql = sqlite3_mprintf("DELETE FROM mandb " 1787 "WHERE rowid = (SELECT id" 1788 " FROM mandb_meta" 1789 " WHERE file = %Q)", 1790 rec->file_path); 1791 sqlite3_exec(db, sql, NULL, NULL, &errmsg); 1792 sqlite3_free(sql); 1793 if (errmsg != NULL) { 1794 if (mflags.verbosity) 1795 warnx("%s", errmsg); 1796 free(errmsg); 1797 } 1798 sqlstr = "UPDATE mandb_meta SET device = :device," 1799 " inode = :inode, mtime = :mtime, id = :id," 1800 " md5_hash = :md5 WHERE file = :file"; 1801 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1802 if (rc != SQLITE_OK) { 1803 if (mflags.verbosity) 1804 warnx("Update failed with error: %s", 1805 sqlite3_errmsg(db)); 1806 close_db(db); 1807 cleanup(rec); 1808 errx(EXIT_FAILURE, 1809 "Consider running makemandb with -f option"); 1810 } 1811 1812 idx = sqlite3_bind_parameter_index(stmt, ":device"); 1813 sqlite3_bind_int64(stmt, idx, rec->device); 1814 idx = sqlite3_bind_parameter_index(stmt, ":inode"); 1815 sqlite3_bind_int64(stmt, idx, rec->inode); 1816 idx = sqlite3_bind_parameter_index(stmt, ":mtime"); 1817 sqlite3_bind_int64(stmt, idx, rec->mtime); 1818 idx = sqlite3_bind_parameter_index(stmt, ":id"); 1819 sqlite3_bind_int64(stmt, idx, mandb_rowid); 1820 idx = sqlite3_bind_parameter_index(stmt, ":md5"); 1821 sqlite3_bind_text(stmt, idx, rec->md5_hash, -1, NULL); 1822 idx = sqlite3_bind_parameter_index(stmt, ":file"); 1823 sqlite3_bind_text(stmt, idx, rec->file_path, -1, NULL); 1824 rc = sqlite3_step(stmt); 1825 sqlite3_finalize(stmt); 1826 1827 if (rc != SQLITE_DONE) { 1828 if (mflags.verbosity) 1829 warnx("%s", sqlite3_errmsg(db)); 1830 close_db(db); 1831 cleanup(rec); 1832 errx(EXIT_FAILURE, 1833 "Consider running makemandb with -f option"); 1834 } 1835 } else if (rc != SQLITE_DONE) { 1836 /* Otherwise make this error fatal */ 1837 warnx("Failed at %s\n%s", rec->file_path, sqlite3_errmsg(db)); 1838 cleanup(rec); 1839 close_db(db); 1840 exit(EXIT_FAILURE); 1841 } 1842 1843/*------------------------ Populate the mandb_links table---------------------*/ 1844 char *str = NULL; 1845 char *links; 1846 if (rec->links && strlen(rec->links)) { 1847 links = rec->links; 1848 for(ln = strtok(links, " "); ln; ln = strtok(NULL, " ")) { 1849 if (ln[0] == ',') 1850 ln++; 1851 if(ln[strlen(ln) - 1] == ',') 1852 ln[strlen(ln) - 1] = 0; 1853 1854 str = sqlite3_mprintf("INSERT INTO mandb_links" 1855 " VALUES (%Q, %Q, %Q, %Q, %Q)", 1856 ln, rec->name, rec->section, 1857 rec->machine, rec->md5_hash); 1858 sqlite3_exec(db, str, NULL, NULL, &errmsg); 1859 sqlite3_free(str); 1860 if (errmsg != NULL) { 1861 warnx("%s", errmsg); 1862 cleanup(rec); 1863 free(errmsg); 1864 return -1; 1865 } 1866 } 1867 } 1868 1869 cleanup(rec); 1870 return 0; 1871 1872 Out: 1873 if (mflags.verbosity) 1874 warnx("%s", sqlite3_errmsg(db)); 1875 cleanup(rec); 1876 return -1; 1877} 1878 1879/* 1880 * check_md5-- 1881 * Generates the md5 hash of the file and checks if it already doesn't exist 1882 * in the table (passed as the 3rd parameter). 1883 * This function is being used to avoid hardlinks. 1884 * On successful completion it will also set the value of the fourth parameter 1885 * to the md5 hash of the file (computed previously). It is the responsibility 1886 * of the caller to free this buffer. 1887 * Return values: 1888 * -1: If an error occurs somewhere and sets the md5 return buffer to NULL. 1889 * 0: If the md5 hash does not exist in the table. 1890 * 1: If the hash exists in the database. 1891 */ 1892static int 1893check_md5(const char *file, sqlite3 *db, const char *table, char **md5sum, 1894 void *buf, size_t buflen) 1895{ 1896 int rc = 0; 1897 int idx = -1; 1898 char *sqlstr = NULL; 1899 sqlite3_stmt *stmt = NULL; 1900 1901 assert(file != NULL); 1902 *md5sum = MD5Data(buf, buflen, NULL); 1903 if (*md5sum == NULL) { 1904 if (mflags.verbosity) 1905 warn("md5 failed: %s", file); 1906 return -1; 1907 } 1908 1909 easprintf(&sqlstr, "SELECT * FROM %s WHERE md5_hash = :md5_hash", 1910 table); 1911 rc = sqlite3_prepare_v2(db, sqlstr, -1, &stmt, NULL); 1912 if (rc != SQLITE_OK) { 1913 free(sqlstr); 1914 free(*md5sum); 1915 *md5sum = NULL; 1916 return -1; 1917 } 1918 1919 idx = sqlite3_bind_parameter_index(stmt, ":md5_hash"); 1920 rc = sqlite3_bind_text(stmt, idx, *md5sum, -1, NULL); 1921 if (rc != SQLITE_OK) { 1922 if (mflags.verbosity) 1923 warnx("%s", sqlite3_errmsg(db)); 1924 sqlite3_finalize(stmt); 1925 free(sqlstr); 1926 free(*md5sum); 1927 *md5sum = NULL; 1928 return -1; 1929 } 1930 1931 if (sqlite3_step(stmt) == SQLITE_ROW) { 1932 sqlite3_finalize(stmt); 1933 free(sqlstr); 1934 return 0; 1935 } 1936 1937 sqlite3_finalize(stmt); 1938 free(sqlstr); 1939 return 1; 1940} 1941 1942/* Optimize the index for faster search */ 1943static void 1944optimize(sqlite3 *db) 1945{ 1946 const char *sqlstr; 1947 char *errmsg = NULL; 1948 1949 if (mflags.verbosity == 2) 1950 printf("Optimizing the database index\n"); 1951 sqlstr = "INSERT INTO mandb(mandb) VALUES (\'optimize\');" 1952 "VACUUM"; 1953 sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg); 1954 if (errmsg != NULL) { 1955 if (mflags.verbosity) 1956 warnx("%s", errmsg); 1957 free(errmsg); 1958 return; 1959 } 1960} 1961 1962/* 1963 * cleanup -- 1964 * cleans up the global buffers 1965 */ 1966static void 1967cleanup(mandb_rec *rec) 1968{ 1969 rec->desc.offset = 0; 1970 rec->lib.offset = 0; 1971 rec->return_vals.offset = 0; 1972 rec->env.offset = 0; 1973 rec->exit_status.offset = 0; 1974 rec->diagnostics.offset = 0; 1975 rec->errors.offset = 0; 1976 rec->files.offset = 0; 1977 1978 free(rec->machine); 1979 rec->machine = NULL; 1980 1981 free(rec->links); 1982 rec->links = NULL; 1983 1984 free(rec->file_path); 1985 rec->file_path = NULL; 1986 1987 free(rec->name); 1988 rec->name = NULL; 1989 1990 free(rec->name_desc); 1991 rec->name_desc = NULL; 1992 1993 free(rec->md5_hash); 1994 rec->md5_hash = NULL; 1995 1996 free(rec->section); 1997 rec->section = NULL; 1998} 1999 2000/* 2001 * init_secbuffs-- 2002 * Sets the value of buflen for all the sec_buff field of rec. And then 2003 * allocate memory to each sec_buff member of rec. 2004 */ 2005static void 2006init_secbuffs(mandb_rec *rec) 2007{ 2008 /* 2009 * Some sec_buff might need more memory, for example desc, 2010 * which stores the data of the DESCRIPTION section, 2011 * while some might need very small amount of memory. 2012 * Therefore explicitly setting the value of buflen field for 2013 * each sec_buff. 2014 */ 2015 rec->desc.buflen = 10 * BUFLEN; 2016 rec->desc.data = emalloc(rec->desc.buflen); 2017 rec->desc.offset = 0; 2018 2019 rec->lib.buflen = BUFLEN / 2; 2020 rec->lib.data = emalloc(rec->lib.buflen); 2021 rec->lib.offset = 0; 2022 2023 rec->return_vals.buflen = BUFLEN; 2024 rec->return_vals.data = emalloc(rec->return_vals.buflen); 2025 rec->return_vals.offset = 0; 2026 2027 rec->exit_status.buflen = BUFLEN; 2028 rec->exit_status.data = emalloc(rec->exit_status.buflen); 2029 rec->exit_status.offset = 0; 2030 2031 rec->env.buflen = BUFLEN; 2032 rec->env.data = emalloc(rec->env.buflen); 2033 rec->env.offset = 0; 2034 2035 rec->files.buflen = BUFLEN; 2036 rec->files.data = emalloc(rec->files.buflen); 2037 rec->files.offset = 0; 2038 2039 rec->diagnostics.buflen = BUFLEN; 2040 rec->diagnostics.data = emalloc(rec->diagnostics.buflen); 2041 rec->diagnostics.offset = 0; 2042 2043 rec->errors.buflen = BUFLEN; 2044 rec->errors.data = emalloc(rec->errors.buflen); 2045 rec->errors.offset = 0; 2046} 2047 2048/* 2049 * free_secbuffs-- 2050 * This function should be called at the end, when all the pages have been 2051 * parsed. 2052 * It frees the memory allocated to sec_buffs by init_secbuffs in the starting. 2053 */ 2054static void 2055free_secbuffs(mandb_rec *rec) 2056{ 2057 free(rec->desc.data); 2058 free(rec->lib.data); 2059 free(rec->return_vals.data); 2060 free(rec->exit_status.data); 2061 free(rec->env.data); 2062 free(rec->files.data); 2063 free(rec->diagnostics.data); 2064 free(rec->errors.data); 2065} 2066 2067static void 2068replace_hyph(char *str) 2069{ 2070 char *iter = str; 2071 while ((iter = strchr(iter, ASCII_HYPH)) != NULL) 2072 *iter = '-'; 2073 2074 iter = str; 2075 while ((iter = strchr(iter, ASCII_NBRSP)) != NULL) 2076 *iter = '-'; 2077} 2078 2079static char * 2080parse_escape(const char *str) 2081{ 2082 const char *backslash, *last_backslash; 2083 char *result, *iter; 2084 size_t len; 2085 2086 assert(str); 2087 2088 last_backslash = str; 2089 backslash = strchr(str, '\\'); 2090 if (backslash == NULL) { 2091 result = estrdup(str); 2092 replace_hyph(result); 2093 return result; 2094 } 2095 2096 result = emalloc(strlen(str) + 1); 2097 iter = result; 2098 2099 do { 2100 len = backslash - last_backslash; 2101 memcpy(iter, last_backslash, len); 2102 iter += len; 2103 if (backslash[1] == '-' || backslash[1] == ' ') { 2104 *iter++ = backslash[1]; 2105 last_backslash = backslash + 2; 2106 backslash = strchr(last_backslash, '\\'); 2107 } else { 2108 ++backslash; 2109 mandoc_escape(&backslash, NULL, NULL); 2110 last_backslash = backslash; 2111 if (backslash == NULL) 2112 break; 2113 backslash = strchr(last_backslash, '\\'); 2114 } 2115 } while (backslash != NULL); 2116 if (last_backslash != NULL) 2117 strcpy(iter, last_backslash); 2118 2119 replace_hyph(result); 2120 return result; 2121} 2122 2123/* 2124 * append-- 2125 * Concatenates a space and src at the end of sbuff->data (much like concat in 2126 * apropos-utils.c). 2127 * Rather than reallocating space for writing data, it uses the value of the 2128 * offset field of sec_buff to write new data at the free space left in the 2129 * buffer. 2130 * In case the size of the data to be appended exceeds the number of bytes left 2131 * in the buffer, it reallocates buflen number of bytes and then continues. 2132 * Value of offset field should be adjusted as new data is written. 2133 * 2134 * NOTE: This function does not write the null byte at the end of the buffers, 2135 * write a null byte at the position pointed to by offset before inserting data 2136 * in the db. 2137 */ 2138static void 2139append(secbuff *sbuff, const char *src) 2140{ 2141 short flag = 0; 2142 size_t srclen, newlen; 2143 char *temp; 2144 2145 assert(src != NULL); 2146 temp = parse_escape(src); 2147 srclen = strlen(temp); 2148 2149 if (sbuff->data == NULL) { 2150 sbuff->data = emalloc(sbuff->buflen); 2151 sbuff->offset = 0; 2152 } 2153 2154 newlen = sbuff->offset + srclen + 2; 2155 if (newlen >= sbuff->buflen) { 2156 while (sbuff->buflen < newlen) 2157 sbuff->buflen += sbuff->buflen; 2158 sbuff->data = erealloc(sbuff->data, sbuff->buflen); 2159 flag = 1; 2160 } 2161 2162 /* Append a space at the end of the buffer. */ 2163 if (sbuff->offset || flag) 2164 sbuff->data[sbuff->offset++] = ' '; 2165 /* Now, copy src at the end of the buffer. */ 2166 memcpy(sbuff->data + sbuff->offset, temp, srclen); 2167 sbuff->offset += srclen; 2168 free(temp); 2169} 2170 2171static void 2172usage(void) 2173{ 2174 fprintf(stderr, "Usage: %s [-floQqv] [-C path]\n", getprogname()); 2175 exit(1); 2176} 2177