1 /* Id: mandocdb.c,v 1.218 2016/07/12 05:18:38 kristaps Exp */ 2 /* 3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2011-2016 Ingo Schwarze <schwarze@openbsd.org> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 */ 18 #include "config.h" 19 20 #include <sys/types.h> 21 #include <sys/stat.h> 22 #include <sys/wait.h> 23 24 #include <assert.h> 25 #include <ctype.h> 26 #if HAVE_ERR 27 #include <err.h> 28 #endif 29 #include <errno.h> 30 #include <fcntl.h> 31 #if HAVE_FTS 32 #include <fts.h> 33 #else 34 #include "compat_fts.h" 35 #endif 36 #include <limits.h> 37 #if HAVE_SANDBOX_INIT 38 #include <sandbox.h> 39 #endif 40 #include <stddef.h> 41 #include <stdio.h> 42 #include <stdint.h> 43 #include <stdlib.h> 44 #include <string.h> 45 #include <unistd.h> 46 47 #include <sqlite3.h> 48 49 #include "mandoc_aux.h" 50 #include "mandoc_ohash.h" 51 #include "mandoc.h" 52 #include "roff.h" 53 #include "mdoc.h" 54 #include "main.h" 55 #include "man.h" 56 #include "manconf.h" 57 #include "mansearch.h" 58 59 extern int mansearch_keymax; 60 extern const char *const mansearch_keynames[]; 61 62 #define SQL_EXEC(_v) \ 63 if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \ 64 say("", "%s: %s", (_v), sqlite3_errmsg(db)) 65 #define SQL_BIND_TEXT(_s, _i, _v) \ 66 if (SQLITE_OK != sqlite3_bind_text \ 67 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \ 68 say(mlink->file, "%s", sqlite3_errmsg(db)) 69 #define SQL_BIND_INT(_s, _i, _v) \ 70 if (SQLITE_OK != sqlite3_bind_int \ 71 ((_s), (_i)++, (_v))) \ 72 say(mlink->file, "%s", sqlite3_errmsg(db)) 73 #define SQL_BIND_INT64(_s, _i, _v) \ 74 if (SQLITE_OK != sqlite3_bind_int64 \ 75 ((_s), (_i)++, (_v))) \ 76 say(mlink->file, "%s", sqlite3_errmsg(db)) 77 #define SQL_STEP(_s) \ 78 if (SQLITE_DONE != sqlite3_step((_s))) \ 79 say(mlink->file, "%s", sqlite3_errmsg(db)) 80 81 enum op { 82 OP_DEFAULT = 0, /* new dbs from dir list or default config */ 83 OP_CONFFILE, /* new databases from custom config file */ 84 OP_UPDATE, /* delete/add entries in existing database */ 85 OP_DELETE, /* delete entries from existing database */ 86 OP_TEST /* change no databases, report potential problems */ 87 }; 88 89 struct str { 90 const struct mpage *mpage; /* if set, the owning parse */ 91 uint64_t mask; /* bitmask in sequence */ 92 char key[]; /* rendered text */ 93 }; 94 95 struct inodev { 96 ino_t st_ino; 97 dev_t st_dev; 98 }; 99 100 struct mpage { 101 struct inodev inodev; /* used for hashing routine */ 102 int64_t pageid; /* pageid in mpages SQL table */ 103 char *sec; /* section from file content */ 104 char *arch; /* architecture from file content */ 105 char *title; /* title from file content */ 106 char *desc; /* description from file content */ 107 struct mlink *mlinks; /* singly linked list */ 108 int form; /* format from file content */ 109 int name_head_done; 110 }; 111 112 struct mlink { 113 char file[PATH_MAX]; /* filename rel. to manpath */ 114 char *dsec; /* section from directory */ 115 char *arch; /* architecture from directory */ 116 char *name; /* name from file name (not empty) */ 117 char *fsec; /* section from file name suffix */ 118 struct mlink *next; /* singly linked list */ 119 struct mpage *mpage; /* parent */ 120 int dform; /* format from directory */ 121 int fform; /* format from file name suffix */ 122 int gzip; /* filename has a .gz suffix */ 123 }; 124 125 enum stmt { 126 STMT_DELETE_PAGE = 0, /* delete mpage */ 127 STMT_INSERT_PAGE, /* insert mpage */ 128 STMT_INSERT_LINK, /* insert mlink */ 129 STMT_INSERT_NAME, /* insert name */ 130 STMT_SELECT_NAME, /* retrieve existing name flags */ 131 STMT_INSERT_KEY, /* insert parsed key */ 132 STMT__MAX 133 }; 134 135 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *, 136 const struct roff_node *); 137 138 struct mdoc_handler { 139 mdoc_fp fp; /* optional handler */ 140 uint64_t mask; /* set unless handler returns 0 */ 141 }; 142 143 static void dbclose(int); 144 static void dbadd(struct mpage *); 145 static void dbadd_mlink(const struct mlink *mlink); 146 static void dbadd_mlink_name(const struct mlink *mlink); 147 static int dbopen(int); 148 static void dbprune(void); 149 static void filescan(const char *); 150 static void mlink_add(struct mlink *, const struct stat *); 151 static void mlink_check(struct mpage *, struct mlink *); 152 static void mlink_free(struct mlink *); 153 static void mlinks_undupe(struct mpage *); 154 static void mpages_free(void); 155 static void mpages_merge(struct mparse *); 156 static void names_check(void); 157 static void parse_cat(struct mpage *, int); 158 static void parse_man(struct mpage *, const struct roff_meta *, 159 const struct roff_node *); 160 static void parse_mdoc(struct mpage *, const struct roff_meta *, 161 const struct roff_node *); 162 static int parse_mdoc_head(struct mpage *, const struct roff_meta *, 163 const struct roff_node *); 164 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *, 165 const struct roff_node *); 166 static void parse_mdoc_fname(struct mpage *, const struct roff_node *); 167 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *, 168 const struct roff_node *); 169 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *, 170 const struct roff_node *); 171 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *, 172 const struct roff_node *); 173 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *, 174 const struct roff_node *); 175 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *, 176 const struct roff_node *); 177 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *, 178 const struct roff_node *); 179 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *, 180 const struct roff_node *); 181 static void putkey(const struct mpage *, char *, uint64_t); 182 static void putkeys(const struct mpage *, char *, size_t, uint64_t); 183 static void putmdockey(const struct mpage *, 184 const struct roff_node *, uint64_t); 185 static int render_string(char **, size_t *); 186 static void say(const char *, const char *, ...); 187 static int set_basedir(const char *, int); 188 static int treescan(void); 189 static size_t utf8(unsigned int, char [7]); 190 191 static char tempfilename[32]; 192 static int nodb; /* no database changes */ 193 static int mparse_options; /* abort the parse early */ 194 static int use_all; /* use all found files */ 195 static int debug; /* print what we're doing */ 196 static int warnings; /* warn about crap */ 197 static int write_utf8; /* write UTF-8 output; else ASCII */ 198 static int exitcode; /* to be returned by main */ 199 static enum op op; /* operational mode */ 200 static char basedir[PATH_MAX]; /* current base directory */ 201 static struct ohash mpages; /* table of distinct manual pages */ 202 static struct ohash mlinks; /* table of directory entries */ 203 static struct ohash names; /* table of all names */ 204 static struct ohash strings; /* table of all strings */ 205 static sqlite3 *db = NULL; /* current database */ 206 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */ 207 static uint64_t name_mask; 208 209 static const struct mdoc_handler mdocs[MDOC_MAX] = { 210 { NULL, 0 }, /* Ap */ 211 { NULL, 0 }, /* Dd */ 212 { NULL, 0 }, /* Dt */ 213 { NULL, 0 }, /* Os */ 214 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */ 215 { parse_mdoc_head, TYPE_Ss }, /* Ss */ 216 { NULL, 0 }, /* Pp */ 217 { NULL, 0 }, /* D1 */ 218 { NULL, 0 }, /* Dl */ 219 { NULL, 0 }, /* Bd */ 220 { NULL, 0 }, /* Ed */ 221 { NULL, 0 }, /* Bl */ 222 { NULL, 0 }, /* El */ 223 { NULL, 0 }, /* It */ 224 { NULL, 0 }, /* Ad */ 225 { NULL, TYPE_An }, /* An */ 226 { NULL, TYPE_Ar }, /* Ar */ 227 { NULL, TYPE_Cd }, /* Cd */ 228 { NULL, TYPE_Cm }, /* Cm */ 229 { NULL, TYPE_Dv }, /* Dv */ 230 { NULL, TYPE_Er }, /* Er */ 231 { NULL, TYPE_Ev }, /* Ev */ 232 { NULL, 0 }, /* Ex */ 233 { NULL, TYPE_Fa }, /* Fa */ 234 { parse_mdoc_Fd, 0 }, /* Fd */ 235 { NULL, TYPE_Fl }, /* Fl */ 236 { parse_mdoc_Fn, 0 }, /* Fn */ 237 { NULL, TYPE_Ft }, /* Ft */ 238 { NULL, TYPE_Ic }, /* Ic */ 239 { NULL, TYPE_In }, /* In */ 240 { NULL, TYPE_Li }, /* Li */ 241 { parse_mdoc_Nd, 0 }, /* Nd */ 242 { parse_mdoc_Nm, 0 }, /* Nm */ 243 { NULL, 0 }, /* Op */ 244 { NULL, 0 }, /* Ot */ 245 { NULL, TYPE_Pa }, /* Pa */ 246 { NULL, 0 }, /* Rv */ 247 { NULL, TYPE_St }, /* St */ 248 { parse_mdoc_Va, TYPE_Va }, /* Va */ 249 { parse_mdoc_Va, TYPE_Vt }, /* Vt */ 250 { parse_mdoc_Xr, 0 }, /* Xr */ 251 { NULL, 0 }, /* %A */ 252 { NULL, 0 }, /* %B */ 253 { NULL, 0 }, /* %D */ 254 { NULL, 0 }, /* %I */ 255 { NULL, 0 }, /* %J */ 256 { NULL, 0 }, /* %N */ 257 { NULL, 0 }, /* %O */ 258 { NULL, 0 }, /* %P */ 259 { NULL, 0 }, /* %R */ 260 { NULL, 0 }, /* %T */ 261 { NULL, 0 }, /* %V */ 262 { NULL, 0 }, /* Ac */ 263 { NULL, 0 }, /* Ao */ 264 { NULL, 0 }, /* Aq */ 265 { NULL, TYPE_At }, /* At */ 266 { NULL, 0 }, /* Bc */ 267 { NULL, 0 }, /* Bf */ 268 { NULL, 0 }, /* Bo */ 269 { NULL, 0 }, /* Bq */ 270 { NULL, TYPE_Bsx }, /* Bsx */ 271 { NULL, TYPE_Bx }, /* Bx */ 272 { NULL, 0 }, /* Db */ 273 { NULL, 0 }, /* Dc */ 274 { NULL, 0 }, /* Do */ 275 { NULL, 0 }, /* Dq */ 276 { NULL, 0 }, /* Ec */ 277 { NULL, 0 }, /* Ef */ 278 { NULL, TYPE_Em }, /* Em */ 279 { NULL, 0 }, /* Eo */ 280 { NULL, TYPE_Fx }, /* Fx */ 281 { NULL, TYPE_Ms }, /* Ms */ 282 { NULL, 0 }, /* No */ 283 { NULL, 0 }, /* Ns */ 284 { NULL, TYPE_Nx }, /* Nx */ 285 { NULL, TYPE_Ox }, /* Ox */ 286 { NULL, 0 }, /* Pc */ 287 { NULL, 0 }, /* Pf */ 288 { NULL, 0 }, /* Po */ 289 { NULL, 0 }, /* Pq */ 290 { NULL, 0 }, /* Qc */ 291 { NULL, 0 }, /* Ql */ 292 { NULL, 0 }, /* Qo */ 293 { NULL, 0 }, /* Qq */ 294 { NULL, 0 }, /* Re */ 295 { NULL, 0 }, /* Rs */ 296 { NULL, 0 }, /* Sc */ 297 { NULL, 0 }, /* So */ 298 { NULL, 0 }, /* Sq */ 299 { NULL, 0 }, /* Sm */ 300 { NULL, 0 }, /* Sx */ 301 { NULL, TYPE_Sy }, /* Sy */ 302 { NULL, TYPE_Tn }, /* Tn */ 303 { NULL, 0 }, /* Ux */ 304 { NULL, 0 }, /* Xc */ 305 { NULL, 0 }, /* Xo */ 306 { parse_mdoc_Fo, 0 }, /* Fo */ 307 { NULL, 0 }, /* Fc */ 308 { NULL, 0 }, /* Oo */ 309 { NULL, 0 }, /* Oc */ 310 { NULL, 0 }, /* Bk */ 311 { NULL, 0 }, /* Ek */ 312 { NULL, 0 }, /* Bt */ 313 { NULL, 0 }, /* Hf */ 314 { NULL, 0 }, /* Fr */ 315 { NULL, 0 }, /* Ud */ 316 { NULL, TYPE_Lb }, /* Lb */ 317 { NULL, 0 }, /* Lp */ 318 { NULL, TYPE_Lk }, /* Lk */ 319 { NULL, TYPE_Mt }, /* Mt */ 320 { NULL, 0 }, /* Brq */ 321 { NULL, 0 }, /* Bro */ 322 { NULL, 0 }, /* Brc */ 323 { NULL, 0 }, /* %C */ 324 { NULL, 0 }, /* Es */ 325 { NULL, 0 }, /* En */ 326 { NULL, TYPE_Dx }, /* Dx */ 327 { NULL, 0 }, /* %Q */ 328 { NULL, 0 }, /* br */ 329 { NULL, 0 }, /* sp */ 330 { NULL, 0 }, /* %U */ 331 { NULL, 0 }, /* Ta */ 332 { NULL, 0 }, /* ll */ 333 }; 334 335 336 int 337 mandocdb(int argc, char *argv[]) 338 { 339 struct manconf conf; 340 struct mparse *mp; 341 const char *path_arg, *progname; 342 size_t j, sz; 343 int ch, i; 344 345 #if HAVE_PLEDGE 346 if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) { 347 warn("pledge"); 348 return (int)MANDOCLEVEL_SYSERR; 349 } 350 #endif 351 352 #if HAVE_SANDBOX_INIT 353 if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) { 354 warnx("sandbox_init"); 355 return (int)MANDOCLEVEL_SYSERR; 356 } 357 #endif 358 359 memset(&conf, 0, sizeof(conf)); 360 memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *)); 361 362 /* 363 * We accept a few different invocations. 364 * The CHECKOP macro makes sure that invocation styles don't 365 * clobber each other. 366 */ 367 #define CHECKOP(_op, _ch) do \ 368 if (OP_DEFAULT != (_op)) { \ 369 warnx("-%c: Conflicting option", (_ch)); \ 370 goto usage; \ 371 } while (/*CONSTCOND*/0) 372 373 path_arg = NULL; 374 op = OP_DEFAULT; 375 376 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v"))) 377 switch (ch) { 378 case 'a': 379 use_all = 1; 380 break; 381 case 'C': 382 CHECKOP(op, ch); 383 path_arg = optarg; 384 op = OP_CONFFILE; 385 break; 386 case 'D': 387 debug++; 388 break; 389 case 'd': 390 CHECKOP(op, ch); 391 path_arg = optarg; 392 op = OP_UPDATE; 393 break; 394 case 'n': 395 nodb = 1; 396 break; 397 case 'p': 398 warnings = 1; 399 break; 400 case 'Q': 401 mparse_options |= MPARSE_QUICK; 402 break; 403 case 'T': 404 if (strcmp(optarg, "utf8")) { 405 warnx("-T%s: Unsupported output format", 406 optarg); 407 goto usage; 408 } 409 write_utf8 = 1; 410 break; 411 case 't': 412 CHECKOP(op, ch); 413 dup2(STDOUT_FILENO, STDERR_FILENO); 414 op = OP_TEST; 415 nodb = warnings = 1; 416 break; 417 case 'u': 418 CHECKOP(op, ch); 419 path_arg = optarg; 420 op = OP_DELETE; 421 break; 422 case 'v': 423 /* Compatibility with espie@'s makewhatis. */ 424 break; 425 default: 426 goto usage; 427 } 428 429 argc -= optind; 430 argv += optind; 431 432 #if HAVE_PLEDGE 433 if (nodb) { 434 if (pledge("stdio rpath", NULL) == -1) { 435 warn("pledge"); 436 return (int)MANDOCLEVEL_SYSERR; 437 } 438 } 439 #endif 440 441 if (OP_CONFFILE == op && argc > 0) { 442 warnx("-C: Too many arguments"); 443 goto usage; 444 } 445 446 exitcode = (int)MANDOCLEVEL_OK; 447 mchars_alloc(); 448 mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL); 449 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev)); 450 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file)); 451 452 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) { 453 454 /* 455 * Most of these deal with a specific directory. 456 * Jump into that directory first. 457 */ 458 if (OP_TEST != op && 0 == set_basedir(path_arg, 1)) 459 goto out; 460 461 if (dbopen(1)) { 462 /* 463 * The existing database is usable. Process 464 * all files specified on the command-line. 465 */ 466 #if HAVE_PLEDGE 467 if (!nodb) { 468 if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) { 469 warn("pledge"); 470 exitcode = (int)MANDOCLEVEL_SYSERR; 471 goto out; 472 } 473 } 474 #endif 475 use_all = 1; 476 for (i = 0; i < argc; i++) 477 filescan(argv[i]); 478 if (OP_TEST != op) 479 dbprune(); 480 } else { 481 /* 482 * Database missing or corrupt. 483 * Recreate from scratch. 484 */ 485 exitcode = (int)MANDOCLEVEL_OK; 486 op = OP_DEFAULT; 487 if (0 == treescan()) 488 goto out; 489 if (0 == dbopen(0)) 490 goto out; 491 } 492 if (OP_DELETE != op) 493 mpages_merge(mp); 494 dbclose(OP_DEFAULT == op ? 0 : 1); 495 } else { 496 /* 497 * If we have arguments, use them as our manpaths. 498 * If we don't, grok from manpath(1) or however else 499 * manconf_parse() wants to do it. 500 */ 501 if (argc > 0) { 502 conf.manpath.paths = mandoc_reallocarray(NULL, 503 argc, sizeof(char *)); 504 conf.manpath.sz = (size_t)argc; 505 for (i = 0; i < argc; i++) 506 conf.manpath.paths[i] = mandoc_strdup(argv[i]); 507 } else 508 manconf_parse(&conf, path_arg, NULL, NULL); 509 510 if (conf.manpath.sz == 0) { 511 exitcode = (int)MANDOCLEVEL_BADARG; 512 say("", "Empty manpath"); 513 } 514 515 /* 516 * First scan the tree rooted at a base directory, then 517 * build a new database and finally move it into place. 518 * Ignore zero-length directories and strip trailing 519 * slashes. 520 */ 521 for (j = 0; j < conf.manpath.sz; j++) { 522 sz = strlen(conf.manpath.paths[j]); 523 if (sz && conf.manpath.paths[j][sz - 1] == '/') 524 conf.manpath.paths[j][--sz] = '\0'; 525 if (0 == sz) 526 continue; 527 528 if (j) { 529 mandoc_ohash_init(&mpages, 6, 530 offsetof(struct mpage, inodev)); 531 mandoc_ohash_init(&mlinks, 6, 532 offsetof(struct mlink, file)); 533 } 534 535 if ( ! set_basedir(conf.manpath.paths[j], argc > 0)) 536 continue; 537 if (0 == treescan()) 538 continue; 539 if (0 == dbopen(0)) 540 continue; 541 542 mpages_merge(mp); 543 if (warnings && !nodb && 544 ! (MPARSE_QUICK & mparse_options)) 545 names_check(); 546 dbclose(0); 547 548 if (j + 1 < conf.manpath.sz) { 549 mpages_free(); 550 ohash_delete(&mpages); 551 ohash_delete(&mlinks); 552 } 553 } 554 } 555 out: 556 manconf_free(&conf); 557 mparse_free(mp); 558 mchars_free(); 559 mpages_free(); 560 ohash_delete(&mpages); 561 ohash_delete(&mlinks); 562 return exitcode; 563 usage: 564 progname = getprogname(); 565 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n" 566 " %s [-aDnpQ] [-Tutf8] dir ...\n" 567 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n" 568 " %s [-Dnp] -u dir [file ...]\n" 569 " %s [-Q] -t file ...\n", 570 progname, progname, progname, progname, progname); 571 572 return (int)MANDOCLEVEL_BADARG; 573 } 574 575 /* 576 * Scan a directory tree rooted at "basedir" for manpages. 577 * We use fts(), scanning directory parts along the way for clues to our 578 * section and architecture. 579 * 580 * If use_all has been specified, grok all files. 581 * If not, sanitise paths to the following: 582 * 583 * [./]man*[/<arch>]/<name>.<section> 584 * or 585 * [./]cat<section>[/<arch>]/<name>.0 586 * 587 * TODO: accommodate for multi-language directories. 588 */ 589 static int 590 treescan(void) 591 { 592 char buf[PATH_MAX]; 593 FTS *f; 594 FTSENT *ff; 595 struct mlink *mlink; 596 int dform, gzip; 597 char *dsec, *arch, *fsec, *cp; 598 const char *path; 599 const char *argv[2]; 600 601 argv[0] = "."; 602 argv[1] = NULL; 603 604 f = fts_open((void *)argv, 605 FTS_PHYSICAL | FTS_NOCHDIR, NULL); 606 if (f == NULL) { 607 exitcode = (int)MANDOCLEVEL_SYSERR; 608 say("", "&fts_open"); 609 return 0; 610 } 611 612 dsec = arch = NULL; 613 dform = FORM_NONE; 614 615 while ((ff = fts_read(f)) != NULL) { 616 path = ff->fts_path + 2; 617 switch (ff->fts_info) { 618 619 /* 620 * Symbolic links require various sanity checks, 621 * then get handled just like regular files. 622 */ 623 case FTS_SL: 624 if (realpath(path, buf) == NULL) { 625 if (warnings) 626 say(path, "&realpath"); 627 continue; 628 } 629 if (strstr(buf, basedir) != buf 630 #ifdef HOMEBREWDIR 631 && strstr(buf, HOMEBREWDIR) != buf 632 #endif 633 ) { 634 if (warnings) say("", 635 "%s: outside base directory", buf); 636 continue; 637 } 638 /* Use logical inode to avoid mpages dupe. */ 639 if (stat(path, ff->fts_statp) == -1) { 640 if (warnings) 641 say(path, "&stat"); 642 continue; 643 } 644 /* FALLTHROUGH */ 645 646 /* 647 * If we're a regular file, add an mlink by using the 648 * stored directory data and handling the filename. 649 */ 650 case FTS_F: 651 if ( ! strcmp(path, MANDOC_DB)) 652 continue; 653 if ( ! use_all && ff->fts_level < 2) { 654 if (warnings) 655 say(path, "Extraneous file"); 656 continue; 657 } 658 gzip = 0; 659 fsec = NULL; 660 while (fsec == NULL) { 661 fsec = strrchr(ff->fts_name, '.'); 662 if (fsec == NULL || strcmp(fsec+1, "gz")) 663 break; 664 gzip = 1; 665 *fsec = '\0'; 666 fsec = NULL; 667 } 668 if (fsec == NULL) { 669 if ( ! use_all) { 670 if (warnings) 671 say(path, 672 "No filename suffix"); 673 continue; 674 } 675 } else if ( ! strcmp(++fsec, "html")) { 676 if (warnings) 677 say(path, "Skip html"); 678 continue; 679 } else if ( ! strcmp(fsec, "ps")) { 680 if (warnings) 681 say(path, "Skip ps"); 682 continue; 683 } else if ( ! strcmp(fsec, "pdf")) { 684 if (warnings) 685 say(path, "Skip pdf"); 686 continue; 687 } else if ( ! use_all && 688 ((dform == FORM_SRC && 689 strncmp(fsec, dsec, strlen(dsec))) || 690 (dform == FORM_CAT && strcmp(fsec, "0")))) { 691 if (warnings) 692 say(path, "Wrong filename suffix"); 693 continue; 694 } else 695 fsec[-1] = '\0'; 696 697 mlink = mandoc_calloc(1, sizeof(struct mlink)); 698 if (strlcpy(mlink->file, path, 699 sizeof(mlink->file)) >= 700 sizeof(mlink->file)) { 701 say(path, "Filename too long"); 702 free(mlink); 703 continue; 704 } 705 mlink->dform = dform; 706 mlink->dsec = dsec; 707 mlink->arch = arch; 708 mlink->name = ff->fts_name; 709 mlink->fsec = fsec; 710 mlink->gzip = gzip; 711 mlink_add(mlink, ff->fts_statp); 712 continue; 713 714 case FTS_D: 715 case FTS_DP: 716 break; 717 718 default: 719 if (warnings) 720 say(path, "Not a regular file"); 721 continue; 722 } 723 724 switch (ff->fts_level) { 725 case 0: 726 /* Ignore the root directory. */ 727 break; 728 case 1: 729 /* 730 * This might contain manX/ or catX/. 731 * Try to infer this from the name. 732 * If we're not in use_all, enforce it. 733 */ 734 cp = ff->fts_name; 735 if (ff->fts_info == FTS_DP) { 736 dform = FORM_NONE; 737 dsec = NULL; 738 break; 739 } 740 741 if ( ! strncmp(cp, "man", 3)) { 742 dform = FORM_SRC; 743 dsec = cp + 3; 744 } else if ( ! strncmp(cp, "cat", 3)) { 745 dform = FORM_CAT; 746 dsec = cp + 3; 747 } else { 748 dform = FORM_NONE; 749 dsec = NULL; 750 } 751 752 if (dsec != NULL || use_all) 753 break; 754 755 if (warnings) 756 say(path, "Unknown directory part"); 757 fts_set(f, ff, FTS_SKIP); 758 break; 759 case 2: 760 /* 761 * Possibly our architecture. 762 * If we're descending, keep tabs on it. 763 */ 764 if (ff->fts_info != FTS_DP && dsec != NULL) 765 arch = ff->fts_name; 766 else 767 arch = NULL; 768 break; 769 default: 770 if (ff->fts_info == FTS_DP || use_all) 771 break; 772 if (warnings) 773 say(path, "Extraneous directory part"); 774 fts_set(f, ff, FTS_SKIP); 775 break; 776 } 777 } 778 779 fts_close(f); 780 return 1; 781 } 782 783 /* 784 * Add a file to the mlinks table. 785 * Do not verify that it's a "valid" looking manpage (we'll do that 786 * later). 787 * 788 * Try to infer the manual section, architecture, and page name from the 789 * path, assuming it looks like 790 * 791 * [./]man*[/<arch>]/<name>.<section> 792 * or 793 * [./]cat<section>[/<arch>]/<name>.0 794 * 795 * See treescan() for the fts(3) version of this. 796 */ 797 static void 798 filescan(const char *file) 799 { 800 char buf[PATH_MAX]; 801 struct stat st; 802 struct mlink *mlink; 803 char *p, *start; 804 805 assert(use_all); 806 807 if (0 == strncmp(file, "./", 2)) 808 file += 2; 809 810 /* 811 * We have to do lstat(2) before realpath(3) loses 812 * the information whether this is a symbolic link. 813 * We need to know that because for symbolic links, 814 * we want to use the orginal file name, while for 815 * regular files, we want to use the real path. 816 */ 817 if (-1 == lstat(file, &st)) { 818 exitcode = (int)MANDOCLEVEL_BADARG; 819 say(file, "&lstat"); 820 return; 821 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) { 822 exitcode = (int)MANDOCLEVEL_BADARG; 823 say(file, "Not a regular file"); 824 return; 825 } 826 827 /* 828 * We have to resolve the file name to the real path 829 * in any case for the base directory check. 830 */ 831 if (NULL == realpath(file, buf)) { 832 exitcode = (int)MANDOCLEVEL_BADARG; 833 say(file, "&realpath"); 834 return; 835 } 836 837 if (OP_TEST == op) 838 start = buf; 839 else if (strstr(buf, basedir) == buf) 840 start = buf + strlen(basedir); 841 #ifdef HOMEBREWDIR 842 else if (strstr(buf, HOMEBREWDIR) == buf) 843 start = buf; 844 #endif 845 else { 846 exitcode = (int)MANDOCLEVEL_BADARG; 847 say("", "%s: outside base directory", buf); 848 return; 849 } 850 851 /* 852 * Now we are sure the file is inside our tree. 853 * If it is a symbolic link, ignore the real path 854 * and use the original name. 855 * This implies passing stuff like "cat1/../man1/foo.1" 856 * on the command line won't work. So don't do that. 857 * Note the stat(2) can still fail if the link target 858 * doesn't exist. 859 */ 860 if (S_IFLNK & st.st_mode) { 861 if (-1 == stat(buf, &st)) { 862 exitcode = (int)MANDOCLEVEL_BADARG; 863 say(file, "&stat"); 864 return; 865 } 866 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) { 867 say(file, "Filename too long"); 868 return; 869 } 870 start = buf; 871 if (OP_TEST != op && strstr(buf, basedir) == buf) 872 start += strlen(basedir); 873 } 874 875 mlink = mandoc_calloc(1, sizeof(struct mlink)); 876 mlink->dform = FORM_NONE; 877 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >= 878 sizeof(mlink->file)) { 879 say(start, "Filename too long"); 880 free(mlink); 881 return; 882 } 883 884 /* 885 * First try to guess our directory structure. 886 * If we find a separator, try to look for man* or cat*. 887 * If we find one of these and what's underneath is a directory, 888 * assume it's an architecture. 889 */ 890 if (NULL != (p = strchr(start, '/'))) { 891 *p++ = '\0'; 892 if (0 == strncmp(start, "man", 3)) { 893 mlink->dform = FORM_SRC; 894 mlink->dsec = start + 3; 895 } else if (0 == strncmp(start, "cat", 3)) { 896 mlink->dform = FORM_CAT; 897 mlink->dsec = start + 3; 898 } 899 900 start = p; 901 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) { 902 *p++ = '\0'; 903 mlink->arch = start; 904 start = p; 905 } 906 } 907 908 /* 909 * Now check the file suffix. 910 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage. 911 */ 912 p = strrchr(start, '\0'); 913 while (p-- > start && '/' != *p && '.' != *p) 914 /* Loop. */ ; 915 916 if ('.' == *p) { 917 *p++ = '\0'; 918 mlink->fsec = p; 919 } 920 921 /* 922 * Now try to parse the name. 923 * Use the filename portion of the path. 924 */ 925 mlink->name = start; 926 if (NULL != (p = strrchr(start, '/'))) { 927 mlink->name = p + 1; 928 *p = '\0'; 929 } 930 mlink_add(mlink, &st); 931 } 932 933 static void 934 mlink_add(struct mlink *mlink, const struct stat *st) 935 { 936 struct inodev inodev; 937 struct mpage *mpage; 938 unsigned int slot; 939 940 assert(NULL != mlink->file); 941 942 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : ""); 943 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : ""); 944 mlink->name = mandoc_strdup(mlink->name ? mlink->name : ""); 945 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : ""); 946 947 if ('0' == *mlink->fsec) { 948 free(mlink->fsec); 949 mlink->fsec = mandoc_strdup(mlink->dsec); 950 mlink->fform = FORM_CAT; 951 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec) 952 mlink->fform = FORM_SRC; 953 else 954 mlink->fform = FORM_NONE; 955 956 slot = ohash_qlookup(&mlinks, mlink->file); 957 assert(NULL == ohash_find(&mlinks, slot)); 958 ohash_insert(&mlinks, slot, mlink); 959 960 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */ 961 inodev.st_ino = st->st_ino; 962 inodev.st_dev = st->st_dev; 963 slot = ohash_lookup_memory(&mpages, (char *)&inodev, 964 sizeof(struct inodev), inodev.st_ino); 965 mpage = ohash_find(&mpages, slot); 966 if (NULL == mpage) { 967 mpage = mandoc_calloc(1, sizeof(struct mpage)); 968 mpage->inodev.st_ino = inodev.st_ino; 969 mpage->inodev.st_dev = inodev.st_dev; 970 ohash_insert(&mpages, slot, mpage); 971 } else 972 mlink->next = mpage->mlinks; 973 mpage->mlinks = mlink; 974 mlink->mpage = mpage; 975 } 976 977 static void 978 mlink_free(struct mlink *mlink) 979 { 980 981 free(mlink->dsec); 982 free(mlink->arch); 983 free(mlink->name); 984 free(mlink->fsec); 985 free(mlink); 986 } 987 988 static void 989 mpages_free(void) 990 { 991 struct mpage *mpage; 992 struct mlink *mlink; 993 unsigned int slot; 994 995 mpage = ohash_first(&mpages, &slot); 996 while (NULL != mpage) { 997 while (NULL != (mlink = mpage->mlinks)) { 998 mpage->mlinks = mlink->next; 999 mlink_free(mlink); 1000 } 1001 free(mpage->sec); 1002 free(mpage->arch); 1003 free(mpage->title); 1004 free(mpage->desc); 1005 free(mpage); 1006 mpage = ohash_next(&mpages, &slot); 1007 } 1008 } 1009 1010 /* 1011 * For each mlink to the mpage, check whether the path looks like 1012 * it is formatted, and if it does, check whether a source manual 1013 * exists by the same name, ignoring the suffix. 1014 * If both conditions hold, drop the mlink. 1015 */ 1016 static void 1017 mlinks_undupe(struct mpage *mpage) 1018 { 1019 char buf[PATH_MAX]; 1020 struct mlink **prev; 1021 struct mlink *mlink; 1022 char *bufp; 1023 1024 mpage->form = FORM_CAT; 1025 prev = &mpage->mlinks; 1026 while (NULL != (mlink = *prev)) { 1027 if (FORM_CAT != mlink->dform) { 1028 mpage->form = FORM_NONE; 1029 goto nextlink; 1030 } 1031 (void)strlcpy(buf, mlink->file, sizeof(buf)); 1032 bufp = strstr(buf, "cat"); 1033 assert(NULL != bufp); 1034 memcpy(bufp, "man", 3); 1035 if (NULL != (bufp = strrchr(buf, '.'))) 1036 *++bufp = '\0'; 1037 (void)strlcat(buf, mlink->dsec, sizeof(buf)); 1038 if (NULL == ohash_find(&mlinks, 1039 ohash_qlookup(&mlinks, buf))) 1040 goto nextlink; 1041 if (warnings) 1042 say(mlink->file, "Man source exists: %s", buf); 1043 if (use_all) 1044 goto nextlink; 1045 *prev = mlink->next; 1046 mlink_free(mlink); 1047 continue; 1048 nextlink: 1049 prev = &(*prev)->next; 1050 } 1051 } 1052 1053 static void 1054 mlink_check(struct mpage *mpage, struct mlink *mlink) 1055 { 1056 struct str *str; 1057 unsigned int slot; 1058 1059 /* 1060 * Check whether the manual section given in a file 1061 * agrees with the directory where the file is located. 1062 * Some manuals have suffixes like (3p) on their 1063 * section number either inside the file or in the 1064 * directory name, some are linked into more than one 1065 * section, like encrypt(1) = makekey(8). 1066 */ 1067 1068 if (FORM_SRC == mpage->form && 1069 strcasecmp(mpage->sec, mlink->dsec)) 1070 say(mlink->file, "Section \"%s\" manual in %s directory", 1071 mpage->sec, mlink->dsec); 1072 1073 /* 1074 * Manual page directories exist for each kernel 1075 * architecture as returned by machine(1). 1076 * However, many manuals only depend on the 1077 * application architecture as returned by arch(1). 1078 * For example, some (2/ARM) manuals are shared 1079 * across the "armish" and "zaurus" kernel 1080 * architectures. 1081 * A few manuals are even shared across completely 1082 * different architectures, for example fdformat(1) 1083 * on amd64, i386, sparc, and sparc64. 1084 */ 1085 1086 if (strcasecmp(mpage->arch, mlink->arch)) 1087 say(mlink->file, "Architecture \"%s\" manual in " 1088 "\"%s\" directory", mpage->arch, mlink->arch); 1089 1090 /* 1091 * XXX 1092 * parse_cat() doesn't set NAME_TITLE yet. 1093 */ 1094 1095 if (FORM_CAT == mpage->form) 1096 return; 1097 1098 /* 1099 * Check whether this mlink 1100 * appears as a name in the NAME section. 1101 */ 1102 1103 slot = ohash_qlookup(&names, mlink->name); 1104 str = ohash_find(&names, slot); 1105 assert(NULL != str); 1106 if ( ! (NAME_TITLE & str->mask)) 1107 say(mlink->file, "Name missing in NAME section"); 1108 } 1109 1110 /* 1111 * Run through the files in the global vector "mpages" 1112 * and add them to the database specified in "basedir". 1113 * 1114 * This handles the parsing scheme itself, using the cues of directory 1115 * and filename to determine whether the file is parsable or not. 1116 */ 1117 static void 1118 mpages_merge(struct mparse *mp) 1119 { 1120 char any[] = "any"; 1121 struct mpage *mpage, *mpage_dest; 1122 struct mlink *mlink, *mlink_dest; 1123 struct roff_man *man; 1124 char *sodest; 1125 char *cp; 1126 int fd; 1127 unsigned int pslot; 1128 1129 if ( ! nodb) 1130 SQL_EXEC("BEGIN TRANSACTION"); 1131 1132 mpage = ohash_first(&mpages, &pslot); 1133 while (mpage != NULL) { 1134 mlinks_undupe(mpage); 1135 if ((mlink = mpage->mlinks) == NULL) { 1136 mpage = ohash_next(&mpages, &pslot); 1137 continue; 1138 } 1139 1140 name_mask = NAME_MASK; 1141 mandoc_ohash_init(&names, 4, offsetof(struct str, key)); 1142 mandoc_ohash_init(&strings, 6, offsetof(struct str, key)); 1143 mparse_reset(mp); 1144 man = NULL; 1145 sodest = NULL; 1146 1147 if ((fd = mparse_open(mp, mlink->file)) == -1) { 1148 say(mlink->file, "&open"); 1149 goto nextpage; 1150 } 1151 1152 /* 1153 * Interpret the file as mdoc(7) or man(7) source 1154 * code, unless it is known to be formatted. 1155 */ 1156 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) { 1157 mparse_readfd(mp, fd, mlink->file); 1158 close(fd); 1159 mparse_result(mp, &man, &sodest); 1160 } 1161 1162 if (sodest != NULL) { 1163 mlink_dest = ohash_find(&mlinks, 1164 ohash_qlookup(&mlinks, sodest)); 1165 if (mlink_dest == NULL) { 1166 mandoc_asprintf(&cp, "%s.gz", sodest); 1167 mlink_dest = ohash_find(&mlinks, 1168 ohash_qlookup(&mlinks, cp)); 1169 free(cp); 1170 } 1171 if (mlink_dest != NULL) { 1172 1173 /* The .so target exists. */ 1174 1175 mpage_dest = mlink_dest->mpage; 1176 while (1) { 1177 mlink->mpage = mpage_dest; 1178 1179 /* 1180 * If the target was already 1181 * processed, add the links 1182 * to the database now. 1183 * Otherwise, this will 1184 * happen when we come 1185 * to the target. 1186 */ 1187 1188 if (mpage_dest->pageid) 1189 dbadd_mlink_name(mlink); 1190 1191 if (mlink->next == NULL) 1192 break; 1193 mlink = mlink->next; 1194 } 1195 1196 /* Move all links to the target. */ 1197 1198 mlink->next = mlink_dest->next; 1199 mlink_dest->next = mpage->mlinks; 1200 mpage->mlinks = NULL; 1201 } 1202 goto nextpage; 1203 } else if (man != NULL && man->macroset == MACROSET_MDOC) { 1204 mdoc_validate(man); 1205 mpage->form = FORM_SRC; 1206 mpage->sec = man->meta.msec; 1207 mpage->sec = mandoc_strdup( 1208 mpage->sec == NULL ? "" : mpage->sec); 1209 mpage->arch = man->meta.arch; 1210 mpage->arch = mandoc_strdup( 1211 mpage->arch == NULL ? "" : mpage->arch); 1212 mpage->title = mandoc_strdup(man->meta.title); 1213 } else if (man != NULL && man->macroset == MACROSET_MAN) { 1214 man_validate(man); 1215 mpage->form = FORM_SRC; 1216 mpage->sec = mandoc_strdup(man->meta.msec); 1217 mpage->arch = mandoc_strdup(mlink->arch); 1218 mpage->title = mandoc_strdup(man->meta.title); 1219 } else { 1220 mpage->form = FORM_CAT; 1221 mpage->sec = mandoc_strdup(mlink->dsec); 1222 mpage->arch = mandoc_strdup(mlink->arch); 1223 mpage->title = mandoc_strdup(mlink->name); 1224 } 1225 putkey(mpage, mpage->sec, TYPE_sec); 1226 if (*mpage->arch != '\0') 1227 putkey(mpage, mpage->arch, TYPE_arch); 1228 1229 for ( ; mlink != NULL; mlink = mlink->next) { 1230 if ('\0' != *mlink->dsec) 1231 putkey(mpage, mlink->dsec, TYPE_sec); 1232 if ('\0' != *mlink->fsec) 1233 putkey(mpage, mlink->fsec, TYPE_sec); 1234 putkey(mpage, '\0' == *mlink->arch ? 1235 any : mlink->arch, TYPE_arch); 1236 putkey(mpage, mlink->name, NAME_FILE); 1237 } 1238 1239 assert(mpage->desc == NULL); 1240 if (man != NULL && man->macroset == MACROSET_MDOC) 1241 parse_mdoc(mpage, &man->meta, man->first); 1242 else if (man != NULL) 1243 parse_man(mpage, &man->meta, man->first); 1244 else 1245 parse_cat(mpage, fd); 1246 if (mpage->desc == NULL) 1247 mpage->desc = mandoc_strdup(mpage->mlinks->name); 1248 1249 if (warnings && !use_all) 1250 for (mlink = mpage->mlinks; mlink; 1251 mlink = mlink->next) 1252 mlink_check(mpage, mlink); 1253 1254 dbadd(mpage); 1255 mlink = mpage->mlinks; 1256 1257 nextpage: 1258 ohash_delete(&strings); 1259 ohash_delete(&names); 1260 mpage = ohash_next(&mpages, &pslot); 1261 } 1262 1263 if (0 == nodb) 1264 SQL_EXEC("END TRANSACTION"); 1265 } 1266 1267 static void 1268 names_check(void) 1269 { 1270 sqlite3_stmt *stmt; 1271 const char *name, *sec, *arch, *key; 1272 1273 sqlite3_prepare_v2(db, 1274 "SELECT name, sec, arch, key FROM (" 1275 "SELECT name AS key, pageid FROM names " 1276 "WHERE bits & ? AND NOT EXISTS (" 1277 "SELECT pageid FROM mlinks " 1278 "WHERE mlinks.pageid == names.pageid " 1279 "AND mlinks.name == names.name" 1280 ")" 1281 ") JOIN (" 1282 "SELECT sec, arch, name, pageid FROM mlinks " 1283 "GROUP BY pageid" 1284 ") USING (pageid);", 1285 -1, &stmt, NULL); 1286 1287 if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK) 1288 say("", "%s", sqlite3_errmsg(db)); 1289 1290 while (sqlite3_step(stmt) == SQLITE_ROW) { 1291 name = (const char *)sqlite3_column_text(stmt, 0); 1292 sec = (const char *)sqlite3_column_text(stmt, 1); 1293 arch = (const char *)sqlite3_column_text(stmt, 2); 1294 key = (const char *)sqlite3_column_text(stmt, 3); 1295 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec, 1296 '\0' == *arch ? "" : "/", 1297 '\0' == *arch ? "" : arch, key); 1298 } 1299 sqlite3_finalize(stmt); 1300 } 1301 1302 static void 1303 parse_cat(struct mpage *mpage, int fd) 1304 { 1305 FILE *stream; 1306 char *line, *p, *title; 1307 size_t linesz, plen, titlesz; 1308 ssize_t len; 1309 int offs; 1310 1311 stream = (-1 == fd) ? 1312 fopen(mpage->mlinks->file, "r") : 1313 fdopen(fd, "r"); 1314 if (NULL == stream) { 1315 if (-1 != fd) 1316 close(fd); 1317 if (warnings) 1318 say(mpage->mlinks->file, "&fopen"); 1319 return; 1320 } 1321 1322 line = NULL; 1323 linesz = 0; 1324 1325 /* Skip to first blank line. */ 1326 1327 while (getline(&line, &linesz, stream) != -1) 1328 if (*line == '\n') 1329 break; 1330 1331 /* 1332 * Assume the first line that is not indented 1333 * is the first section header. Skip to it. 1334 */ 1335 1336 while (getline(&line, &linesz, stream) != -1) 1337 if (*line != '\n' && *line != ' ') 1338 break; 1339 1340 /* 1341 * Read up until the next section into a buffer. 1342 * Strip the leading and trailing newline from each read line, 1343 * appending a trailing space. 1344 * Ignore empty (whitespace-only) lines. 1345 */ 1346 1347 titlesz = 0; 1348 title = NULL; 1349 1350 while ((len = getline(&line, &linesz, stream)) != -1) { 1351 if (*line != ' ') 1352 break; 1353 offs = 0; 1354 while (isspace((unsigned char)line[offs])) 1355 offs++; 1356 if (line[offs] == '\0') 1357 continue; 1358 title = mandoc_realloc(title, titlesz + len - offs); 1359 memcpy(title + titlesz, line + offs, len - offs); 1360 titlesz += len - offs; 1361 title[titlesz - 1] = ' '; 1362 } 1363 free(line); 1364 1365 /* 1366 * If no page content can be found, or the input line 1367 * is already the next section header, or there is no 1368 * trailing newline, reuse the page title as the page 1369 * description. 1370 */ 1371 1372 if (NULL == title || '\0' == *title) { 1373 if (warnings) 1374 say(mpage->mlinks->file, 1375 "Cannot find NAME section"); 1376 fclose(stream); 1377 free(title); 1378 return; 1379 } 1380 1381 title[titlesz - 1] = '\0'; 1382 1383 /* 1384 * Skip to the first dash. 1385 * Use the remaining line as the description (no more than 70 1386 * bytes). 1387 */ 1388 1389 if (NULL != (p = strstr(title, "- "))) { 1390 for (p += 2; ' ' == *p || '\b' == *p; p++) 1391 /* Skip to next word. */ ; 1392 } else { 1393 if (warnings) 1394 say(mpage->mlinks->file, 1395 "No dash in title line"); 1396 p = title; 1397 } 1398 1399 plen = strlen(p); 1400 1401 /* Strip backspace-encoding from line. */ 1402 1403 while (NULL != (line = memchr(p, '\b', plen))) { 1404 len = line - p; 1405 if (0 == len) { 1406 memmove(line, line + 1, plen--); 1407 continue; 1408 } 1409 memmove(line - 1, line + 1, plen - len); 1410 plen -= 2; 1411 } 1412 1413 mpage->desc = mandoc_strdup(p); 1414 fclose(stream); 1415 free(title); 1416 } 1417 1418 /* 1419 * Put a type/word pair into the word database for this particular file. 1420 */ 1421 static void 1422 putkey(const struct mpage *mpage, char *value, uint64_t type) 1423 { 1424 char *cp; 1425 1426 assert(NULL != value); 1427 if (TYPE_arch == type) 1428 for (cp = value; *cp; cp++) 1429 if (isupper((unsigned char)*cp)) 1430 *cp = _tolower((unsigned char)*cp); 1431 putkeys(mpage, value, strlen(value), type); 1432 } 1433 1434 /* 1435 * Grok all nodes at or below a certain mdoc node into putkey(). 1436 */ 1437 static void 1438 putmdockey(const struct mpage *mpage, 1439 const struct roff_node *n, uint64_t m) 1440 { 1441 1442 for ( ; NULL != n; n = n->next) { 1443 if (NULL != n->child) 1444 putmdockey(mpage, n->child, m); 1445 if (n->type == ROFFT_TEXT) 1446 putkey(mpage, n->string, m); 1447 } 1448 } 1449 1450 static void 1451 parse_man(struct mpage *mpage, const struct roff_meta *meta, 1452 const struct roff_node *n) 1453 { 1454 const struct roff_node *head, *body; 1455 char *start, *title; 1456 char byte; 1457 size_t sz; 1458 1459 if (n == NULL) 1460 return; 1461 1462 /* 1463 * We're only searching for one thing: the first text child in 1464 * the BODY of a NAME section. Since we don't keep track of 1465 * sections in -man, run some hoops to find out whether we're in 1466 * the correct section or not. 1467 */ 1468 1469 if (n->type == ROFFT_BODY && n->tok == MAN_SH) { 1470 body = n; 1471 if ((head = body->parent->head) != NULL && 1472 (head = head->child) != NULL && 1473 head->next == NULL && 1474 head->type == ROFFT_TEXT && 1475 strcmp(head->string, "NAME") == 0 && 1476 body->child != NULL) { 1477 1478 /* 1479 * Suck the entire NAME section into memory. 1480 * Yes, we might run away. 1481 * But too many manuals have big, spread-out 1482 * NAME sections over many lines. 1483 */ 1484 1485 title = NULL; 1486 deroff(&title, body); 1487 if (NULL == title) 1488 return; 1489 1490 /* 1491 * Go through a special heuristic dance here. 1492 * Conventionally, one or more manual names are 1493 * comma-specified prior to a whitespace, then a 1494 * dash, then a description. Try to puzzle out 1495 * the name parts here. 1496 */ 1497 1498 start = title; 1499 for ( ;; ) { 1500 sz = strcspn(start, " ,"); 1501 if ('\0' == start[sz]) 1502 break; 1503 1504 byte = start[sz]; 1505 start[sz] = '\0'; 1506 1507 /* 1508 * Assume a stray trailing comma in the 1509 * name list if a name begins with a dash. 1510 */ 1511 1512 if ('-' == start[0] || 1513 ('\\' == start[0] && '-' == start[1])) 1514 break; 1515 1516 putkey(mpage, start, NAME_TITLE); 1517 if ( ! (mpage->name_head_done || 1518 strcasecmp(start, meta->title))) { 1519 putkey(mpage, start, NAME_HEAD); 1520 mpage->name_head_done = 1; 1521 } 1522 1523 if (' ' == byte) { 1524 start += sz + 1; 1525 break; 1526 } 1527 1528 assert(',' == byte); 1529 start += sz + 1; 1530 while (' ' == *start) 1531 start++; 1532 } 1533 1534 if (start == title) { 1535 putkey(mpage, start, NAME_TITLE); 1536 if ( ! (mpage->name_head_done || 1537 strcasecmp(start, meta->title))) { 1538 putkey(mpage, start, NAME_HEAD); 1539 mpage->name_head_done = 1; 1540 } 1541 free(title); 1542 return; 1543 } 1544 1545 while (isspace((unsigned char)*start)) 1546 start++; 1547 1548 if (0 == strncmp(start, "-", 1)) 1549 start += 1; 1550 else if (0 == strncmp(start, "\\-\\-", 4)) 1551 start += 4; 1552 else if (0 == strncmp(start, "\\-", 2)) 1553 start += 2; 1554 else if (0 == strncmp(start, "\\(en", 4)) 1555 start += 4; 1556 else if (0 == strncmp(start, "\\(em", 4)) 1557 start += 4; 1558 1559 while (' ' == *start) 1560 start++; 1561 1562 mpage->desc = mandoc_strdup(start); 1563 free(title); 1564 return; 1565 } 1566 } 1567 1568 for (n = n->child; n; n = n->next) { 1569 if (NULL != mpage->desc) 1570 break; 1571 parse_man(mpage, meta, n); 1572 } 1573 } 1574 1575 static void 1576 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta, 1577 const struct roff_node *n) 1578 { 1579 1580 assert(NULL != n); 1581 for (n = n->child; NULL != n; n = n->next) { 1582 switch (n->type) { 1583 case ROFFT_ELEM: 1584 case ROFFT_BLOCK: 1585 case ROFFT_HEAD: 1586 case ROFFT_BODY: 1587 case ROFFT_TAIL: 1588 if (NULL != mdocs[n->tok].fp) 1589 if (0 == (*mdocs[n->tok].fp)(mpage, meta, n)) 1590 break; 1591 if (mdocs[n->tok].mask) 1592 putmdockey(mpage, n->child, 1593 mdocs[n->tok].mask); 1594 break; 1595 default: 1596 assert(n->type != ROFFT_ROOT); 1597 continue; 1598 } 1599 if (NULL != n->child) 1600 parse_mdoc(mpage, meta, n); 1601 } 1602 } 1603 1604 static int 1605 parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta, 1606 const struct roff_node *n) 1607 { 1608 char *start, *end; 1609 size_t sz; 1610 1611 if (SEC_SYNOPSIS != n->sec || 1612 NULL == (n = n->child) || 1613 n->type != ROFFT_TEXT) 1614 return 0; 1615 1616 /* 1617 * Only consider those `Fd' macro fields that begin with an 1618 * "inclusion" token (versus, e.g., #define). 1619 */ 1620 1621 if (strcmp("#include", n->string)) 1622 return 0; 1623 1624 if ((n = n->next) == NULL || n->type != ROFFT_TEXT) 1625 return 0; 1626 1627 /* 1628 * Strip away the enclosing angle brackets and make sure we're 1629 * not zero-length. 1630 */ 1631 1632 start = n->string; 1633 if ('<' == *start || '"' == *start) 1634 start++; 1635 1636 if (0 == (sz = strlen(start))) 1637 return 0; 1638 1639 end = &start[(int)sz - 1]; 1640 if ('>' == *end || '"' == *end) 1641 end--; 1642 1643 if (end > start) 1644 putkeys(mpage, start, end - start + 1, TYPE_In); 1645 return 0; 1646 } 1647 1648 static void 1649 parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n) 1650 { 1651 char *cp; 1652 size_t sz; 1653 1654 if (n->type != ROFFT_TEXT) 1655 return; 1656 1657 /* Skip function pointer punctuation. */ 1658 1659 cp = n->string; 1660 while (*cp == '(' || *cp == '*') 1661 cp++; 1662 sz = strcspn(cp, "()"); 1663 1664 putkeys(mpage, cp, sz, TYPE_Fn); 1665 if (n->sec == SEC_SYNOPSIS) 1666 putkeys(mpage, cp, sz, NAME_SYN); 1667 } 1668 1669 static int 1670 parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta, 1671 const struct roff_node *n) 1672 { 1673 1674 if (n->child == NULL) 1675 return 0; 1676 1677 parse_mdoc_fname(mpage, n->child); 1678 1679 for (n = n->child->next; n != NULL; n = n->next) 1680 if (n->type == ROFFT_TEXT) 1681 putkey(mpage, n->string, TYPE_Fa); 1682 1683 return 0; 1684 } 1685 1686 static int 1687 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta, 1688 const struct roff_node *n) 1689 { 1690 1691 if (n->type != ROFFT_HEAD) 1692 return 1; 1693 1694 if (n->child != NULL) 1695 parse_mdoc_fname(mpage, n->child); 1696 1697 return 0; 1698 } 1699 1700 static int 1701 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta, 1702 const struct roff_node *n) 1703 { 1704 char *cp; 1705 1706 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY) 1707 return 0; 1708 1709 if (n->child != NULL && 1710 n->child->next == NULL && 1711 n->child->type == ROFFT_TEXT) 1712 return 1; 1713 1714 cp = NULL; 1715 deroff(&cp, n); 1716 if (cp != NULL) { 1717 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va || 1718 n->type == ROFFT_BODY ? TYPE_Va : 0)); 1719 free(cp); 1720 } 1721 1722 return 0; 1723 } 1724 1725 static int 1726 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta, 1727 const struct roff_node *n) 1728 { 1729 char *cp; 1730 1731 if (NULL == (n = n->child)) 1732 return 0; 1733 1734 if (NULL == n->next) { 1735 putkey(mpage, n->string, TYPE_Xr); 1736 return 0; 1737 } 1738 1739 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); 1740 putkey(mpage, cp, TYPE_Xr); 1741 free(cp); 1742 return 0; 1743 } 1744 1745 static int 1746 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta, 1747 const struct roff_node *n) 1748 { 1749 1750 if (n->type == ROFFT_BODY) 1751 deroff(&mpage->desc, n); 1752 return 0; 1753 } 1754 1755 static int 1756 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta, 1757 const struct roff_node *n) 1758 { 1759 1760 if (SEC_NAME == n->sec) 1761 putmdockey(mpage, n->child, NAME_TITLE); 1762 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) { 1763 if (n->child == NULL) 1764 putkey(mpage, meta->name, NAME_SYN); 1765 else 1766 putmdockey(mpage, n->child, NAME_SYN); 1767 } 1768 if ( ! (mpage->name_head_done || 1769 n->child == NULL || n->child->string == NULL || 1770 strcasecmp(n->child->string, meta->title))) { 1771 putkey(mpage, n->child->string, ROFFT_HEAD); 1772 mpage->name_head_done = 1; 1773 } 1774 return 0; 1775 } 1776 1777 static int 1778 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta, 1779 const struct roff_node *n) 1780 { 1781 1782 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD; 1783 } 1784 1785 static int 1786 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta, 1787 const struct roff_node *n) 1788 { 1789 1790 return n->type == ROFFT_HEAD; 1791 } 1792 1793 /* 1794 * Add a string to the hash table for the current manual. 1795 * Each string has a bitmask telling which macros it belongs to. 1796 * When we finish the manual, we'll dump the table. 1797 */ 1798 static void 1799 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v) 1800 { 1801 struct ohash *htab; 1802 struct str *s; 1803 const char *end; 1804 unsigned int slot; 1805 int i, mustfree; 1806 1807 if (0 == sz) 1808 return; 1809 1810 mustfree = render_string(&cp, &sz); 1811 1812 if (TYPE_Nm & v) { 1813 htab = &names; 1814 v &= name_mask; 1815 if (v & NAME_FIRST) 1816 name_mask &= ~NAME_FIRST; 1817 if (debug > 1) 1818 say(mpage->mlinks->file, 1819 "Adding name %*s, bits=%d", sz, cp, v); 1820 } else { 1821 htab = &strings; 1822 if (debug > 1) 1823 for (i = 0; i < mansearch_keymax; i++) 1824 if ((uint64_t)1 << i & v) 1825 say(mpage->mlinks->file, 1826 "Adding key %s=%*s", 1827 mansearch_keynames[i], sz, cp); 1828 } 1829 1830 end = cp + sz; 1831 slot = ohash_qlookupi(htab, cp, &end); 1832 s = ohash_find(htab, slot); 1833 1834 if (NULL != s && mpage == s->mpage) { 1835 s->mask |= v; 1836 return; 1837 } else if (NULL == s) { 1838 s = mandoc_calloc(1, sizeof(struct str) + sz + 1); 1839 memcpy(s->key, cp, sz); 1840 ohash_insert(htab, slot, s); 1841 } 1842 s->mpage = mpage; 1843 s->mask = v; 1844 1845 if (mustfree) 1846 free(cp); 1847 } 1848 1849 /* 1850 * Take a Unicode codepoint and produce its UTF-8 encoding. 1851 * This isn't the best way to do this, but it works. 1852 * The magic numbers are from the UTF-8 packaging. 1853 * They're not as scary as they seem: read the UTF-8 spec for details. 1854 */ 1855 static size_t 1856 utf8(unsigned int cp, char out[7]) 1857 { 1858 size_t rc; 1859 1860 rc = 0; 1861 if (cp <= 0x0000007F) { 1862 rc = 1; 1863 out[0] = (char)cp; 1864 } else if (cp <= 0x000007FF) { 1865 rc = 2; 1866 out[0] = (cp >> 6 & 31) | 192; 1867 out[1] = (cp & 63) | 128; 1868 } else if (cp <= 0x0000FFFF) { 1869 rc = 3; 1870 out[0] = (cp >> 12 & 15) | 224; 1871 out[1] = (cp >> 6 & 63) | 128; 1872 out[2] = (cp & 63) | 128; 1873 } else if (cp <= 0x001FFFFF) { 1874 rc = 4; 1875 out[0] = (cp >> 18 & 7) | 240; 1876 out[1] = (cp >> 12 & 63) | 128; 1877 out[2] = (cp >> 6 & 63) | 128; 1878 out[3] = (cp & 63) | 128; 1879 } else if (cp <= 0x03FFFFFF) { 1880 rc = 5; 1881 out[0] = (cp >> 24 & 3) | 248; 1882 out[1] = (cp >> 18 & 63) | 128; 1883 out[2] = (cp >> 12 & 63) | 128; 1884 out[3] = (cp >> 6 & 63) | 128; 1885 out[4] = (cp & 63) | 128; 1886 } else if (cp <= 0x7FFFFFFF) { 1887 rc = 6; 1888 out[0] = (cp >> 30 & 1) | 252; 1889 out[1] = (cp >> 24 & 63) | 128; 1890 out[2] = (cp >> 18 & 63) | 128; 1891 out[3] = (cp >> 12 & 63) | 128; 1892 out[4] = (cp >> 6 & 63) | 128; 1893 out[5] = (cp & 63) | 128; 1894 } else 1895 return 0; 1896 1897 out[rc] = '\0'; 1898 return rc; 1899 } 1900 1901 /* 1902 * If the string contains escape sequences, 1903 * replace it with an allocated rendering and return 1, 1904 * such that the caller can free it after use. 1905 * Otherwise, do nothing and return 0. 1906 */ 1907 static int 1908 render_string(char **public, size_t *psz) 1909 { 1910 const char *src, *scp, *addcp, *seq; 1911 char *dst; 1912 size_t ssz, dsz, addsz; 1913 char utfbuf[7], res[6]; 1914 int seqlen, unicode; 1915 1916 res[0] = '\\'; 1917 res[1] = '\t'; 1918 res[2] = ASCII_NBRSP; 1919 res[3] = ASCII_HYPH; 1920 res[4] = ASCII_BREAK; 1921 res[5] = '\0'; 1922 1923 src = scp = *public; 1924 ssz = *psz; 1925 dst = NULL; 1926 dsz = 0; 1927 1928 while (scp < src + *psz) { 1929 1930 /* Leave normal characters unchanged. */ 1931 1932 if (strchr(res, *scp) == NULL) { 1933 if (dst != NULL) 1934 dst[dsz++] = *scp; 1935 scp++; 1936 continue; 1937 } 1938 1939 /* 1940 * Found something that requires replacing, 1941 * make sure we have a destination buffer. 1942 */ 1943 1944 if (dst == NULL) { 1945 dst = mandoc_malloc(ssz + 1); 1946 dsz = scp - src; 1947 memcpy(dst, src, dsz); 1948 } 1949 1950 /* Handle single-char special characters. */ 1951 1952 switch (*scp) { 1953 case '\\': 1954 break; 1955 case '\t': 1956 case ASCII_NBRSP: 1957 dst[dsz++] = ' '; 1958 scp++; 1959 continue; 1960 case ASCII_HYPH: 1961 dst[dsz++] = '-'; 1962 /* FALLTHROUGH */ 1963 case ASCII_BREAK: 1964 scp++; 1965 continue; 1966 default: 1967 abort(); 1968 } 1969 1970 /* 1971 * Found an escape sequence. 1972 * Read past the slash, then parse it. 1973 * Ignore everything except characters. 1974 */ 1975 1976 scp++; 1977 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL) 1978 continue; 1979 1980 /* 1981 * Render the special character 1982 * as either UTF-8 or ASCII. 1983 */ 1984 1985 if (write_utf8) { 1986 unicode = mchars_spec2cp(seq, seqlen); 1987 if (unicode <= 0) 1988 continue; 1989 addsz = utf8(unicode, utfbuf); 1990 if (addsz == 0) 1991 continue; 1992 addcp = utfbuf; 1993 } else { 1994 addcp = mchars_spec2str(seq, seqlen, &addsz); 1995 if (addcp == NULL) 1996 continue; 1997 if (*addcp == ASCII_NBRSP) { 1998 addcp = " "; 1999 addsz = 1; 2000 } 2001 } 2002 2003 /* Copy the rendered glyph into the stream. */ 2004 2005 ssz += addsz; 2006 dst = mandoc_realloc(dst, ssz + 1); 2007 memcpy(dst + dsz, addcp, addsz); 2008 dsz += addsz; 2009 } 2010 if (dst != NULL) { 2011 *public = dst; 2012 *psz = dsz; 2013 } 2014 2015 /* Trim trailing whitespace and NUL-terminate. */ 2016 2017 while (*psz > 0 && (*public)[*psz - 1] == ' ') 2018 --*psz; 2019 if (dst != NULL) { 2020 (*public)[*psz] = '\0'; 2021 return 1; 2022 } else 2023 return 0; 2024 } 2025 2026 static void 2027 dbadd_mlink(const struct mlink *mlink) 2028 { 2029 size_t i; 2030 2031 i = 1; 2032 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec); 2033 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch); 2034 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name); 2035 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid); 2036 SQL_STEP(stmts[STMT_INSERT_LINK]); 2037 sqlite3_reset(stmts[STMT_INSERT_LINK]); 2038 } 2039 2040 static void 2041 dbadd_mlink_name(const struct mlink *mlink) 2042 { 2043 uint64_t bits; 2044 size_t i; 2045 2046 dbadd_mlink(mlink); 2047 2048 i = 1; 2049 SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid); 2050 bits = NAME_FILE & NAME_MASK; 2051 if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) { 2052 bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0); 2053 sqlite3_reset(stmts[STMT_SELECT_NAME]); 2054 } 2055 2056 i = 1; 2057 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits); 2058 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name); 2059 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid); 2060 SQL_STEP(stmts[STMT_INSERT_NAME]); 2061 sqlite3_reset(stmts[STMT_INSERT_NAME]); 2062 } 2063 2064 /* 2065 * Flush the current page's terms (and their bits) into the database. 2066 * Wrap the entire set of additions in a transaction to make sqlite be a 2067 * little faster. 2068 * Also, handle escape sequences at the last possible moment. 2069 */ 2070 static void 2071 dbadd(struct mpage *mpage) 2072 { 2073 struct mlink *mlink; 2074 struct str *key; 2075 char *cp; 2076 size_t i; 2077 unsigned int slot; 2078 int mustfree; 2079 2080 mlink = mpage->mlinks; 2081 2082 if (nodb) { 2083 for (key = ohash_first(&names, &slot); NULL != key; 2084 key = ohash_next(&names, &slot)) 2085 free(key); 2086 for (key = ohash_first(&strings, &slot); NULL != key; 2087 key = ohash_next(&strings, &slot)) 2088 free(key); 2089 if (0 == debug) 2090 return; 2091 while (NULL != mlink) { 2092 fputs(mlink->name, stdout); 2093 if (NULL == mlink->next || 2094 strcmp(mlink->dsec, mlink->next->dsec) || 2095 strcmp(mlink->fsec, mlink->next->fsec) || 2096 strcmp(mlink->arch, mlink->next->arch)) { 2097 putchar('('); 2098 if ('\0' == *mlink->dsec) 2099 fputs(mlink->fsec, stdout); 2100 else 2101 fputs(mlink->dsec, stdout); 2102 if ('\0' != *mlink->arch) 2103 printf("/%s", mlink->arch); 2104 putchar(')'); 2105 } 2106 mlink = mlink->next; 2107 if (NULL != mlink) 2108 fputs(", ", stdout); 2109 } 2110 printf(" - %s\n", mpage->desc); 2111 return; 2112 } 2113 2114 if (debug) 2115 say(mlink->file, "Adding to database"); 2116 2117 cp = mpage->desc; 2118 i = strlen(cp); 2119 mustfree = render_string(&cp, &i); 2120 i = 1; 2121 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp); 2122 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form); 2123 SQL_STEP(stmts[STMT_INSERT_PAGE]); 2124 mpage->pageid = sqlite3_last_insert_rowid(db); 2125 sqlite3_reset(stmts[STMT_INSERT_PAGE]); 2126 if (mustfree) 2127 free(cp); 2128 2129 while (NULL != mlink) { 2130 dbadd_mlink(mlink); 2131 mlink = mlink->next; 2132 } 2133 mlink = mpage->mlinks; 2134 2135 for (key = ohash_first(&names, &slot); NULL != key; 2136 key = ohash_next(&names, &slot)) { 2137 assert(key->mpage == mpage); 2138 i = 1; 2139 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask); 2140 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key); 2141 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid); 2142 SQL_STEP(stmts[STMT_INSERT_NAME]); 2143 sqlite3_reset(stmts[STMT_INSERT_NAME]); 2144 free(key); 2145 } 2146 for (key = ohash_first(&strings, &slot); NULL != key; 2147 key = ohash_next(&strings, &slot)) { 2148 assert(key->mpage == mpage); 2149 i = 1; 2150 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask); 2151 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key); 2152 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid); 2153 SQL_STEP(stmts[STMT_INSERT_KEY]); 2154 sqlite3_reset(stmts[STMT_INSERT_KEY]); 2155 free(key); 2156 } 2157 } 2158 2159 static void 2160 dbprune(void) 2161 { 2162 struct mpage *mpage; 2163 struct mlink *mlink; 2164 size_t i; 2165 unsigned int slot; 2166 2167 if (0 == nodb) 2168 SQL_EXEC("BEGIN TRANSACTION"); 2169 2170 for (mpage = ohash_first(&mpages, &slot); NULL != mpage; 2171 mpage = ohash_next(&mpages, &slot)) { 2172 mlink = mpage->mlinks; 2173 if (debug) 2174 say(mlink->file, "Deleting from database"); 2175 if (nodb) 2176 continue; 2177 for ( ; NULL != mlink; mlink = mlink->next) { 2178 i = 1; 2179 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], 2180 i, mlink->dsec); 2181 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], 2182 i, mlink->arch); 2183 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE], 2184 i, mlink->name); 2185 SQL_STEP(stmts[STMT_DELETE_PAGE]); 2186 sqlite3_reset(stmts[STMT_DELETE_PAGE]); 2187 } 2188 } 2189 2190 if (0 == nodb) 2191 SQL_EXEC("END TRANSACTION"); 2192 } 2193 2194 /* 2195 * Close an existing database and its prepared statements. 2196 * If "real" is not set, rename the temporary file into the real one. 2197 */ 2198 static void 2199 dbclose(int real) 2200 { 2201 size_t i; 2202 int status; 2203 pid_t child; 2204 2205 if (nodb) 2206 return; 2207 2208 for (i = 0; i < STMT__MAX; i++) { 2209 sqlite3_finalize(stmts[i]); 2210 stmts[i] = NULL; 2211 } 2212 2213 sqlite3_close(db); 2214 db = NULL; 2215 2216 if (real) 2217 return; 2218 2219 if ('\0' == *tempfilename) { 2220 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) { 2221 exitcode = (int)MANDOCLEVEL_SYSERR; 2222 say(MANDOC_DB, "&rename"); 2223 } 2224 return; 2225 } 2226 2227 switch (child = fork()) { 2228 case -1: 2229 exitcode = (int)MANDOCLEVEL_SYSERR; 2230 say("", "&fork cmp"); 2231 return; 2232 case 0: 2233 execlp("cmp", "cmp", "-s", 2234 tempfilename, MANDOC_DB, (char *)NULL); 2235 say("", "&exec cmp"); 2236 exit(0); 2237 default: 2238 break; 2239 } 2240 if (-1 == waitpid(child, &status, 0)) { 2241 exitcode = (int)MANDOCLEVEL_SYSERR; 2242 say("", "&wait cmp"); 2243 } else if (WIFSIGNALED(status)) { 2244 exitcode = (int)MANDOCLEVEL_SYSERR; 2245 say("", "cmp died from signal %d", WTERMSIG(status)); 2246 } else if (WEXITSTATUS(status)) { 2247 exitcode = (int)MANDOCLEVEL_SYSERR; 2248 say(MANDOC_DB, 2249 "Data changed, but cannot replace database"); 2250 } 2251 2252 *strrchr(tempfilename, '/') = '\0'; 2253 switch (child = fork()) { 2254 case -1: 2255 exitcode = (int)MANDOCLEVEL_SYSERR; 2256 say("", "&fork rm"); 2257 return; 2258 case 0: 2259 execlp("rm", "rm", "-rf", tempfilename, (char *)NULL); 2260 say("", "&exec rm"); 2261 exit((int)MANDOCLEVEL_SYSERR); 2262 default: 2263 break; 2264 } 2265 if (-1 == waitpid(child, &status, 0)) { 2266 exitcode = (int)MANDOCLEVEL_SYSERR; 2267 say("", "&wait rm"); 2268 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) { 2269 exitcode = (int)MANDOCLEVEL_SYSERR; 2270 say("", "%s: Cannot remove temporary directory", 2271 tempfilename); 2272 } 2273 } 2274 2275 /* 2276 * This is straightforward stuff. 2277 * Open a database connection to a "temporary" database, then open a set 2278 * of prepared statements we'll use over and over again. 2279 * If "real" is set, we use the existing database; if not, we truncate a 2280 * temporary one. 2281 * Must be matched by dbclose(). 2282 */ 2283 static int 2284 dbopen(int real) 2285 { 2286 const char *sql; 2287 int rc, ofl; 2288 2289 if (nodb) 2290 return 1; 2291 2292 *tempfilename = '\0'; 2293 ofl = SQLITE_OPEN_READWRITE; 2294 2295 if (real) { 2296 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL); 2297 if (SQLITE_OK != rc) { 2298 exitcode = (int)MANDOCLEVEL_SYSERR; 2299 if (SQLITE_CANTOPEN != rc) 2300 say(MANDOC_DB, "%s", sqlite3_errstr(rc)); 2301 return 0; 2302 } 2303 goto prepare_statements; 2304 } 2305 2306 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE; 2307 2308 remove(MANDOC_DB "~"); 2309 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL); 2310 if (SQLITE_OK == rc) 2311 goto create_tables; 2312 if (MPARSE_QUICK & mparse_options) { 2313 exitcode = (int)MANDOCLEVEL_SYSERR; 2314 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc)); 2315 return 0; 2316 } 2317 2318 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX", 2319 sizeof(tempfilename)); 2320 if (NULL == mkdtemp(tempfilename)) { 2321 exitcode = (int)MANDOCLEVEL_SYSERR; 2322 say("", "&%s", tempfilename); 2323 return 0; 2324 } 2325 (void)strlcat(tempfilename, "/" MANDOC_DB, 2326 sizeof(tempfilename)); 2327 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL); 2328 if (SQLITE_OK != rc) { 2329 exitcode = (int)MANDOCLEVEL_SYSERR; 2330 say("", "%s: %s", tempfilename, sqlite3_errstr(rc)); 2331 return 0; 2332 } 2333 2334 create_tables: 2335 sql = "CREATE TABLE \"mpages\" (\n" 2336 " \"desc\" TEXT NOT NULL,\n" 2337 " \"form\" INTEGER NOT NULL,\n" 2338 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n" 2339 ");\n" 2340 "\n" 2341 "CREATE TABLE \"mlinks\" (\n" 2342 " \"sec\" TEXT NOT NULL,\n" 2343 " \"arch\" TEXT NOT NULL,\n" 2344 " \"name\" TEXT NOT NULL,\n" 2345 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " 2346 "ON DELETE CASCADE\n" 2347 ");\n" 2348 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n" 2349 "\n" 2350 "CREATE TABLE \"names\" (\n" 2351 " \"bits\" INTEGER NOT NULL,\n" 2352 " \"name\" TEXT NOT NULL,\n" 2353 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " 2354 "ON DELETE CASCADE,\n" 2355 " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n" 2356 ");\n" 2357 "\n" 2358 "CREATE TABLE \"keys\" (\n" 2359 " \"bits\" INTEGER NOT NULL,\n" 2360 " \"key\" TEXT NOT NULL,\n" 2361 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) " 2362 "ON DELETE CASCADE\n" 2363 ");\n" 2364 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n"; 2365 2366 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) { 2367 exitcode = (int)MANDOCLEVEL_SYSERR; 2368 say(MANDOC_DB, "%s", sqlite3_errmsg(db)); 2369 sqlite3_close(db); 2370 return 0; 2371 } 2372 2373 prepare_statements: 2374 if (SQLITE_OK != sqlite3_exec(db, 2375 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) { 2376 exitcode = (int)MANDOCLEVEL_SYSERR; 2377 say(MANDOC_DB, "PRAGMA foreign_keys: %s", 2378 sqlite3_errmsg(db)); 2379 sqlite3_close(db); 2380 return 0; 2381 } 2382 2383 sql = "DELETE FROM mpages WHERE pageid IN " 2384 "(SELECT pageid FROM mlinks WHERE " 2385 "sec=? AND arch=? AND name=?)"; 2386 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL); 2387 sql = "INSERT INTO mpages " 2388 "(desc,form) VALUES (?,?)"; 2389 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL); 2390 sql = "INSERT INTO mlinks " 2391 "(sec,arch,name,pageid) VALUES (?,?,?,?)"; 2392 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL); 2393 sql = "SELECT bits FROM names where pageid = ?"; 2394 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL); 2395 sql = "INSERT INTO names " 2396 "(bits,name,pageid) VALUES (?,?,?)"; 2397 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL); 2398 sql = "INSERT INTO keys " 2399 "(bits,key,pageid) VALUES (?,?,?)"; 2400 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL); 2401 2402 #ifndef __APPLE__ 2403 /* 2404 * When opening a new database, we can turn off 2405 * synchronous mode for much better performance. 2406 */ 2407 2408 if (real && SQLITE_OK != sqlite3_exec(db, 2409 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) { 2410 exitcode = (int)MANDOCLEVEL_SYSERR; 2411 say(MANDOC_DB, "PRAGMA synchronous: %s", 2412 sqlite3_errmsg(db)); 2413 sqlite3_close(db); 2414 return 0; 2415 } 2416 #endif 2417 2418 return 1; 2419 } 2420 2421 static int 2422 set_basedir(const char *targetdir, int report_baddir) 2423 { 2424 static char startdir[PATH_MAX]; 2425 static int getcwd_status; /* 1 = ok, 2 = failure */ 2426 static int chdir_status; /* 1 = changed directory */ 2427 char *cp; 2428 2429 /* 2430 * Remember the original working directory, if possible. 2431 * This will be needed if the second or a later directory 2432 * on the command line is given as a relative path. 2433 * Do not error out if the current directory is not 2434 * searchable: Maybe it won't be needed after all. 2435 */ 2436 if (0 == getcwd_status) { 2437 if (NULL == getcwd(startdir, sizeof(startdir))) { 2438 getcwd_status = 2; 2439 (void)strlcpy(startdir, strerror(errno), 2440 sizeof(startdir)); 2441 } else 2442 getcwd_status = 1; 2443 } 2444 2445 /* 2446 * We are leaving the old base directory. 2447 * Do not use it any longer, not even for messages. 2448 */ 2449 *basedir = '\0'; 2450 2451 /* 2452 * If and only if the directory was changed earlier and 2453 * the next directory to process is given as a relative path, 2454 * first go back, or bail out if that is impossible. 2455 */ 2456 if (chdir_status && '/' != *targetdir) { 2457 if (2 == getcwd_status) { 2458 exitcode = (int)MANDOCLEVEL_SYSERR; 2459 say("", "getcwd: %s", startdir); 2460 return 0; 2461 } 2462 if (-1 == chdir(startdir)) { 2463 exitcode = (int)MANDOCLEVEL_SYSERR; 2464 say("", "&chdir %s", startdir); 2465 return 0; 2466 } 2467 } 2468 2469 /* 2470 * Always resolve basedir to the canonicalized absolute 2471 * pathname and append a trailing slash, such that 2472 * we can reliably check whether files are inside. 2473 */ 2474 if (NULL == realpath(targetdir, basedir)) { 2475 if (report_baddir || errno != ENOENT) { 2476 exitcode = (int)MANDOCLEVEL_BADARG; 2477 say("", "&%s: realpath", targetdir); 2478 } 2479 return 0; 2480 } else if (-1 == chdir(basedir)) { 2481 if (report_baddir || errno != ENOENT) { 2482 exitcode = (int)MANDOCLEVEL_BADARG; 2483 say("", "&chdir"); 2484 } 2485 return 0; 2486 } 2487 chdir_status = 1; 2488 cp = strchr(basedir, '\0'); 2489 if ('/' != cp[-1]) { 2490 if (cp - basedir >= PATH_MAX - 1) { 2491 exitcode = (int)MANDOCLEVEL_SYSERR; 2492 say("", "Filename too long"); 2493 return 0; 2494 } 2495 *cp++ = '/'; 2496 *cp = '\0'; 2497 } 2498 return 1; 2499 } 2500 2501 static void 2502 say(const char *file, const char *format, ...) 2503 { 2504 va_list ap; 2505 int use_errno; 2506 2507 if ('\0' != *basedir) 2508 fprintf(stderr, "%s", basedir); 2509 if ('\0' != *basedir && '\0' != *file) 2510 fputc('/', stderr); 2511 if ('\0' != *file) 2512 fprintf(stderr, "%s", file); 2513 2514 use_errno = 1; 2515 if (NULL != format) { 2516 switch (*format) { 2517 case '&': 2518 format++; 2519 break; 2520 case '\0': 2521 format = NULL; 2522 break; 2523 default: 2524 use_errno = 0; 2525 break; 2526 } 2527 } 2528 if (NULL != format) { 2529 if ('\0' != *basedir || '\0' != *file) 2530 fputs(": ", stderr); 2531 va_start(ap, format); 2532 vfprintf(stderr, format, ap); 2533 va_end(ap); 2534 } 2535 if (use_errno) { 2536 if ('\0' != *basedir || '\0' != *file || NULL != format) 2537 fputs(": ", stderr); 2538 perror(NULL); 2539 } else 2540 fputc('\n', stderr); 2541 } 2542