1 /* $OpenBSD: mandocdb.c,v 1.216 2020/04/03 11:34:19 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2011-2020 Ingo Schwarze <schwarze@openbsd.org> 4 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv> 5 * Copyright (c) 2016 Ed Maste <emaste@freebsd.org> 6 * 7 * Permission to use, copy, modify, and distribute this software for any 8 * purpose with or without fee is hereby granted, provided that the above 9 * copyright notice and this permission notice appear in all copies. 10 * 11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 18 * 19 * Implementation of the makewhatis(8) program. 20 */ 21 #include <sys/types.h> 22 #include <sys/mman.h> 23 #include <sys/stat.h> 24 25 #include <assert.h> 26 #include <ctype.h> 27 #include <err.h> 28 #include <errno.h> 29 #include <fcntl.h> 30 #include <fts.h> 31 #include <limits.h> 32 #include <stdarg.h> 33 #include <stddef.h> 34 #include <stdio.h> 35 #include <stdint.h> 36 #include <stdlib.h> 37 #include <string.h> 38 #include <unistd.h> 39 40 #include "mandoc_aux.h" 41 #include "mandoc_ohash.h" 42 #include "mandoc.h" 43 #include "roff.h" 44 #include "mdoc.h" 45 #include "man.h" 46 #include "mandoc_parse.h" 47 #include "manconf.h" 48 #include "mansearch.h" 49 #include "dba_array.h" 50 #include "dba.h" 51 52 extern const char *const mansearch_keynames[]; 53 54 enum op { 55 OP_DEFAULT = 0, /* new dbs from dir list or default config */ 56 OP_CONFFILE, /* new databases from custom config file */ 57 OP_UPDATE, /* delete/add entries in existing database */ 58 OP_DELETE, /* delete entries from existing database */ 59 OP_TEST /* change no databases, report potential problems */ 60 }; 61 62 struct str { 63 const struct mpage *mpage; /* if set, the owning parse */ 64 uint64_t mask; /* bitmask in sequence */ 65 char key[]; /* rendered text */ 66 }; 67 68 struct inodev { 69 ino_t st_ino; 70 dev_t st_dev; 71 }; 72 73 struct mpage { 74 struct inodev inodev; /* used for hashing routine */ 75 struct dba_array *dba; 76 char *sec; /* section from file content */ 77 char *arch; /* architecture from file content */ 78 char *title; /* title from file content */ 79 char *desc; /* description from file content */ 80 struct mpage *next; /* singly linked list */ 81 struct mlink *mlinks; /* singly linked list */ 82 int name_head_done; 83 enum form form; /* format from file content */ 84 }; 85 86 struct mlink { 87 char file[PATH_MAX]; /* filename rel. to manpath */ 88 char *dsec; /* section from directory */ 89 char *arch; /* architecture from directory */ 90 char *name; /* name from file name (not empty) */ 91 char *fsec; /* section from file name suffix */ 92 struct mlink *next; /* singly linked list */ 93 struct mpage *mpage; /* parent */ 94 int gzip; /* filename has a .gz suffix */ 95 enum form dform; /* format from directory */ 96 enum form fform; /* format from file name suffix */ 97 }; 98 99 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *, 100 const struct roff_node *); 101 102 struct mdoc_handler { 103 mdoc_fp fp; /* optional handler */ 104 uint64_t mask; /* set unless handler returns 0 */ 105 int taboo; /* node flags that must not be set */ 106 }; 107 108 109 int mandocdb(int, char *[]); 110 111 static void dbadd(struct dba *, struct mpage *); 112 static void dbadd_mlink(const struct mlink *); 113 static void dbprune(struct dba *); 114 static void dbwrite(struct dba *); 115 static void filescan(const char *); 116 static int fts_compare(const FTSENT **, const FTSENT **); 117 static void mlink_add(struct mlink *, const struct stat *); 118 static void mlink_check(struct mpage *, struct mlink *); 119 static void mlink_free(struct mlink *); 120 static void mlinks_undupe(struct mpage *); 121 static void mpages_free(void); 122 static void mpages_merge(struct dba *, struct mparse *); 123 static void parse_cat(struct mpage *, int); 124 static void parse_man(struct mpage *, const struct roff_meta *, 125 const struct roff_node *); 126 static void parse_mdoc(struct mpage *, const struct roff_meta *, 127 const struct roff_node *); 128 static int parse_mdoc_head(struct mpage *, const struct roff_meta *, 129 const struct roff_node *); 130 static int parse_mdoc_Fa(struct mpage *, const struct roff_meta *, 131 const struct roff_node *); 132 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *, 133 const struct roff_node *); 134 static void parse_mdoc_fname(struct mpage *, const struct roff_node *); 135 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *, 136 const struct roff_node *); 137 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *, 138 const struct roff_node *); 139 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *, 140 const struct roff_node *); 141 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *, 142 const struct roff_node *); 143 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *, 144 const struct roff_node *); 145 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *, 146 const struct roff_node *); 147 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *, 148 const struct roff_node *); 149 static void putkey(const struct mpage *, char *, uint64_t); 150 static void putkeys(const struct mpage *, char *, size_t, uint64_t); 151 static void putmdockey(const struct mpage *, 152 const struct roff_node *, uint64_t, int); 153 static int render_string(char **, size_t *); 154 static void say(const char *, const char *, ...) 155 __attribute__((__format__ (__printf__, 2, 3))); 156 static int set_basedir(const char *, int); 157 static int treescan(void); 158 static size_t utf8(unsigned int, char [7]); 159 160 static int nodb; /* no database changes */ 161 static int mparse_options; /* abort the parse early */ 162 static int use_all; /* use all found files */ 163 static int debug; /* print what we're doing */ 164 static int warnings; /* warn about crap */ 165 static int write_utf8; /* write UTF-8 output; else ASCII */ 166 static int exitcode; /* to be returned by main */ 167 static enum op op; /* operational mode */ 168 static char basedir[PATH_MAX]; /* current base directory */ 169 static size_t basedir_len; /* strlen(basedir) */ 170 static struct mpage *mpage_head; /* list of distinct manual pages */ 171 static struct ohash mpages; /* table of distinct manual pages */ 172 static struct ohash mlinks; /* table of directory entries */ 173 static struct ohash names; /* table of all names */ 174 static struct ohash strings; /* table of all strings */ 175 static uint64_t name_mask; 176 177 static const struct mdoc_handler mdoc_handlers[MDOC_MAX - MDOC_Dd] = { 178 { NULL, 0, NODE_NOPRT }, /* Dd */ 179 { NULL, 0, NODE_NOPRT }, /* Dt */ 180 { NULL, 0, NODE_NOPRT }, /* Os */ 181 { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */ 182 { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */ 183 { NULL, 0, 0 }, /* Pp */ 184 { NULL, 0, 0 }, /* D1 */ 185 { NULL, 0, 0 }, /* Dl */ 186 { NULL, 0, 0 }, /* Bd */ 187 { NULL, 0, 0 }, /* Ed */ 188 { NULL, 0, 0 }, /* Bl */ 189 { NULL, 0, 0 }, /* El */ 190 { NULL, 0, 0 }, /* It */ 191 { NULL, 0, 0 }, /* Ad */ 192 { NULL, TYPE_An, 0 }, /* An */ 193 { NULL, 0, 0 }, /* Ap */ 194 { NULL, TYPE_Ar, 0 }, /* Ar */ 195 { NULL, TYPE_Cd, 0 }, /* Cd */ 196 { NULL, TYPE_Cm, 0 }, /* Cm */ 197 { NULL, TYPE_Dv, 0 }, /* Dv */ 198 { NULL, TYPE_Er, 0 }, /* Er */ 199 { NULL, TYPE_Ev, 0 }, /* Ev */ 200 { NULL, 0, 0 }, /* Ex */ 201 { parse_mdoc_Fa, 0, 0 }, /* Fa */ 202 { parse_mdoc_Fd, 0, 0 }, /* Fd */ 203 { NULL, TYPE_Fl, 0 }, /* Fl */ 204 { parse_mdoc_Fn, 0, 0 }, /* Fn */ 205 { NULL, TYPE_Ft | TYPE_Vt, 0 }, /* Ft */ 206 { NULL, TYPE_Ic, 0 }, /* Ic */ 207 { NULL, TYPE_In, 0 }, /* In */ 208 { NULL, TYPE_Li, 0 }, /* Li */ 209 { parse_mdoc_Nd, 0, 0 }, /* Nd */ 210 { parse_mdoc_Nm, 0, 0 }, /* Nm */ 211 { NULL, 0, 0 }, /* Op */ 212 { NULL, 0, 0 }, /* Ot */ 213 { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */ 214 { NULL, 0, 0 }, /* Rv */ 215 { NULL, TYPE_St, 0 }, /* St */ 216 { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */ 217 { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */ 218 { parse_mdoc_Xr, 0, 0 }, /* Xr */ 219 { NULL, 0, 0 }, /* %A */ 220 { NULL, 0, 0 }, /* %B */ 221 { NULL, 0, 0 }, /* %D */ 222 { NULL, 0, 0 }, /* %I */ 223 { NULL, 0, 0 }, /* %J */ 224 { NULL, 0, 0 }, /* %N */ 225 { NULL, 0, 0 }, /* %O */ 226 { NULL, 0, 0 }, /* %P */ 227 { NULL, 0, 0 }, /* %R */ 228 { NULL, 0, 0 }, /* %T */ 229 { NULL, 0, 0 }, /* %V */ 230 { NULL, 0, 0 }, /* Ac */ 231 { NULL, 0, 0 }, /* Ao */ 232 { NULL, 0, 0 }, /* Aq */ 233 { NULL, TYPE_At, 0 }, /* At */ 234 { NULL, 0, 0 }, /* Bc */ 235 { NULL, 0, 0 }, /* Bf */ 236 { NULL, 0, 0 }, /* Bo */ 237 { NULL, 0, 0 }, /* Bq */ 238 { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */ 239 { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */ 240 { NULL, 0, 0 }, /* Db */ 241 { NULL, 0, 0 }, /* Dc */ 242 { NULL, 0, 0 }, /* Do */ 243 { NULL, 0, 0 }, /* Dq */ 244 { NULL, 0, 0 }, /* Ec */ 245 { NULL, 0, 0 }, /* Ef */ 246 { NULL, TYPE_Em, 0 }, /* Em */ 247 { NULL, 0, 0 }, /* Eo */ 248 { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */ 249 { NULL, TYPE_Ms, 0 }, /* Ms */ 250 { NULL, 0, 0 }, /* No */ 251 { NULL, 0, 0 }, /* Ns */ 252 { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */ 253 { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */ 254 { NULL, 0, 0 }, /* Pc */ 255 { NULL, 0, 0 }, /* Pf */ 256 { NULL, 0, 0 }, /* Po */ 257 { NULL, 0, 0 }, /* Pq */ 258 { NULL, 0, 0 }, /* Qc */ 259 { NULL, 0, 0 }, /* Ql */ 260 { NULL, 0, 0 }, /* Qo */ 261 { NULL, 0, 0 }, /* Qq */ 262 { NULL, 0, 0 }, /* Re */ 263 { NULL, 0, 0 }, /* Rs */ 264 { NULL, 0, 0 }, /* Sc */ 265 { NULL, 0, 0 }, /* So */ 266 { NULL, 0, 0 }, /* Sq */ 267 { NULL, 0, 0 }, /* Sm */ 268 { NULL, 0, 0 }, /* Sx */ 269 { NULL, TYPE_Sy, 0 }, /* Sy */ 270 { NULL, TYPE_Tn, 0 }, /* Tn */ 271 { NULL, 0, NODE_NOSRC }, /* Ux */ 272 { NULL, 0, 0 }, /* Xc */ 273 { NULL, 0, 0 }, /* Xo */ 274 { parse_mdoc_Fo, 0, 0 }, /* Fo */ 275 { NULL, 0, 0 }, /* Fc */ 276 { NULL, 0, 0 }, /* Oo */ 277 { NULL, 0, 0 }, /* Oc */ 278 { NULL, 0, 0 }, /* Bk */ 279 { NULL, 0, 0 }, /* Ek */ 280 { NULL, 0, 0 }, /* Bt */ 281 { NULL, 0, 0 }, /* Hf */ 282 { NULL, 0, 0 }, /* Fr */ 283 { NULL, 0, 0 }, /* Ud */ 284 { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */ 285 { NULL, 0, 0 }, /* Lp */ 286 { NULL, TYPE_Lk, 0 }, /* Lk */ 287 { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */ 288 { NULL, 0, 0 }, /* Brq */ 289 { NULL, 0, 0 }, /* Bro */ 290 { NULL, 0, 0 }, /* Brc */ 291 { NULL, 0, 0 }, /* %C */ 292 { NULL, 0, 0 }, /* Es */ 293 { NULL, 0, 0 }, /* En */ 294 { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */ 295 { NULL, 0, 0 }, /* %Q */ 296 { NULL, 0, 0 }, /* %U */ 297 { NULL, 0, 0 }, /* Ta */ 298 }; 299 300 301 int 302 mandocdb(int argc, char *argv[]) 303 { 304 struct manconf conf; 305 struct mparse *mp; 306 struct dba *dba; 307 const char *path_arg, *progname; 308 size_t j, sz; 309 int ch, i; 310 311 if (pledge("stdio rpath wpath cpath", NULL) == -1) { 312 warn("pledge"); 313 return (int)MANDOCLEVEL_SYSERR; 314 } 315 316 memset(&conf, 0, sizeof(conf)); 317 318 /* 319 * We accept a few different invocations. 320 * The CHECKOP macro makes sure that invocation styles don't 321 * clobber each other. 322 */ 323 #define CHECKOP(_op, _ch) do \ 324 if ((_op) != OP_DEFAULT) { \ 325 warnx("-%c: Conflicting option", (_ch)); \ 326 goto usage; \ 327 } while (/*CONSTCOND*/0) 328 329 mparse_options = MPARSE_VALIDATE; 330 path_arg = NULL; 331 op = OP_DEFAULT; 332 333 while ((ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")) != -1) 334 switch (ch) { 335 case 'a': 336 use_all = 1; 337 break; 338 case 'C': 339 CHECKOP(op, ch); 340 path_arg = optarg; 341 op = OP_CONFFILE; 342 break; 343 case 'D': 344 debug++; 345 break; 346 case 'd': 347 CHECKOP(op, ch); 348 path_arg = optarg; 349 op = OP_UPDATE; 350 break; 351 case 'n': 352 nodb = 1; 353 break; 354 case 'p': 355 warnings = 1; 356 break; 357 case 'Q': 358 mparse_options |= MPARSE_QUICK; 359 break; 360 case 'T': 361 if (strcmp(optarg, "utf8") != 0) { 362 warnx("-T%s: Unsupported output format", 363 optarg); 364 goto usage; 365 } 366 write_utf8 = 1; 367 break; 368 case 't': 369 CHECKOP(op, ch); 370 dup2(STDOUT_FILENO, STDERR_FILENO); 371 op = OP_TEST; 372 nodb = warnings = 1; 373 break; 374 case 'u': 375 CHECKOP(op, ch); 376 path_arg = optarg; 377 op = OP_DELETE; 378 break; 379 case 'v': 380 /* Compatibility with espie@'s makewhatis. */ 381 break; 382 default: 383 goto usage; 384 } 385 386 argc -= optind; 387 argv += optind; 388 389 if (nodb) { 390 if (pledge("stdio rpath", NULL) == -1) { 391 warn("pledge"); 392 return (int)MANDOCLEVEL_SYSERR; 393 } 394 } 395 396 if (op == OP_CONFFILE && argc > 0) { 397 warnx("-C: Too many arguments"); 398 goto usage; 399 } 400 401 exitcode = (int)MANDOCLEVEL_OK; 402 mchars_alloc(); 403 mp = mparse_alloc(mparse_options, MANDOC_OS_OTHER, NULL); 404 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev)); 405 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file)); 406 407 if (op == OP_UPDATE || op == OP_DELETE || op == OP_TEST) { 408 409 /* 410 * Most of these deal with a specific directory. 411 * Jump into that directory first. 412 */ 413 if (op != OP_TEST && set_basedir(path_arg, 1) == 0) 414 goto out; 415 416 dba = nodb ? dba_new(128) : dba_read(MANDOC_DB); 417 if (dba != NULL) { 418 /* 419 * The existing database is usable. Process 420 * all files specified on the command-line. 421 */ 422 use_all = 1; 423 for (i = 0; i < argc; i++) 424 filescan(argv[i]); 425 if (nodb == 0) 426 dbprune(dba); 427 } else { 428 /* Database missing or corrupt. */ 429 if (op != OP_UPDATE || errno != ENOENT) 430 say(MANDOC_DB, "%s: Automatically recreating" 431 " from scratch", strerror(errno)); 432 exitcode = (int)MANDOCLEVEL_OK; 433 op = OP_DEFAULT; 434 if (treescan() == 0) 435 goto out; 436 dba = dba_new(128); 437 } 438 if (op != OP_DELETE) 439 mpages_merge(dba, mp); 440 if (nodb == 0) 441 dbwrite(dba); 442 dba_free(dba); 443 } else { 444 /* 445 * If we have arguments, use them as our manpaths. 446 * If we don't, use man.conf(5). 447 */ 448 if (argc > 0) { 449 conf.manpath.paths = mandoc_reallocarray(NULL, 450 argc, sizeof(char *)); 451 conf.manpath.sz = (size_t)argc; 452 for (i = 0; i < argc; i++) 453 conf.manpath.paths[i] = mandoc_strdup(argv[i]); 454 } else 455 manconf_parse(&conf, path_arg, NULL, NULL); 456 457 if (conf.manpath.sz == 0) { 458 exitcode = (int)MANDOCLEVEL_BADARG; 459 say("", "Empty manpath"); 460 } 461 462 /* 463 * First scan the tree rooted at a base directory, then 464 * build a new database and finally move it into place. 465 * Ignore zero-length directories and strip trailing 466 * slashes. 467 */ 468 for (j = 0; j < conf.manpath.sz; j++) { 469 sz = strlen(conf.manpath.paths[j]); 470 if (sz && conf.manpath.paths[j][sz - 1] == '/') 471 conf.manpath.paths[j][--sz] = '\0'; 472 if (sz == 0) 473 continue; 474 475 if (j) { 476 mandoc_ohash_init(&mpages, 6, 477 offsetof(struct mpage, inodev)); 478 mandoc_ohash_init(&mlinks, 6, 479 offsetof(struct mlink, file)); 480 } 481 482 if (set_basedir(conf.manpath.paths[j], argc > 0) == 0) 483 continue; 484 if (treescan() == 0) 485 continue; 486 dba = dba_new(128); 487 mpages_merge(dba, mp); 488 if (nodb == 0) 489 dbwrite(dba); 490 dba_free(dba); 491 492 if (j + 1 < conf.manpath.sz) { 493 mpages_free(); 494 ohash_delete(&mpages); 495 ohash_delete(&mlinks); 496 } 497 } 498 } 499 out: 500 manconf_free(&conf); 501 mparse_free(mp); 502 mchars_free(); 503 mpages_free(); 504 ohash_delete(&mpages); 505 ohash_delete(&mlinks); 506 return exitcode; 507 usage: 508 progname = getprogname(); 509 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n" 510 " %s [-aDnpQ] [-Tutf8] dir ...\n" 511 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n" 512 " %s [-Dnp] -u dir [file ...]\n" 513 " %s [-Q] -t file ...\n", 514 progname, progname, progname, progname, progname); 515 516 return (int)MANDOCLEVEL_BADARG; 517 } 518 519 /* 520 * To get a singly linked list in alpha order while inserting entries 521 * at the beginning, process directory entries in reverse alpha order. 522 */ 523 static int 524 fts_compare(const FTSENT **a, const FTSENT **b) 525 { 526 return -strcmp((*a)->fts_name, (*b)->fts_name); 527 } 528 529 /* 530 * Scan a directory tree rooted at "basedir" for manpages. 531 * We use fts(), scanning directory parts along the way for clues to our 532 * section and architecture. 533 * 534 * If use_all has been specified, grok all files. 535 * If not, sanitise paths to the following: 536 * 537 * [./]man*[/<arch>]/<name>.<section> 538 * or 539 * [./]cat<section>[/<arch>]/<name>.0 540 * 541 * TODO: accommodate for multi-language directories. 542 */ 543 static int 544 treescan(void) 545 { 546 char buf[PATH_MAX]; 547 FTS *f; 548 FTSENT *ff; 549 struct mlink *mlink; 550 int gzip; 551 enum form dform; 552 char *dsec, *arch, *fsec, *cp; 553 const char *path; 554 const char *argv[2]; 555 556 argv[0] = "."; 557 argv[1] = NULL; 558 559 f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR, 560 fts_compare); 561 if (f == NULL) { 562 exitcode = (int)MANDOCLEVEL_SYSERR; 563 say("", "&fts_open"); 564 return 0; 565 } 566 567 dsec = arch = NULL; 568 dform = FORM_NONE; 569 570 while ((ff = fts_read(f)) != NULL) { 571 path = ff->fts_path + 2; 572 switch (ff->fts_info) { 573 574 /* 575 * Symbolic links require various sanity checks, 576 * then get handled just like regular files. 577 */ 578 case FTS_SL: 579 if (realpath(path, buf) == NULL) { 580 if (warnings) 581 say(path, "&realpath"); 582 continue; 583 } 584 if (strncmp(buf, basedir, basedir_len) != 0) { 585 if (warnings) say("", 586 "%s: outside base directory", buf); 587 continue; 588 } 589 /* Use logical inode to avoid mpages dupe. */ 590 if (stat(path, ff->fts_statp) == -1) { 591 if (warnings) 592 say(path, "&stat"); 593 continue; 594 } 595 /* FALLTHROUGH */ 596 597 /* 598 * If we're a regular file, add an mlink by using the 599 * stored directory data and handling the filename. 600 */ 601 case FTS_F: 602 if ( ! strcmp(path, MANDOC_DB)) 603 continue; 604 if ( ! use_all && ff->fts_level < 2) { 605 if (warnings) 606 say(path, "Extraneous file"); 607 continue; 608 } 609 gzip = 0; 610 fsec = NULL; 611 while (fsec == NULL) { 612 fsec = strrchr(ff->fts_name, '.'); 613 if (fsec == NULL || strcmp(fsec+1, "gz")) 614 break; 615 gzip = 1; 616 *fsec = '\0'; 617 fsec = NULL; 618 } 619 if (fsec == NULL) { 620 if ( ! use_all) { 621 if (warnings) 622 say(path, 623 "No filename suffix"); 624 continue; 625 } 626 } else if ( ! strcmp(++fsec, "html")) { 627 if (warnings) 628 say(path, "Skip html"); 629 continue; 630 } else if ( ! strcmp(fsec, "ps")) { 631 if (warnings) 632 say(path, "Skip ps"); 633 continue; 634 } else if ( ! strcmp(fsec, "pdf")) { 635 if (warnings) 636 say(path, "Skip pdf"); 637 continue; 638 } else if ( ! use_all && 639 ((dform == FORM_SRC && 640 strncmp(fsec, dsec, strlen(dsec))) || 641 (dform == FORM_CAT && strcmp(fsec, "0")))) { 642 if (warnings) 643 say(path, "Wrong filename suffix"); 644 continue; 645 } else 646 fsec[-1] = '\0'; 647 648 mlink = mandoc_calloc(1, sizeof(struct mlink)); 649 if (strlcpy(mlink->file, path, 650 sizeof(mlink->file)) >= 651 sizeof(mlink->file)) { 652 say(path, "Filename too long"); 653 free(mlink); 654 continue; 655 } 656 mlink->dform = dform; 657 mlink->dsec = dsec; 658 mlink->arch = arch; 659 mlink->name = ff->fts_name; 660 mlink->fsec = fsec; 661 mlink->gzip = gzip; 662 mlink_add(mlink, ff->fts_statp); 663 continue; 664 665 case FTS_D: 666 case FTS_DP: 667 break; 668 669 default: 670 if (warnings) 671 say(path, "Not a regular file"); 672 continue; 673 } 674 675 switch (ff->fts_level) { 676 case 0: 677 /* Ignore the root directory. */ 678 break; 679 case 1: 680 /* 681 * This might contain manX/ or catX/. 682 * Try to infer this from the name. 683 * If we're not in use_all, enforce it. 684 */ 685 cp = ff->fts_name; 686 if (ff->fts_info == FTS_DP) { 687 dform = FORM_NONE; 688 dsec = NULL; 689 break; 690 } 691 692 if ( ! strncmp(cp, "man", 3)) { 693 dform = FORM_SRC; 694 dsec = cp + 3; 695 } else if ( ! strncmp(cp, "cat", 3)) { 696 dform = FORM_CAT; 697 dsec = cp + 3; 698 } else { 699 dform = FORM_NONE; 700 dsec = NULL; 701 } 702 703 if (dsec != NULL || use_all) 704 break; 705 706 if (warnings) 707 say(path, "Unknown directory part"); 708 fts_set(f, ff, FTS_SKIP); 709 break; 710 case 2: 711 /* 712 * Possibly our architecture. 713 * If we're descending, keep tabs on it. 714 */ 715 if (ff->fts_info != FTS_DP && dsec != NULL) 716 arch = ff->fts_name; 717 else 718 arch = NULL; 719 break; 720 default: 721 if (ff->fts_info == FTS_DP || use_all) 722 break; 723 if (warnings) 724 say(path, "Extraneous directory part"); 725 fts_set(f, ff, FTS_SKIP); 726 break; 727 } 728 } 729 730 fts_close(f); 731 return 1; 732 } 733 734 /* 735 * Add a file to the mlinks table. 736 * Do not verify that it's a "valid" looking manpage (we'll do that 737 * later). 738 * 739 * Try to infer the manual section, architecture, and page name from the 740 * path, assuming it looks like 741 * 742 * [./]man*[/<arch>]/<name>.<section> 743 * or 744 * [./]cat<section>[/<arch>]/<name>.0 745 * 746 * See treescan() for the fts(3) version of this. 747 */ 748 static void 749 filescan(const char *infile) 750 { 751 struct stat st; 752 struct mlink *mlink; 753 char *linkfile, *p, *realdir, *start, *usefile; 754 size_t realdir_len; 755 756 assert(use_all); 757 758 if (strncmp(infile, "./", 2) == 0) 759 infile += 2; 760 761 /* 762 * We have to do lstat(2) before realpath(3) loses 763 * the information whether this is a symbolic link. 764 * We need to know that because for symbolic links, 765 * we want to use the orginal file name, while for 766 * regular files, we want to use the real path. 767 */ 768 if (lstat(infile, &st) == -1) { 769 exitcode = (int)MANDOCLEVEL_BADARG; 770 say(infile, "&lstat"); 771 return; 772 } else if (S_ISREG(st.st_mode) == 0 && S_ISLNK(st.st_mode) == 0) { 773 exitcode = (int)MANDOCLEVEL_BADARG; 774 say(infile, "Not a regular file"); 775 return; 776 } 777 778 /* 779 * We have to resolve the file name to the real path 780 * in any case for the base directory check. 781 */ 782 if ((usefile = realpath(infile, NULL)) == NULL) { 783 exitcode = (int)MANDOCLEVEL_BADARG; 784 say(infile, "&realpath"); 785 return; 786 } 787 788 if (op == OP_TEST) 789 start = usefile; 790 else if (strncmp(usefile, basedir, basedir_len) == 0) 791 start = usefile + basedir_len; 792 else { 793 exitcode = (int)MANDOCLEVEL_BADARG; 794 say("", "%s: outside base directory", infile); 795 free(usefile); 796 return; 797 } 798 799 /* 800 * Now we are sure the file is inside our tree. 801 * If it is a symbolic link, ignore the real path 802 * and use the original name. 803 */ 804 do { 805 if (S_ISLNK(st.st_mode) == 0) 806 break; 807 808 /* 809 * Some implementations of realpath(3) may succeed 810 * even if the target of the link does not exist, 811 * so check again for extra safety. 812 */ 813 if (stat(usefile, &st) == -1) { 814 exitcode = (int)MANDOCLEVEL_BADARG; 815 say(infile, "&stat"); 816 free(usefile); 817 return; 818 } 819 linkfile = mandoc_strdup(infile); 820 if (op == OP_TEST) { 821 free(usefile); 822 start = usefile = linkfile; 823 break; 824 } 825 if (strncmp(infile, basedir, basedir_len) == 0) { 826 free(usefile); 827 usefile = linkfile; 828 start = usefile + basedir_len; 829 break; 830 } 831 832 /* 833 * This symbolic link points into the basedir 834 * from the outside. Let's see whether any of 835 * the parent directories resolve to the basedir. 836 */ 837 p = strchr(linkfile, '\0'); 838 do { 839 while (*--p != '/') 840 continue; 841 *p = '\0'; 842 if ((realdir = realpath(linkfile, NULL)) == NULL) { 843 exitcode = (int)MANDOCLEVEL_BADARG; 844 say(infile, "&realpath"); 845 free(linkfile); 846 free(usefile); 847 return; 848 } 849 realdir_len = strlen(realdir) + 1; 850 free(realdir); 851 *p = '/'; 852 } while (realdir_len > basedir_len); 853 854 /* 855 * If one of the directories resolves to the basedir, 856 * use the rest of the original name. 857 * Otherwise, the best we can do 858 * is to use the filename pointed to. 859 */ 860 if (realdir_len == basedir_len) { 861 free(usefile); 862 usefile = linkfile; 863 start = p + 1; 864 } else { 865 free(linkfile); 866 start = usefile + basedir_len; 867 } 868 } while (/* CONSTCOND */ 0); 869 870 mlink = mandoc_calloc(1, sizeof(struct mlink)); 871 mlink->dform = FORM_NONE; 872 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >= 873 sizeof(mlink->file)) { 874 say(start, "Filename too long"); 875 free(mlink); 876 free(usefile); 877 return; 878 } 879 880 /* 881 * In test mode or when the original name is absolute 882 * but outside our tree, guess the base directory. 883 */ 884 885 if (op == OP_TEST || (start == usefile && *start == '/')) { 886 if (strncmp(usefile, "man/", 4) == 0) 887 start = usefile + 4; 888 else if ((start = strstr(usefile, "/man/")) != NULL) 889 start += 5; 890 else 891 start = usefile; 892 } 893 894 /* 895 * First try to guess our directory structure. 896 * If we find a separator, try to look for man* or cat*. 897 * If we find one of these and what's underneath is a directory, 898 * assume it's an architecture. 899 */ 900 if ((p = strchr(start, '/')) != NULL) { 901 *p++ = '\0'; 902 if (strncmp(start, "man", 3) == 0) { 903 mlink->dform = FORM_SRC; 904 mlink->dsec = start + 3; 905 } else if (strncmp(start, "cat", 3) == 0) { 906 mlink->dform = FORM_CAT; 907 mlink->dsec = start + 3; 908 } 909 910 start = p; 911 if (mlink->dsec != NULL && (p = strchr(start, '/')) != NULL) { 912 *p++ = '\0'; 913 mlink->arch = start; 914 start = p; 915 } 916 } 917 918 /* 919 * Now check the file suffix. 920 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage. 921 */ 922 p = strrchr(start, '\0'); 923 while (p-- > start && *p != '/' && *p != '.') 924 continue; 925 926 if (*p == '.') { 927 *p++ = '\0'; 928 mlink->fsec = p; 929 } 930 931 /* 932 * Now try to parse the name. 933 * Use the filename portion of the path. 934 */ 935 mlink->name = start; 936 if ((p = strrchr(start, '/')) != NULL) { 937 mlink->name = p + 1; 938 *p = '\0'; 939 } 940 mlink_add(mlink, &st); 941 free(usefile); 942 } 943 944 static void 945 mlink_add(struct mlink *mlink, const struct stat *st) 946 { 947 struct inodev inodev; 948 struct mpage *mpage; 949 unsigned int slot; 950 951 assert(NULL != mlink->file); 952 953 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : ""); 954 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : ""); 955 mlink->name = mandoc_strdup(mlink->name ? mlink->name : ""); 956 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : ""); 957 958 if ('0' == *mlink->fsec) { 959 free(mlink->fsec); 960 mlink->fsec = mandoc_strdup(mlink->dsec); 961 mlink->fform = FORM_CAT; 962 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec) 963 mlink->fform = FORM_SRC; 964 else 965 mlink->fform = FORM_NONE; 966 967 slot = ohash_qlookup(&mlinks, mlink->file); 968 assert(NULL == ohash_find(&mlinks, slot)); 969 ohash_insert(&mlinks, slot, mlink); 970 971 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */ 972 inodev.st_ino = st->st_ino; 973 inodev.st_dev = st->st_dev; 974 slot = ohash_lookup_memory(&mpages, (char *)&inodev, 975 sizeof(struct inodev), inodev.st_ino); 976 mpage = ohash_find(&mpages, slot); 977 if (NULL == mpage) { 978 mpage = mandoc_calloc(1, sizeof(struct mpage)); 979 mpage->inodev.st_ino = inodev.st_ino; 980 mpage->inodev.st_dev = inodev.st_dev; 981 mpage->form = FORM_NONE; 982 mpage->next = mpage_head; 983 mpage_head = mpage; 984 ohash_insert(&mpages, slot, mpage); 985 } else 986 mlink->next = mpage->mlinks; 987 mpage->mlinks = mlink; 988 mlink->mpage = mpage; 989 } 990 991 static void 992 mlink_free(struct mlink *mlink) 993 { 994 995 free(mlink->dsec); 996 free(mlink->arch); 997 free(mlink->name); 998 free(mlink->fsec); 999 free(mlink); 1000 } 1001 1002 static void 1003 mpages_free(void) 1004 { 1005 struct mpage *mpage; 1006 struct mlink *mlink; 1007 1008 while ((mpage = mpage_head) != NULL) { 1009 while ((mlink = mpage->mlinks) != NULL) { 1010 mpage->mlinks = mlink->next; 1011 mlink_free(mlink); 1012 } 1013 mpage_head = mpage->next; 1014 free(mpage->sec); 1015 free(mpage->arch); 1016 free(mpage->title); 1017 free(mpage->desc); 1018 free(mpage); 1019 } 1020 } 1021 1022 /* 1023 * For each mlink to the mpage, check whether the path looks like 1024 * it is formatted, and if it does, check whether a source manual 1025 * exists by the same name, ignoring the suffix. 1026 * If both conditions hold, drop the mlink. 1027 */ 1028 static void 1029 mlinks_undupe(struct mpage *mpage) 1030 { 1031 char buf[PATH_MAX]; 1032 struct mlink **prev; 1033 struct mlink *mlink; 1034 char *bufp; 1035 1036 mpage->form = FORM_CAT; 1037 prev = &mpage->mlinks; 1038 while (NULL != (mlink = *prev)) { 1039 if (FORM_CAT != mlink->dform) { 1040 mpage->form = FORM_NONE; 1041 goto nextlink; 1042 } 1043 (void)strlcpy(buf, mlink->file, sizeof(buf)); 1044 bufp = strstr(buf, "cat"); 1045 assert(NULL != bufp); 1046 memcpy(bufp, "man", 3); 1047 if (NULL != (bufp = strrchr(buf, '.'))) 1048 *++bufp = '\0'; 1049 (void)strlcat(buf, mlink->dsec, sizeof(buf)); 1050 if (NULL == ohash_find(&mlinks, 1051 ohash_qlookup(&mlinks, buf))) 1052 goto nextlink; 1053 if (warnings) 1054 say(mlink->file, "Man source exists: %s", buf); 1055 if (use_all) 1056 goto nextlink; 1057 *prev = mlink->next; 1058 mlink_free(mlink); 1059 continue; 1060 nextlink: 1061 prev = &(*prev)->next; 1062 } 1063 } 1064 1065 static void 1066 mlink_check(struct mpage *mpage, struct mlink *mlink) 1067 { 1068 struct str *str; 1069 unsigned int slot; 1070 1071 /* 1072 * Check whether the manual section given in a file 1073 * agrees with the directory where the file is located. 1074 * Some manuals have suffixes like (3p) on their 1075 * section number either inside the file or in the 1076 * directory name, some are linked into more than one 1077 * section, like encrypt(1) = makekey(8). 1078 */ 1079 1080 if (FORM_SRC == mpage->form && 1081 strcasecmp(mpage->sec, mlink->dsec)) 1082 say(mlink->file, "Section \"%s\" manual in %s directory", 1083 mpage->sec, mlink->dsec); 1084 1085 /* 1086 * Manual page directories exist for each kernel 1087 * architecture as returned by machine(1). 1088 * However, many manuals only depend on the 1089 * application architecture as returned by arch(1). 1090 * For example, some (2/ARM) manuals are shared 1091 * across the "armish" and "zaurus" kernel 1092 * architectures. 1093 * A few manuals are even shared across completely 1094 * different architectures, for example fdformat(1) 1095 * on amd64, i386, and sparc64. 1096 */ 1097 1098 if (strcasecmp(mpage->arch, mlink->arch)) 1099 say(mlink->file, "Architecture \"%s\" manual in " 1100 "\"%s\" directory", mpage->arch, mlink->arch); 1101 1102 /* 1103 * XXX 1104 * parse_cat() doesn't set NAME_TITLE yet. 1105 */ 1106 1107 if (FORM_CAT == mpage->form) 1108 return; 1109 1110 /* 1111 * Check whether this mlink 1112 * appears as a name in the NAME section. 1113 */ 1114 1115 slot = ohash_qlookup(&names, mlink->name); 1116 str = ohash_find(&names, slot); 1117 assert(NULL != str); 1118 if ( ! (NAME_TITLE & str->mask)) 1119 say(mlink->file, "Name missing in NAME section"); 1120 } 1121 1122 /* 1123 * Run through the files in the global vector "mpages" 1124 * and add them to the database specified in "basedir". 1125 * 1126 * This handles the parsing scheme itself, using the cues of directory 1127 * and filename to determine whether the file is parsable or not. 1128 */ 1129 static void 1130 mpages_merge(struct dba *dba, struct mparse *mp) 1131 { 1132 struct mpage *mpage, *mpage_dest; 1133 struct mlink *mlink, *mlink_dest; 1134 struct roff_meta *meta; 1135 char *cp; 1136 int fd; 1137 1138 for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) { 1139 mlinks_undupe(mpage); 1140 if ((mlink = mpage->mlinks) == NULL) 1141 continue; 1142 1143 name_mask = NAME_MASK; 1144 mandoc_ohash_init(&names, 4, offsetof(struct str, key)); 1145 mandoc_ohash_init(&strings, 6, offsetof(struct str, key)); 1146 mparse_reset(mp); 1147 meta = NULL; 1148 1149 if ((fd = mparse_open(mp, mlink->file)) == -1) { 1150 say(mlink->file, "&open"); 1151 goto nextpage; 1152 } 1153 1154 /* 1155 * Interpret the file as mdoc(7) or man(7) source 1156 * code, unless it is known to be formatted. 1157 */ 1158 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) { 1159 mparse_readfd(mp, fd, mlink->file); 1160 close(fd); 1161 fd = -1; 1162 meta = mparse_result(mp); 1163 } 1164 1165 if (meta != NULL && meta->sodest != NULL) { 1166 mlink_dest = ohash_find(&mlinks, 1167 ohash_qlookup(&mlinks, meta->sodest)); 1168 if (mlink_dest == NULL) { 1169 mandoc_asprintf(&cp, "%s.gz", meta->sodest); 1170 mlink_dest = ohash_find(&mlinks, 1171 ohash_qlookup(&mlinks, cp)); 1172 free(cp); 1173 } 1174 if (mlink_dest != NULL) { 1175 1176 /* The .so target exists. */ 1177 1178 mpage_dest = mlink_dest->mpage; 1179 while (1) { 1180 mlink->mpage = mpage_dest; 1181 1182 /* 1183 * If the target was already 1184 * processed, add the links 1185 * to the database now. 1186 * Otherwise, this will 1187 * happen when we come 1188 * to the target. 1189 */ 1190 1191 if (mpage_dest->dba != NULL) 1192 dbadd_mlink(mlink); 1193 1194 if (mlink->next == NULL) 1195 break; 1196 mlink = mlink->next; 1197 } 1198 1199 /* Move all links to the target. */ 1200 1201 mlink->next = mlink_dest->next; 1202 mlink_dest->next = mpage->mlinks; 1203 mpage->mlinks = NULL; 1204 goto nextpage; 1205 } 1206 meta->macroset = MACROSET_NONE; 1207 } 1208 if (meta != NULL && meta->macroset == MACROSET_MDOC) { 1209 mpage->form = FORM_SRC; 1210 mpage->sec = meta->msec; 1211 mpage->sec = mandoc_strdup( 1212 mpage->sec == NULL ? "" : mpage->sec); 1213 mpage->arch = meta->arch; 1214 mpage->arch = mandoc_strdup( 1215 mpage->arch == NULL ? "" : mpage->arch); 1216 mpage->title = mandoc_strdup(meta->title); 1217 } else if (meta != NULL && meta->macroset == MACROSET_MAN) { 1218 if (*meta->msec != '\0' || *meta->title != '\0') { 1219 mpage->form = FORM_SRC; 1220 mpage->sec = mandoc_strdup(meta->msec); 1221 mpage->arch = mandoc_strdup(mlink->arch); 1222 mpage->title = mandoc_strdup(meta->title); 1223 } else 1224 meta = NULL; 1225 } 1226 1227 assert(mpage->desc == NULL); 1228 if (meta == NULL || meta->sodest != NULL) { 1229 mpage->sec = mandoc_strdup(mlink->dsec); 1230 mpage->arch = mandoc_strdup(mlink->arch); 1231 mpage->title = mandoc_strdup(mlink->name); 1232 if (meta == NULL) { 1233 mpage->form = FORM_CAT; 1234 parse_cat(mpage, fd); 1235 } else 1236 mpage->form = FORM_SRC; 1237 } else if (meta->macroset == MACROSET_MDOC) 1238 parse_mdoc(mpage, meta, meta->first); 1239 else 1240 parse_man(mpage, meta, meta->first); 1241 if (mpage->desc == NULL) { 1242 mpage->desc = mandoc_strdup(mlink->name); 1243 if (warnings) 1244 say(mlink->file, "No one-line description, " 1245 "using filename \"%s\"", mlink->name); 1246 } 1247 1248 for (mlink = mpage->mlinks; 1249 mlink != NULL; 1250 mlink = mlink->next) { 1251 putkey(mpage, mlink->name, NAME_FILE); 1252 if (warnings && !use_all) 1253 mlink_check(mpage, mlink); 1254 } 1255 1256 dbadd(dba, mpage); 1257 1258 nextpage: 1259 ohash_delete(&strings); 1260 ohash_delete(&names); 1261 } 1262 } 1263 1264 static void 1265 parse_cat(struct mpage *mpage, int fd) 1266 { 1267 FILE *stream; 1268 struct mlink *mlink; 1269 char *line, *p, *title, *sec; 1270 size_t linesz, plen, titlesz; 1271 ssize_t len; 1272 int offs; 1273 1274 mlink = mpage->mlinks; 1275 stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r"); 1276 if (stream == NULL) { 1277 if (fd != -1) 1278 close(fd); 1279 if (warnings) 1280 say(mlink->file, "&fopen"); 1281 return; 1282 } 1283 1284 line = NULL; 1285 linesz = 0; 1286 1287 /* Parse the section number from the header line. */ 1288 1289 while (getline(&line, &linesz, stream) != -1) { 1290 if (*line == '\n') 1291 continue; 1292 if ((sec = strchr(line, '(')) == NULL) 1293 break; 1294 if ((p = strchr(++sec, ')')) == NULL) 1295 break; 1296 free(mpage->sec); 1297 mpage->sec = mandoc_strndup(sec, p - sec); 1298 if (warnings && *mlink->dsec != '\0' && 1299 strcasecmp(mpage->sec, mlink->dsec)) 1300 say(mlink->file, 1301 "Section \"%s\" manual in %s directory", 1302 mpage->sec, mlink->dsec); 1303 break; 1304 } 1305 1306 /* Skip to first blank line. */ 1307 1308 while (line == NULL || *line != '\n') 1309 if (getline(&line, &linesz, stream) == -1) 1310 break; 1311 1312 /* 1313 * Assume the first line that is not indented 1314 * is the first section header. Skip to it. 1315 */ 1316 1317 while (getline(&line, &linesz, stream) != -1) 1318 if (*line != '\n' && *line != ' ') 1319 break; 1320 1321 /* 1322 * Read up until the next section into a buffer. 1323 * Strip the leading and trailing newline from each read line, 1324 * appending a trailing space. 1325 * Ignore empty (whitespace-only) lines. 1326 */ 1327 1328 titlesz = 0; 1329 title = NULL; 1330 1331 while ((len = getline(&line, &linesz, stream)) != -1) { 1332 if (*line != ' ') 1333 break; 1334 offs = 0; 1335 while (isspace((unsigned char)line[offs])) 1336 offs++; 1337 if (line[offs] == '\0') 1338 continue; 1339 title = mandoc_realloc(title, titlesz + len - offs); 1340 memcpy(title + titlesz, line + offs, len - offs); 1341 titlesz += len - offs; 1342 title[titlesz - 1] = ' '; 1343 } 1344 free(line); 1345 1346 /* 1347 * If no page content can be found, or the input line 1348 * is already the next section header, or there is no 1349 * trailing newline, reuse the page title as the page 1350 * description. 1351 */ 1352 1353 if (NULL == title || '\0' == *title) { 1354 if (warnings) 1355 say(mlink->file, "Cannot find NAME section"); 1356 fclose(stream); 1357 free(title); 1358 return; 1359 } 1360 1361 title[titlesz - 1] = '\0'; 1362 1363 /* 1364 * Skip to the first dash. 1365 * Use the remaining line as the description (no more than 70 1366 * bytes). 1367 */ 1368 1369 if (NULL != (p = strstr(title, "- "))) { 1370 for (p += 2; ' ' == *p || '\b' == *p; p++) 1371 /* Skip to next word. */ ; 1372 } else { 1373 if (warnings) 1374 say(mlink->file, "No dash in title line, " 1375 "reusing \"%s\" as one-line description", title); 1376 p = title; 1377 } 1378 1379 plen = strlen(p); 1380 1381 /* Strip backspace-encoding from line. */ 1382 1383 while (NULL != (line = memchr(p, '\b', plen))) { 1384 len = line - p; 1385 if (0 == len) { 1386 memmove(line, line + 1, plen--); 1387 continue; 1388 } 1389 memmove(line - 1, line + 1, plen - len); 1390 plen -= 2; 1391 } 1392 1393 /* 1394 * Cut off excessive one-line descriptions. 1395 * Bad pages are not worth better heuristics. 1396 */ 1397 1398 mpage->desc = mandoc_strndup(p, 150); 1399 fclose(stream); 1400 free(title); 1401 } 1402 1403 /* 1404 * Put a type/word pair into the word database for this particular file. 1405 */ 1406 static void 1407 putkey(const struct mpage *mpage, char *value, uint64_t type) 1408 { 1409 putkeys(mpage, value, strlen(value), type); 1410 } 1411 1412 /* 1413 * Grok all nodes at or below a certain mdoc node into putkey(). 1414 */ 1415 static void 1416 putmdockey(const struct mpage *mpage, 1417 const struct roff_node *n, uint64_t m, int taboo) 1418 { 1419 1420 for ( ; NULL != n; n = n->next) { 1421 if (n->flags & taboo) 1422 continue; 1423 if (NULL != n->child) 1424 putmdockey(mpage, n->child, m, taboo); 1425 if (n->type == ROFFT_TEXT) 1426 putkey(mpage, n->string, m); 1427 } 1428 } 1429 1430 static void 1431 parse_man(struct mpage *mpage, const struct roff_meta *meta, 1432 const struct roff_node *n) 1433 { 1434 const struct roff_node *head, *body; 1435 char *start, *title; 1436 char byte; 1437 size_t sz; 1438 1439 if (n == NULL) 1440 return; 1441 1442 /* 1443 * We're only searching for one thing: the first text child in 1444 * the BODY of a NAME section. Since we don't keep track of 1445 * sections in -man, run some hoops to find out whether we're in 1446 * the correct section or not. 1447 */ 1448 1449 if (n->type == ROFFT_BODY && n->tok == MAN_SH) { 1450 body = n; 1451 if ((head = body->parent->head) != NULL && 1452 (head = head->child) != NULL && 1453 head->next == NULL && 1454 head->type == ROFFT_TEXT && 1455 strcmp(head->string, "NAME") == 0 && 1456 body->child != NULL) { 1457 1458 /* 1459 * Suck the entire NAME section into memory. 1460 * Yes, we might run away. 1461 * But too many manuals have big, spread-out 1462 * NAME sections over many lines. 1463 */ 1464 1465 title = NULL; 1466 deroff(&title, body); 1467 if (NULL == title) 1468 return; 1469 1470 /* 1471 * Go through a special heuristic dance here. 1472 * Conventionally, one or more manual names are 1473 * comma-specified prior to a whitespace, then a 1474 * dash, then a description. Try to puzzle out 1475 * the name parts here. 1476 */ 1477 1478 start = title; 1479 for ( ;; ) { 1480 sz = strcspn(start, " ,"); 1481 if ('\0' == start[sz]) 1482 break; 1483 1484 byte = start[sz]; 1485 start[sz] = '\0'; 1486 1487 /* 1488 * Assume a stray trailing comma in the 1489 * name list if a name begins with a dash. 1490 */ 1491 1492 if ('-' == start[0] || 1493 ('\\' == start[0] && '-' == start[1])) 1494 break; 1495 1496 putkey(mpage, start, NAME_TITLE); 1497 if ( ! (mpage->name_head_done || 1498 strcasecmp(start, meta->title))) { 1499 putkey(mpage, start, NAME_HEAD); 1500 mpage->name_head_done = 1; 1501 } 1502 1503 if (' ' == byte) { 1504 start += sz + 1; 1505 break; 1506 } 1507 1508 assert(',' == byte); 1509 start += sz + 1; 1510 while (' ' == *start) 1511 start++; 1512 } 1513 1514 if (start == title) { 1515 putkey(mpage, start, NAME_TITLE); 1516 if ( ! (mpage->name_head_done || 1517 strcasecmp(start, meta->title))) { 1518 putkey(mpage, start, NAME_HEAD); 1519 mpage->name_head_done = 1; 1520 } 1521 free(title); 1522 return; 1523 } 1524 1525 while (isspace((unsigned char)*start)) 1526 start++; 1527 1528 if (0 == strncmp(start, "-", 1)) 1529 start += 1; 1530 else if (0 == strncmp(start, "\\-\\-", 4)) 1531 start += 4; 1532 else if (0 == strncmp(start, "\\-", 2)) 1533 start += 2; 1534 else if (0 == strncmp(start, "\\(en", 4)) 1535 start += 4; 1536 else if (0 == strncmp(start, "\\(em", 4)) 1537 start += 4; 1538 1539 while (' ' == *start) 1540 start++; 1541 1542 /* 1543 * Cut off excessive one-line descriptions. 1544 * Bad pages are not worth better heuristics. 1545 */ 1546 1547 mpage->desc = mandoc_strndup(start, 150); 1548 free(title); 1549 return; 1550 } 1551 } 1552 1553 for (n = n->child; n; n = n->next) { 1554 if (NULL != mpage->desc) 1555 break; 1556 parse_man(mpage, meta, n); 1557 } 1558 } 1559 1560 static void 1561 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta, 1562 const struct roff_node *n) 1563 { 1564 const struct mdoc_handler *handler; 1565 1566 for (n = n->child; n != NULL; n = n->next) { 1567 if (n->tok == TOKEN_NONE || n->tok < ROFF_MAX) 1568 continue; 1569 assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX); 1570 handler = mdoc_handlers + (n->tok - MDOC_Dd); 1571 if (n->flags & handler->taboo) 1572 continue; 1573 1574 switch (n->type) { 1575 case ROFFT_ELEM: 1576 case ROFFT_BLOCK: 1577 case ROFFT_HEAD: 1578 case ROFFT_BODY: 1579 case ROFFT_TAIL: 1580 if (handler->fp != NULL && 1581 (*handler->fp)(mpage, meta, n) == 0) 1582 break; 1583 if (handler->mask) 1584 putmdockey(mpage, n->child, 1585 handler->mask, handler->taboo); 1586 break; 1587 default: 1588 continue; 1589 } 1590 if (NULL != n->child) 1591 parse_mdoc(mpage, meta, n); 1592 } 1593 } 1594 1595 static int 1596 parse_mdoc_Fa(struct mpage *mpage, const struct roff_meta *meta, 1597 const struct roff_node *n) 1598 { 1599 uint64_t mask; 1600 1601 mask = TYPE_Fa; 1602 if (n->sec == SEC_SYNOPSIS) 1603 mask |= TYPE_Vt; 1604 1605 putmdockey(mpage, n->child, mask, 0); 1606 return 0; 1607 } 1608 1609 static int 1610 parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta, 1611 const struct roff_node *n) 1612 { 1613 char *start, *end; 1614 size_t sz; 1615 1616 if (SEC_SYNOPSIS != n->sec || 1617 NULL == (n = n->child) || 1618 n->type != ROFFT_TEXT) 1619 return 0; 1620 1621 /* 1622 * Only consider those `Fd' macro fields that begin with an 1623 * "inclusion" token (versus, e.g., #define). 1624 */ 1625 1626 if (strcmp("#include", n->string)) 1627 return 0; 1628 1629 if ((n = n->next) == NULL || n->type != ROFFT_TEXT) 1630 return 0; 1631 1632 /* 1633 * Strip away the enclosing angle brackets and make sure we're 1634 * not zero-length. 1635 */ 1636 1637 start = n->string; 1638 if ('<' == *start || '"' == *start) 1639 start++; 1640 1641 if (0 == (sz = strlen(start))) 1642 return 0; 1643 1644 end = &start[(int)sz - 1]; 1645 if ('>' == *end || '"' == *end) 1646 end--; 1647 1648 if (end > start) 1649 putkeys(mpage, start, end - start + 1, TYPE_In); 1650 return 0; 1651 } 1652 1653 static void 1654 parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n) 1655 { 1656 char *cp; 1657 size_t sz; 1658 1659 if (n->type != ROFFT_TEXT) 1660 return; 1661 1662 /* Skip function pointer punctuation. */ 1663 1664 cp = n->string; 1665 while (*cp == '(' || *cp == '*') 1666 cp++; 1667 sz = strcspn(cp, "()"); 1668 1669 putkeys(mpage, cp, sz, TYPE_Fn); 1670 if (n->sec == SEC_SYNOPSIS) 1671 putkeys(mpage, cp, sz, NAME_SYN); 1672 } 1673 1674 static int 1675 parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta, 1676 const struct roff_node *n) 1677 { 1678 uint64_t mask; 1679 1680 if (n->child == NULL) 1681 return 0; 1682 1683 parse_mdoc_fname(mpage, n->child); 1684 1685 n = n->child->next; 1686 if (n != NULL && n->type == ROFFT_TEXT) { 1687 mask = TYPE_Fa; 1688 if (n->sec == SEC_SYNOPSIS) 1689 mask |= TYPE_Vt; 1690 putmdockey(mpage, n, mask, 0); 1691 } 1692 1693 return 0; 1694 } 1695 1696 static int 1697 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta, 1698 const struct roff_node *n) 1699 { 1700 1701 if (n->type != ROFFT_HEAD) 1702 return 1; 1703 1704 if (n->child != NULL) 1705 parse_mdoc_fname(mpage, n->child); 1706 1707 return 0; 1708 } 1709 1710 static int 1711 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta, 1712 const struct roff_node *n) 1713 { 1714 char *cp; 1715 1716 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY) 1717 return 0; 1718 1719 if (n->child != NULL && 1720 n->child->next == NULL && 1721 n->child->type == ROFFT_TEXT) 1722 return 1; 1723 1724 cp = NULL; 1725 deroff(&cp, n); 1726 if (cp != NULL) { 1727 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va || 1728 n->type == ROFFT_BODY ? TYPE_Va : 0)); 1729 free(cp); 1730 } 1731 1732 return 0; 1733 } 1734 1735 static int 1736 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta, 1737 const struct roff_node *n) 1738 { 1739 char *cp; 1740 1741 if (NULL == (n = n->child)) 1742 return 0; 1743 1744 if (NULL == n->next) { 1745 putkey(mpage, n->string, TYPE_Xr); 1746 return 0; 1747 } 1748 1749 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string); 1750 putkey(mpage, cp, TYPE_Xr); 1751 free(cp); 1752 return 0; 1753 } 1754 1755 static int 1756 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta, 1757 const struct roff_node *n) 1758 { 1759 1760 if (n->type == ROFFT_BODY) 1761 deroff(&mpage->desc, n); 1762 return 0; 1763 } 1764 1765 static int 1766 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta, 1767 const struct roff_node *n) 1768 { 1769 1770 if (SEC_NAME == n->sec) 1771 putmdockey(mpage, n->child, NAME_TITLE, 0); 1772 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) { 1773 if (n->child == NULL) 1774 putkey(mpage, meta->name, NAME_SYN); 1775 else 1776 putmdockey(mpage, n->child, NAME_SYN, 0); 1777 } 1778 if ( ! (mpage->name_head_done || 1779 n->child == NULL || n->child->string == NULL || 1780 strcasecmp(n->child->string, meta->title))) { 1781 putkey(mpage, n->child->string, NAME_HEAD); 1782 mpage->name_head_done = 1; 1783 } 1784 return 0; 1785 } 1786 1787 static int 1788 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta, 1789 const struct roff_node *n) 1790 { 1791 1792 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD; 1793 } 1794 1795 static int 1796 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta, 1797 const struct roff_node *n) 1798 { 1799 1800 return n->type == ROFFT_HEAD; 1801 } 1802 1803 /* 1804 * Add a string to the hash table for the current manual. 1805 * Each string has a bitmask telling which macros it belongs to. 1806 * When we finish the manual, we'll dump the table. 1807 */ 1808 static void 1809 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v) 1810 { 1811 struct ohash *htab; 1812 struct str *s; 1813 const char *end; 1814 unsigned int slot; 1815 int i, mustfree; 1816 1817 if (0 == sz) 1818 return; 1819 1820 mustfree = render_string(&cp, &sz); 1821 1822 if (TYPE_Nm & v) { 1823 htab = &names; 1824 v &= name_mask; 1825 if (v & NAME_FIRST) 1826 name_mask &= ~NAME_FIRST; 1827 if (debug > 1) 1828 say(mpage->mlinks->file, 1829 "Adding name %*s, bits=0x%llx", (int)sz, cp, 1830 (unsigned long long)v); 1831 } else { 1832 htab = &strings; 1833 if (debug > 1) 1834 for (i = 0; i < KEY_MAX; i++) 1835 if ((uint64_t)1 << i & v) 1836 say(mpage->mlinks->file, 1837 "Adding key %s=%*s", 1838 mansearch_keynames[i], (int)sz, cp); 1839 } 1840 1841 end = cp + sz; 1842 slot = ohash_qlookupi(htab, cp, &end); 1843 s = ohash_find(htab, slot); 1844 1845 if (NULL != s && mpage == s->mpage) { 1846 s->mask |= v; 1847 return; 1848 } else if (NULL == s) { 1849 s = mandoc_calloc(1, sizeof(struct str) + sz + 1); 1850 memcpy(s->key, cp, sz); 1851 ohash_insert(htab, slot, s); 1852 } 1853 s->mpage = mpage; 1854 s->mask = v; 1855 1856 if (mustfree) 1857 free(cp); 1858 } 1859 1860 /* 1861 * Take a Unicode codepoint and produce its UTF-8 encoding. 1862 * This isn't the best way to do this, but it works. 1863 * The magic numbers are from the UTF-8 packaging. 1864 * They're not as scary as they seem: read the UTF-8 spec for details. 1865 */ 1866 static size_t 1867 utf8(unsigned int cp, char out[7]) 1868 { 1869 size_t rc; 1870 1871 rc = 0; 1872 if (cp <= 0x0000007F) { 1873 rc = 1; 1874 out[0] = (char)cp; 1875 } else if (cp <= 0x000007FF) { 1876 rc = 2; 1877 out[0] = (cp >> 6 & 31) | 192; 1878 out[1] = (cp & 63) | 128; 1879 } else if (cp <= 0x0000FFFF) { 1880 rc = 3; 1881 out[0] = (cp >> 12 & 15) | 224; 1882 out[1] = (cp >> 6 & 63) | 128; 1883 out[2] = (cp & 63) | 128; 1884 } else if (cp <= 0x001FFFFF) { 1885 rc = 4; 1886 out[0] = (cp >> 18 & 7) | 240; 1887 out[1] = (cp >> 12 & 63) | 128; 1888 out[2] = (cp >> 6 & 63) | 128; 1889 out[3] = (cp & 63) | 128; 1890 } else if (cp <= 0x03FFFFFF) { 1891 rc = 5; 1892 out[0] = (cp >> 24 & 3) | 248; 1893 out[1] = (cp >> 18 & 63) | 128; 1894 out[2] = (cp >> 12 & 63) | 128; 1895 out[3] = (cp >> 6 & 63) | 128; 1896 out[4] = (cp & 63) | 128; 1897 } else if (cp <= 0x7FFFFFFF) { 1898 rc = 6; 1899 out[0] = (cp >> 30 & 1) | 252; 1900 out[1] = (cp >> 24 & 63) | 128; 1901 out[2] = (cp >> 18 & 63) | 128; 1902 out[3] = (cp >> 12 & 63) | 128; 1903 out[4] = (cp >> 6 & 63) | 128; 1904 out[5] = (cp & 63) | 128; 1905 } else 1906 return 0; 1907 1908 out[rc] = '\0'; 1909 return rc; 1910 } 1911 1912 /* 1913 * If the string contains escape sequences, 1914 * replace it with an allocated rendering and return 1, 1915 * such that the caller can free it after use. 1916 * Otherwise, do nothing and return 0. 1917 */ 1918 static int 1919 render_string(char **public, size_t *psz) 1920 { 1921 const char *src, *scp, *addcp, *seq; 1922 char *dst; 1923 size_t ssz, dsz, addsz; 1924 char utfbuf[7], res[6]; 1925 int seqlen, unicode; 1926 1927 res[0] = '\\'; 1928 res[1] = '\t'; 1929 res[2] = ASCII_NBRSP; 1930 res[3] = ASCII_HYPH; 1931 res[4] = ASCII_BREAK; 1932 res[5] = '\0'; 1933 1934 src = scp = *public; 1935 ssz = *psz; 1936 dst = NULL; 1937 dsz = 0; 1938 1939 while (scp < src + *psz) { 1940 1941 /* Leave normal characters unchanged. */ 1942 1943 if (strchr(res, *scp) == NULL) { 1944 if (dst != NULL) 1945 dst[dsz++] = *scp; 1946 scp++; 1947 continue; 1948 } 1949 1950 /* 1951 * Found something that requires replacing, 1952 * make sure we have a destination buffer. 1953 */ 1954 1955 if (dst == NULL) { 1956 dst = mandoc_malloc(ssz + 1); 1957 dsz = scp - src; 1958 memcpy(dst, src, dsz); 1959 } 1960 1961 /* Handle single-char special characters. */ 1962 1963 switch (*scp) { 1964 case '\\': 1965 break; 1966 case '\t': 1967 case ASCII_NBRSP: 1968 dst[dsz++] = ' '; 1969 scp++; 1970 continue; 1971 case ASCII_HYPH: 1972 dst[dsz++] = '-'; 1973 /* FALLTHROUGH */ 1974 case ASCII_BREAK: 1975 scp++; 1976 continue; 1977 default: 1978 abort(); 1979 } 1980 1981 /* 1982 * Found an escape sequence. 1983 * Read past the slash, then parse it. 1984 * Ignore everything except characters. 1985 */ 1986 1987 scp++; 1988 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL) 1989 continue; 1990 1991 /* 1992 * Render the special character 1993 * as either UTF-8 or ASCII. 1994 */ 1995 1996 if (write_utf8) { 1997 unicode = mchars_spec2cp(seq, seqlen); 1998 if (unicode <= 0) 1999 continue; 2000 addsz = utf8(unicode, utfbuf); 2001 if (addsz == 0) 2002 continue; 2003 addcp = utfbuf; 2004 } else { 2005 addcp = mchars_spec2str(seq, seqlen, &addsz); 2006 if (addcp == NULL) 2007 continue; 2008 if (*addcp == ASCII_NBRSP) { 2009 addcp = " "; 2010 addsz = 1; 2011 } 2012 } 2013 2014 /* Copy the rendered glyph into the stream. */ 2015 2016 ssz += addsz; 2017 dst = mandoc_realloc(dst, ssz + 1); 2018 memcpy(dst + dsz, addcp, addsz); 2019 dsz += addsz; 2020 } 2021 if (dst != NULL) { 2022 *public = dst; 2023 *psz = dsz; 2024 } 2025 2026 /* Trim trailing whitespace and NUL-terminate. */ 2027 2028 while (*psz > 0 && (*public)[*psz - 1] == ' ') 2029 --*psz; 2030 if (dst != NULL) { 2031 (*public)[*psz] = '\0'; 2032 return 1; 2033 } else 2034 return 0; 2035 } 2036 2037 static void 2038 dbadd_mlink(const struct mlink *mlink) 2039 { 2040 dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE); 2041 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec); 2042 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec); 2043 dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch); 2044 dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file); 2045 } 2046 2047 /* 2048 * Flush the current page's terms (and their bits) into the database. 2049 * Also, handle escape sequences at the last possible moment. 2050 */ 2051 static void 2052 dbadd(struct dba *dba, struct mpage *mpage) 2053 { 2054 struct mlink *mlink; 2055 struct str *key; 2056 char *cp; 2057 uint64_t mask; 2058 size_t i; 2059 unsigned int slot; 2060 int mustfree; 2061 2062 mlink = mpage->mlinks; 2063 2064 if (nodb) { 2065 for (key = ohash_first(&names, &slot); NULL != key; 2066 key = ohash_next(&names, &slot)) 2067 free(key); 2068 for (key = ohash_first(&strings, &slot); NULL != key; 2069 key = ohash_next(&strings, &slot)) 2070 free(key); 2071 if (0 == debug) 2072 return; 2073 while (NULL != mlink) { 2074 fputs(mlink->name, stdout); 2075 if (NULL == mlink->next || 2076 strcmp(mlink->dsec, mlink->next->dsec) || 2077 strcmp(mlink->fsec, mlink->next->fsec) || 2078 strcmp(mlink->arch, mlink->next->arch)) { 2079 putchar('('); 2080 if ('\0' == *mlink->dsec) 2081 fputs(mlink->fsec, stdout); 2082 else 2083 fputs(mlink->dsec, stdout); 2084 if ('\0' != *mlink->arch) 2085 printf("/%s", mlink->arch); 2086 putchar(')'); 2087 } 2088 mlink = mlink->next; 2089 if (NULL != mlink) 2090 fputs(", ", stdout); 2091 } 2092 printf(" - %s\n", mpage->desc); 2093 return; 2094 } 2095 2096 if (debug) 2097 say(mlink->file, "Adding to database"); 2098 2099 cp = mpage->desc; 2100 i = strlen(cp); 2101 mustfree = render_string(&cp, &i); 2102 mpage->dba = dba_page_new(dba->pages, 2103 *mpage->arch == '\0' ? mlink->arch : mpage->arch, 2104 cp, mlink->file, mpage->form); 2105 if (mustfree) 2106 free(cp); 2107 dba_page_add(mpage->dba, DBP_SECT, mpage->sec); 2108 2109 while (mlink != NULL) { 2110 dbadd_mlink(mlink); 2111 mlink = mlink->next; 2112 } 2113 2114 for (key = ohash_first(&names, &slot); NULL != key; 2115 key = ohash_next(&names, &slot)) { 2116 assert(key->mpage == mpage); 2117 dba_page_alias(mpage->dba, key->key, key->mask); 2118 free(key); 2119 } 2120 for (key = ohash_first(&strings, &slot); NULL != key; 2121 key = ohash_next(&strings, &slot)) { 2122 assert(key->mpage == mpage); 2123 i = 0; 2124 for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) { 2125 if (key->mask & mask) 2126 dba_macro_add(dba->macros, i, 2127 key->key, mpage->dba); 2128 i++; 2129 } 2130 free(key); 2131 } 2132 } 2133 2134 static void 2135 dbprune(struct dba *dba) 2136 { 2137 struct dba_array *page, *files; 2138 char *file; 2139 2140 dba_array_FOREACH(dba->pages, page) { 2141 files = dba_array_get(page, DBP_FILE); 2142 dba_array_FOREACH(files, file) { 2143 if (*file < ' ') 2144 file++; 2145 if (ohash_find(&mlinks, ohash_qlookup(&mlinks, 2146 file)) != NULL) { 2147 if (debug) 2148 say(file, "Deleting from database"); 2149 dba_array_del(dba->pages); 2150 break; 2151 } 2152 } 2153 } 2154 } 2155 2156 /* 2157 * Write the database from memory to disk. 2158 */ 2159 static void 2160 dbwrite(struct dba *dba) 2161 { 2162 struct stat sb1, sb2; 2163 char tfn[33], *cp1, *cp2; 2164 off_t i; 2165 int fd1, fd2; 2166 2167 /* 2168 * Do not write empty databases, and delete existing ones 2169 * when makewhatis -u causes them to become empty. 2170 */ 2171 2172 dba_array_start(dba->pages); 2173 if (dba_array_next(dba->pages) == NULL) { 2174 if (unlink(MANDOC_DB) == -1 && errno != ENOENT) 2175 say(MANDOC_DB, "&unlink"); 2176 return; 2177 } 2178 2179 /* 2180 * Build the database in a temporary file, 2181 * then atomically move it into place. 2182 */ 2183 2184 if (dba_write(MANDOC_DB "~", dba) != -1) { 2185 if (rename(MANDOC_DB "~", MANDOC_DB) == -1) { 2186 exitcode = (int)MANDOCLEVEL_SYSERR; 2187 say(MANDOC_DB, "&rename"); 2188 unlink(MANDOC_DB "~"); 2189 } 2190 return; 2191 } 2192 2193 /* 2194 * We lack write permission and cannot replace the database 2195 * file, but let's at least check whether the data changed. 2196 */ 2197 2198 (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn)); 2199 if (mkdtemp(tfn) == NULL) { 2200 exitcode = (int)MANDOCLEVEL_SYSERR; 2201 say("", "&%s", tfn); 2202 return; 2203 } 2204 cp1 = cp2 = MAP_FAILED; 2205 fd1 = fd2 = -1; 2206 (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn)); 2207 if (dba_write(tfn, dba) == -1) { 2208 say(tfn, "&dba_write"); 2209 goto err; 2210 } 2211 if ((fd1 = open(MANDOC_DB, O_RDONLY, 0)) == -1) { 2212 say(MANDOC_DB, "&open"); 2213 goto err; 2214 } 2215 if ((fd2 = open(tfn, O_RDONLY, 0)) == -1) { 2216 say(tfn, "&open"); 2217 goto err; 2218 } 2219 if (fstat(fd1, &sb1) == -1) { 2220 say(MANDOC_DB, "&fstat"); 2221 goto err; 2222 } 2223 if (fstat(fd2, &sb2) == -1) { 2224 say(tfn, "&fstat"); 2225 goto err; 2226 } 2227 if (sb1.st_size != sb2.st_size) 2228 goto err; 2229 if ((cp1 = mmap(NULL, sb1.st_size, PROT_READ, MAP_PRIVATE, 2230 fd1, 0)) == MAP_FAILED) { 2231 say(MANDOC_DB, "&mmap"); 2232 goto err; 2233 } 2234 if ((cp2 = mmap(NULL, sb2.st_size, PROT_READ, MAP_PRIVATE, 2235 fd2, 0)) == MAP_FAILED) { 2236 say(tfn, "&mmap"); 2237 goto err; 2238 } 2239 for (i = 0; i < sb1.st_size; i++) 2240 if (cp1[i] != cp2[i]) 2241 goto err; 2242 goto out; 2243 2244 err: 2245 exitcode = (int)MANDOCLEVEL_SYSERR; 2246 say(MANDOC_DB, "Data changed, but cannot replace database"); 2247 2248 out: 2249 if (cp1 != MAP_FAILED) 2250 munmap(cp1, sb1.st_size); 2251 if (cp2 != MAP_FAILED) 2252 munmap(cp2, sb2.st_size); 2253 if (fd1 != -1) 2254 close(fd1); 2255 if (fd2 != -1) 2256 close(fd2); 2257 unlink(tfn); 2258 *strrchr(tfn, '/') = '\0'; 2259 rmdir(tfn); 2260 } 2261 2262 static int 2263 set_basedir(const char *targetdir, int report_baddir) 2264 { 2265 static char startdir[PATH_MAX]; 2266 static int getcwd_status; /* 1 = ok, 2 = failure */ 2267 static int chdir_status; /* 1 = changed directory */ 2268 2269 /* 2270 * Remember the original working directory, if possible. 2271 * This will be needed if the second or a later directory 2272 * on the command line is given as a relative path. 2273 * Do not error out if the current directory is not 2274 * searchable: Maybe it won't be needed after all. 2275 */ 2276 if (getcwd_status == 0) { 2277 if (getcwd(startdir, sizeof(startdir)) == NULL) { 2278 getcwd_status = 2; 2279 (void)strlcpy(startdir, strerror(errno), 2280 sizeof(startdir)); 2281 } else 2282 getcwd_status = 1; 2283 } 2284 2285 /* 2286 * We are leaving the old base directory. 2287 * Do not use it any longer, not even for messages. 2288 */ 2289 *basedir = '\0'; 2290 basedir_len = 0; 2291 2292 /* 2293 * If and only if the directory was changed earlier and 2294 * the next directory to process is given as a relative path, 2295 * first go back, or bail out if that is impossible. 2296 */ 2297 if (chdir_status && *targetdir != '/') { 2298 if (getcwd_status == 2) { 2299 exitcode = (int)MANDOCLEVEL_SYSERR; 2300 say("", "getcwd: %s", startdir); 2301 return 0; 2302 } 2303 if (chdir(startdir) == -1) { 2304 exitcode = (int)MANDOCLEVEL_SYSERR; 2305 say("", "&chdir %s", startdir); 2306 return 0; 2307 } 2308 } 2309 2310 /* 2311 * Always resolve basedir to the canonicalized absolute 2312 * pathname and append a trailing slash, such that 2313 * we can reliably check whether files are inside. 2314 */ 2315 if (realpath(targetdir, basedir) == NULL) { 2316 if (report_baddir || errno != ENOENT) { 2317 exitcode = (int)MANDOCLEVEL_BADARG; 2318 say("", "&%s: realpath", targetdir); 2319 } 2320 *basedir = '\0'; 2321 return 0; 2322 } else if (chdir(basedir) == -1) { 2323 if (report_baddir || errno != ENOENT) { 2324 exitcode = (int)MANDOCLEVEL_BADARG; 2325 say("", "&chdir"); 2326 } 2327 *basedir = '\0'; 2328 return 0; 2329 } 2330 chdir_status = 1; 2331 basedir_len = strlen(basedir); 2332 if (basedir[basedir_len - 1] != '/') { 2333 if (basedir_len >= PATH_MAX - 1) { 2334 exitcode = (int)MANDOCLEVEL_SYSERR; 2335 say("", "Filename too long"); 2336 *basedir = '\0'; 2337 basedir_len = 0; 2338 return 0; 2339 } 2340 basedir[basedir_len++] = '/'; 2341 basedir[basedir_len] = '\0'; 2342 } 2343 return 1; 2344 } 2345 2346 static void 2347 say(const char *file, const char *format, ...) 2348 { 2349 va_list ap; 2350 int use_errno; 2351 2352 if (*basedir != '\0') 2353 fprintf(stderr, "%s", basedir); 2354 if (*basedir != '\0' && *file != '\0') 2355 fputc('/', stderr); 2356 if (*file != '\0') 2357 fprintf(stderr, "%s", file); 2358 2359 use_errno = 1; 2360 if (format != NULL) { 2361 switch (*format) { 2362 case '&': 2363 format++; 2364 break; 2365 case '\0': 2366 format = NULL; 2367 break; 2368 default: 2369 use_errno = 0; 2370 break; 2371 } 2372 } 2373 if (format != NULL) { 2374 if (*basedir != '\0' || *file != '\0') 2375 fputs(": ", stderr); 2376 va_start(ap, format); 2377 vfprintf(stderr, format, ap); 2378 va_end(ap); 2379 } 2380 if (use_errno) { 2381 if (*basedir != '\0' || *file != '\0' || format != NULL) 2382 fputs(": ", stderr); 2383 perror(NULL); 2384 } else 2385 fputc('\n', stderr); 2386 } 2387