1 /* $Id: mandocdb.c,v 1.274 2024/05/14 21:19:12 schwarze Exp $ */
2 /*
3 * Copyright (c) 2011-2021, 2024 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 "config.h"
22
23 #include <sys/types.h>
24 #include <sys/mman.h>
25 #include <sys/stat.h>
26
27 #include <assert.h>
28 #include <ctype.h>
29 #if HAVE_ERR
30 #include <err.h>
31 #endif
32 #include <errno.h>
33 #include <fcntl.h>
34 #if HAVE_FTS
35 #include <fts.h>
36 #else
37 #include "compat_fts.h"
38 #endif
39 #include <limits.h>
40 #if HAVE_SANDBOX_INIT
41 #include <sandbox.h>
42 #endif
43 #include <stdarg.h>
44 #include <stddef.h>
45 #include <stdio.h>
46 #include <stdint.h>
47 #include <stdlib.h>
48 #include <string.h>
49 #include <unistd.h>
50
51 #include "mandoc_aux.h"
52 #include "mandoc_ohash.h"
53 #include "mandoc.h"
54 #include "roff.h"
55 #include "mdoc.h"
56 #include "man.h"
57 #include "mandoc_parse.h"
58 #include "manconf.h"
59 #include "mansearch.h"
60 #include "dba_array.h"
61 #include "dba.h"
62
63 extern const char *const mansearch_keynames[];
64
65 enum op {
66 OP_DEFAULT = 0, /* new dbs from dir list or default config */
67 OP_CONFFILE, /* new databases from custom config file */
68 OP_UPDATE, /* delete/add entries in existing database */
69 OP_DELETE, /* delete entries from existing database */
70 OP_TEST /* change no databases, report potential problems */
71 };
72
73 struct str {
74 const struct mpage *mpage; /* if set, the owning parse */
75 uint64_t mask; /* bitmask in sequence */
76 char key[]; /* rendered text */
77 };
78
79 struct inodev {
80 ino_t st_ino;
81 dev_t st_dev;
82 };
83
84 struct mpage {
85 struct inodev inodev; /* used for hashing routine */
86 struct dba_array *dba;
87 char *sec; /* section from file content */
88 char *arch; /* architecture from file content */
89 char *title; /* title from file content */
90 char *desc; /* description from file content */
91 struct mpage *next; /* singly linked list */
92 struct mlink *mlinks; /* singly linked list */
93 int name_head_done;
94 enum form form; /* format from file content */
95 };
96
97 struct mlink {
98 char file[PATH_MAX]; /* filename rel. to manpath */
99 char *dsec; /* section from directory */
100 char *arch; /* architecture from directory */
101 char *name; /* name from file name (not empty) */
102 char *fsec; /* section from file name suffix */
103 struct mlink *next; /* singly linked list */
104 struct mpage *mpage; /* parent */
105 int gzip; /* filename has a .gz suffix */
106 enum form dform; /* format from directory */
107 enum form fform; /* format from file name suffix */
108 };
109
110 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
111 const struct roff_node *);
112
113 struct mdoc_handler {
114 mdoc_fp fp; /* optional handler */
115 uint64_t mask; /* set unless handler returns 0 */
116 int taboo; /* node flags that must not be set */
117 };
118
119
120 int mandocdb(int, char *[]);
121
122 static void dbadd(struct dba *, struct mpage *);
123 static void dbadd_mlink(const struct mlink *);
124 static void dbprune(struct dba *);
125 static void dbwrite(struct dba *);
126 static void filescan(const char *);
127 #if HAVE_FTS_COMPARE_CONST
128 static int fts_compare(const FTSENT *const *, const FTSENT *const *);
129 #else
130 static int fts_compare(const FTSENT **, const FTSENT **);
131 #endif
132 static void mlink_add(struct mlink *, const struct stat *);
133 static void mlink_check(struct mpage *, struct mlink *);
134 static void mlink_free(struct mlink *);
135 static void mlinks_undupe(struct mpage *);
136 static void mpages_free(void);
137 static void mpages_merge(struct dba *, struct mparse *);
138 static void parse_cat(struct mpage *, int);
139 static void parse_man(struct mpage *, const struct roff_meta *,
140 const struct roff_node *);
141 static void parse_mdoc(struct mpage *, const struct roff_meta *,
142 const struct roff_node *);
143 static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
144 const struct roff_node *);
145 static int parse_mdoc_Fa(struct mpage *, const struct roff_meta *,
146 const struct roff_node *);
147 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
148 const struct roff_node *);
149 static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
150 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
151 const struct roff_node *);
152 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
153 const struct roff_node *);
154 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
155 const struct roff_node *);
156 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
157 const struct roff_node *);
158 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
159 const struct roff_node *);
160 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
161 const struct roff_node *);
162 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
163 const struct roff_node *);
164 static void putkey(const struct mpage *, char *, uint64_t);
165 static void putkeys(const struct mpage *, char *, size_t, uint64_t);
166 static void putmdockey(const struct mpage *,
167 const struct roff_node *, uint64_t, int);
168 #ifdef READ_ALLOWED_PATH
169 static int read_allowed(const char *);
170 #endif
171 static int render_string(char **, size_t *);
172 static void say(const char *, const char *, ...)
173 __attribute__((__format__ (__printf__, 2, 3)));
174 static int set_basedir(const char *, int);
175 static int treescan(void);
176 static size_t utf8(unsigned int, char[5]);
177
178 static int nodb; /* no database changes */
179 static int mparse_options; /* abort the parse early */
180 static int use_all; /* use all found files */
181 static int debug; /* print what we're doing */
182 static int warnings; /* warn about crap */
183 static int write_utf8; /* write UTF-8 output; else ASCII */
184 static int exitcode; /* to be returned by main */
185 static enum op op; /* operational mode */
186 static char basedir[PATH_MAX]; /* current base directory */
187 static size_t basedir_len; /* strlen(basedir) */
188 static struct mpage *mpage_head; /* list of distinct manual pages */
189 static struct ohash mpages; /* table of distinct manual pages */
190 static struct ohash mlinks; /* table of directory entries */
191 static struct ohash names; /* table of all names */
192 static struct ohash strings; /* table of all strings */
193 static uint64_t name_mask;
194
195 static const struct mdoc_handler mdoc_handlers[MDOC_MAX - MDOC_Dd] = {
196 { NULL, 0, NODE_NOPRT }, /* Dd */
197 { NULL, 0, NODE_NOPRT }, /* Dt */
198 { NULL, 0, NODE_NOPRT }, /* Os */
199 { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
200 { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
201 { NULL, 0, 0 }, /* Pp */
202 { NULL, 0, 0 }, /* D1 */
203 { NULL, 0, 0 }, /* Dl */
204 { NULL, 0, 0 }, /* Bd */
205 { NULL, 0, 0 }, /* Ed */
206 { NULL, 0, 0 }, /* Bl */
207 { NULL, 0, 0 }, /* El */
208 { NULL, 0, 0 }, /* It */
209 { NULL, 0, 0 }, /* Ad */
210 { NULL, TYPE_An, 0 }, /* An */
211 { NULL, 0, 0 }, /* Ap */
212 { NULL, TYPE_Ar, 0 }, /* Ar */
213 { NULL, TYPE_Cd, 0 }, /* Cd */
214 { NULL, TYPE_Cm, 0 }, /* Cm */
215 { NULL, TYPE_Dv, 0 }, /* Dv */
216 { NULL, TYPE_Er, 0 }, /* Er */
217 { NULL, TYPE_Ev, 0 }, /* Ev */
218 { NULL, 0, 0 }, /* Ex */
219 { parse_mdoc_Fa, 0, 0 }, /* Fa */
220 { parse_mdoc_Fd, 0, 0 }, /* Fd */
221 { NULL, TYPE_Fl, 0 }, /* Fl */
222 { parse_mdoc_Fn, 0, 0 }, /* Fn */
223 { NULL, TYPE_Ft | TYPE_Vt, 0 }, /* Ft */
224 { NULL, TYPE_Ic, 0 }, /* Ic */
225 { NULL, TYPE_In, 0 }, /* In */
226 { NULL, TYPE_Li, 0 }, /* Li */
227 { parse_mdoc_Nd, 0, 0 }, /* Nd */
228 { parse_mdoc_Nm, 0, 0 }, /* Nm */
229 { NULL, 0, 0 }, /* Op */
230 { NULL, 0, 0 }, /* Ot */
231 { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
232 { NULL, 0, 0 }, /* Rv */
233 { NULL, TYPE_St, 0 }, /* St */
234 { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
235 { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
236 { parse_mdoc_Xr, 0, 0 }, /* Xr */
237 { NULL, 0, 0 }, /* %A */
238 { NULL, 0, 0 }, /* %B */
239 { NULL, 0, 0 }, /* %D */
240 { NULL, 0, 0 }, /* %I */
241 { NULL, 0, 0 }, /* %J */
242 { NULL, 0, 0 }, /* %N */
243 { NULL, 0, 0 }, /* %O */
244 { NULL, 0, 0 }, /* %P */
245 { NULL, 0, 0 }, /* %R */
246 { NULL, 0, 0 }, /* %T */
247 { NULL, 0, 0 }, /* %V */
248 { NULL, 0, 0 }, /* Ac */
249 { NULL, 0, 0 }, /* Ao */
250 { NULL, 0, 0 }, /* Aq */
251 { NULL, TYPE_At, 0 }, /* At */
252 { NULL, 0, 0 }, /* Bc */
253 { NULL, 0, 0 }, /* Bf */
254 { NULL, 0, 0 }, /* Bo */
255 { NULL, 0, 0 }, /* Bq */
256 { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
257 { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
258 { NULL, 0, 0 }, /* Db */
259 { NULL, 0, 0 }, /* Dc */
260 { NULL, 0, 0 }, /* Do */
261 { NULL, 0, 0 }, /* Dq */
262 { NULL, 0, 0 }, /* Ec */
263 { NULL, 0, 0 }, /* Ef */
264 { NULL, TYPE_Em, 0 }, /* Em */
265 { NULL, 0, 0 }, /* Eo */
266 { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
267 { NULL, TYPE_Ms, 0 }, /* Ms */
268 { NULL, 0, 0 }, /* No */
269 { NULL, 0, 0 }, /* Ns */
270 { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
271 { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
272 { NULL, 0, 0 }, /* Pc */
273 { NULL, 0, 0 }, /* Pf */
274 { NULL, 0, 0 }, /* Po */
275 { NULL, 0, 0 }, /* Pq */
276 { NULL, 0, 0 }, /* Qc */
277 { NULL, 0, 0 }, /* Ql */
278 { NULL, 0, 0 }, /* Qo */
279 { NULL, 0, 0 }, /* Qq */
280 { NULL, 0, 0 }, /* Re */
281 { NULL, 0, 0 }, /* Rs */
282 { NULL, 0, 0 }, /* Sc */
283 { NULL, 0, 0 }, /* So */
284 { NULL, 0, 0 }, /* Sq */
285 { NULL, 0, 0 }, /* Sm */
286 { NULL, 0, 0 }, /* Sx */
287 { NULL, TYPE_Sy, 0 }, /* Sy */
288 { NULL, TYPE_Tn, 0 }, /* Tn */
289 { NULL, 0, NODE_NOSRC }, /* Ux */
290 { NULL, 0, 0 }, /* Xc */
291 { NULL, 0, 0 }, /* Xo */
292 { parse_mdoc_Fo, 0, 0 }, /* Fo */
293 { NULL, 0, 0 }, /* Fc */
294 { NULL, 0, 0 }, /* Oo */
295 { NULL, 0, 0 }, /* Oc */
296 { NULL, 0, 0 }, /* Bk */
297 { NULL, 0, 0 }, /* Ek */
298 { NULL, 0, 0 }, /* Bt */
299 { NULL, 0, 0 }, /* Hf */
300 { NULL, 0, 0 }, /* Fr */
301 { NULL, 0, 0 }, /* Ud */
302 { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
303 { NULL, 0, 0 }, /* Lp */
304 { NULL, TYPE_Lk, 0 }, /* Lk */
305 { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
306 { NULL, 0, 0 }, /* Brq */
307 { NULL, 0, 0 }, /* Bro */
308 { NULL, 0, 0 }, /* Brc */
309 { NULL, 0, 0 }, /* %C */
310 { NULL, 0, 0 }, /* Es */
311 { NULL, 0, 0 }, /* En */
312 { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
313 { NULL, 0, 0 }, /* %Q */
314 { NULL, 0, 0 }, /* %U */
315 { NULL, 0, 0 }, /* Ta */
316 };
317
318
319 int
mandocdb(int argc,char * argv[])320 mandocdb(int argc, char *argv[])
321 {
322 struct manconf conf;
323 struct mparse *mp;
324 struct dba *dba;
325 const char *path_arg, *progname;
326 size_t j, sz;
327 int ch, i;
328
329 #if HAVE_PLEDGE
330 if (pledge("stdio rpath wpath cpath", NULL) == -1) {
331 warn("pledge");
332 return (int)MANDOCLEVEL_SYSERR;
333 }
334 #endif
335
336 #if HAVE_SANDBOX_INIT
337 if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
338 warnx("sandbox_init");
339 return (int)MANDOCLEVEL_SYSERR;
340 }
341 #endif
342
343 memset(&conf, 0, sizeof(conf));
344
345 /*
346 * We accept a few different invocations.
347 * The CHECKOP macro makes sure that invocation styles don't
348 * clobber each other.
349 */
350 #define CHECKOP(_op, _ch) do \
351 if ((_op) != OP_DEFAULT) { \
352 warnx("-%c: Conflicting option", (_ch)); \
353 goto usage; \
354 } while (/*CONSTCOND*/0)
355
356 mparse_options = MPARSE_UTF8 | MPARSE_LATIN1 | MPARSE_VALIDATE;
357 path_arg = NULL;
358 op = OP_DEFAULT;
359
360 while ((ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")) != -1)
361 switch (ch) {
362 case 'a':
363 use_all = 1;
364 break;
365 case 'C':
366 CHECKOP(op, ch);
367 path_arg = optarg;
368 op = OP_CONFFILE;
369 break;
370 case 'D':
371 debug++;
372 break;
373 case 'd':
374 CHECKOP(op, ch);
375 path_arg = optarg;
376 op = OP_UPDATE;
377 break;
378 case 'n':
379 nodb = 1;
380 break;
381 case 'p':
382 warnings = 1;
383 break;
384 case 'Q':
385 mparse_options |= MPARSE_QUICK;
386 break;
387 case 'T':
388 if (strcmp(optarg, "utf8") != 0) {
389 warnx("-T%s: Unsupported output format",
390 optarg);
391 goto usage;
392 }
393 write_utf8 = 1;
394 break;
395 case 't':
396 CHECKOP(op, ch);
397 dup2(STDOUT_FILENO, STDERR_FILENO);
398 op = OP_TEST;
399 nodb = warnings = 1;
400 break;
401 case 'u':
402 CHECKOP(op, ch);
403 path_arg = optarg;
404 op = OP_DELETE;
405 break;
406 case 'v':
407 /* Compatibility with espie@'s makewhatis. */
408 break;
409 default:
410 goto usage;
411 }
412
413 argc -= optind;
414 argv += optind;
415
416 #if HAVE_PLEDGE
417 if (nodb) {
418 if (pledge("stdio rpath", NULL) == -1) {
419 warn("pledge");
420 return (int)MANDOCLEVEL_SYSERR;
421 }
422 }
423 #endif
424
425 if (op == OP_CONFFILE && argc > 0) {
426 warnx("-C: Too many arguments");
427 goto usage;
428 }
429
430 exitcode = (int)MANDOCLEVEL_OK;
431 mchars_alloc();
432 mp = mparse_alloc(mparse_options, MANDOC_OS_OTHER, NULL);
433 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
434 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
435
436 if (op == OP_UPDATE || op == OP_DELETE || op == OP_TEST) {
437
438 /*
439 * Most of these deal with a specific directory.
440 * Jump into that directory first.
441 */
442 if (op != OP_TEST && set_basedir(path_arg, 1) == 0)
443 goto out;
444
445 dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
446 if (dba != NULL) {
447 /*
448 * The existing database is usable. Process
449 * all files specified on the command-line.
450 */
451 use_all = 1;
452 for (i = 0; i < argc; i++)
453 filescan(argv[i]);
454 if (nodb == 0)
455 dbprune(dba);
456 } else {
457 /* Database missing or corrupt. */
458 if (op != OP_UPDATE || errno != ENOENT)
459 say(MANDOC_DB, "%s: Automatically recreating"
460 " from scratch", strerror(errno));
461 exitcode = (int)MANDOCLEVEL_OK;
462 op = OP_DEFAULT;
463 if (treescan() == 0)
464 goto out;
465 dba = dba_new(128);
466 }
467 if (op != OP_DELETE)
468 mpages_merge(dba, mp);
469 if (nodb == 0)
470 dbwrite(dba);
471 dba_free(dba);
472 } else {
473 /*
474 * If we have arguments, use them as our manpaths.
475 * If we don't, use man.conf(5).
476 */
477 if (argc > 0) {
478 conf.manpath.paths = mandoc_reallocarray(NULL,
479 argc, sizeof(char *));
480 conf.manpath.sz = (size_t)argc;
481 for (i = 0; i < argc; i++)
482 conf.manpath.paths[i] = mandoc_strdup(argv[i]);
483 } else
484 manconf_parse(&conf, path_arg, NULL, NULL);
485
486 if (conf.manpath.sz == 0) {
487 exitcode = (int)MANDOCLEVEL_BADARG;
488 say("", "Empty manpath");
489 }
490
491 /*
492 * First scan the tree rooted at a base directory, then
493 * build a new database and finally move it into place.
494 * Ignore zero-length directories and strip trailing
495 * slashes.
496 */
497 for (j = 0; j < conf.manpath.sz; j++) {
498 sz = strlen(conf.manpath.paths[j]);
499 if (sz && conf.manpath.paths[j][sz - 1] == '/')
500 conf.manpath.paths[j][--sz] = '\0';
501 if (sz == 0)
502 continue;
503
504 if (j) {
505 mandoc_ohash_init(&mpages, 6,
506 offsetof(struct mpage, inodev));
507 mandoc_ohash_init(&mlinks, 6,
508 offsetof(struct mlink, file));
509 }
510
511 if (set_basedir(conf.manpath.paths[j], argc > 0) == 0)
512 continue;
513 if (treescan() == 0)
514 continue;
515 dba = dba_new(128);
516 mpages_merge(dba, mp);
517 if (nodb == 0)
518 dbwrite(dba);
519 dba_free(dba);
520
521 if (j + 1 < conf.manpath.sz) {
522 mpages_free();
523 ohash_delete(&mpages);
524 ohash_delete(&mlinks);
525 }
526 }
527 }
528 out:
529 manconf_free(&conf);
530 mparse_free(mp);
531 mchars_free();
532 mpages_free();
533 ohash_delete(&mpages);
534 ohash_delete(&mlinks);
535 #if DEBUG_MEMORY
536 mandoc_dbg_finish();
537 #endif
538 return exitcode;
539 usage:
540 progname = getprogname();
541 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
542 " %s [-aDnpQ] [-Tutf8] dir ...\n"
543 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
544 " %s [-Dnp] -u dir [file ...]\n"
545 " %s [-Q] -t file ...\n",
546 progname, progname, progname, progname, progname);
547
548 return (int)MANDOCLEVEL_BADARG;
549 }
550
551 /*
552 * To get a singly linked list in alpha order while inserting entries
553 * at the beginning, process directory entries in reverse alpha order.
554 */
555 static int
556 #if HAVE_FTS_COMPARE_CONST
fts_compare(const FTSENT * const * a,const FTSENT * const * b)557 fts_compare(const FTSENT *const *a, const FTSENT *const *b)
558 #else
559 fts_compare(const FTSENT **a, const FTSENT **b)
560 #endif
561 {
562 return -strcmp((*a)->fts_name, (*b)->fts_name);
563 }
564
565 /*
566 * Scan a directory tree rooted at "basedir" for manpages.
567 * We use fts(), scanning directory parts along the way for clues to our
568 * section and architecture.
569 *
570 * If use_all has been specified, grok all files.
571 * If not, sanitise paths to the following:
572 *
573 * [./]man*[/<arch>]/<name>.<section>
574 * or
575 * [./]cat<section>[/<arch>]/<name>.0
576 *
577 * TODO: accommodate for multi-language directories.
578 */
579 static int
treescan(void)580 treescan(void)
581 {
582 char buf[PATH_MAX];
583 FTS *f;
584 FTSENT *ff;
585 struct mlink *mlink;
586 int gzip;
587 enum form dform;
588 char *dsec, *arch, *fsec, *cp;
589 const char *path;
590 const char *argv[2];
591
592 argv[0] = ".";
593 argv[1] = NULL;
594
595 f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
596 fts_compare);
597 if (f == NULL) {
598 exitcode = (int)MANDOCLEVEL_SYSERR;
599 say("", "&fts_open");
600 return 0;
601 }
602
603 dsec = arch = NULL;
604 dform = FORM_NONE;
605
606 while ((ff = fts_read(f)) != NULL) {
607 path = ff->fts_path + 2;
608 switch (ff->fts_info) {
609
610 /*
611 * Symbolic links require various sanity checks,
612 * then get handled just like regular files.
613 */
614 case FTS_SL:
615 if (realpath(path, buf) == NULL) {
616 if (warnings)
617 say(path, "&realpath");
618 continue;
619 }
620 if (strncmp(buf, basedir, basedir_len) != 0
621 #ifdef READ_ALLOWED_PATH
622 && !read_allowed(buf)
623 #endif
624 ) {
625 if (warnings) say("",
626 "%s: outside base directory", buf);
627 continue;
628 }
629 /* Use logical inode to avoid mpages dupe. */
630 if (stat(path, ff->fts_statp) == -1) {
631 if (warnings)
632 say(path, "&stat");
633 continue;
634 }
635 if ((ff->fts_statp->st_mode & S_IFMT) != S_IFREG)
636 continue;
637 /* FALLTHROUGH */
638
639 /*
640 * If we're a regular file, add an mlink by using the
641 * stored directory data and handling the filename.
642 */
643 case FTS_F:
644 if ( ! strcmp(path, MANDOC_DB))
645 continue;
646 if ( ! use_all && ff->fts_level < 2) {
647 if (warnings)
648 say(path, "Extraneous file");
649 continue;
650 }
651 gzip = 0;
652 fsec = NULL;
653 while (fsec == NULL) {
654 fsec = strrchr(ff->fts_name, '.');
655 if (fsec == NULL || strcmp(fsec+1, "gz"))
656 break;
657 gzip = 1;
658 *fsec = '\0';
659 fsec = NULL;
660 }
661 if (fsec == NULL) {
662 if ( ! use_all) {
663 if (warnings)
664 say(path,
665 "No filename suffix");
666 continue;
667 }
668 } else if ( ! strcmp(++fsec, "html")) {
669 if (warnings)
670 say(path, "Skip html");
671 continue;
672 } else if ( ! strcmp(fsec, "ps")) {
673 if (warnings)
674 say(path, "Skip ps");
675 continue;
676 } else if ( ! strcmp(fsec, "pdf")) {
677 if (warnings)
678 say(path, "Skip pdf");
679 continue;
680 } else if ( ! use_all &&
681 ((dform == FORM_SRC &&
682 strncmp(fsec, dsec, strlen(dsec))) ||
683 (dform == FORM_CAT && strcmp(fsec, "0")))) {
684 if (warnings)
685 say(path, "Wrong filename suffix");
686 continue;
687 } else
688 fsec[-1] = '\0';
689
690 mlink = mandoc_calloc(1, sizeof(struct mlink));
691 if (strlcpy(mlink->file, path,
692 sizeof(mlink->file)) >=
693 sizeof(mlink->file)) {
694 say(path, "Filename too long");
695 free(mlink);
696 continue;
697 }
698 mlink->dform = dform;
699 mlink->dsec = dsec;
700 mlink->arch = arch;
701 mlink->name = ff->fts_name;
702 mlink->fsec = fsec;
703 mlink->gzip = gzip;
704 mlink_add(mlink, ff->fts_statp);
705 continue;
706
707 case FTS_D:
708 case FTS_DP:
709 break;
710
711 default:
712 if (warnings)
713 say(path, "Not a regular file");
714 continue;
715 }
716
717 switch (ff->fts_level) {
718 case 0:
719 /* Ignore the root directory. */
720 break;
721 case 1:
722 /*
723 * This might contain manX/ or catX/.
724 * Try to infer this from the name.
725 * If we're not in use_all, enforce it.
726 */
727 cp = ff->fts_name;
728 if (ff->fts_info == FTS_DP) {
729 dform = FORM_NONE;
730 dsec = NULL;
731 break;
732 }
733
734 if ( ! strncmp(cp, "man", 3)) {
735 dform = FORM_SRC;
736 dsec = cp + 3;
737 } else if ( ! strncmp(cp, "cat", 3)) {
738 dform = FORM_CAT;
739 dsec = cp + 3;
740 } else {
741 dform = FORM_NONE;
742 dsec = NULL;
743 }
744
745 if (dsec != NULL || use_all)
746 break;
747
748 if (warnings)
749 say(path, "Unknown directory part");
750 fts_set(f, ff, FTS_SKIP);
751 break;
752 case 2:
753 /*
754 * Possibly our architecture.
755 * If we're descending, keep tabs on it.
756 */
757 if (ff->fts_info != FTS_DP && dsec != NULL)
758 arch = ff->fts_name;
759 else
760 arch = NULL;
761 break;
762 default:
763 if (ff->fts_info == FTS_DP || use_all)
764 break;
765 if (warnings)
766 say(path, "Extraneous directory part");
767 fts_set(f, ff, FTS_SKIP);
768 break;
769 }
770 }
771
772 fts_close(f);
773 return 1;
774 }
775
776 /*
777 * Add a file to the mlinks table.
778 * Do not verify that it's a "valid" looking manpage (we'll do that
779 * later).
780 *
781 * Try to infer the manual section, architecture, and page name from the
782 * path, assuming it looks like
783 *
784 * [./]man*[/<arch>]/<name>.<section>
785 * or
786 * [./]cat<section>[/<arch>]/<name>.0
787 *
788 * See treescan() for the fts(3) version of this.
789 */
790 static void
filescan(const char * infile)791 filescan(const char *infile)
792 {
793 struct stat st;
794 struct mlink *mlink;
795 char *linkfile, *p, *realdir, *start, *usefile;
796 size_t realdir_len;
797
798 assert(use_all);
799
800 if (strncmp(infile, "./", 2) == 0)
801 infile += 2;
802
803 /*
804 * We have to do lstat(2) before realpath(3) loses
805 * the information whether this is a symbolic link.
806 * We need to know that because for symbolic links,
807 * we want to use the original file name, while for
808 * regular files, we want to use the real path.
809 */
810 if (lstat(infile, &st) == -1) {
811 exitcode = (int)MANDOCLEVEL_BADARG;
812 say(infile, "&lstat");
813 return;
814 } else if (S_ISREG(st.st_mode) == 0 && S_ISLNK(st.st_mode) == 0) {
815 exitcode = (int)MANDOCLEVEL_BADARG;
816 say(infile, "Not a regular file");
817 return;
818 }
819
820 /*
821 * We have to resolve the file name to the real path
822 * in any case for the base directory check.
823 */
824 if ((usefile = realpath(infile, NULL)) == NULL) {
825 exitcode = (int)MANDOCLEVEL_BADARG;
826 say(infile, "&realpath");
827 return;
828 }
829
830 if (op == OP_TEST)
831 start = usefile;
832 else if (strncmp(usefile, basedir, basedir_len) == 0)
833 start = usefile + basedir_len;
834 #ifdef READ_ALLOWED_PATH
835 else if (read_allowed(usefile))
836 start = usefile;
837 #endif
838 else {
839 exitcode = (int)MANDOCLEVEL_BADARG;
840 say("", "%s: outside base directory", infile);
841 free(usefile);
842 return;
843 }
844
845 /*
846 * Now we are sure the file is inside our tree.
847 * If it is a symbolic link, ignore the real path
848 * and use the original name.
849 */
850 do {
851 if (S_ISLNK(st.st_mode) == 0)
852 break;
853
854 /*
855 * Some implementations of realpath(3) may succeed
856 * even if the target of the link does not exist,
857 * so check again for extra safety.
858 */
859 if (stat(usefile, &st) == -1) {
860 exitcode = (int)MANDOCLEVEL_BADARG;
861 say(infile, "&stat");
862 free(usefile);
863 return;
864 }
865 linkfile = mandoc_strdup(infile);
866 if (op == OP_TEST) {
867 free(usefile);
868 start = usefile = linkfile;
869 break;
870 }
871 if (strncmp(infile, basedir, basedir_len) == 0) {
872 free(usefile);
873 usefile = linkfile;
874 start = usefile + basedir_len;
875 break;
876 }
877
878 /*
879 * This symbolic link points into the basedir
880 * from the outside. Let's see whether any of
881 * the parent directories resolve to the basedir.
882 */
883 p = strchr(linkfile, '\0');
884 do {
885 while (*--p != '/')
886 continue;
887 *p = '\0';
888 if ((realdir = realpath(linkfile, NULL)) == NULL) {
889 exitcode = (int)MANDOCLEVEL_BADARG;
890 say(infile, "&realpath");
891 free(linkfile);
892 free(usefile);
893 return;
894 }
895 realdir_len = strlen(realdir) + 1;
896 free(realdir);
897 *p = '/';
898 } while (realdir_len > basedir_len);
899
900 /*
901 * If one of the directories resolves to the basedir,
902 * use the rest of the original name.
903 * Otherwise, the best we can do
904 * is to use the filename pointed to.
905 */
906 if (realdir_len == basedir_len) {
907 free(usefile);
908 usefile = linkfile;
909 start = p + 1;
910 } else {
911 free(linkfile);
912 start = usefile + basedir_len;
913 }
914 } while (/* CONSTCOND */ 0);
915
916 mlink = mandoc_calloc(1, sizeof(struct mlink));
917 mlink->dform = FORM_NONE;
918 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
919 sizeof(mlink->file)) {
920 say(start, "Filename too long");
921 free(mlink);
922 free(usefile);
923 return;
924 }
925
926 /*
927 * In test mode or when the original name is absolute
928 * but outside our tree, guess the base directory.
929 */
930
931 if (op == OP_TEST || (start == usefile && *start == '/')) {
932 if (strncmp(usefile, "man/", 4) == 0)
933 start = usefile + 4;
934 else if ((start = strstr(usefile, "/man/")) != NULL)
935 start += 5;
936 else
937 start = usefile;
938 }
939
940 /*
941 * First try to guess our directory structure.
942 * If we find a separator, try to look for man* or cat*.
943 * If we find one of these and what's underneath is a directory,
944 * assume it's an architecture.
945 */
946 if ((p = strchr(start, '/')) != NULL) {
947 *p++ = '\0';
948 if (strncmp(start, "man", 3) == 0) {
949 mlink->dform = FORM_SRC;
950 mlink->dsec = start + 3;
951 } else if (strncmp(start, "cat", 3) == 0) {
952 mlink->dform = FORM_CAT;
953 mlink->dsec = start + 3;
954 }
955
956 start = p;
957 if (mlink->dsec != NULL && (p = strchr(start, '/')) != NULL) {
958 *p++ = '\0';
959 mlink->arch = start;
960 start = p;
961 }
962 }
963
964 /*
965 * Now check the file suffix.
966 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
967 */
968 p = strrchr(start, '\0');
969 while (p-- > start && *p != '/' && *p != '.')
970 continue;
971
972 if (*p == '.') {
973 *p++ = '\0';
974 mlink->fsec = p;
975 }
976
977 /*
978 * Now try to parse the name.
979 * Use the filename portion of the path.
980 */
981 mlink->name = start;
982 if ((p = strrchr(start, '/')) != NULL) {
983 mlink->name = p + 1;
984 *p = '\0';
985 }
986 mlink_add(mlink, &st);
987 free(usefile);
988 }
989
990 static void
mlink_add(struct mlink * mlink,const struct stat * st)991 mlink_add(struct mlink *mlink, const struct stat *st)
992 {
993 struct inodev inodev;
994 struct mpage *mpage;
995 unsigned int slot;
996
997 assert(NULL != mlink->file);
998
999 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
1000 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
1001 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
1002 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
1003
1004 if ('0' == *mlink->fsec) {
1005 free(mlink->fsec);
1006 mlink->fsec = mandoc_strdup(mlink->dsec);
1007 mlink->fform = FORM_CAT;
1008 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
1009 mlink->fform = FORM_SRC;
1010 else
1011 mlink->fform = FORM_NONE;
1012
1013 slot = ohash_qlookup(&mlinks, mlink->file);
1014 assert(NULL == ohash_find(&mlinks, slot));
1015 ohash_insert(&mlinks, slot, mlink);
1016
1017 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
1018 inodev.st_ino = st->st_ino;
1019 inodev.st_dev = st->st_dev;
1020 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
1021 sizeof(struct inodev), inodev.st_ino);
1022 mpage = ohash_find(&mpages, slot);
1023 if (NULL == mpage) {
1024 mpage = mandoc_calloc(1, sizeof(struct mpage));
1025 mpage->inodev.st_ino = inodev.st_ino;
1026 mpage->inodev.st_dev = inodev.st_dev;
1027 mpage->form = FORM_NONE;
1028 mpage->next = mpage_head;
1029 mpage_head = mpage;
1030 ohash_insert(&mpages, slot, mpage);
1031 } else
1032 mlink->next = mpage->mlinks;
1033 mpage->mlinks = mlink;
1034 mlink->mpage = mpage;
1035 }
1036
1037 static void
mlink_free(struct mlink * mlink)1038 mlink_free(struct mlink *mlink)
1039 {
1040
1041 free(mlink->dsec);
1042 free(mlink->arch);
1043 free(mlink->name);
1044 free(mlink->fsec);
1045 free(mlink);
1046 }
1047
1048 static void
mpages_free(void)1049 mpages_free(void)
1050 {
1051 struct mpage *mpage;
1052 struct mlink *mlink;
1053
1054 while ((mpage = mpage_head) != NULL) {
1055 while ((mlink = mpage->mlinks) != NULL) {
1056 mpage->mlinks = mlink->next;
1057 mlink_free(mlink);
1058 }
1059 mpage_head = mpage->next;
1060 free(mpage->sec);
1061 free(mpage->arch);
1062 free(mpage->title);
1063 free(mpage->desc);
1064 free(mpage);
1065 }
1066 }
1067
1068 /*
1069 * For each mlink to the mpage, check whether the path looks like
1070 * it is formatted, and if it does, check whether a source manual
1071 * exists by the same name, ignoring the suffix.
1072 * If both conditions hold, drop the mlink.
1073 */
1074 static void
mlinks_undupe(struct mpage * mpage)1075 mlinks_undupe(struct mpage *mpage)
1076 {
1077 char buf[PATH_MAX];
1078 struct mlink **prev;
1079 struct mlink *mlink;
1080 char *bufp;
1081
1082 mpage->form = FORM_CAT;
1083 prev = &mpage->mlinks;
1084 while (NULL != (mlink = *prev)) {
1085 if (FORM_CAT != mlink->dform) {
1086 mpage->form = FORM_NONE;
1087 goto nextlink;
1088 }
1089 (void)strlcpy(buf, mlink->file, sizeof(buf));
1090 bufp = strstr(buf, "cat");
1091 assert(NULL != bufp);
1092 memcpy(bufp, "man", 3);
1093 if (NULL != (bufp = strrchr(buf, '.')))
1094 *++bufp = '\0';
1095 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1096 if (NULL == ohash_find(&mlinks,
1097 ohash_qlookup(&mlinks, buf)))
1098 goto nextlink;
1099 if (warnings)
1100 say(mlink->file, "Man source exists: %s", buf);
1101 if (use_all)
1102 goto nextlink;
1103 *prev = mlink->next;
1104 mlink_free(mlink);
1105 continue;
1106 nextlink:
1107 prev = &(*prev)->next;
1108 }
1109 }
1110
1111 static void
mlink_check(struct mpage * mpage,struct mlink * mlink)1112 mlink_check(struct mpage *mpage, struct mlink *mlink)
1113 {
1114 struct str *str;
1115 unsigned int slot;
1116
1117 /*
1118 * Check whether the manual section given in a file
1119 * agrees with the directory where the file is located.
1120 * Some manuals have suffixes like (3p) on their
1121 * section number either inside the file or in the
1122 * directory name, some are linked into more than one
1123 * section, like encrypt(1) = makekey(8).
1124 */
1125
1126 if (FORM_SRC == mpage->form &&
1127 strcasecmp(mpage->sec, mlink->dsec))
1128 say(mlink->file, "Section \"%s\" manual in %s directory",
1129 mpage->sec, mlink->dsec);
1130
1131 /*
1132 * Manual page directories exist for each kernel
1133 * architecture as returned by machine(1).
1134 * However, many manuals only depend on the
1135 * application architecture as returned by arch(1).
1136 * For example, some (2/ARM) manuals are shared
1137 * across the "armish" and "zaurus" kernel
1138 * architectures.
1139 * A few manuals are even shared across completely
1140 * different architectures, for example fdformat(1)
1141 * on amd64, i386, and sparc64.
1142 */
1143
1144 if (strcasecmp(mpage->arch, mlink->arch))
1145 say(mlink->file, "Architecture \"%s\" manual in "
1146 "\"%s\" directory", mpage->arch, mlink->arch);
1147
1148 /*
1149 * XXX
1150 * parse_cat() doesn't set NAME_TITLE yet.
1151 */
1152
1153 if (FORM_CAT == mpage->form)
1154 return;
1155
1156 /*
1157 * Check whether this mlink
1158 * appears as a name in the NAME section.
1159 */
1160
1161 slot = ohash_qlookup(&names, mlink->name);
1162 str = ohash_find(&names, slot);
1163 assert(NULL != str);
1164 if ( ! (NAME_TITLE & str->mask))
1165 say(mlink->file, "Name missing in NAME section");
1166 }
1167
1168 /*
1169 * Run through the files in the global vector "mpages"
1170 * and add them to the database specified in "basedir".
1171 *
1172 * This handles the parsing scheme itself, using the cues of directory
1173 * and filename to determine whether the file is parsable or not.
1174 */
1175 static void
mpages_merge(struct dba * dba,struct mparse * mp)1176 mpages_merge(struct dba *dba, struct mparse *mp)
1177 {
1178 struct mpage *mpage, *mpage_dest;
1179 struct mlink *mlink, *mlink_dest;
1180 struct roff_meta *meta;
1181 char *cp;
1182 int fd;
1183
1184 for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1185 mlinks_undupe(mpage);
1186 if ((mlink = mpage->mlinks) == NULL)
1187 continue;
1188
1189 name_mask = NAME_MASK;
1190 mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1191 mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1192 mparse_reset(mp);
1193 meta = NULL;
1194
1195 if ((fd = mparse_open(mp, mlink->file)) == -1) {
1196 say(mlink->file, "&open");
1197 goto nextpage;
1198 }
1199
1200 /*
1201 * Interpret the file as mdoc(7) or man(7) source
1202 * code, unless it is known to be formatted.
1203 */
1204 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1205 mparse_readfd(mp, fd, mlink->file);
1206 close(fd);
1207 fd = -1;
1208 meta = mparse_result(mp);
1209 }
1210
1211 if (meta != NULL && meta->sodest != NULL) {
1212 mlink_dest = ohash_find(&mlinks,
1213 ohash_qlookup(&mlinks, meta->sodest));
1214 if (mlink_dest == NULL) {
1215 mandoc_asprintf(&cp, "%s.gz", meta->sodest);
1216 mlink_dest = ohash_find(&mlinks,
1217 ohash_qlookup(&mlinks, cp));
1218 free(cp);
1219 }
1220 if (mlink_dest != NULL) {
1221
1222 /* The .so target exists. */
1223
1224 mpage_dest = mlink_dest->mpage;
1225 while (1) {
1226 mlink->mpage = mpage_dest;
1227
1228 /*
1229 * If the target was already
1230 * processed, add the links
1231 * to the database now.
1232 * Otherwise, this will
1233 * happen when we come
1234 * to the target.
1235 */
1236
1237 if (mpage_dest->dba != NULL)
1238 dbadd_mlink(mlink);
1239
1240 if (mlink->next == NULL)
1241 break;
1242 mlink = mlink->next;
1243 }
1244
1245 /* Move all links to the target. */
1246
1247 mlink->next = mlink_dest->next;
1248 mlink_dest->next = mpage->mlinks;
1249 mpage->mlinks = NULL;
1250 goto nextpage;
1251 }
1252 meta->macroset = MACROSET_NONE;
1253 }
1254 if (meta != NULL && meta->macroset == MACROSET_MDOC) {
1255 mpage->form = FORM_SRC;
1256 mpage->sec = meta->msec;
1257 mpage->sec = mandoc_strdup(
1258 mpage->sec == NULL ? "" : mpage->sec);
1259 mpage->arch = meta->arch;
1260 mpage->arch = mandoc_strdup(
1261 mpage->arch == NULL ? "" : mpage->arch);
1262 mpage->title = mandoc_strdup(meta->title);
1263 } else if (meta != NULL && meta->macroset == MACROSET_MAN) {
1264 if (*meta->msec != '\0' || *meta->title != '\0') {
1265 mpage->form = FORM_SRC;
1266 mpage->sec = mandoc_strdup(meta->msec);
1267 mpage->arch = mandoc_strdup(mlink->arch);
1268 mpage->title = mandoc_strdup(meta->title);
1269 } else
1270 meta = NULL;
1271 }
1272
1273 assert(mpage->desc == NULL);
1274 if (meta == NULL || meta->sodest != NULL) {
1275 mpage->sec = mandoc_strdup(mlink->dsec);
1276 mpage->arch = mandoc_strdup(mlink->arch);
1277 mpage->title = mandoc_strdup(mlink->name);
1278 if (meta == NULL) {
1279 mpage->form = FORM_CAT;
1280 parse_cat(mpage, fd);
1281 } else
1282 mpage->form = FORM_SRC;
1283 } else if (meta->macroset == MACROSET_MDOC)
1284 parse_mdoc(mpage, meta, meta->first);
1285 else
1286 parse_man(mpage, meta, meta->first);
1287 if (mpage->desc == NULL) {
1288 mpage->desc = mandoc_strdup(mlink->name);
1289 if (warnings)
1290 say(mlink->file, "No one-line description, "
1291 "using filename \"%s\"", mlink->name);
1292 }
1293
1294 for (mlink = mpage->mlinks;
1295 mlink != NULL;
1296 mlink = mlink->next) {
1297 putkey(mpage, mlink->name, NAME_FILE);
1298 if (warnings && !use_all)
1299 mlink_check(mpage, mlink);
1300 }
1301
1302 dbadd(dba, mpage);
1303
1304 nextpage:
1305 ohash_delete(&strings);
1306 ohash_delete(&names);
1307 }
1308 }
1309
1310 static void
parse_cat(struct mpage * mpage,int fd)1311 parse_cat(struct mpage *mpage, int fd)
1312 {
1313 FILE *stream;
1314 struct mlink *mlink;
1315 char *line, *p, *title, *sec;
1316 size_t linesz, plen, titlesz;
1317 ssize_t len;
1318 int offs;
1319
1320 mlink = mpage->mlinks;
1321 stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1322 if (stream == NULL) {
1323 if (fd != -1)
1324 close(fd);
1325 if (warnings)
1326 say(mlink->file, "&fopen");
1327 return;
1328 }
1329
1330 line = NULL;
1331 linesz = 0;
1332
1333 /* Parse the section number from the header line. */
1334
1335 while (getline(&line, &linesz, stream) != -1) {
1336 if (*line == '\n')
1337 continue;
1338 if ((sec = strchr(line, '(')) == NULL)
1339 break;
1340 if ((p = strchr(++sec, ')')) == NULL)
1341 break;
1342 free(mpage->sec);
1343 mpage->sec = mandoc_strndup(sec, p - sec);
1344 if (warnings && *mlink->dsec != '\0' &&
1345 strcasecmp(mpage->sec, mlink->dsec))
1346 say(mlink->file,
1347 "Section \"%s\" manual in %s directory",
1348 mpage->sec, mlink->dsec);
1349 break;
1350 }
1351
1352 /* Skip to first blank line. */
1353
1354 while (line == NULL || *line != '\n')
1355 if (getline(&line, &linesz, stream) == -1)
1356 break;
1357
1358 /*
1359 * Assume the first line that is not indented
1360 * is the first section header. Skip to it.
1361 */
1362
1363 while (getline(&line, &linesz, stream) != -1)
1364 if (*line != '\n' && *line != ' ')
1365 break;
1366
1367 /*
1368 * Read up until the next section into a buffer.
1369 * Strip the leading and trailing newline from each read line,
1370 * appending a trailing space.
1371 * Ignore empty (whitespace-only) lines.
1372 */
1373
1374 titlesz = 0;
1375 title = NULL;
1376
1377 while ((len = getline(&line, &linesz, stream)) != -1) {
1378 if (*line != ' ')
1379 break;
1380 offs = 0;
1381 while (isspace((unsigned char)line[offs]))
1382 offs++;
1383 if (line[offs] == '\0')
1384 continue;
1385 title = mandoc_realloc(title, titlesz + len - offs);
1386 memcpy(title + titlesz, line + offs, len - offs);
1387 titlesz += len - offs;
1388 title[titlesz - 1] = ' ';
1389 }
1390 free(line);
1391
1392 /*
1393 * If no page content can be found, or the input line
1394 * is already the next section header, or there is no
1395 * trailing newline, reuse the page title as the page
1396 * description.
1397 */
1398
1399 if (NULL == title || '\0' == *title) {
1400 if (warnings)
1401 say(mlink->file, "Cannot find NAME section");
1402 fclose(stream);
1403 free(title);
1404 return;
1405 }
1406
1407 title[titlesz - 1] = '\0';
1408
1409 /*
1410 * Skip to the first dash.
1411 * Use the remaining line as the description (no more than 70
1412 * bytes).
1413 */
1414
1415 if (NULL != (p = strstr(title, "- "))) {
1416 for (p += 2; ' ' == *p || '\b' == *p; p++)
1417 /* Skip to next word. */ ;
1418 } else {
1419 if (warnings)
1420 say(mlink->file, "No dash in title line, "
1421 "reusing \"%s\" as one-line description", title);
1422 p = title;
1423 }
1424
1425 plen = strlen(p);
1426
1427 /* Strip backspace-encoding from line. */
1428
1429 while (NULL != (line = memchr(p, '\b', plen))) {
1430 len = line - p;
1431 if (0 == len) {
1432 memmove(line, line + 1, plen--);
1433 continue;
1434 }
1435 memmove(line - 1, line + 1, plen - len);
1436 plen -= 2;
1437 }
1438
1439 /*
1440 * Cut off excessive one-line descriptions.
1441 * Bad pages are not worth better heuristics.
1442 */
1443
1444 mpage->desc = mandoc_strndup(p, 150);
1445 fclose(stream);
1446 free(title);
1447 }
1448
1449 /*
1450 * Put a type/word pair into the word database for this particular file.
1451 */
1452 static void
putkey(const struct mpage * mpage,char * value,uint64_t type)1453 putkey(const struct mpage *mpage, char *value, uint64_t type)
1454 {
1455 putkeys(mpage, value, strlen(value), type);
1456 }
1457
1458 /*
1459 * Grok all nodes at or below a certain mdoc node into putkey().
1460 */
1461 static void
putmdockey(const struct mpage * mpage,const struct roff_node * n,uint64_t m,int taboo)1462 putmdockey(const struct mpage *mpage,
1463 const struct roff_node *n, uint64_t m, int taboo)
1464 {
1465
1466 for ( ; NULL != n; n = n->next) {
1467 if (n->flags & taboo)
1468 continue;
1469 if (NULL != n->child)
1470 putmdockey(mpage, n->child, m, taboo);
1471 if (n->type == ROFFT_TEXT)
1472 putkey(mpage, n->string, m);
1473 }
1474 }
1475
1476 static void
parse_man(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1477 parse_man(struct mpage *mpage, const struct roff_meta *meta,
1478 const struct roff_node *n)
1479 {
1480 const struct roff_node *head, *body;
1481 char *start, *title;
1482 char byte;
1483 size_t sz;
1484
1485 if (n == NULL)
1486 return;
1487
1488 /*
1489 * We're only searching for one thing: the first text child in
1490 * the BODY of a NAME section. Since we don't keep track of
1491 * sections in -man, run some hoops to find out whether we're in
1492 * the correct section or not.
1493 */
1494
1495 if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1496 body = n;
1497 if ((head = body->parent->head) != NULL &&
1498 (head = head->child) != NULL &&
1499 head->next == NULL &&
1500 head->type == ROFFT_TEXT &&
1501 strcmp(head->string, "NAME") == 0 &&
1502 body->child != NULL) {
1503
1504 /*
1505 * Suck the entire NAME section into memory.
1506 * Yes, we might run away.
1507 * But too many manuals have big, spread-out
1508 * NAME sections over many lines.
1509 */
1510
1511 title = NULL;
1512 deroff(&title, body);
1513 if (NULL == title)
1514 return;
1515
1516 /*
1517 * Go through a special heuristic dance here.
1518 * Conventionally, one or more manual names are
1519 * comma-specified prior to a whitespace, then a
1520 * dash, then a description. Try to puzzle out
1521 * the name parts here.
1522 */
1523
1524 start = title;
1525 for ( ;; ) {
1526 sz = strcspn(start, " ,");
1527 if ('\0' == start[sz])
1528 break;
1529
1530 byte = start[sz];
1531 start[sz] = '\0';
1532
1533 /*
1534 * Assume a stray trailing comma in the
1535 * name list if a name begins with a dash.
1536 */
1537
1538 if ('-' == start[0] ||
1539 ('\\' == start[0] && '-' == start[1]))
1540 break;
1541
1542 putkey(mpage, start, NAME_TITLE);
1543 if ( ! (mpage->name_head_done ||
1544 strcasecmp(start, meta->title))) {
1545 putkey(mpage, start, NAME_HEAD);
1546 mpage->name_head_done = 1;
1547 }
1548
1549 if (' ' == byte) {
1550 start += sz + 1;
1551 break;
1552 }
1553
1554 assert(',' == byte);
1555 start += sz + 1;
1556 while (' ' == *start)
1557 start++;
1558 }
1559
1560 if (start == title) {
1561 putkey(mpage, start, NAME_TITLE);
1562 if ( ! (mpage->name_head_done ||
1563 strcasecmp(start, meta->title))) {
1564 putkey(mpage, start, NAME_HEAD);
1565 mpage->name_head_done = 1;
1566 }
1567 free(title);
1568 return;
1569 }
1570
1571 while (isspace((unsigned char)*start))
1572 start++;
1573
1574 if (0 == strncmp(start, "-", 1))
1575 start += 1;
1576 else if (0 == strncmp(start, "\\-\\-", 4))
1577 start += 4;
1578 else if (0 == strncmp(start, "\\-", 2))
1579 start += 2;
1580 else if (0 == strncmp(start, "\\(en", 4))
1581 start += 4;
1582 else if (0 == strncmp(start, "\\(em", 4))
1583 start += 4;
1584
1585 while (' ' == *start)
1586 start++;
1587
1588 /*
1589 * Cut off excessive one-line descriptions.
1590 * Bad pages are not worth better heuristics.
1591 */
1592
1593 mpage->desc = mandoc_strndup(start, 150);
1594 free(title);
1595 return;
1596 }
1597 }
1598
1599 for (n = n->child; n; n = n->next) {
1600 if (NULL != mpage->desc)
1601 break;
1602 parse_man(mpage, meta, n);
1603 }
1604 }
1605
1606 static void
parse_mdoc(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1607 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1608 const struct roff_node *n)
1609 {
1610 const struct mdoc_handler *handler;
1611
1612 for (n = n->child; n != NULL; n = n->next) {
1613 if (n->tok == TOKEN_NONE || n->tok < ROFF_MAX)
1614 continue;
1615 assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1616 handler = mdoc_handlers + (n->tok - MDOC_Dd);
1617 if (n->flags & handler->taboo)
1618 continue;
1619
1620 switch (n->type) {
1621 case ROFFT_ELEM:
1622 case ROFFT_BLOCK:
1623 case ROFFT_HEAD:
1624 case ROFFT_BODY:
1625 case ROFFT_TAIL:
1626 if (handler->fp != NULL &&
1627 (*handler->fp)(mpage, meta, n) == 0)
1628 break;
1629 if (handler->mask)
1630 putmdockey(mpage, n->child,
1631 handler->mask, handler->taboo);
1632 break;
1633 default:
1634 continue;
1635 }
1636 if (NULL != n->child)
1637 parse_mdoc(mpage, meta, n);
1638 }
1639 }
1640
1641 static int
parse_mdoc_Fa(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1642 parse_mdoc_Fa(struct mpage *mpage, const struct roff_meta *meta,
1643 const struct roff_node *n)
1644 {
1645 uint64_t mask;
1646
1647 mask = TYPE_Fa;
1648 if (n->sec == SEC_SYNOPSIS)
1649 mask |= TYPE_Vt;
1650
1651 putmdockey(mpage, n->child, mask, 0);
1652 return 0;
1653 }
1654
1655 static int
parse_mdoc_Fd(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1656 parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1657 const struct roff_node *n)
1658 {
1659 char *start, *end;
1660 size_t sz;
1661
1662 if (SEC_SYNOPSIS != n->sec ||
1663 NULL == (n = n->child) ||
1664 n->type != ROFFT_TEXT)
1665 return 0;
1666
1667 /*
1668 * Only consider those `Fd' macro fields that begin with an
1669 * "inclusion" token (versus, e.g., #define).
1670 */
1671
1672 if (strcmp("#include", n->string))
1673 return 0;
1674
1675 if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1676 return 0;
1677
1678 /*
1679 * Strip away the enclosing angle brackets and make sure we're
1680 * not zero-length.
1681 */
1682
1683 start = n->string;
1684 if ('<' == *start || '"' == *start)
1685 start++;
1686
1687 if (0 == (sz = strlen(start)))
1688 return 0;
1689
1690 end = &start[(int)sz - 1];
1691 if ('>' == *end || '"' == *end)
1692 end--;
1693
1694 if (end > start)
1695 putkeys(mpage, start, end - start + 1, TYPE_In);
1696 return 0;
1697 }
1698
1699 static void
parse_mdoc_fname(struct mpage * mpage,const struct roff_node * n)1700 parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1701 {
1702 char *cp;
1703 size_t sz;
1704
1705 if (n->type != ROFFT_TEXT)
1706 return;
1707
1708 /* Skip function pointer punctuation. */
1709
1710 cp = n->string;
1711 while (*cp == '(' || *cp == '*')
1712 cp++;
1713 sz = strcspn(cp, "()");
1714
1715 putkeys(mpage, cp, sz, TYPE_Fn);
1716 if (n->sec == SEC_SYNOPSIS)
1717 putkeys(mpage, cp, sz, NAME_SYN);
1718 }
1719
1720 static int
parse_mdoc_Fn(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1721 parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1722 const struct roff_node *n)
1723 {
1724 uint64_t mask;
1725
1726 if (n->child == NULL)
1727 return 0;
1728
1729 parse_mdoc_fname(mpage, n->child);
1730
1731 n = n->child->next;
1732 if (n != NULL && n->type == ROFFT_TEXT) {
1733 mask = TYPE_Fa;
1734 if (n->sec == SEC_SYNOPSIS)
1735 mask |= TYPE_Vt;
1736 putmdockey(mpage, n, mask, 0);
1737 }
1738
1739 return 0;
1740 }
1741
1742 static int
parse_mdoc_Fo(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1743 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1744 const struct roff_node *n)
1745 {
1746
1747 if (n->type != ROFFT_HEAD)
1748 return 1;
1749
1750 if (n->child != NULL)
1751 parse_mdoc_fname(mpage, n->child);
1752
1753 return 0;
1754 }
1755
1756 static int
parse_mdoc_Va(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1757 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1758 const struct roff_node *n)
1759 {
1760 char *cp;
1761
1762 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1763 return 0;
1764
1765 if (n->child != NULL &&
1766 n->child->next == NULL &&
1767 n->child->type == ROFFT_TEXT)
1768 return 1;
1769
1770 cp = NULL;
1771 deroff(&cp, n);
1772 if (cp != NULL) {
1773 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1774 n->type == ROFFT_BODY ? TYPE_Va : 0));
1775 free(cp);
1776 }
1777
1778 return 0;
1779 }
1780
1781 static int
parse_mdoc_Xr(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1782 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1783 const struct roff_node *n)
1784 {
1785 char *cp;
1786
1787 if (NULL == (n = n->child))
1788 return 0;
1789
1790 if (NULL == n->next) {
1791 putkey(mpage, n->string, TYPE_Xr);
1792 return 0;
1793 }
1794
1795 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1796 putkey(mpage, cp, TYPE_Xr);
1797 free(cp);
1798 return 0;
1799 }
1800
1801 static int
parse_mdoc_Nd(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1802 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1803 const struct roff_node *n)
1804 {
1805
1806 if (n->type == ROFFT_BODY)
1807 deroff(&mpage->desc, n);
1808 return 0;
1809 }
1810
1811 static int
parse_mdoc_Nm(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1812 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1813 const struct roff_node *n)
1814 {
1815
1816 if (SEC_NAME == n->sec)
1817 putmdockey(mpage, n->child, NAME_TITLE, 0);
1818 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1819 if (n->child == NULL)
1820 putkey(mpage, meta->name, NAME_SYN);
1821 else
1822 putmdockey(mpage, n->child, NAME_SYN, 0);
1823 }
1824 if ( ! (mpage->name_head_done ||
1825 n->child == NULL || n->child->string == NULL ||
1826 strcasecmp(n->child->string, meta->title))) {
1827 putkey(mpage, n->child->string, NAME_HEAD);
1828 mpage->name_head_done = 1;
1829 }
1830 return 0;
1831 }
1832
1833 static int
parse_mdoc_Sh(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1834 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1835 const struct roff_node *n)
1836 {
1837
1838 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1839 }
1840
1841 static int
parse_mdoc_head(struct mpage * mpage,const struct roff_meta * meta,const struct roff_node * n)1842 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1843 const struct roff_node *n)
1844 {
1845
1846 return n->type == ROFFT_HEAD;
1847 }
1848
1849 /*
1850 * Add a string to the hash table for the current manual.
1851 * Each string has a bitmask telling which macros it belongs to.
1852 * When we finish the manual, we'll dump the table.
1853 */
1854 static void
putkeys(const struct mpage * mpage,char * cp,size_t sz,uint64_t v)1855 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1856 {
1857 struct ohash *htab;
1858 struct str *s;
1859 const char *end;
1860 unsigned int slot;
1861 int i, mustfree;
1862
1863 if (0 == sz)
1864 return;
1865
1866 mustfree = render_string(&cp, &sz);
1867
1868 if (TYPE_Nm & v) {
1869 htab = &names;
1870 v &= name_mask;
1871 if (v & NAME_FIRST)
1872 name_mask &= ~NAME_FIRST;
1873 if (debug > 1)
1874 say(mpage->mlinks->file,
1875 "Adding name %*s, bits=0x%llx", (int)sz, cp,
1876 (unsigned long long)v);
1877 } else {
1878 htab = &strings;
1879 if (debug > 1)
1880 for (i = 0; i < KEY_MAX; i++)
1881 if ((uint64_t)1 << i & v)
1882 say(mpage->mlinks->file,
1883 "Adding key %s=%*s",
1884 mansearch_keynames[i], (int)sz, cp);
1885 }
1886
1887 end = cp + sz;
1888 slot = ohash_qlookupi(htab, cp, &end);
1889 s = ohash_find(htab, slot);
1890
1891 if (NULL != s && mpage == s->mpage) {
1892 s->mask |= v;
1893 return;
1894 } else if (NULL == s) {
1895 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1896 memcpy(s->key, cp, sz);
1897 ohash_insert(htab, slot, s);
1898 }
1899 s->mpage = mpage;
1900 s->mask = v;
1901
1902 if (mustfree)
1903 free(cp);
1904 }
1905
1906 /*
1907 * Take a Unicode codepoint and produce its UTF-8 encoding.
1908 * This isn't the best way to do this, but it works.
1909 * The magic numbers are from the UTF-8 packaging.
1910 * Read the UTF-8 spec or the utf8(7) manual page for details.
1911 */
1912 static size_t
utf8(unsigned int cp,char out[5])1913 utf8(unsigned int cp, char out[5])
1914 {
1915 size_t rc;
1916
1917 if (cp <= 0x7f) {
1918 rc = 1;
1919 out[0] = (char)cp;
1920 } else if (cp <= 0x7ff) {
1921 rc = 2;
1922 out[0] = (cp >> 6 & 31) | 192;
1923 out[1] = (cp & 63) | 128;
1924 } else if (cp >= 0xd800 && cp <= 0xdfff) {
1925 rc = 0; /* reject UTF-16 surrogate */
1926 } else if (cp <= 0xffff) {
1927 rc = 3;
1928 out[0] = (cp >> 12 & 15) | 224;
1929 out[1] = (cp >> 6 & 63) | 128;
1930 out[2] = (cp & 63) | 128;
1931 } else if (cp <= 0x10ffff) {
1932 rc = 4;
1933 out[0] = (cp >> 18 & 7) | 240;
1934 out[1] = (cp >> 12 & 63) | 128;
1935 out[2] = (cp >> 6 & 63) | 128;
1936 out[3] = (cp & 63) | 128;
1937 } else
1938 rc = 0;
1939
1940 out[rc] = '\0';
1941 return rc;
1942 }
1943
1944 /*
1945 * If the string contains escape sequences,
1946 * replace it with an allocated rendering and return 1,
1947 * such that the caller can free it after use.
1948 * Otherwise, do nothing and return 0.
1949 */
1950 static int
render_string(char ** public,size_t * psz)1951 render_string(char **public, size_t *psz)
1952 {
1953 const char *src, *scp, *addcp, *seq;
1954 char *dst;
1955 size_t ssz, dsz, addsz;
1956 char utfbuf[7], res[6];
1957 int seqlen, unicode;
1958
1959 res[0] = '\\';
1960 res[1] = '\t';
1961 res[2] = ASCII_NBRSP;
1962 res[3] = ASCII_HYPH;
1963 res[4] = ASCII_BREAK;
1964 res[5] = '\0';
1965
1966 src = scp = *public;
1967 ssz = *psz;
1968 dst = NULL;
1969 dsz = 0;
1970
1971 while (scp < src + *psz) {
1972
1973 /* Leave normal characters unchanged. */
1974
1975 if (strchr(res, *scp) == NULL) {
1976 if (dst != NULL)
1977 dst[dsz++] = *scp;
1978 scp++;
1979 continue;
1980 }
1981
1982 /*
1983 * Found something that requires replacing,
1984 * make sure we have a destination buffer.
1985 */
1986
1987 if (dst == NULL) {
1988 dst = mandoc_malloc(ssz + 1);
1989 dsz = scp - src;
1990 memcpy(dst, src, dsz);
1991 }
1992
1993 /* Handle single-char special characters. */
1994
1995 switch (*scp) {
1996 case '\\':
1997 break;
1998 case '\t':
1999 case ASCII_NBRSP:
2000 dst[dsz++] = ' ';
2001 scp++;
2002 continue;
2003 case ASCII_HYPH:
2004 dst[dsz++] = '-';
2005 /* FALLTHROUGH */
2006 case ASCII_BREAK:
2007 scp++;
2008 continue;
2009 default:
2010 abort();
2011 }
2012
2013 /*
2014 * Found an escape sequence.
2015 * Read past the slash, then parse it.
2016 * Ignore everything except characters.
2017 */
2018
2019 scp++;
2020 switch (mandoc_escape(&scp, &seq, &seqlen)) {
2021 case ESCAPE_UNICODE:
2022 unicode = mchars_num2uc(seq + 1, seqlen - 1);
2023 break;
2024 case ESCAPE_NUMBERED:
2025 unicode = mchars_num2char(seq, seqlen);
2026 break;
2027 case ESCAPE_SPECIAL:
2028 unicode = mchars_spec2cp(seq, seqlen);
2029 break;
2030 default:
2031 unicode = -1;
2032 break;
2033 }
2034 if (unicode <= 0)
2035 continue;
2036
2037 /*
2038 * Render the special character
2039 * as either UTF-8 or ASCII.
2040 */
2041
2042 if (write_utf8) {
2043 addsz = utf8(unicode, utfbuf);
2044 if (addsz == 0)
2045 continue;
2046 addcp = utfbuf;
2047 } else {
2048 addcp = mchars_uc2str(unicode);
2049 if (addcp == NULL)
2050 continue;
2051 if (*addcp == ASCII_NBRSP)
2052 addcp = " ";
2053 addsz = strlen(addcp);
2054 }
2055
2056 /* Copy the rendered glyph into the stream. */
2057
2058 ssz += addsz;
2059 dst = mandoc_realloc(dst, ssz + 1);
2060 memcpy(dst + dsz, addcp, addsz);
2061 dsz += addsz;
2062 }
2063 if (dst != NULL) {
2064 *public = dst;
2065 *psz = dsz;
2066 }
2067
2068 /* Trim trailing whitespace and NUL-terminate. */
2069
2070 while (*psz > 0 && (*public)[*psz - 1] == ' ')
2071 --*psz;
2072 if (dst != NULL) {
2073 (*public)[*psz] = '\0';
2074 return 1;
2075 } else
2076 return 0;
2077 }
2078
2079 static void
dbadd_mlink(const struct mlink * mlink)2080 dbadd_mlink(const struct mlink *mlink)
2081 {
2082 dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
2083 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
2084 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
2085 dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
2086 dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
2087 }
2088
2089 /*
2090 * Flush the current page's terms (and their bits) into the database.
2091 * Also, handle escape sequences at the last possible moment.
2092 */
2093 static void
dbadd(struct dba * dba,struct mpage * mpage)2094 dbadd(struct dba *dba, struct mpage *mpage)
2095 {
2096 struct mlink *mlink;
2097 struct str *key;
2098 char *cp;
2099 uint64_t mask;
2100 size_t i;
2101 unsigned int slot;
2102 int mustfree;
2103
2104 mlink = mpage->mlinks;
2105
2106 if (nodb) {
2107 for (key = ohash_first(&names, &slot); NULL != key;
2108 key = ohash_next(&names, &slot))
2109 free(key);
2110 for (key = ohash_first(&strings, &slot); NULL != key;
2111 key = ohash_next(&strings, &slot))
2112 free(key);
2113 if (0 == debug)
2114 return;
2115 while (NULL != mlink) {
2116 fputs(mlink->name, stdout);
2117 if (NULL == mlink->next ||
2118 strcmp(mlink->dsec, mlink->next->dsec) ||
2119 strcmp(mlink->fsec, mlink->next->fsec) ||
2120 strcmp(mlink->arch, mlink->next->arch)) {
2121 putchar('(');
2122 if ('\0' == *mlink->dsec)
2123 fputs(mlink->fsec, stdout);
2124 else
2125 fputs(mlink->dsec, stdout);
2126 if ('\0' != *mlink->arch)
2127 printf("/%s", mlink->arch);
2128 putchar(')');
2129 }
2130 mlink = mlink->next;
2131 if (NULL != mlink)
2132 fputs(", ", stdout);
2133 }
2134 printf(" - %s\n", mpage->desc);
2135 return;
2136 }
2137
2138 if (debug)
2139 say(mlink->file, "Adding to database");
2140
2141 cp = mpage->desc;
2142 i = strlen(cp);
2143 mustfree = render_string(&cp, &i);
2144 mpage->dba = dba_page_new(dba->pages,
2145 *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2146 cp, mlink->file, mpage->form);
2147 if (mustfree)
2148 free(cp);
2149 dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
2150
2151 while (mlink != NULL) {
2152 dbadd_mlink(mlink);
2153 mlink = mlink->next;
2154 }
2155
2156 for (key = ohash_first(&names, &slot); NULL != key;
2157 key = ohash_next(&names, &slot)) {
2158 assert(key->mpage == mpage);
2159 dba_page_alias(mpage->dba, key->key, key->mask);
2160 free(key);
2161 }
2162 for (key = ohash_first(&strings, &slot); NULL != key;
2163 key = ohash_next(&strings, &slot)) {
2164 assert(key->mpage == mpage);
2165 i = 0;
2166 for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2167 if (key->mask & mask)
2168 dba_macro_add(dba->macros, i,
2169 key->key, mpage->dba);
2170 i++;
2171 }
2172 free(key);
2173 }
2174 }
2175
2176 static void
dbprune(struct dba * dba)2177 dbprune(struct dba *dba)
2178 {
2179 struct dba_array *page, *files;
2180 char *file;
2181
2182 dba_array_FOREACH(dba->pages, page) {
2183 files = dba_array_get(page, DBP_FILE);
2184 dba_array_FOREACH(files, file) {
2185 if (*file < ' ')
2186 file++;
2187 if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2188 file)) != NULL) {
2189 if (debug)
2190 say(file, "Deleting from database");
2191 dba_array_del(dba->pages);
2192 break;
2193 }
2194 }
2195 }
2196 }
2197
2198 /*
2199 * Write the database from memory to disk.
2200 */
2201 static void
dbwrite(struct dba * dba)2202 dbwrite(struct dba *dba)
2203 {
2204 struct stat sb1, sb2;
2205 char tfn[33], *cp1, *cp2;
2206 off_t i;
2207 int fd1, fd2;
2208
2209 /*
2210 * Do not write empty databases, and delete existing ones
2211 * when makewhatis -u causes them to become empty.
2212 */
2213
2214 dba_array_start(dba->pages);
2215 if (dba_array_next(dba->pages) == NULL) {
2216 if (unlink(MANDOC_DB) == -1 && errno != ENOENT)
2217 say(MANDOC_DB, "&unlink");
2218 return;
2219 }
2220
2221 /*
2222 * Build the database in a temporary file,
2223 * then atomically move it into place.
2224 */
2225
2226 if (dba_write(MANDOC_DB "~", dba) != -1) {
2227 if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2228 exitcode = (int)MANDOCLEVEL_SYSERR;
2229 say(MANDOC_DB, "&rename");
2230 unlink(MANDOC_DB "~");
2231 }
2232 return;
2233 }
2234
2235 /*
2236 * We lack write permission and cannot replace the database
2237 * file, but let's at least check whether the data changed.
2238 */
2239
2240 (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2241 if (mkdtemp(tfn) == NULL) {
2242 exitcode = (int)MANDOCLEVEL_SYSERR;
2243 say("", "&%s", tfn);
2244 return;
2245 }
2246 cp1 = cp2 = MAP_FAILED;
2247 fd1 = fd2 = -1;
2248 (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2249 if (dba_write(tfn, dba) == -1) {
2250 say(tfn, "&dba_write");
2251 goto err;
2252 }
2253 if ((fd1 = open(MANDOC_DB, O_RDONLY)) == -1) {
2254 say(MANDOC_DB, "&open");
2255 goto err;
2256 }
2257 if ((fd2 = open(tfn, O_RDONLY)) == -1) {
2258 say(tfn, "&open");
2259 goto err;
2260 }
2261 if (fstat(fd1, &sb1) == -1) {
2262 say(MANDOC_DB, "&fstat");
2263 goto err;
2264 }
2265 if (fstat(fd2, &sb2) == -1) {
2266 say(tfn, "&fstat");
2267 goto err;
2268 }
2269 if (sb1.st_size != sb2.st_size)
2270 goto err;
2271 if ((cp1 = mmap(NULL, sb1.st_size, PROT_READ, MAP_PRIVATE,
2272 fd1, 0)) == MAP_FAILED) {
2273 say(MANDOC_DB, "&mmap");
2274 goto err;
2275 }
2276 if ((cp2 = mmap(NULL, sb2.st_size, PROT_READ, MAP_PRIVATE,
2277 fd2, 0)) == MAP_FAILED) {
2278 say(tfn, "&mmap");
2279 goto err;
2280 }
2281 for (i = 0; i < sb1.st_size; i++)
2282 if (cp1[i] != cp2[i])
2283 goto err;
2284 goto out;
2285
2286 err:
2287 exitcode = (int)MANDOCLEVEL_SYSERR;
2288 say(MANDOC_DB, "Data changed, but cannot replace database");
2289
2290 out:
2291 if (cp1 != MAP_FAILED)
2292 munmap(cp1, sb1.st_size);
2293 if (cp2 != MAP_FAILED)
2294 munmap(cp2, sb2.st_size);
2295 if (fd1 != -1)
2296 close(fd1);
2297 if (fd2 != -1)
2298 close(fd2);
2299 unlink(tfn);
2300 *strrchr(tfn, '/') = '\0';
2301 rmdir(tfn);
2302 }
2303
2304 static int
set_basedir(const char * targetdir,int report_baddir)2305 set_basedir(const char *targetdir, int report_baddir)
2306 {
2307 static char startdir[PATH_MAX];
2308 static int getcwd_status; /* 1 = ok, 2 = failure */
2309 static int chdir_status; /* 1 = changed directory */
2310
2311 /*
2312 * Remember the original working directory, if possible.
2313 * This will be needed if the second or a later directory
2314 * on the command line is given as a relative path.
2315 * Do not error out if the current directory is not
2316 * searchable: Maybe it won't be needed after all.
2317 */
2318 if (getcwd_status == 0) {
2319 if (getcwd(startdir, sizeof(startdir)) == NULL) {
2320 getcwd_status = 2;
2321 (void)strlcpy(startdir, strerror(errno),
2322 sizeof(startdir));
2323 } else
2324 getcwd_status = 1;
2325 }
2326
2327 /*
2328 * We are leaving the old base directory.
2329 * Do not use it any longer, not even for messages.
2330 */
2331 *basedir = '\0';
2332 basedir_len = 0;
2333
2334 /*
2335 * If and only if the directory was changed earlier and
2336 * the next directory to process is given as a relative path,
2337 * first go back, or bail out if that is impossible.
2338 */
2339 if (chdir_status && *targetdir != '/') {
2340 if (getcwd_status == 2) {
2341 exitcode = (int)MANDOCLEVEL_SYSERR;
2342 say("", "getcwd: %s", startdir);
2343 return 0;
2344 }
2345 if (chdir(startdir) == -1) {
2346 exitcode = (int)MANDOCLEVEL_SYSERR;
2347 say("", "&chdir %s", startdir);
2348 return 0;
2349 }
2350 }
2351
2352 /*
2353 * Always resolve basedir to the canonicalized absolute
2354 * pathname and append a trailing slash, such that
2355 * we can reliably check whether files are inside.
2356 */
2357 if (realpath(targetdir, basedir) == NULL) {
2358 if (report_baddir || errno != ENOENT) {
2359 exitcode = (int)MANDOCLEVEL_BADARG;
2360 say("", "&%s: realpath", targetdir);
2361 }
2362 *basedir = '\0';
2363 return 0;
2364 } else if (chdir(basedir) == -1) {
2365 if (report_baddir || errno != ENOENT) {
2366 exitcode = (int)MANDOCLEVEL_BADARG;
2367 say("", "&chdir");
2368 }
2369 *basedir = '\0';
2370 return 0;
2371 }
2372 chdir_status = 1;
2373 basedir_len = strlen(basedir);
2374 if (basedir[basedir_len - 1] != '/') {
2375 if (basedir_len >= PATH_MAX - 1) {
2376 exitcode = (int)MANDOCLEVEL_SYSERR;
2377 say("", "Filename too long");
2378 *basedir = '\0';
2379 basedir_len = 0;
2380 return 0;
2381 }
2382 basedir[basedir_len++] = '/';
2383 basedir[basedir_len] = '\0';
2384 }
2385 return 1;
2386 }
2387
2388 #ifdef READ_ALLOWED_PATH
2389 static int
read_allowed(const char * candidate)2390 read_allowed(const char *candidate)
2391 {
2392 const char *cp;
2393 size_t len;
2394
2395 for (cp = READ_ALLOWED_PATH;; cp += len) {
2396 while (*cp == ':')
2397 cp++;
2398 if (*cp == '\0')
2399 return 0;
2400 len = strcspn(cp, ":");
2401 if (strncmp(candidate, cp, len) == 0)
2402 return 1;
2403 }
2404 }
2405 #endif
2406
2407 static void
say(const char * file,const char * format,...)2408 say(const char *file, const char *format, ...)
2409 {
2410 va_list ap;
2411 int use_errno;
2412
2413 if (*basedir != '\0')
2414 fprintf(stderr, "%s", basedir);
2415 if (*basedir != '\0' && *file != '\0')
2416 fputc('/', stderr);
2417 if (*file != '\0')
2418 fprintf(stderr, "%s", file);
2419
2420 use_errno = 1;
2421 if (format != NULL) {
2422 switch (*format) {
2423 case '&':
2424 format++;
2425 break;
2426 case '\0':
2427 format = NULL;
2428 break;
2429 default:
2430 use_errno = 0;
2431 break;
2432 }
2433 }
2434 if (format != NULL) {
2435 if (*basedir != '\0' || *file != '\0')
2436 fputs(": ", stderr);
2437 va_start(ap, format);
2438 vfprintf(stderr, format, ap);
2439 va_end(ap);
2440 }
2441 if (use_errno) {
2442 if (*basedir != '\0' || *file != '\0' || format != NULL)
2443 fputs(": ", stderr);
2444 perror(NULL);
2445 } else
2446 fputc('\n', stderr);
2447 }
2448