?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
mandoc: update to 1.14.1
[unleashed.git] / bin / mandoc / mandocdb.c
blob3b26ca964964929af2d8e678be72b4a44cb205c4
1 /* $Id: mandocdb.c,v 1.244 2017/02/17 14:45:55 schwarze Exp $ */
2 /*
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2017 Ingo Schwarze <schwarze@openbsd.org>
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 #include "config.h"
20
21 #include <sys/types.h>
22 #include <sys/stat.h>
23 #include <sys/wait.h>
24
25 #include <assert.h>
26 #include <ctype.h>
27 #if HAVE_ERR
28 #include <err.h>
29 #endif
30 #include <errno.h>
31 #include <fcntl.h>
32 #if HAVE_FTS
33 #include <fts.h>
34 #else
35 #include "compat_fts.h"
36 #endif
37 #include <limits.h>
38 #if HAVE_SANDBOX_INIT
39 #include <sandbox.h>
40 #endif
41 #include <stdarg.h>
42 #include <stddef.h>
43 #include <stdio.h>
44 #include <stdint.h>
45 #include <stdlib.h>
46 #include <string.h>
47 #include <unistd.h>
48
49 #include "mandoc_aux.h"
50 #include "mandoc_ohash.h"
51 #include "mandoc.h"
52 #include "roff.h"
53 #include "mdoc.h"
54 #include "man.h"
55 #include "manconf.h"
56 #include "mansearch.h"
57 #include "dba_array.h"
58 #include "dba.h"
59
60 extern const char *const mansearch_keynames[];
61
62 enum op {
63 OP_DEFAULT = 0, /* new dbs from dir list or default config */
64 OP_CONFFILE, /* new databases from custom config file */
65 OP_UPDATE, /* delete/add entries in existing database */
66 OP_DELETE, /* delete entries from existing database */
67 OP_TEST /* change no databases, report potential problems */
68 };
69
70 struct str {
71 const struct mpage *mpage; /* if set, the owning parse */
72 uint64_t mask; /* bitmask in sequence */
73 char key[]; /* rendered text */
74 };
75
76 struct inodev {
77 ino_t st_ino;
78 dev_t st_dev;
79 };
80
81 struct mpage {
82 struct inodev inodev; /* used for hashing routine */
83 struct dba_array *dba;
84 char *sec; /* section from file content */
85 char *arch; /* architecture from file content */
86 char *title; /* title from file content */
87 char *desc; /* description from file content */
88 struct mpage *next; /* singly linked list */
89 struct mlink *mlinks; /* singly linked list */
90 int name_head_done;
91 enum form form; /* format from file content */
92 };
93
94 struct mlink {
95 char file[PATH_MAX]; /* filename rel. to manpath */
96 char *dsec; /* section from directory */
97 char *arch; /* architecture from directory */
98 char *name; /* name from file name (not empty) */
99 char *fsec; /* section from file name suffix */
100 struct mlink *next; /* singly linked list */
101 struct mpage *mpage; /* parent */
102 int gzip; /* filename has a .gz suffix */
103 enum form dform; /* format from directory */
104 enum form fform; /* format from file name suffix */
105 };
106
107 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
108 const struct roff_node *);
109
110 struct mdoc_handler {
111 mdoc_fp fp; /* optional handler */
112 uint64_t mask; /* set unless handler returns 0 */
113 int taboo; /* node flags that must not be set */
114 };
115
116
117 int mandocdb(int, char *[]);
118
119 static void dbadd(struct dba *, struct mpage *);
120 static void dbadd_mlink(const struct mlink *mlink);
121 static void dbprune(struct dba *);
122 static void dbwrite(struct dba *);
123 static void filescan(const char *);
124 #if HAVE_FTS_COMPARE_CONST
125 static int fts_compare(const FTSENT *const *, const FTSENT *const *);
126 #else
127 static int fts_compare(const FTSENT **, const FTSENT **);
128 #endif
129 static void mlink_add(struct mlink *, const struct stat *);
130 static void mlink_check(struct mpage *, struct mlink *);
131 static void mlink_free(struct mlink *);
132 static void mlinks_undupe(struct mpage *);
133 static void mpages_free(void);
134 static void mpages_merge(struct dba *, struct mparse *);
135 static void parse_cat(struct mpage *, int);
136 static void parse_man(struct mpage *, const struct roff_meta *,
137 const struct roff_node *);
138 static void parse_mdoc(struct mpage *, const struct roff_meta *,
139 const struct roff_node *);
140 static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
141 const struct roff_node *);
142 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
143 const struct roff_node *);
144 static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
145 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
146 const struct roff_node *);
147 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
148 const struct roff_node *);
149 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
150 const struct roff_node *);
151 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
152 const struct roff_node *);
153 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
154 const struct roff_node *);
155 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
156 const struct roff_node *);
157 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
158 const struct roff_node *);
159 static void putkey(const struct mpage *, char *, uint64_t);
160 static void putkeys(const struct mpage *, char *, size_t, uint64_t);
161 static void putmdockey(const struct mpage *,
162 const struct roff_node *, uint64_t, int);
163 static int render_string(char **, size_t *);
164 static void say(const char *, const char *, ...)
165 __attribute__((__format__ (__printf__, 2, 3)));
166 static int set_basedir(const char *, int);
167 static int treescan(void);
168 static size_t utf8(unsigned int, char [7]);
169
170 static int nodb; /* no database changes */
171 static int mparse_options; /* abort the parse early */
172 static int use_all; /* use all found files */
173 static int debug; /* print what we're doing */
174 static int warnings; /* warn about crap */
175 static int write_utf8; /* write UTF-8 output; else ASCII */
176 static int exitcode; /* to be returned by main */
177 static enum op op; /* operational mode */
178 static char basedir[PATH_MAX]; /* current base directory */
179 static struct mpage *mpage_head; /* list of distinct manual pages */
180 static struct ohash mpages; /* table of distinct manual pages */
181 static struct ohash mlinks; /* table of directory entries */
182 static struct ohash names; /* table of all names */
183 static struct ohash strings; /* table of all strings */
184 static uint64_t name_mask;
185
186 static const struct mdoc_handler mdocs[MDOC_MAX] = {
187 { NULL, 0, 0 }, /* Ap */
188 { NULL, 0, NODE_NOPRT }, /* Dd */
189 { NULL, 0, NODE_NOPRT }, /* Dt */
190 { NULL, 0, NODE_NOPRT }, /* Os */
191 { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
192 { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
193 { NULL, 0, 0 }, /* Pp */
194 { NULL, 0, 0 }, /* D1 */
195 { NULL, 0, 0 }, /* Dl */
196 { NULL, 0, 0 }, /* Bd */
197 { NULL, 0, 0 }, /* Ed */
198 { NULL, 0, 0 }, /* Bl */
199 { NULL, 0, 0 }, /* El */
200 { NULL, 0, 0 }, /* It */
201 { NULL, 0, 0 }, /* Ad */
202 { NULL, TYPE_An, 0 }, /* An */
203 { NULL, TYPE_Ar, 0 }, /* Ar */
204 { NULL, TYPE_Cd, 0 }, /* Cd */
205 { NULL, TYPE_Cm, 0 }, /* Cm */
206 { NULL, TYPE_Dv, 0 }, /* Dv */
207 { NULL, TYPE_Er, 0 }, /* Er */
208 { NULL, TYPE_Ev, 0 }, /* Ev */
209 { NULL, 0, 0 }, /* Ex */
210 { NULL, TYPE_Fa, 0 }, /* Fa */
211 { parse_mdoc_Fd, 0, 0 }, /* Fd */
212 { NULL, TYPE_Fl, 0 }, /* Fl */
213 { parse_mdoc_Fn, 0, 0 }, /* Fn */
214 { NULL, TYPE_Ft, 0 }, /* Ft */
215 { NULL, TYPE_Ic, 0 }, /* Ic */
216 { NULL, TYPE_In, 0 }, /* In */
217 { NULL, TYPE_Li, 0 }, /* Li */
218 { parse_mdoc_Nd, 0, 0 }, /* Nd */
219 { parse_mdoc_Nm, 0, 0 }, /* Nm */
220 { NULL, 0, 0 }, /* Op */
221 { NULL, 0, 0 }, /* Ot */
222 { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
223 { NULL, 0, 0 }, /* Rv */
224 { NULL, TYPE_St, 0 }, /* St */
225 { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
226 { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
227 { parse_mdoc_Xr, 0, 0 }, /* Xr */
228 { NULL, 0, 0 }, /* %A */
229 { NULL, 0, 0 }, /* %B */
230 { NULL, 0, 0 }, /* %D */
231 { NULL, 0, 0 }, /* %I */
232 { NULL, 0, 0 }, /* %J */
233 { NULL, 0, 0 }, /* %N */
234 { NULL, 0, 0 }, /* %O */
235 { NULL, 0, 0 }, /* %P */
236 { NULL, 0, 0 }, /* %R */
237 { NULL, 0, 0 }, /* %T */
238 { NULL, 0, 0 }, /* %V */
239 { NULL, 0, 0 }, /* Ac */
240 { NULL, 0, 0 }, /* Ao */
241 { NULL, 0, 0 }, /* Aq */
242 { NULL, TYPE_At, 0 }, /* At */
243 { NULL, 0, 0 }, /* Bc */
244 { NULL, 0, 0 }, /* Bf */
245 { NULL, 0, 0 }, /* Bo */
246 { NULL, 0, 0 }, /* Bq */
247 { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
248 { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
249 { NULL, 0, 0 }, /* Db */
250 { NULL, 0, 0 }, /* Dc */
251 { NULL, 0, 0 }, /* Do */
252 { NULL, 0, 0 }, /* Dq */
253 { NULL, 0, 0 }, /* Ec */
254 { NULL, 0, 0 }, /* Ef */
255 { NULL, TYPE_Em, 0 }, /* Em */
256 { NULL, 0, 0 }, /* Eo */
257 { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
258 { NULL, TYPE_Ms, 0 }, /* Ms */
259 { NULL, 0, 0 }, /* No */
260 { NULL, 0, 0 }, /* Ns */
261 { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
262 { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
263 { NULL, 0, 0 }, /* Pc */
264 { NULL, 0, 0 }, /* Pf */
265 { NULL, 0, 0 }, /* Po */
266 { NULL, 0, 0 }, /* Pq */
267 { NULL, 0, 0 }, /* Qc */
268 { NULL, 0, 0 }, /* Ql */
269 { NULL, 0, 0 }, /* Qo */
270 { NULL, 0, 0 }, /* Qq */
271 { NULL, 0, 0 }, /* Re */
272 { NULL, 0, 0 }, /* Rs */
273 { NULL, 0, 0 }, /* Sc */
274 { NULL, 0, 0 }, /* So */
275 { NULL, 0, 0 }, /* Sq */
276 { NULL, 0, 0 }, /* Sm */
277 { NULL, 0, 0 }, /* Sx */
278 { NULL, TYPE_Sy, 0 }, /* Sy */
279 { NULL, TYPE_Tn, 0 }, /* Tn */
280 { NULL, 0, NODE_NOSRC }, /* Ux */
281 { NULL, 0, 0 }, /* Xc */
282 { NULL, 0, 0 }, /* Xo */
283 { parse_mdoc_Fo, 0, 0 }, /* Fo */
284 { NULL, 0, 0 }, /* Fc */
285 { NULL, 0, 0 }, /* Oo */
286 { NULL, 0, 0 }, /* Oc */
287 { NULL, 0, 0 }, /* Bk */
288 { NULL, 0, 0 }, /* Ek */
289 { NULL, 0, 0 }, /* Bt */
290 { NULL, 0, 0 }, /* Hf */
291 { NULL, 0, 0 }, /* Fr */
292 { NULL, 0, 0 }, /* Ud */
293 { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
294 { NULL, 0, 0 }, /* Lp */
295 { NULL, TYPE_Lk, 0 }, /* Lk */
296 { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
297 { NULL, 0, 0 }, /* Brq */
298 { NULL, 0, 0 }, /* Bro */
299 { NULL, 0, 0 }, /* Brc */
300 { NULL, 0, 0 }, /* %C */
301 { NULL, 0, 0 }, /* Es */
302 { NULL, 0, 0 }, /* En */
303 { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
304 { NULL, 0, 0 }, /* %Q */
305 { NULL, 0, 0 }, /* br */
306 { NULL, 0, 0 }, /* sp */
307 { NULL, 0, 0 }, /* %U */
308 { NULL, 0, 0 }, /* Ta */
309 { NULL, 0, 0 }, /* ll */
310 };
311
312
313 int
314 mandocdb(int argc, char *argv[])
315 {
316 struct manconf conf;
317 struct mparse *mp;
318 struct dba *dba;
319 const char *path_arg, *progname;
320 size_t j, sz;
321 int ch, i;
322
323 #if HAVE_PLEDGE
324 if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
325 warn("pledge");
326 return (int)MANDOCLEVEL_SYSERR;
327 }
328 #endif
329
330 #if HAVE_SANDBOX_INIT
331 if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
332 warnx("sandbox_init");
333 return (int)MANDOCLEVEL_SYSERR;
334 }
335 #endif
336
337 memset(&conf, 0, sizeof(conf));
338
339 /*
340 * We accept a few different invocations.
341 * The CHECKOP macro makes sure that invocation styles don't
342 * clobber each other.
343 */
344 #define CHECKOP(_op, _ch) do \
345 if (OP_DEFAULT != (_op)) { \
346 warnx("-%c: Conflicting option", (_ch)); \
347 goto usage; \
348 } while (/*CONSTCOND*/0)
349
350 path_arg = NULL;
351 op = OP_DEFAULT;
352
353 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
354 switch (ch) {
355 case 'a':
356 use_all = 1;
357 break;
358 case 'C':
359 CHECKOP(op, ch);
360 path_arg = optarg;
361 op = OP_CONFFILE;
362 break;
363 case 'D':
364 debug++;
365 break;
366 case 'd':
367 CHECKOP(op, ch);
368 path_arg = optarg;
369 op = OP_UPDATE;
370 break;
371 case 'n':
372 nodb = 1;
373 break;
374 case 'p':
375 warnings = 1;
376 break;
377 case 'Q':
378 mparse_options |= MPARSE_QUICK;
379 break;
380 case 'T':
381 if (strcmp(optarg, "utf8")) {
382 warnx("-T%s: Unsupported output format",
383 optarg);
384 goto usage;
385 }
386 write_utf8 = 1;
387 break;
388 case 't':
389 CHECKOP(op, ch);
390 dup2(STDOUT_FILENO, STDERR_FILENO);
391 op = OP_TEST;
392 nodb = warnings = 1;
393 break;
394 case 'u':
395 CHECKOP(op, ch);
396 path_arg = optarg;
397 op = OP_DELETE;
398 break;
399 case 'v':
400 /* Compatibility with espie@'s makewhatis. */
401 break;
402 default:
403 goto usage;
404 }
405
406 argc -= optind;
407 argv += optind;
408
409 #if HAVE_PLEDGE
410 if (nodb) {
411 if (pledge("stdio rpath", NULL) == -1) {
412 warn("pledge");
413 return (int)MANDOCLEVEL_SYSERR;
414 }
415 }
416 #endif
417
418 if (OP_CONFFILE == op && argc > 0) {
419 warnx("-C: Too many arguments");
420 goto usage;
421 }
422
423 exitcode = (int)MANDOCLEVEL_OK;
424 mchars_alloc();
425 mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
426 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
427 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
428
429 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
430
431 /*
432 * Most of these deal with a specific directory.
433 * Jump into that directory first.
434 */
435 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
436 goto out;
437
438 dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
439 if (dba != NULL) {
440 /*
441 * The existing database is usable. Process
442 * all files specified on the command-line.
443 */
444 #if HAVE_PLEDGE
445 if (!nodb) {
446 if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
447 warn("pledge");
448 exitcode = (int)MANDOCLEVEL_SYSERR;
449 goto out;
450 }
451 }
452 #endif
453 use_all = 1;
454 for (i = 0; i < argc; i++)
455 filescan(argv[i]);
456 if (nodb == 0)
457 dbprune(dba);
458 } else {
459 /* Database missing or corrupt. */
460 if (op != OP_UPDATE || errno != ENOENT)
461 say(MANDOC_DB, "%s: Automatically recreating"
462 " from scratch", strerror(errno));
463 exitcode = (int)MANDOCLEVEL_OK;
464 op = OP_DEFAULT;
465 if (0 == treescan())
466 goto out;
467 dba = dba_new(128);
468 }
469 if (OP_DELETE != op)
470 mpages_merge(dba, mp);
471 if (nodb == 0)
472 dbwrite(dba);
473 dba_free(dba);
474 } else {
475 /*
476 * If we have arguments, use them as our manpaths.
477 * If we don't, use man.conf(5).
478 */
479 if (argc > 0) {
480 conf.manpath.paths = mandoc_reallocarray(NULL,
481 argc, sizeof(char *));
482 conf.manpath.sz = (size_t)argc;
483 for (i = 0; i < argc; i++)
484 conf.manpath.paths[i] = mandoc_strdup(argv[i]);
485 } else
486 manconf_parse(&conf, path_arg, NULL, NULL);
487
488 if (conf.manpath.sz == 0) {
489 exitcode = (int)MANDOCLEVEL_BADARG;
490 say("", "Empty manpath");
491 }
492
493 /*
494 * First scan the tree rooted at a base directory, then
495 * build a new database and finally move it into place.
496 * Ignore zero-length directories and strip trailing
497 * slashes.
498 */
499 for (j = 0; j < conf.manpath.sz; j++) {
500 sz = strlen(conf.manpath.paths[j]);
501 if (sz && conf.manpath.paths[j][sz - 1] == '/')
502 conf.manpath.paths[j][--sz] = '\0';
503 if (0 == sz)
504 continue;
505
506 if (j) {
507 mandoc_ohash_init(&mpages, 6,
508 offsetof(struct mpage, inodev));
509 mandoc_ohash_init(&mlinks, 6,
510 offsetof(struct mlink, file));
511 }
512
513 if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
514 continue;
515 if (0 == treescan())
516 continue;
517 dba = dba_new(128);
518 mpages_merge(dba, mp);
519 if (nodb == 0)
520 dbwrite(dba);
521 dba_free(dba);
522
523 if (j + 1 < conf.manpath.sz) {
524 mpages_free();
525 ohash_delete(&mpages);
526 ohash_delete(&mlinks);
527 }
528 }
529 }
530 out:
531 manconf_free(&conf);
532 mparse_free(mp);
533 mchars_free();
534 mpages_free();
535 ohash_delete(&mpages);
536 ohash_delete(&mlinks);
537 return exitcode;
538 usage:
539 progname = getprogname();
540 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
541 " %s [-aDnpQ] [-Tutf8] dir ...\n"
542 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
543 " %s [-Dnp] -u dir [file ...]\n"
544 " %s [-Q] -t file ...\n",
545 progname, progname, progname, progname, progname);
546
547 return (int)MANDOCLEVEL_BADARG;
548 }
549
550 /*
551 * To get a singly linked list in alpha order while inserting entries
552 * at the beginning, process directory entries in reverse alpha order.
553 */
554 static int
555 #if HAVE_FTS_COMPARE_CONST
556 fts_compare(const FTSENT *const *a, const FTSENT *const *b)
557 #else
558 fts_compare(const FTSENT **a, const FTSENT **b)
559 #endif
560 {
561 return -strcmp((*a)->fts_name, (*b)->fts_name);
562 }
563
564 /*
565 * Scan a directory tree rooted at "basedir" for manpages.
566 * We use fts(), scanning directory parts along the way for clues to our
567 * section and architecture.
568 *
569 * If use_all has been specified, grok all files.
570 * If not, sanitise paths to the following:
571 *
572 * [./]man*[/<arch>]/<name>.<section>
573 * or
574 * [./]cat<section>[/<arch>]/<name>.0
575 *
576 * TODO: accommodate for multi-language directories.
577 */
578 static int
579 treescan(void)
580 {
581 char buf[PATH_MAX];
582 FTS *f;
583 FTSENT *ff;
584 struct mlink *mlink;
585 int gzip;
586 enum form dform;
587 char *dsec, *arch, *fsec, *cp;
588 const char *path;
589 const char *argv[2];
590
591 argv[0] = ".";
592 argv[1] = NULL;
593
594 f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
595 fts_compare);
596 if (f == NULL) {
597 exitcode = (int)MANDOCLEVEL_SYSERR;
598 say("", "&fts_open");
599 return 0;
600 }
601
602 dsec = arch = NULL;
603 dform = FORM_NONE;
604
605 while ((ff = fts_read(f)) != NULL) {
606 path = ff->fts_path + 2;
607 switch (ff->fts_info) {
608
609 /*
610 * Symbolic links require various sanity checks,
611 * then get handled just like regular files.
612 */
613 case FTS_SL:
614 if (realpath(path, buf) == NULL) {
615 if (warnings)
616 say(path, "&realpath");
617 continue;
618 }
619 if (strstr(buf, basedir) != buf
620 #ifdef HOMEBREWDIR
621 && strstr(buf, HOMEBREWDIR) != buf
622 #endif
623 ) {
624 if (warnings) say("",
625 "%s: outside base directory", buf);
626 continue;
627 }
628 /* Use logical inode to avoid mpages dupe. */
629 if (stat(path, ff->fts_statp) == -1) {
630 if (warnings)
631 say(path, "&stat");
632 continue;
633 }
634 /* FALLTHROUGH */
635
636 /*
637 * If we're a regular file, add an mlink by using the
638 * stored directory data and handling the filename.
639 */
640 case FTS_F:
641 if ( ! strcmp(path, MANDOC_DB))
642 continue;
643 if ( ! use_all && ff->fts_level < 2) {
644 if (warnings)
645 say(path, "Extraneous file");
646 continue;
647 }
648 gzip = 0;
649 fsec = NULL;
650 while (fsec == NULL) {
651 fsec = strrchr(ff->fts_name, '.');
652 if (fsec == NULL || strcmp(fsec+1, "gz"))
653 break;
654 gzip = 1;
655 *fsec = '\0';
656 fsec = NULL;
657 }
658 if (fsec == NULL) {
659 if ( ! use_all) {
660 if (warnings)
661 say(path,
662 "No filename suffix");
663 continue;
664 }
665 } else if ( ! strcmp(++fsec, "html")) {
666 if (warnings)
667 say(path, "Skip html");
668 continue;
669 } else if ( ! strcmp(fsec, "ps")) {
670 if (warnings)
671 say(path, "Skip ps");
672 continue;
673 } else if ( ! strcmp(fsec, "pdf")) {
674 if (warnings)
675 say(path, "Skip pdf");
676 continue;
677 } else if ( ! use_all &&
678 ((dform == FORM_SRC &&
679 strncmp(fsec, dsec, strlen(dsec))) ||
680 (dform == FORM_CAT && strcmp(fsec, "0")))) {
681 if (warnings)
682 say(path, "Wrong filename suffix");
683 continue;
684 } else
685 fsec[-1] = '\0';
686
687 mlink = mandoc_calloc(1, sizeof(struct mlink));
688 if (strlcpy(mlink->file, path,
689 sizeof(mlink->file)) >=
690 sizeof(mlink->file)) {
691 say(path, "Filename too long");
692 free(mlink);
693 continue;
694 }
695 mlink->dform = dform;
696 mlink->dsec = dsec;
697 mlink->arch = arch;
698 mlink->name = ff->fts_name;
699 mlink->fsec = fsec;
700 mlink->gzip = gzip;
701 mlink_add(mlink, ff->fts_statp);
702 continue;
703
704 case FTS_D:
705 case FTS_DP:
706 break;
707
708 default:
709 if (warnings)
710 say(path, "Not a regular file");
711 continue;
712 }
713
714 switch (ff->fts_level) {
715 case 0:
716 /* Ignore the root directory. */
717 break;
718 case 1:
719 /*
720 * This might contain manX/ or catX/.
721 * Try to infer this from the name.
722 * If we're not in use_all, enforce it.
723 */
724 cp = ff->fts_name;
725 if (ff->fts_info == FTS_DP) {
726 dform = FORM_NONE;
727 dsec = NULL;
728 break;
729 }
730
731 if ( ! strncmp(cp, "man", 3)) {
732 dform = FORM_SRC;
733 dsec = cp + 3;
734 } else if ( ! strncmp(cp, "cat", 3)) {
735 dform = FORM_CAT;
736 dsec = cp + 3;
737 } else {
738 dform = FORM_NONE;
739 dsec = NULL;
740 }
741
742 if (dsec != NULL || use_all)
743 break;
744
745 if (warnings)
746 say(path, "Unknown directory part");
747 fts_set(f, ff, FTS_SKIP);
748 break;
749 case 2:
750 /*
751 * Possibly our architecture.
752 * If we're descending, keep tabs on it.
753 */
754 if (ff->fts_info != FTS_DP && dsec != NULL)
755 arch = ff->fts_name;
756 else
757 arch = NULL;
758 break;
759 default:
760 if (ff->fts_info == FTS_DP || use_all)
761 break;
762 if (warnings)
763 say(path, "Extraneous directory part");
764 fts_set(f, ff, FTS_SKIP);
765 break;
766 }
767 }
768
769 fts_close(f);
770 return 1;
771 }
772
773 /*
774 * Add a file to the mlinks table.
775 * Do not verify that it's a "valid" looking manpage (we'll do that
776 * later).
777 *
778 * Try to infer the manual section, architecture, and page name from the
779 * path, assuming it looks like
780 *
781 * [./]man*[/<arch>]/<name>.<section>
782 * or
783 * [./]cat<section>[/<arch>]/<name>.0
784 *
785 * See treescan() for the fts(3) version of this.
786 */
787 static void
788 filescan(const char *file)
789 {
790 char buf[PATH_MAX];
791 struct stat st;
792 struct mlink *mlink;
793 char *p, *start;
794
795 assert(use_all);
796
797 if (0 == strncmp(file, "./", 2))
798 file += 2;
799
800 /*
801 * We have to do lstat(2) before realpath(3) loses
802 * the information whether this is a symbolic link.
803 * We need to know that because for symbolic links,
804 * we want to use the orginal file name, while for
805 * regular files, we want to use the real path.
806 */
807 if (-1 == lstat(file, &st)) {
808 exitcode = (int)MANDOCLEVEL_BADARG;
809 say(file, "&lstat");
810 return;
811 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
812 exitcode = (int)MANDOCLEVEL_BADARG;
813 say(file, "Not a regular file");
814 return;
815 }
816
817 /*
818 * We have to resolve the file name to the real path
819 * in any case for the base directory check.
820 */
821 if (NULL == realpath(file, buf)) {
822 exitcode = (int)MANDOCLEVEL_BADARG;
823 say(file, "&realpath");
824 return;
825 }
826
827 if (OP_TEST == op)
828 start = buf;
829 else if (strstr(buf, basedir) == buf)
830 start = buf + strlen(basedir);
831 #ifdef HOMEBREWDIR
832 else if (strstr(buf, HOMEBREWDIR) == buf)
833 start = buf;
834 #endif
835 else {
836 exitcode = (int)MANDOCLEVEL_BADARG;
837 say("", "%s: outside base directory", buf);
838 return;
839 }
840
841 /*
842 * Now we are sure the file is inside our tree.
843 * If it is a symbolic link, ignore the real path
844 * and use the original name.
845 * This implies passing stuff like "cat1/../man1/foo.1"
846 * on the command line won't work. So don't do that.
847 * Note the stat(2) can still fail if the link target
848 * doesn't exist.
849 */
850 if (S_IFLNK & st.st_mode) {
851 if (-1 == stat(buf, &st)) {
852 exitcode = (int)MANDOCLEVEL_BADARG;
853 say(file, "&stat");
854 return;
855 }
856 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
857 say(file, "Filename too long");
858 return;
859 }
860 start = buf;
861 if (OP_TEST != op && strstr(buf, basedir) == buf)
862 start += strlen(basedir);
863 }
864
865 mlink = mandoc_calloc(1, sizeof(struct mlink));
866 mlink->dform = FORM_NONE;
867 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
868 sizeof(mlink->file)) {
869 say(start, "Filename too long");
870 free(mlink);
871 return;
872 }
873
874 /*
875 * In test mode or when the original name is absolute
876 * but outside our tree, guess the base directory.
877 */
878
879 if (op == OP_TEST || (start == buf && *start == '/')) {
880 if (strncmp(buf, "man/", 4) == 0)
881 start = buf + 4;
882 else if ((start = strstr(buf, "/man/")) != NULL)
883 start += 5;
884 else
885 start = buf;
886 }
887
888 /*
889 * First try to guess our directory structure.
890 * If we find a separator, try to look for man* or cat*.
891 * If we find one of these and what's underneath is a directory,
892 * assume it's an architecture.
893 */
894 if (NULL != (p = strchr(start, '/'))) {
895 *p++ = '\0';
896 if (0 == strncmp(start, "man", 3)) {
897 mlink->dform = FORM_SRC;
898 mlink->dsec = start + 3;
899 } else if (0 == strncmp(start, "cat", 3)) {
900 mlink->dform = FORM_CAT;
901 mlink->dsec = start + 3;
902 }
903
904 start = p;
905 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
906 *p++ = '\0';
907 mlink->arch = start;
908 start = p;
909 }
910 }
911
912 /*
913 * Now check the file suffix.
914 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
915 */
916 p = strrchr(start, '\0');
917 while (p-- > start && '/' != *p && '.' != *p)
918 /* Loop. */ ;
919
920 if ('.' == *p) {
921 *p++ = '\0';
922 mlink->fsec = p;
923 }
924
925 /*
926 * Now try to parse the name.
927 * Use the filename portion of the path.
928 */
929 mlink->name = start;
930 if (NULL != (p = strrchr(start, '/'))) {
931 mlink->name = p + 1;
932 *p = '\0';
933 }
934 mlink_add(mlink, &st);
935 }
936
937 static void
938 mlink_add(struct mlink *mlink, const struct stat *st)
939 {
940 struct inodev inodev;
941 struct mpage *mpage;
942 unsigned int slot;
943
944 assert(NULL != mlink->file);
945
946 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
947 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
948 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
949 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
950
951 if ('0' == *mlink->fsec) {
952 free(mlink->fsec);
953 mlink->fsec = mandoc_strdup(mlink->dsec);
954 mlink->fform = FORM_CAT;
955 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
956 mlink->fform = FORM_SRC;
957 else
958 mlink->fform = FORM_NONE;
959
960 slot = ohash_qlookup(&mlinks, mlink->file);
961 assert(NULL == ohash_find(&mlinks, slot));
962 ohash_insert(&mlinks, slot, mlink);
963
964 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
965 inodev.st_ino = st->st_ino;
966 inodev.st_dev = st->st_dev;
967 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
968 sizeof(struct inodev), inodev.st_ino);
969 mpage = ohash_find(&mpages, slot);
970 if (NULL == mpage) {
971 mpage = mandoc_calloc(1, sizeof(struct mpage));
972 mpage->inodev.st_ino = inodev.st_ino;
973 mpage->inodev.st_dev = inodev.st_dev;
974 mpage->form = FORM_NONE;
975 mpage->next = mpage_head;
976 mpage_head = mpage;
977 ohash_insert(&mpages, slot, mpage);
978 } else
979 mlink->next = mpage->mlinks;
980 mpage->mlinks = mlink;
981 mlink->mpage = mpage;
982 }
983
984 static void
985 mlink_free(struct mlink *mlink)
986 {
987
988 free(mlink->dsec);
989 free(mlink->arch);
990 free(mlink->name);
991 free(mlink->fsec);
992 free(mlink);
993 }
994
995 static void
996 mpages_free(void)
997 {
998 struct mpage *mpage;
999 struct mlink *mlink;
1000
1001 while ((mpage = mpage_head) != NULL) {
1002 while ((mlink = mpage->mlinks) != NULL) {
1003 mpage->mlinks = mlink->next;
1004 mlink_free(mlink);
1005 }
1006 mpage_head = mpage->next;
1007 free(mpage->sec);
1008 free(mpage->arch);
1009 free(mpage->title);
1010 free(mpage->desc);
1011 free(mpage);
1012 }
1013 }
1014
1015 /*
1016 * For each mlink to the mpage, check whether the path looks like
1017 * it is formatted, and if it does, check whether a source manual
1018 * exists by the same name, ignoring the suffix.
1019 * If both conditions hold, drop the mlink.
1020 */
1021 static void
1022 mlinks_undupe(struct mpage *mpage)
1023 {
1024 char buf[PATH_MAX];
1025 struct mlink **prev;
1026 struct mlink *mlink;
1027 char *bufp;
1028
1029 mpage->form = FORM_CAT;
1030 prev = &mpage->mlinks;
1031 while (NULL != (mlink = *prev)) {
1032 if (FORM_CAT != mlink->dform) {
1033 mpage->form = FORM_NONE;
1034 goto nextlink;
1035 }
1036 (void)strlcpy(buf, mlink->file, sizeof(buf));
1037 bufp = strstr(buf, "cat");
1038 assert(NULL != bufp);
1039 memcpy(bufp, "man", 3);
1040 if (NULL != (bufp = strrchr(buf, '.')))
1041 *++bufp = '\0';
1042 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1043 if (NULL == ohash_find(&mlinks,
1044 ohash_qlookup(&mlinks, buf)))
1045 goto nextlink;
1046 if (warnings)
1047 say(mlink->file, "Man source exists: %s", buf);
1048 if (use_all)
1049 goto nextlink;
1050 *prev = mlink->next;
1051 mlink_free(mlink);
1052 continue;
1053 nextlink:
1054 prev = &(*prev)->next;
1055 }
1056 }
1057
1058 static void
1059 mlink_check(struct mpage *mpage, struct mlink *mlink)
1060 {
1061 struct str *str;
1062 unsigned int slot;
1063
1064 /*
1065 * Check whether the manual section given in a file
1066 * agrees with the directory where the file is located.
1067 * Some manuals have suffixes like (3p) on their
1068 * section number either inside the file or in the
1069 * directory name, some are linked into more than one
1070 * section, like encrypt(1) = makekey(8).
1071 */
1072
1073 if (FORM_SRC == mpage->form &&
1074 strcasecmp(mpage->sec, mlink->dsec))
1075 say(mlink->file, "Section \"%s\" manual in %s directory",
1076 mpage->sec, mlink->dsec);
1077
1078 /*
1079 * Manual page directories exist for each kernel
1080 * architecture as returned by machine(1).
1081 * However, many manuals only depend on the
1082 * application architecture as returned by arch(1).
1083 * For example, some (2/ARM) manuals are shared
1084 * across the "armish" and "zaurus" kernel
1085 * architectures.
1086 * A few manuals are even shared across completely
1087 * different architectures, for example fdformat(1)
1088 * on amd64, i386, and sparc64.
1089 */
1090
1091 if (strcasecmp(mpage->arch, mlink->arch))
1092 say(mlink->file, "Architecture \"%s\" manual in "
1093 "\"%s\" directory", mpage->arch, mlink->arch);
1094
1095 /*
1096 * XXX
1097 * parse_cat() doesn't set NAME_TITLE yet.
1098 */
1099
1100 if (FORM_CAT == mpage->form)
1101 return;
1102
1103 /*
1104 * Check whether this mlink
1105 * appears as a name in the NAME section.
1106 */
1107
1108 slot = ohash_qlookup(&names, mlink->name);
1109 str = ohash_find(&names, slot);
1110 assert(NULL != str);
1111 if ( ! (NAME_TITLE & str->mask))
1112 say(mlink->file, "Name missing in NAME section");
1113 }
1114
1115 /*
1116 * Run through the files in the global vector "mpages"
1117 * and add them to the database specified in "basedir".
1118 *
1119 * This handles the parsing scheme itself, using the cues of directory
1120 * and filename to determine whether the file is parsable or not.
1121 */
1122 static void
1123 mpages_merge(struct dba *dba, struct mparse *mp)
1124 {
1125 struct mpage *mpage, *mpage_dest;
1126 struct mlink *mlink, *mlink_dest;
1127 struct roff_man *man;
1128 char *sodest;
1129 char *cp;
1130 int fd;
1131
1132 for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1133 mlinks_undupe(mpage);
1134 if ((mlink = mpage->mlinks) == NULL)
1135 continue;
1136
1137 name_mask = NAME_MASK;
1138 mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1139 mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1140 mparse_reset(mp);
1141 man = NULL;
1142 sodest = NULL;
1143
1144 if ((fd = mparse_open(mp, mlink->file)) == -1) {
1145 say(mlink->file, "&open");
1146 goto nextpage;
1147 }
1148
1149 /*
1150 * Interpret the file as mdoc(7) or man(7) source
1151 * code, unless it is known to be formatted.
1152 */
1153 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1154 mparse_readfd(mp, fd, mlink->file);
1155 close(fd);
1156 fd = -1;
1157 mparse_result(mp, &man, &sodest);
1158 }
1159
1160 if (sodest != NULL) {
1161 mlink_dest = ohash_find(&mlinks,
1162 ohash_qlookup(&mlinks, sodest));
1163 if (mlink_dest == NULL) {
1164 mandoc_asprintf(&cp, "%s.gz", sodest);
1165 mlink_dest = ohash_find(&mlinks,
1166 ohash_qlookup(&mlinks, cp));
1167 free(cp);
1168 }
1169 if (mlink_dest != NULL) {
1170
1171 /* The .so target exists. */
1172
1173 mpage_dest = mlink_dest->mpage;
1174 while (1) {
1175 mlink->mpage = mpage_dest;
1176
1177 /*
1178 * If the target was already
1179 * processed, add the links
1180 * to the database now.
1181 * Otherwise, this will
1182 * happen when we come
1183 * to the target.
1184 */
1185
1186 if (mpage_dest->dba != NULL)
1187 dbadd_mlink(mlink);
1188
1189 if (mlink->next == NULL)
1190 break;
1191 mlink = mlink->next;
1192 }
1193
1194 /* Move all links to the target. */
1195
1196 mlink->next = mlink_dest->next;
1197 mlink_dest->next = mpage->mlinks;
1198 mpage->mlinks = NULL;
1199 }
1200 goto nextpage;
1201 } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1202 mdoc_validate(man);
1203 mpage->form = FORM_SRC;
1204 mpage->sec = man->meta.msec;
1205 mpage->sec = mandoc_strdup(
1206 mpage->sec == NULL ? "" : mpage->sec);
1207 mpage->arch = man->meta.arch;
1208 mpage->arch = mandoc_strdup(
1209 mpage->arch == NULL ? "" : mpage->arch);
1210 mpage->title = mandoc_strdup(man->meta.title);
1211 } else if (man != NULL && man->macroset == MACROSET_MAN) {
1212 man_validate(man);
1213 if (*man->meta.msec != '\0' ||
1214 *man->meta.msec != '\0') {
1215 mpage->form = FORM_SRC;
1216 mpage->sec = mandoc_strdup(man->meta.msec);
1217 mpage->arch = mandoc_strdup(mlink->arch);
1218 mpage->title = mandoc_strdup(man->meta.title);
1219 } else
1220 man = NULL;
1221 }
1222
1223 assert(mpage->desc == NULL);
1224 if (man == NULL) {
1225 mpage->form = FORM_CAT;
1226 mpage->sec = mandoc_strdup(mlink->dsec);
1227 mpage->arch = mandoc_strdup(mlink->arch);
1228 mpage->title = mandoc_strdup(mlink->name);
1229 parse_cat(mpage, fd);
1230 } else if (man->macroset == MACROSET_MDOC)
1231 parse_mdoc(mpage, &man->meta, man->first);
1232 else
1233 parse_man(mpage, &man->meta, man->first);
1234 if (mpage->desc == NULL) {
1235 mpage->desc = mandoc_strdup(mlink->name);
1236 if (warnings)
1237 say(mlink->file, "No one-line description, "
1238 "using filename \"%s\"", mlink->name);
1239 }
1240
1241 for (mlink = mpage->mlinks;
1242 mlink != NULL;
1243 mlink = mlink->next) {
1244 putkey(mpage, mlink->name, NAME_FILE);
1245 if (warnings && !use_all)
1246 mlink_check(mpage, mlink);
1247 }
1248
1249 dbadd(dba, mpage);
1250
1251 nextpage:
1252 ohash_delete(&strings);
1253 ohash_delete(&names);
1254 }
1255 }
1256
1257 static void
1258 parse_cat(struct mpage *mpage, int fd)
1259 {
1260 FILE *stream;
1261 struct mlink *mlink;
1262 char *line, *p, *title, *sec;
1263 size_t linesz, plen, titlesz;
1264 ssize_t len;
1265 int offs;
1266
1267 mlink = mpage->mlinks;
1268 stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1269 if (stream == NULL) {
1270 if (fd != -1)
1271 close(fd);
1272 if (warnings)
1273 say(mlink->file, "&fopen");
1274 return;
1275 }
1276
1277 line = NULL;
1278 linesz = 0;
1279
1280 /* Parse the section number from the header line. */
1281
1282 while (getline(&line, &linesz, stream) != -1) {
1283 if (*line == '\n')
1284 continue;
1285 if ((sec = strchr(line, '(')) == NULL)
1286 break;
1287 if ((p = strchr(++sec, ')')) == NULL)
1288 break;
1289 free(mpage->sec);
1290 mpage->sec = mandoc_strndup(sec, p - sec);
1291 if (warnings && *mlink->dsec != '\0' &&
1292 strcasecmp(mpage->sec, mlink->dsec))
1293 say(mlink->file,
1294 "Section \"%s\" manual in %s directory",
1295 mpage->sec, mlink->dsec);
1296 break;
1297 }
1298
1299 /* Skip to first blank line. */
1300
1301 while (line == NULL || *line != '\n')
1302 if (getline(&line, &linesz, stream) == -1)
1303 break;
1304
1305 /*
1306 * Assume the first line that is not indented
1307 * is the first section header. Skip to it.
1308 */
1309
1310 while (getline(&line, &linesz, stream) != -1)
1311 if (*line != '\n' && *line != ' ')
1312 break;
1313
1314 /*
1315 * Read up until the next section into a buffer.
1316 * Strip the leading and trailing newline from each read line,
1317 * appending a trailing space.
1318 * Ignore empty (whitespace-only) lines.
1319 */
1320
1321 titlesz = 0;
1322 title = NULL;
1323
1324 while ((len = getline(&line, &linesz, stream)) != -1) {
1325 if (*line != ' ')
1326 break;
1327 offs = 0;
1328 while (isspace((unsigned char)line[offs]))
1329 offs++;
1330 if (line[offs] == '\0')
1331 continue;
1332 title = mandoc_realloc(title, titlesz + len - offs);
1333 memcpy(title + titlesz, line + offs, len - offs);
1334 titlesz += len - offs;
1335 title[titlesz - 1] = ' ';
1336 }
1337 free(line);
1338
1339 /*
1340 * If no page content can be found, or the input line
1341 * is already the next section header, or there is no
1342 * trailing newline, reuse the page title as the page
1343 * description.
1344 */
1345
1346 if (NULL == title || '\0' == *title) {
1347 if (warnings)
1348 say(mlink->file, "Cannot find NAME section");
1349 fclose(stream);
1350 free(title);
1351 return;
1352 }
1353
1354 title[titlesz - 1] = '\0';
1355
1356 /*
1357 * Skip to the first dash.
1358 * Use the remaining line as the description (no more than 70
1359 * bytes).
1360 */
1361
1362 if (NULL != (p = strstr(title, "- "))) {
1363 for (p += 2; ' ' == *p || '\b' == *p; p++)
1364 /* Skip to next word. */ ;
1365 } else {
1366 if (warnings)
1367 say(mlink->file, "No dash in title line, "
1368 "reusing \"%s\" as one-line description", title);
1369 p = title;
1370 }
1371
1372 plen = strlen(p);
1373
1374 /* Strip backspace-encoding from line. */
1375
1376 while (NULL != (line = memchr(p, '\b', plen))) {
1377 len = line - p;
1378 if (0 == len) {
1379 memmove(line, line + 1, plen--);
1380 continue;
1381 }
1382 memmove(line - 1, line + 1, plen - len);
1383 plen -= 2;
1384 }
1385
1386 mpage->desc = mandoc_strdup(p);
1387 fclose(stream);
1388 free(title);
1389 }
1390
1391 /*
1392 * Put a type/word pair into the word database for this particular file.
1393 */
1394 static void
1395 putkey(const struct mpage *mpage, char *value, uint64_t type)
1396 {
1397 putkeys(mpage, value, strlen(value), type);
1398 }
1399
1400 /*
1401 * Grok all nodes at or below a certain mdoc node into putkey().
1402 */
1403 static void
1404 putmdockey(const struct mpage *mpage,
1405 const struct roff_node *n, uint64_t m, int taboo)
1406 {
1407
1408 for ( ; NULL != n; n = n->next) {
1409 if (n->flags & taboo)
1410 continue;
1411 if (NULL != n->child)
1412 putmdockey(mpage, n->child, m, taboo);
1413 if (n->type == ROFFT_TEXT)
1414 putkey(mpage, n->string, m);
1415 }
1416 }
1417
1418 static void
1419 parse_man(struct mpage *mpage, const struct roff_meta *meta,
1420 const struct roff_node *n)
1421 {
1422 const struct roff_node *head, *body;
1423 char *start, *title;
1424 char byte;
1425 size_t sz;
1426
1427 if (n == NULL)
1428 return;
1429
1430 /*
1431 * We're only searching for one thing: the first text child in
1432 * the BODY of a NAME section. Since we don't keep track of
1433 * sections in -man, run some hoops to find out whether we're in
1434 * the correct section or not.
1435 */
1436
1437 if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1438 body = n;
1439 if ((head = body->parent->head) != NULL &&
1440 (head = head->child) != NULL &&
1441 head->next == NULL &&
1442 head->type == ROFFT_TEXT &&
1443 strcmp(head->string, "NAME") == 0 &&
1444 body->child != NULL) {
1445
1446 /*
1447 * Suck the entire NAME section into memory.
1448 * Yes, we might run away.
1449 * But too many manuals have big, spread-out
1450 * NAME sections over many lines.
1451 */
1452
1453 title = NULL;
1454 deroff(&title, body);
1455 if (NULL == title)
1456 return;
1457
1458 /*
1459 * Go through a special heuristic dance here.
1460 * Conventionally, one or more manual names are
1461 * comma-specified prior to a whitespace, then a
1462 * dash, then a description. Try to puzzle out
1463 * the name parts here.
1464 */
1465
1466 start = title;
1467 for ( ;; ) {
1468 sz = strcspn(start, " ,");
1469 if ('\0' == start[sz])
1470 break;
1471
1472 byte = start[sz];
1473 start[sz] = '\0';
1474
1475 /*
1476 * Assume a stray trailing comma in the
1477 * name list if a name begins with a dash.
1478 */
1479
1480 if ('-' == start[0] ||
1481 ('\\' == start[0] && '-' == start[1]))
1482 break;
1483
1484 putkey(mpage, start, NAME_TITLE);
1485 if ( ! (mpage->name_head_done ||
1486 strcasecmp(start, meta->title))) {
1487 putkey(mpage, start, NAME_HEAD);
1488 mpage->name_head_done = 1;
1489 }
1490
1491 if (' ' == byte) {
1492 start += sz + 1;
1493 break;
1494 }
1495
1496 assert(',' == byte);
1497 start += sz + 1;
1498 while (' ' == *start)
1499 start++;
1500 }
1501
1502 if (start == title) {
1503 putkey(mpage, start, NAME_TITLE);
1504 if ( ! (mpage->name_head_done ||
1505 strcasecmp(start, meta->title))) {
1506 putkey(mpage, start, NAME_HEAD);
1507 mpage->name_head_done = 1;
1508 }
1509 free(title);
1510 return;
1511 }
1512
1513 while (isspace((unsigned char)*start))
1514 start++;
1515
1516 if (0 == strncmp(start, "-", 1))
1517 start += 1;
1518 else if (0 == strncmp(start, "\\-\\-", 4))
1519 start += 4;
1520 else if (0 == strncmp(start, "\\-", 2))
1521 start += 2;
1522 else if (0 == strncmp(start, "\\(en", 4))
1523 start += 4;
1524 else if (0 == strncmp(start, "\\(em", 4))
1525 start += 4;
1526
1527 while (' ' == *start)
1528 start++;
1529
1530 mpage->desc = mandoc_strdup(start);
1531 free(title);
1532 return;
1533 }
1534 }
1535
1536 for (n = n->child; n; n = n->next) {
1537 if (NULL != mpage->desc)
1538 break;
1539 parse_man(mpage, meta, n);
1540 }
1541 }
1542
1543 static void
1544 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1545 const struct roff_node *n)
1546 {
1547
1548 assert(NULL != n);
1549 for (n = n->child; NULL != n; n = n->next) {
1550 if (n->flags & mdocs[n->tok].taboo)
1551 continue;
1552 switch (n->type) {
1553 case ROFFT_ELEM:
1554 case ROFFT_BLOCK:
1555 case ROFFT_HEAD:
1556 case ROFFT_BODY:
1557 case ROFFT_TAIL:
1558 if (NULL != mdocs[n->tok].fp)
1559 if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1560 break;
1561 if (mdocs[n->tok].mask)
1562 putmdockey(mpage, n->child,
1563 mdocs[n->tok].mask, mdocs[n->tok].taboo);
1564 break;
1565 default:
1566 assert(n->type != ROFFT_ROOT);
1567 continue;
1568 }
1569 if (NULL != n->child)
1570 parse_mdoc(mpage, meta, n);
1571 }
1572 }
1573
1574 static int
1575 parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1576 const struct roff_node *n)
1577 {
1578 char *start, *end;
1579 size_t sz;
1580
1581 if (SEC_SYNOPSIS != n->sec ||
1582 NULL == (n = n->child) ||
1583 n->type != ROFFT_TEXT)
1584 return 0;
1585
1586 /*
1587 * Only consider those `Fd' macro fields that begin with an
1588 * "inclusion" token (versus, e.g., #define).
1589 */
1590
1591 if (strcmp("#include", n->string))
1592 return 0;
1593
1594 if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1595 return 0;
1596
1597 /*
1598 * Strip away the enclosing angle brackets and make sure we're
1599 * not zero-length.
1600 */
1601
1602 start = n->string;
1603 if ('<' == *start || '"' == *start)
1604 start++;
1605
1606 if (0 == (sz = strlen(start)))
1607 return 0;
1608
1609 end = &start[(int)sz - 1];
1610 if ('>' == *end || '"' == *end)
1611 end--;
1612
1613 if (end > start)
1614 putkeys(mpage, start, end - start + 1, TYPE_In);
1615 return 0;
1616 }
1617
1618 static void
1619 parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1620 {
1621 char *cp;
1622 size_t sz;
1623
1624 if (n->type != ROFFT_TEXT)
1625 return;
1626
1627 /* Skip function pointer punctuation. */
1628
1629 cp = n->string;
1630 while (*cp == '(' || *cp == '*')
1631 cp++;
1632 sz = strcspn(cp, "()");
1633
1634 putkeys(mpage, cp, sz, TYPE_Fn);
1635 if (n->sec == SEC_SYNOPSIS)
1636 putkeys(mpage, cp, sz, NAME_SYN);
1637 }
1638
1639 static int
1640 parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1641 const struct roff_node *n)
1642 {
1643
1644 if (n->child == NULL)
1645 return 0;
1646
1647 parse_mdoc_fname(mpage, n->child);
1648
1649 for (n = n->child->next; n != NULL; n = n->next)
1650 if (n->type == ROFFT_TEXT)
1651 putkey(mpage, n->string, TYPE_Fa);
1652
1653 return 0;
1654 }
1655
1656 static int
1657 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1658 const struct roff_node *n)
1659 {
1660
1661 if (n->type != ROFFT_HEAD)
1662 return 1;
1663
1664 if (n->child != NULL)
1665 parse_mdoc_fname(mpage, n->child);
1666
1667 return 0;
1668 }
1669
1670 static int
1671 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1672 const struct roff_node *n)
1673 {
1674 char *cp;
1675
1676 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1677 return 0;
1678
1679 if (n->child != NULL &&
1680 n->child->next == NULL &&
1681 n->child->type == ROFFT_TEXT)
1682 return 1;
1683
1684 cp = NULL;
1685 deroff(&cp, n);
1686 if (cp != NULL) {
1687 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1688 n->type == ROFFT_BODY ? TYPE_Va : 0));
1689 free(cp);
1690 }
1691
1692 return 0;
1693 }
1694
1695 static int
1696 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1697 const struct roff_node *n)
1698 {
1699 char *cp;
1700
1701 if (NULL == (n = n->child))
1702 return 0;
1703
1704 if (NULL == n->next) {
1705 putkey(mpage, n->string, TYPE_Xr);
1706 return 0;
1707 }
1708
1709 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1710 putkey(mpage, cp, TYPE_Xr);
1711 free(cp);
1712 return 0;
1713 }
1714
1715 static int
1716 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1717 const struct roff_node *n)
1718 {
1719
1720 if (n->type == ROFFT_BODY)
1721 deroff(&mpage->desc, n);
1722 return 0;
1723 }
1724
1725 static int
1726 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1727 const struct roff_node *n)
1728 {
1729
1730 if (SEC_NAME == n->sec)
1731 putmdockey(mpage, n->child, NAME_TITLE, 0);
1732 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1733 if (n->child == NULL)
1734 putkey(mpage, meta->name, NAME_SYN);
1735 else
1736 putmdockey(mpage, n->child, NAME_SYN, 0);
1737 }
1738 if ( ! (mpage->name_head_done ||
1739 n->child == NULL || n->child->string == NULL ||
1740 strcasecmp(n->child->string, meta->title))) {
1741 putkey(mpage, n->child->string, NAME_HEAD);
1742 mpage->name_head_done = 1;
1743 }
1744 return 0;
1745 }
1746
1747 static int
1748 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1749 const struct roff_node *n)
1750 {
1751
1752 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1753 }
1754
1755 static int
1756 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1757 const struct roff_node *n)
1758 {
1759
1760 return n->type == ROFFT_HEAD;
1761 }
1762
1763 /*
1764 * Add a string to the hash table for the current manual.
1765 * Each string has a bitmask telling which macros it belongs to.
1766 * When we finish the manual, we'll dump the table.
1767 */
1768 static void
1769 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1770 {
1771 struct ohash *htab;
1772 struct str *s;
1773 const char *end;
1774 unsigned int slot;
1775 int i, mustfree;
1776
1777 if (0 == sz)
1778 return;
1779
1780 mustfree = render_string(&cp, &sz);
1781
1782 if (TYPE_Nm & v) {
1783 htab = &names;
1784 v &= name_mask;
1785 if (v & NAME_FIRST)
1786 name_mask &= ~NAME_FIRST;
1787 if (debug > 1)
1788 say(mpage->mlinks->file,
1789 "Adding name %*s, bits=0x%llx", (int)sz, cp,
1790 (unsigned long long)v);
1791 } else {
1792 htab = &strings;
1793 if (debug > 1)
1794 for (i = 0; i < KEY_MAX; i++)
1795 if ((uint64_t)1 << i & v)
1796 say(mpage->mlinks->file,
1797 "Adding key %s=%*s",
1798 mansearch_keynames[i], (int)sz, cp);
1799 }
1800
1801 end = cp + sz;
1802 slot = ohash_qlookupi(htab, cp, &end);
1803 s = ohash_find(htab, slot);
1804
1805 if (NULL != s && mpage == s->mpage) {
1806 s->mask |= v;
1807 return;
1808 } else if (NULL == s) {
1809 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1810 memcpy(s->key, cp, sz);
1811 ohash_insert(htab, slot, s);
1812 }
1813 s->mpage = mpage;
1814 s->mask = v;
1815
1816 if (mustfree)
1817 free(cp);
1818 }
1819
1820 /*
1821 * Take a Unicode codepoint and produce its UTF-8 encoding.
1822 * This isn't the best way to do this, but it works.
1823 * The magic numbers are from the UTF-8 packaging.
1824 * They're not as scary as they seem: read the UTF-8 spec for details.
1825 */
1826 static size_t
1827 utf8(unsigned int cp, char out[7])
1828 {
1829 size_t rc;
1830
1831 rc = 0;
1832 if (cp <= 0x0000007F) {
1833 rc = 1;
1834 out[0] = (char)cp;
1835 } else if (cp <= 0x000007FF) {
1836 rc = 2;
1837 out[0] = (cp >> 6 & 31) | 192;
1838 out[1] = (cp & 63) | 128;
1839 } else if (cp <= 0x0000FFFF) {
1840 rc = 3;
1841 out[0] = (cp >> 12 & 15) | 224;
1842 out[1] = (cp >> 6 & 63) | 128;
1843 out[2] = (cp & 63) | 128;
1844 } else if (cp <= 0x001FFFFF) {
1845 rc = 4;
1846 out[0] = (cp >> 18 & 7) | 240;
1847 out[1] = (cp >> 12 & 63) | 128;
1848 out[2] = (cp >> 6 & 63) | 128;
1849 out[3] = (cp & 63) | 128;
1850 } else if (cp <= 0x03FFFFFF) {
1851 rc = 5;
1852 out[0] = (cp >> 24 & 3) | 248;
1853 out[1] = (cp >> 18 & 63) | 128;
1854 out[2] = (cp >> 12 & 63) | 128;
1855 out[3] = (cp >> 6 & 63) | 128;
1856 out[4] = (cp & 63) | 128;
1857 } else if (cp <= 0x7FFFFFFF) {
1858 rc = 6;
1859 out[0] = (cp >> 30 & 1) | 252;
1860 out[1] = (cp >> 24 & 63) | 128;
1861 out[2] = (cp >> 18 & 63) | 128;
1862 out[3] = (cp >> 12 & 63) | 128;
1863 out[4] = (cp >> 6 & 63) | 128;
1864 out[5] = (cp & 63) | 128;
1865 } else
1866 return 0;
1867
1868 out[rc] = '\0';
1869 return rc;
1870 }
1871
1872 /*
1873 * If the string contains escape sequences,
1874 * replace it with an allocated rendering and return 1,
1875 * such that the caller can free it after use.
1876 * Otherwise, do nothing and return 0.
1877 */
1878 static int
1879 render_string(char **public, size_t *psz)
1880 {
1881 const char *src, *scp, *addcp, *seq;
1882 char *dst;
1883 size_t ssz, dsz, addsz;
1884 char utfbuf[7], res[6];
1885 int seqlen, unicode;
1886
1887 res[0] = '\\';
1888 res[1] = '\t';
1889 res[2] = ASCII_NBRSP;
1890 res[3] = ASCII_HYPH;
1891 res[4] = ASCII_BREAK;
1892 res[5] = '\0';
1893
1894 src = scp = *public;
1895 ssz = *psz;
1896 dst = NULL;
1897 dsz = 0;
1898
1899 while (scp < src + *psz) {
1900
1901 /* Leave normal characters unchanged. */
1902
1903 if (strchr(res, *scp) == NULL) {
1904 if (dst != NULL)
1905 dst[dsz++] = *scp;
1906 scp++;
1907 continue;
1908 }
1909
1910 /*
1911 * Found something that requires replacing,
1912 * make sure we have a destination buffer.
1913 */
1914
1915 if (dst == NULL) {
1916 dst = mandoc_malloc(ssz + 1);
1917 dsz = scp - src;
1918 memcpy(dst, src, dsz);
1919 }
1920
1921 /* Handle single-char special characters. */
1922
1923 switch (*scp) {
1924 case '\\':
1925 break;
1926 case '\t':
1927 case ASCII_NBRSP:
1928 dst[dsz++] = ' ';
1929 scp++;
1930 continue;
1931 case ASCII_HYPH:
1932 dst[dsz++] = '-';
1933 /* FALLTHROUGH */
1934 case ASCII_BREAK:
1935 scp++;
1936 continue;
1937 default:
1938 abort();
1939 }
1940
1941 /*
1942 * Found an escape sequence.
1943 * Read past the slash, then parse it.
1944 * Ignore everything except characters.
1945 */
1946
1947 scp++;
1948 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1949 continue;
1950
1951 /*
1952 * Render the special character
1953 * as either UTF-8 or ASCII.
1954 */
1955
1956 if (write_utf8) {
1957 unicode = mchars_spec2cp(seq, seqlen);
1958 if (unicode <= 0)
1959 continue;
1960 addsz = utf8(unicode, utfbuf);
1961 if (addsz == 0)
1962 continue;
1963 addcp = utfbuf;
1964 } else {
1965 addcp = mchars_spec2str(seq, seqlen, &addsz);
1966 if (addcp == NULL)
1967 continue;
1968 if (*addcp == ASCII_NBRSP) {
1969 addcp = " ";
1970 addsz = 1;
1971 }
1972 }
1973
1974 /* Copy the rendered glyph into the stream. */
1975
1976 ssz += addsz;
1977 dst = mandoc_realloc(dst, ssz + 1);
1978 memcpy(dst + dsz, addcp, addsz);
1979 dsz += addsz;
1980 }
1981 if (dst != NULL) {
1982 *public = dst;
1983 *psz = dsz;
1984 }
1985
1986 /* Trim trailing whitespace and NUL-terminate. */
1987
1988 while (*psz > 0 && (*public)[*psz - 1] == ' ')
1989 --*psz;
1990 if (dst != NULL) {
1991 (*public)[*psz] = '\0';
1992 return 1;
1993 } else
1994 return 0;
1995 }
1996
1997 static void
1998 dbadd_mlink(const struct mlink *mlink)
1999 {
2000 dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
2001 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
2002 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
2003 dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
2004 dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
2005 }
2006
2007 /*
2008 * Flush the current page's terms (and their bits) into the database.
2009 * Also, handle escape sequences at the last possible moment.
2010 */
2011 static void
2012 dbadd(struct dba *dba, struct mpage *mpage)
2013 {
2014 struct mlink *mlink;
2015 struct str *key;
2016 char *cp;
2017 uint64_t mask;
2018 size_t i;
2019 unsigned int slot;
2020 int mustfree;
2021
2022 mlink = mpage->mlinks;
2023
2024 if (nodb) {
2025 for (key = ohash_first(&names, &slot); NULL != key;
2026 key = ohash_next(&names, &slot))
2027 free(key);
2028 for (key = ohash_first(&strings, &slot); NULL != key;
2029 key = ohash_next(&strings, &slot))
2030 free(key);
2031 if (0 == debug)
2032 return;
2033 while (NULL != mlink) {
2034 fputs(mlink->name, stdout);
2035 if (NULL == mlink->next ||
2036 strcmp(mlink->dsec, mlink->next->dsec) ||
2037 strcmp(mlink->fsec, mlink->next->fsec) ||
2038 strcmp(mlink->arch, mlink->next->arch)) {
2039 putchar('(');
2040 if ('\0' == *mlink->dsec)
2041 fputs(mlink->fsec, stdout);
2042 else
2043 fputs(mlink->dsec, stdout);
2044 if ('\0' != *mlink->arch)
2045 printf("/%s", mlink->arch);
2046 putchar(')');
2047 }
2048 mlink = mlink->next;
2049 if (NULL != mlink)
2050 fputs(", ", stdout);
2051 }
2052 printf(" - %s\n", mpage->desc);
2053 return;
2054 }
2055
2056 if (debug)
2057 say(mlink->file, "Adding to database");
2058
2059 cp = mpage->desc;
2060 i = strlen(cp);
2061 mustfree = render_string(&cp, &i);
2062 mpage->dba = dba_page_new(dba->pages,
2063 *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2064 cp, mlink->file, mpage->form);
2065 if (mustfree)
2066 free(cp);
2067 dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
2068
2069 while (mlink != NULL) {
2070 dbadd_mlink(mlink);
2071 mlink = mlink->next;
2072 }
2073
2074 for (key = ohash_first(&names, &slot); NULL != key;
2075 key = ohash_next(&names, &slot)) {
2076 assert(key->mpage == mpage);
2077 dba_page_alias(mpage->dba, key->key, key->mask);
2078 free(key);
2079 }
2080 for (key = ohash_first(&strings, &slot); NULL != key;
2081 key = ohash_next(&strings, &slot)) {
2082 assert(key->mpage == mpage);
2083 i = 0;
2084 for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2085 if (key->mask & mask)
2086 dba_macro_add(dba->macros, i,
2087 key->key, mpage->dba);
2088 i++;
2089 }
2090 free(key);
2091 }
2092 }
2093
2094 static void
2095 dbprune(struct dba *dba)
2096 {
2097 struct dba_array *page, *files;
2098 char *file;
2099
2100 dba_array_FOREACH(dba->pages, page) {
2101 files = dba_array_get(page, DBP_FILE);
2102 dba_array_FOREACH(files, file) {
2103 if (*file < ' ')
2104 file++;
2105 if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2106 file)) != NULL) {
2107 if (debug)
2108 say(file, "Deleting from database");
2109 dba_array_del(dba->pages);
2110 break;
2111 }
2112 }
2113 }
2114 }
2115
2116 /*
2117 * Write the database from memory to disk.
2118 */
2119 static void
2120 dbwrite(struct dba *dba)
2121 {
2122 char tfn[32];
2123 int status;
2124 pid_t child;
2125
2126 if (dba_write(MANDOC_DB "~", dba) != -1) {
2127 if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2128 exitcode = (int)MANDOCLEVEL_SYSERR;
2129 say(MANDOC_DB, "&rename");
2130 unlink(MANDOC_DB "~");
2131 }
2132 return;
2133 }
2134
2135 (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2136 if (mkdtemp(tfn) == NULL) {
2137 exitcode = (int)MANDOCLEVEL_SYSERR;
2138 say("", "&%s", tfn);
2139 return;
2140 }
2141
2142 (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2143 if (dba_write(tfn, dba) == -1) {
2144 exitcode = (int)MANDOCLEVEL_SYSERR;
2145 say(tfn, "&dba_write");
2146 goto out;
2147 }
2148
2149 switch (child = fork()) {
2150 case -1:
2151 exitcode = (int)MANDOCLEVEL_SYSERR;
2152 say("", "&fork cmp");
2153 return;
2154 case 0:
2155 execlp("cmp", "cmp", "-s", tfn, MANDOC_DB, (char *)NULL);
2156 say("", "&exec cmp");
2157 exit(0);
2158 default:
2159 break;
2160 }
2161 if (waitpid(child, &status, 0) == -1) {
2162 exitcode = (int)MANDOCLEVEL_SYSERR;
2163 say("", "&wait cmp");
2164 } else if (WIFSIGNALED(status)) {
2165 exitcode = (int)MANDOCLEVEL_SYSERR;
2166 say("", "cmp died from signal %d", WTERMSIG(status));
2167 } else if (WEXITSTATUS(status)) {
2168 exitcode = (int)MANDOCLEVEL_SYSERR;
2169 say(MANDOC_DB,
2170 "Data changed, but cannot replace database");
2171 }
2172
2173 out:
2174 *strrchr(tfn, '/') = '\0';
2175 switch (child = fork()) {
2176 case -1:
2177 exitcode = (int)MANDOCLEVEL_SYSERR;
2178 say("", "&fork rm");
2179 return;
2180 case 0:
2181 execlp("rm", "rm", "-rf", tfn, (char *)NULL);
2182 say("", "&exec rm");
2183 exit((int)MANDOCLEVEL_SYSERR);
2184 default:
2185 break;
2186 }
2187 if (waitpid(child, &status, 0) == -1) {
2188 exitcode = (int)MANDOCLEVEL_SYSERR;
2189 say("", "&wait rm");
2190 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2191 exitcode = (int)MANDOCLEVEL_SYSERR;
2192 say("", "%s: Cannot remove temporary directory", tfn);
2193 }
2194 }
2195
2196 static int
2197 set_basedir(const char *targetdir, int report_baddir)
2198 {
2199 static char startdir[PATH_MAX];
2200 static int getcwd_status; /* 1 = ok, 2 = failure */
2201 static int chdir_status; /* 1 = changed directory */
2202 char *cp;
2203
2204 /*
2205 * Remember the original working directory, if possible.
2206 * This will be needed if the second or a later directory
2207 * on the command line is given as a relative path.
2208 * Do not error out if the current directory is not
2209 * searchable: Maybe it won't be needed after all.
2210 */
2211 if (0 == getcwd_status) {
2212 if (NULL == getcwd(startdir, sizeof(startdir))) {
2213 getcwd_status = 2;
2214 (void)strlcpy(startdir, strerror(errno),
2215 sizeof(startdir));
2216 } else
2217 getcwd_status = 1;
2218 }
2219
2220 /*
2221 * We are leaving the old base directory.
2222 * Do not use it any longer, not even for messages.
2223 */
2224 *basedir = '\0';
2225
2226 /*
2227 * If and only if the directory was changed earlier and
2228 * the next directory to process is given as a relative path,
2229 * first go back, or bail out if that is impossible.
2230 */
2231 if (chdir_status && '/' != *targetdir) {
2232 if (2 == getcwd_status) {
2233 exitcode = (int)MANDOCLEVEL_SYSERR;
2234 say("", "getcwd: %s", startdir);
2235 return 0;
2236 }
2237 if (-1 == chdir(startdir)) {
2238 exitcode = (int)MANDOCLEVEL_SYSERR;
2239 say("", "&chdir %s", startdir);
2240 return 0;
2241 }
2242 }
2243
2244 /*
2245 * Always resolve basedir to the canonicalized absolute
2246 * pathname and append a trailing slash, such that
2247 * we can reliably check whether files are inside.
2248 */
2249 if (NULL == realpath(targetdir, basedir)) {
2250 if (report_baddir || errno != ENOENT) {
2251 exitcode = (int)MANDOCLEVEL_BADARG;
2252 say("", "&%s: realpath", targetdir);
2253 }
2254 return 0;
2255 } else if (-1 == chdir(basedir)) {
2256 if (report_baddir || errno != ENOENT) {
2257 exitcode = (int)MANDOCLEVEL_BADARG;
2258 say("", "&chdir");
2259 }
2260 return 0;
2261 }
2262 chdir_status = 1;
2263 cp = strchr(basedir, '\0');
2264 if ('/' != cp[-1]) {
2265 if (cp - basedir >= PATH_MAX - 1) {
2266 exitcode = (int)MANDOCLEVEL_SYSERR;
2267 say("", "Filename too long");
2268 return 0;
2269 }
2270 *cp++ = '/';
2271 *cp = '\0';
2272 }
2273 return 1;
2274 }
2275
2276 static void
2277 say(const char *file, const char *format, ...)
2278 {
2279 va_list ap;
2280 int use_errno;
2281
2282 if ('\0' != *basedir)
2283 fprintf(stderr, "%s", basedir);
2284 if ('\0' != *basedir && '\0' != *file)
2285 fputc('/', stderr);
2286 if ('\0' != *file)
2287 fprintf(stderr, "%s", file);
2288
2289 use_errno = 1;
2290 if (NULL != format) {
2291 switch (*format) {
2292 case '&':
2293 format++;
2294 break;
2295 case '\0':
2296 format = NULL;
2297 break;
2298 default:
2299 use_errno = 0;
2300 break;
2301 }
2302 }
2303 if (NULL != format) {
2304 if ('\0' != *basedir || '\0' != *file)
2305 fputs(": ", stderr);
2306 va_start(ap, format);
2307 vfprintf(stderr, format, ap);
2308 va_end(ap);
2309 }
2310 if (use_errno) {
2311 if ('\0' != *basedir || '\0' != *file || NULL != format)
2312 fputs(": ", stderr);
2313 perror(NULL);
2314 } else
2315 fputc('\n', stderr);
2316 }
Unleashed OS