]> git.cameronkatri.com Git - mandoc.git/blob - mandocdb.c
git.ckatri.com / mandoc.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
history | raw | HEAD
Provide a helper function macro_or_word() and use it to prune the
[mandoc.git] / mandocdb.c
1 /* $Id: mandocdb.c,v 1.171 2014/11/27 01:58:21 schwarze Exp $ */
2 /*
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 */
18 #include "config.h"
19
20 #include <sys/types.h>
21 #include <sys/stat.h>
22 #include <sys/wait.h>
23
24 #include <assert.h>
25 #include <ctype.h>
26 #include <errno.h>
27 #include <fcntl.h>
28 #if HAVE_FTS
29 #include <fts.h>
30 #else
31 #include "compat_fts.h"
32 #endif
33 #include <getopt.h>
34 #include <limits.h>
35 #include <stddef.h>
36 #include <stdio.h>
37 #include <stdint.h>
38 #include <stdlib.h>
39 #include <string.h>
40 #include <unistd.h>
41
42 #if HAVE_OHASH
43 #include <ohash.h>
44 #else
45 #include "compat_ohash.h"
46 #endif
47 #include <sqlite3.h>
48
49 #include "mdoc.h"
50 #include "man.h"
51 #include "mandoc.h"
52 #include "mandoc_aux.h"
53 #include "manpath.h"
54 #include "mansearch.h"
55
56 extern int mansearch_keymax;
57 extern const char *const mansearch_keynames[];
58
59 #define SQL_EXEC(_v) \
60 if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
61 say("", "%s: %s", (_v), sqlite3_errmsg(db))
62 #define SQL_BIND_TEXT(_s, _i, _v) \
63 if (SQLITE_OK != sqlite3_bind_text \
64 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
65 say(mlink->file, "%s", sqlite3_errmsg(db))
66 #define SQL_BIND_INT(_s, _i, _v) \
67 if (SQLITE_OK != sqlite3_bind_int \
68 ((_s), (_i)++, (_v))) \
69 say(mlink->file, "%s", sqlite3_errmsg(db))
70 #define SQL_BIND_INT64(_s, _i, _v) \
71 if (SQLITE_OK != sqlite3_bind_int64 \
72 ((_s), (_i)++, (_v))) \
73 say(mlink->file, "%s", sqlite3_errmsg(db))
74 #define SQL_STEP(_s) \
75 if (SQLITE_DONE != sqlite3_step((_s))) \
76 say(mlink->file, "%s", sqlite3_errmsg(db))
77
78 enum op {
79 OP_DEFAULT = 0, /* new dbs from dir list or default config */
80 OP_CONFFILE, /* new databases from custom config file */
81 OP_UPDATE, /* delete/add entries in existing database */
82 OP_DELETE, /* delete entries from existing database */
83 OP_TEST /* change no databases, report potential problems */
84 };
85
86 struct str {
87 char *rendered; /* key in UTF-8 or ASCII form */
88 const struct mpage *mpage; /* if set, the owning parse */
89 uint64_t mask; /* bitmask in sequence */
90 char key[]; /* may contain escape sequences */
91 };
92
93 struct inodev {
94 ino_t st_ino;
95 dev_t st_dev;
96 };
97
98 struct mpage {
99 struct inodev inodev; /* used for hashing routine */
100 int64_t pageid; /* pageid in mpages SQL table */
101 char *sec; /* section from file content */
102 char *arch; /* architecture from file content */
103 char *title; /* title from file content */
104 char *desc; /* description from file content */
105 struct mlink *mlinks; /* singly linked list */
106 int form; /* format from file content */
107 };
108
109 struct mlink {
110 char file[PATH_MAX]; /* filename rel. to manpath */
111 char *dsec; /* section from directory */
112 char *arch; /* architecture from directory */
113 char *name; /* name from file name (not empty) */
114 char *fsec; /* section from file name suffix */
115 struct mlink *next; /* singly linked list */
116 struct mpage *mpage; /* parent */
117 int dform; /* format from directory */
118 int fform; /* format from file name suffix */
119 int gzip; /* filename has a .gz suffix */
120 };
121
122 enum stmt {
123 STMT_DELETE_PAGE = 0, /* delete mpage */
124 STMT_INSERT_PAGE, /* insert mpage */
125 STMT_INSERT_LINK, /* insert mlink */
126 STMT_INSERT_NAME, /* insert name */
127 STMT_INSERT_KEY, /* insert parsed key */
128 STMT__MAX
129 };
130
131 typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
132
133 struct mdoc_handler {
134 mdoc_fp fp; /* optional handler */
135 uint64_t mask; /* set unless handler returns 0 */
136 };
137
138 static void dbclose(int);
139 static void dbadd(struct mpage *, struct mchars *);
140 static void dbadd_mlink(const struct mlink *mlink);
141 static void dbadd_mlink_name(const struct mlink *mlink);
142 static int dbopen(int);
143 static void dbprune(void);
144 static void filescan(const char *);
145 static void *hash_alloc(size_t, void *);
146 static void hash_free(void *, void *);
147 static void *hash_calloc(size_t, size_t, void *);
148 static void mlink_add(struct mlink *, const struct stat *);
149 static void mlink_check(struct mpage *, struct mlink *);
150 static void mlink_free(struct mlink *);
151 static void mlinks_undupe(struct mpage *);
152 static void mpages_free(void);
153 static void mpages_merge(struct mchars *, struct mparse *);
154 static void names_check(void);
155 static void parse_cat(struct mpage *, int);
156 static void parse_man(struct mpage *, const struct man_node *);
157 static void parse_mdoc(struct mpage *, const struct mdoc_node *);
158 static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
159 static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
160 static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
161 static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
162 static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
163 static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
164 static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
165 static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
166 static void putkey(const struct mpage *, char *, uint64_t);
167 static void putkeys(const struct mpage *,
168 const char *, size_t, uint64_t);
169 static void putmdockey(const struct mpage *,
170 const struct mdoc_node *, uint64_t);
171 static void render_key(struct mchars *, struct str *);
172 static void say(const char *, const char *, ...);
173 static int set_basedir(const char *, int);
174 static int treescan(void);
175 static size_t utf8(unsigned int, char [7]);
176
177 static char tempfilename[32];
178 static char *progname;
179 static int nodb; /* no database changes */
180 static int mparse_options; /* abort the parse early */
181 static int use_all; /* use all found files */
182 static int debug; /* print what we're doing */
183 static int warnings; /* warn about crap */
184 static int write_utf8; /* write UTF-8 output; else ASCII */
185 static int exitcode; /* to be returned by main */
186 static enum op op; /* operational mode */
187 static char basedir[PATH_MAX]; /* current base directory */
188 static struct ohash mpages; /* table of distinct manual pages */
189 static struct ohash mlinks; /* table of directory entries */
190 static struct ohash names; /* table of all names */
191 static struct ohash strings; /* table of all strings */
192 static sqlite3 *db = NULL; /* current database */
193 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
194 static uint64_t name_mask;
195
196 static const struct mdoc_handler mdocs[MDOC_MAX] = {
197 { NULL, 0 }, /* Ap */
198 { NULL, 0 }, /* Dd */
199 { NULL, 0 }, /* Dt */
200 { NULL, 0 }, /* Os */
201 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
202 { parse_mdoc_head, TYPE_Ss }, /* Ss */
203 { NULL, 0 }, /* Pp */
204 { NULL, 0 }, /* D1 */
205 { NULL, 0 }, /* Dl */
206 { NULL, 0 }, /* Bd */
207 { NULL, 0 }, /* Ed */
208 { NULL, 0 }, /* Bl */
209 { NULL, 0 }, /* El */
210 { NULL, 0 }, /* It */
211 { NULL, 0 }, /* Ad */
212 { NULL, TYPE_An }, /* An */
213 { NULL, TYPE_Ar }, /* Ar */
214 { NULL, TYPE_Cd }, /* Cd */
215 { NULL, TYPE_Cm }, /* Cm */
216 { NULL, TYPE_Dv }, /* Dv */
217 { NULL, TYPE_Er }, /* Er */
218 { NULL, TYPE_Ev }, /* Ev */
219 { NULL, 0 }, /* Ex */
220 { NULL, TYPE_Fa }, /* Fa */
221 { parse_mdoc_Fd, 0 }, /* Fd */
222 { NULL, TYPE_Fl }, /* Fl */
223 { parse_mdoc_Fn, 0 }, /* Fn */
224 { NULL, TYPE_Ft }, /* Ft */
225 { NULL, TYPE_Ic }, /* Ic */
226 { NULL, TYPE_In }, /* In */
227 { NULL, TYPE_Li }, /* Li */
228 { parse_mdoc_Nd, 0 }, /* Nd */
229 { parse_mdoc_Nm, 0 }, /* Nm */
230 { NULL, 0 }, /* Op */
231 { NULL, 0 }, /* Ot */
232 { NULL, TYPE_Pa }, /* Pa */
233 { NULL, 0 }, /* Rv */
234 { NULL, TYPE_St }, /* St */
235 { NULL, TYPE_Va }, /* Va */
236 { parse_mdoc_body, TYPE_Va }, /* Vt */
237 { parse_mdoc_Xr, 0 }, /* Xr */
238 { NULL, 0 }, /* %A */
239 { NULL, 0 }, /* %B */
240 { NULL, 0 }, /* %D */
241 { NULL, 0 }, /* %I */
242 { NULL, 0 }, /* %J */
243 { NULL, 0 }, /* %N */
244 { NULL, 0 }, /* %O */
245 { NULL, 0 }, /* %P */
246 { NULL, 0 }, /* %R */
247 { NULL, 0 }, /* %T */
248 { NULL, 0 }, /* %V */
249 { NULL, 0 }, /* Ac */
250 { NULL, 0 }, /* Ao */
251 { NULL, 0 }, /* Aq */
252 { NULL, TYPE_At }, /* At */
253 { NULL, 0 }, /* Bc */
254 { NULL, 0 }, /* Bf */
255 { NULL, 0 }, /* Bo */
256 { NULL, 0 }, /* Bq */
257 { NULL, TYPE_Bsx }, /* Bsx */
258 { NULL, TYPE_Bx }, /* Bx */
259 { NULL, 0 }, /* Db */
260 { NULL, 0 }, /* Dc */
261 { NULL, 0 }, /* Do */
262 { NULL, 0 }, /* Dq */
263 { NULL, 0 }, /* Ec */
264 { NULL, 0 }, /* Ef */
265 { NULL, TYPE_Em }, /* Em */
266 { NULL, 0 }, /* Eo */
267 { NULL, TYPE_Fx }, /* Fx */
268 { NULL, TYPE_Ms }, /* Ms */
269 { NULL, 0 }, /* No */
270 { NULL, 0 }, /* Ns */
271 { NULL, TYPE_Nx }, /* Nx */
272 { NULL, TYPE_Ox }, /* Ox */
273 { NULL, 0 }, /* Pc */
274 { NULL, 0 }, /* Pf */
275 { NULL, 0 }, /* Po */
276 { NULL, 0 }, /* Pq */
277 { NULL, 0 }, /* Qc */
278 { NULL, 0 }, /* Ql */
279 { NULL, 0 }, /* Qo */
280 { NULL, 0 }, /* Qq */
281 { NULL, 0 }, /* Re */
282 { NULL, 0 }, /* Rs */
283 { NULL, 0 }, /* Sc */
284 { NULL, 0 }, /* So */
285 { NULL, 0 }, /* Sq */
286 { NULL, 0 }, /* Sm */
287 { NULL, 0 }, /* Sx */
288 { NULL, TYPE_Sy }, /* Sy */
289 { NULL, TYPE_Tn }, /* Tn */
290 { NULL, 0 }, /* Ux */
291 { NULL, 0 }, /* Xc */
292 { NULL, 0 }, /* Xo */
293 { parse_mdoc_head, 0 }, /* Fo */
294 { NULL, 0 }, /* Fc */
295 { NULL, 0 }, /* Oo */
296 { NULL, 0 }, /* Oc */
297 { NULL, 0 }, /* Bk */
298 { NULL, 0 }, /* Ek */
299 { NULL, 0 }, /* Bt */
300 { NULL, 0 }, /* Hf */
301 { NULL, 0 }, /* Fr */
302 { NULL, 0 }, /* Ud */
303 { NULL, TYPE_Lb }, /* Lb */
304 { NULL, 0 }, /* Lp */
305 { NULL, TYPE_Lk }, /* Lk */
306 { NULL, TYPE_Mt }, /* Mt */
307 { NULL, 0 }, /* Brq */
308 { NULL, 0 }, /* Bro */
309 { NULL, 0 }, /* Brc */
310 { NULL, 0 }, /* %C */
311 { NULL, 0 }, /* Es */
312 { NULL, 0 }, /* En */
313 { NULL, TYPE_Dx }, /* Dx */
314 { NULL, 0 }, /* %Q */
315 { NULL, 0 }, /* br */
316 { NULL, 0 }, /* sp */
317 { NULL, 0 }, /* %U */
318 { NULL, 0 }, /* Ta */
319 { NULL, 0 }, /* ll */
320 };
321
322
323 int
324 main(int argc, char *argv[])
325 {
326 int ch, i;
327 size_t j, sz;
328 const char *path_arg;
329 struct mchars *mc;
330 struct manpaths dirs;
331 struct mparse *mp;
332 struct ohash_info mpages_info, mlinks_info;
333
334 memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
335 memset(&dirs, 0, sizeof(struct manpaths));
336
337 mpages_info.alloc = mlinks_info.alloc = hash_alloc;
338 mpages_info.calloc = mlinks_info.calloc = hash_calloc;
339 mpages_info.free = mlinks_info.free = hash_free;
340
341 mpages_info.key_offset = offsetof(struct mpage, inodev);
342 mlinks_info.key_offset = offsetof(struct mlink, file);
343
344 progname = strrchr(argv[0], '/');
345 if (progname == NULL)
346 progname = argv[0];
347 else
348 ++progname;
349
350 /*
351 * We accept a few different invocations.
352 * The CHECKOP macro makes sure that invocation styles don't
353 * clobber each other.
354 */
355 #define CHECKOP(_op, _ch) do \
356 if (OP_DEFAULT != (_op)) { \
357 fprintf(stderr, "%s: -%c: Conflicting option\n", \
358 progname, (_ch)); \
359 goto usage; \
360 } while (/*CONSTCOND*/0)
361
362 path_arg = NULL;
363 op = OP_DEFAULT;
364
365 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
366 switch (ch) {
367 case 'a':
368 use_all = 1;
369 break;
370 case 'C':
371 CHECKOP(op, ch);
372 path_arg = optarg;
373 op = OP_CONFFILE;
374 break;
375 case 'D':
376 debug++;
377 break;
378 case 'd':
379 CHECKOP(op, ch);
380 path_arg = optarg;
381 op = OP_UPDATE;
382 break;
383 case 'n':
384 nodb = 1;
385 break;
386 case 'p':
387 warnings = 1;
388 break;
389 case 'Q':
390 mparse_options |= MPARSE_QUICK;
391 break;
392 case 'T':
393 if (strcmp(optarg, "utf8")) {
394 fprintf(stderr, "%s: -T%s: "
395 "Unsupported output format\n",
396 progname, optarg);
397 goto usage;
398 }
399 write_utf8 = 1;
400 break;
401 case 't':
402 CHECKOP(op, ch);
403 dup2(STDOUT_FILENO, STDERR_FILENO);
404 op = OP_TEST;
405 nodb = warnings = 1;
406 break;
407 case 'u':
408 CHECKOP(op, ch);
409 path_arg = optarg;
410 op = OP_DELETE;
411 break;
412 case 'v':
413 /* Compatibility with espie@'s makewhatis. */
414 break;
415 default:
416 goto usage;
417 }
418
419 argc -= optind;
420 argv += optind;
421
422 if (OP_CONFFILE == op && argc > 0) {
423 fprintf(stderr, "%s: -C: Too many arguments\n",
424 progname);
425 goto usage;
426 }
427
428 exitcode = (int)MANDOCLEVEL_OK;
429 mc = mchars_alloc();
430 mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
431 mc, NULL);
432 ohash_init(&mpages, 6, &mpages_info);
433 ohash_init(&mlinks, 6, &mlinks_info);
434
435 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
436
437 /*
438 * Most of these deal with a specific directory.
439 * Jump into that directory first.
440 */
441 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
442 goto out;
443
444 if (dbopen(1)) {
445 /*
446 * The existing database is usable. Process
447 * all files specified on the command-line.
448 */
449 use_all = 1;
450 for (i = 0; i < argc; i++)
451 filescan(argv[i]);
452 if (OP_TEST != op)
453 dbprune();
454 } else {
455 /*
456 * Database missing or corrupt.
457 * Recreate from scratch.
458 */
459 exitcode = (int)MANDOCLEVEL_OK;
460 op = OP_DEFAULT;
461 if (0 == treescan())
462 goto out;
463 if (0 == dbopen(0))
464 goto out;
465 }
466 if (OP_DELETE != op)
467 mpages_merge(mc, mp);
468 dbclose(OP_DEFAULT == op ? 0 : 1);
469 } else {
470 /*
471 * If we have arguments, use them as our manpaths.
472 * If we don't, grok from manpath(1) or however else
473 * manpath_parse() wants to do it.
474 */
475 if (argc > 0) {
476 dirs.paths = mandoc_reallocarray(NULL,
477 argc, sizeof(char *));
478 dirs.sz = (size_t)argc;
479 for (i = 0; i < argc; i++)
480 dirs.paths[i] = mandoc_strdup(argv[i]);
481 } else
482 manpath_parse(&dirs, path_arg, NULL, NULL);
483
484 if (0 == dirs.sz) {
485 exitcode = (int)MANDOCLEVEL_BADARG;
486 say("", "Empty manpath");
487 }
488
489 /*
490 * First scan the tree rooted at a base directory, then
491 * build a new database and finally move it into place.
492 * Ignore zero-length directories and strip trailing
493 * slashes.
494 */
495 for (j = 0; j < dirs.sz; j++) {
496 sz = strlen(dirs.paths[j]);
497 if (sz && '/' == dirs.paths[j][sz - 1])
498 dirs.paths[j][--sz] = '\0';
499 if (0 == sz)
500 continue;
501
502 if (j) {
503 ohash_init(&mpages, 6, &mpages_info);
504 ohash_init(&mlinks, 6, &mlinks_info);
505 }
506
507 if (0 == set_basedir(dirs.paths[j], argc > 0))
508 continue;
509 if (0 == treescan())
510 continue;
511 if (0 == dbopen(0))
512 continue;
513
514 mpages_merge(mc, mp);
515 if (warnings && !nodb &&
516 ! (MPARSE_QUICK & mparse_options))
517 names_check();
518 dbclose(0);
519
520 if (j + 1 < dirs.sz) {
521 mpages_free();
522 ohash_delete(&mpages);
523 ohash_delete(&mlinks);
524 }
525 }
526 }
527 out:
528 manpath_free(&dirs);
529 mparse_free(mp);
530 mchars_free(mc);
531 mpages_free();
532 ohash_delete(&mpages);
533 ohash_delete(&mlinks);
534 return(exitcode);
535 usage:
536 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
537 " %s [-aDnpQ] [-Tutf8] dir ...\n"
538 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
539 " %s [-Dnp] -u dir [file ...]\n"
540 " %s [-Q] -t file ...\n",
541 progname, progname, progname,
542 progname, progname);
543
544 return((int)MANDOCLEVEL_BADARG);
545 }
546
547 /*
548 * Scan a directory tree rooted at "basedir" for manpages.
549 * We use fts(), scanning directory parts along the way for clues to our
550 * section and architecture.
551 *
552 * If use_all has been specified, grok all files.
553 * If not, sanitise paths to the following:
554 *
555 * [./]man*[/<arch>]/<name>.<section>
556 * or
557 * [./]cat<section>[/<arch>]/<name>.0
558 *
559 * TODO: accomodate for multi-language directories.
560 */
561 static int
562 treescan(void)
563 {
564 char buf[PATH_MAX];
565 FTS *f;
566 FTSENT *ff;
567 struct mlink *mlink;
568 int dform, gzip;
569 char *dsec, *arch, *fsec, *cp;
570 const char *path;
571 const char *argv[2];
572
573 argv[0] = ".";
574 argv[1] = (char *)NULL;
575
576 f = fts_open((char * const *)argv,
577 FTS_PHYSICAL | FTS_NOCHDIR, NULL);
578 if (NULL == f) {
579 exitcode = (int)MANDOCLEVEL_SYSERR;
580 say("", "&fts_open");
581 return(0);
582 }
583
584 dsec = arch = NULL;
585 dform = FORM_NONE;
586
587 while (NULL != (ff = fts_read(f))) {
588 path = ff->fts_path + 2;
589 switch (ff->fts_info) {
590
591 /*
592 * Symbolic links require various sanity checks,
593 * then get handled just like regular files.
594 */
595 case FTS_SL:
596 if (NULL == realpath(path, buf)) {
597 if (warnings)
598 say(path, "&realpath");
599 continue;
600 }
601 if (strstr(buf, basedir) != buf) {
602 if (warnings) say("",
603 "%s: outside base directory", buf);
604 continue;
605 }
606 /* Use logical inode to avoid mpages dupe. */
607 if (-1 == stat(path, ff->fts_statp)) {
608 if (warnings)
609 say(path, "&stat");
610 continue;
611 }
612 /* FALLTHROUGH */
613
614 /*
615 * If we're a regular file, add an mlink by using the
616 * stored directory data and handling the filename.
617 */
618 case FTS_F:
619 if (0 == strcmp(path, MANDOC_DB))
620 continue;
621 if ( ! use_all && ff->fts_level < 2) {
622 if (warnings)
623 say(path, "Extraneous file");
624 continue;
625 }
626 gzip = 0;
627 fsec = NULL;
628 while (NULL == fsec) {
629 fsec = strrchr(ff->fts_name, '.');
630 if (NULL == fsec || strcmp(fsec+1, "gz"))
631 break;
632 gzip = 1;
633 *fsec = '\0';
634 fsec = NULL;
635 }
636 if (NULL == fsec) {
637 if ( ! use_all) {
638 if (warnings)
639 say(path,
640 "No filename suffix");
641 continue;
642 }
643 } else if (0 == strcmp(++fsec, "html")) {
644 if (warnings)
645 say(path, "Skip html");
646 continue;
647 } else if (0 == strcmp(fsec, "ps")) {
648 if (warnings)
649 say(path, "Skip ps");
650 continue;
651 } else if (0 == strcmp(fsec, "pdf")) {
652 if (warnings)
653 say(path, "Skip pdf");
654 continue;
655 } else if ( ! use_all &&
656 ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
657 (FORM_CAT == dform && strcmp(fsec, "0")))) {
658 if (warnings)
659 say(path, "Wrong filename suffix");
660 continue;
661 } else
662 fsec[-1] = '\0';
663
664 mlink = mandoc_calloc(1, sizeof(struct mlink));
665 if (strlcpy(mlink->file, path,
666 sizeof(mlink->file)) >=
667 sizeof(mlink->file)) {
668 say(path, "Filename too long");
669 free(mlink);
670 continue;
671 }
672 mlink->dform = dform;
673 mlink->dsec = dsec;
674 mlink->arch = arch;
675 mlink->name = ff->fts_name;
676 mlink->fsec = fsec;
677 mlink->gzip = gzip;
678 mlink_add(mlink, ff->fts_statp);
679 continue;
680
681 case FTS_D:
682 /* FALLTHROUGH */
683 case FTS_DP:
684 break;
685
686 default:
687 if (warnings)
688 say(path, "Not a regular file");
689 continue;
690 }
691
692 switch (ff->fts_level) {
693 case 0:
694 /* Ignore the root directory. */
695 break;
696 case 1:
697 /*
698 * This might contain manX/ or catX/.
699 * Try to infer this from the name.
700 * If we're not in use_all, enforce it.
701 */
702 cp = ff->fts_name;
703 if (FTS_DP == ff->fts_info)
704 break;
705
706 if (0 == strncmp(cp, "man", 3)) {
707 dform = FORM_SRC;
708 dsec = cp + 3;
709 } else if (0 == strncmp(cp, "cat", 3)) {
710 dform = FORM_CAT;
711 dsec = cp + 3;
712 } else {
713 dform = FORM_NONE;
714 dsec = NULL;
715 }
716
717 if (NULL != dsec || use_all)
718 break;
719
720 if (warnings)
721 say(path, "Unknown directory part");
722 fts_set(f, ff, FTS_SKIP);
723 break;
724 case 2:
725 /*
726 * Possibly our architecture.
727 * If we're descending, keep tabs on it.
728 */
729 if (FTS_DP != ff->fts_info && NULL != dsec)
730 arch = ff->fts_name;
731 else
732 arch = NULL;
733 break;
734 default:
735 if (FTS_DP == ff->fts_info || use_all)
736 break;
737 if (warnings)
738 say(path, "Extraneous directory part");
739 fts_set(f, ff, FTS_SKIP);
740 break;
741 }
742 }
743
744 fts_close(f);
745 return(1);
746 }
747
748 /*
749 * Add a file to the mlinks table.
750 * Do not verify that it's a "valid" looking manpage (we'll do that
751 * later).
752 *
753 * Try to infer the manual section, architecture, and page name from the
754 * path, assuming it looks like
755 *
756 * [./]man*[/<arch>]/<name>.<section>
757 * or
758 * [./]cat<section>[/<arch>]/<name>.0
759 *
760 * See treescan() for the fts(3) version of this.
761 */
762 static void
763 filescan(const char *file)
764 {
765 char buf[PATH_MAX];
766 struct stat st;
767 struct mlink *mlink;
768 char *p, *start;
769
770 assert(use_all);
771
772 if (0 == strncmp(file, "./", 2))
773 file += 2;
774
775 /*
776 * We have to do lstat(2) before realpath(3) loses
777 * the information whether this is a symbolic link.
778 * We need to know that because for symbolic links,
779 * we want to use the orginal file name, while for
780 * regular files, we want to use the real path.
781 */
782 if (-1 == lstat(file, &st)) {
783 exitcode = (int)MANDOCLEVEL_BADARG;
784 say(file, "&lstat");
785 return;
786 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
787 exitcode = (int)MANDOCLEVEL_BADARG;
788 say(file, "Not a regular file");
789 return;
790 }
791
792 /*
793 * We have to resolve the file name to the real path
794 * in any case for the base directory check.
795 */
796 if (NULL == realpath(file, buf)) {
797 exitcode = (int)MANDOCLEVEL_BADARG;
798 say(file, "&realpath");
799 return;
800 }
801
802 if (OP_TEST == op)
803 start = buf;
804 else if (strstr(buf, basedir) == buf)
805 start = buf + strlen(basedir);
806 else {
807 exitcode = (int)MANDOCLEVEL_BADARG;
808 say("", "%s: outside base directory", buf);
809 return;
810 }
811
812 /*
813 * Now we are sure the file is inside our tree.
814 * If it is a symbolic link, ignore the real path
815 * and use the original name.
816 * This implies passing stuff like "cat1/../man1/foo.1"
817 * on the command line won't work. So don't do that.
818 * Note the stat(2) can still fail if the link target
819 * doesn't exist.
820 */
821 if (S_IFLNK & st.st_mode) {
822 if (-1 == stat(buf, &st)) {
823 exitcode = (int)MANDOCLEVEL_BADARG;
824 say(file, "&stat");
825 return;
826 }
827 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
828 say(file, "Filename too long");
829 return;
830 }
831 start = buf;
832 if (OP_TEST != op && strstr(buf, basedir) == buf)
833 start += strlen(basedir);
834 }
835
836 mlink = mandoc_calloc(1, sizeof(struct mlink));
837 mlink->dform = FORM_NONE;
838 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
839 sizeof(mlink->file)) {
840 say(start, "Filename too long");
841 return;
842 }
843
844 /*
845 * First try to guess our directory structure.
846 * If we find a separator, try to look for man* or cat*.
847 * If we find one of these and what's underneath is a directory,
848 * assume it's an architecture.
849 */
850 if (NULL != (p = strchr(start, '/'))) {
851 *p++ = '\0';
852 if (0 == strncmp(start, "man", 3)) {
853 mlink->dform = FORM_SRC;
854 mlink->dsec = start + 3;
855 } else if (0 == strncmp(start, "cat", 3)) {
856 mlink->dform = FORM_CAT;
857 mlink->dsec = start + 3;
858 }
859
860 start = p;
861 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
862 *p++ = '\0';
863 mlink->arch = start;
864 start = p;
865 }
866 }
867
868 /*
869 * Now check the file suffix.
870 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
871 */
872 p = strrchr(start, '\0');
873 while (p-- > start && '/' != *p && '.' != *p)
874 /* Loop. */ ;
875
876 if ('.' == *p) {
877 *p++ = '\0';
878 mlink->fsec = p;
879 }
880
881 /*
882 * Now try to parse the name.
883 * Use the filename portion of the path.
884 */
885 mlink->name = start;
886 if (NULL != (p = strrchr(start, '/'))) {
887 mlink->name = p + 1;
888 *p = '\0';
889 }
890 mlink_add(mlink, &st);
891 }
892
893 static void
894 mlink_add(struct mlink *mlink, const struct stat *st)
895 {
896 struct inodev inodev;
897 struct mpage *mpage;
898 unsigned int slot;
899
900 assert(NULL != mlink->file);
901
902 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
903 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
904 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
905 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
906
907 if ('0' == *mlink->fsec) {
908 free(mlink->fsec);
909 mlink->fsec = mandoc_strdup(mlink->dsec);
910 mlink->fform = FORM_CAT;
911 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
912 mlink->fform = FORM_SRC;
913 else
914 mlink->fform = FORM_NONE;
915
916 slot = ohash_qlookup(&mlinks, mlink->file);
917 assert(NULL == ohash_find(&mlinks, slot));
918 ohash_insert(&mlinks, slot, mlink);
919
920 inodev.st_ino = st->st_ino;
921 inodev.st_dev = st->st_dev;
922 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
923 sizeof(struct inodev), inodev.st_ino);
924 mpage = ohash_find(&mpages, slot);
925 if (NULL == mpage) {
926 mpage = mandoc_calloc(1, sizeof(struct mpage));
927 mpage->inodev.st_ino = inodev.st_ino;
928 mpage->inodev.st_dev = inodev.st_dev;
929 ohash_insert(&mpages, slot, mpage);
930 } else
931 mlink->next = mpage->mlinks;
932 mpage->mlinks = mlink;
933 mlink->mpage = mpage;
934 }
935
936 static void
937 mlink_free(struct mlink *mlink)
938 {
939
940 free(mlink->dsec);
941 free(mlink->arch);
942 free(mlink->name);
943 free(mlink->fsec);
944 free(mlink);
945 }
946
947 static void
948 mpages_free(void)
949 {
950 struct mpage *mpage;
951 struct mlink *mlink;
952 unsigned int slot;
953
954 mpage = ohash_first(&mpages, &slot);
955 while (NULL != mpage) {
956 while (NULL != (mlink = mpage->mlinks)) {
957 mpage->mlinks = mlink->next;
958 mlink_free(mlink);
959 }
960 free(mpage->sec);
961 free(mpage->arch);
962 free(mpage->title);
963 free(mpage->desc);
964 free(mpage);
965 mpage = ohash_next(&mpages, &slot);
966 }
967 }
968
969 /*
970 * For each mlink to the mpage, check whether the path looks like
971 * it is formatted, and if it does, check whether a source manual
972 * exists by the same name, ignoring the suffix.
973 * If both conditions hold, drop the mlink.
974 */
975 static void
976 mlinks_undupe(struct mpage *mpage)
977 {
978 char buf[PATH_MAX];
979 struct mlink **prev;
980 struct mlink *mlink;
981 char *bufp;
982
983 mpage->form = FORM_CAT;
984 prev = &mpage->mlinks;
985 while (NULL != (mlink = *prev)) {
986 if (FORM_CAT != mlink->dform) {
987 mpage->form = FORM_NONE;
988 goto nextlink;
989 }
990 (void)strlcpy(buf, mlink->file, sizeof(buf));
991 bufp = strstr(buf, "cat");
992 assert(NULL != bufp);
993 memcpy(bufp, "man", 3);
994 if (NULL != (bufp = strrchr(buf, '.')))
995 *++bufp = '\0';
996 (void)strlcat(buf, mlink->dsec, sizeof(buf));
997 if (NULL == ohash_find(&mlinks,
998 ohash_qlookup(&mlinks, buf)))
999 goto nextlink;
1000 if (warnings)
1001 say(mlink->file, "Man source exists: %s", buf);
1002 if (use_all)
1003 goto nextlink;
1004 *prev = mlink->next;
1005 mlink_free(mlink);
1006 continue;
1007 nextlink:
1008 prev = &(*prev)->next;
1009 }
1010 }
1011
1012 static void
1013 mlink_check(struct mpage *mpage, struct mlink *mlink)
1014 {
1015 struct str *str;
1016 unsigned int slot;
1017
1018 /*
1019 * Check whether the manual section given in a file
1020 * agrees with the directory where the file is located.
1021 * Some manuals have suffixes like (3p) on their
1022 * section number either inside the file or in the
1023 * directory name, some are linked into more than one
1024 * section, like encrypt(1) = makekey(8).
1025 */
1026
1027 if (FORM_SRC == mpage->form &&
1028 strcasecmp(mpage->sec, mlink->dsec))
1029 say(mlink->file, "Section \"%s\" manual in %s directory",
1030 mpage->sec, mlink->dsec);
1031
1032 /*
1033 * Manual page directories exist for each kernel
1034 * architecture as returned by machine(1).
1035 * However, many manuals only depend on the
1036 * application architecture as returned by arch(1).
1037 * For example, some (2/ARM) manuals are shared
1038 * across the "armish" and "zaurus" kernel
1039 * architectures.
1040 * A few manuals are even shared across completely
1041 * different architectures, for example fdformat(1)
1042 * on amd64, i386, sparc, and sparc64.
1043 */
1044
1045 if (strcasecmp(mpage->arch, mlink->arch))
1046 say(mlink->file, "Architecture \"%s\" manual in "
1047 "\"%s\" directory", mpage->arch, mlink->arch);
1048
1049 /*
1050 * XXX
1051 * parse_cat() doesn't set NAME_TITLE yet.
1052 */
1053
1054 if (FORM_CAT == mpage->form)
1055 return;
1056
1057 /*
1058 * Check whether this mlink
1059 * appears as a name in the NAME section.
1060 */
1061
1062 slot = ohash_qlookup(&names, mlink->name);
1063 str = ohash_find(&names, slot);
1064 assert(NULL != str);
1065 if ( ! (NAME_TITLE & str->mask))
1066 say(mlink->file, "Name missing in NAME section");
1067 }
1068
1069 /*
1070 * Run through the files in the global vector "mpages"
1071 * and add them to the database specified in "basedir".
1072 *
1073 * This handles the parsing scheme itself, using the cues of directory
1074 * and filename to determine whether the file is parsable or not.
1075 */
1076 static void
1077 mpages_merge(struct mchars *mc, struct mparse *mp)
1078 {
1079 char any[] = "any";
1080 struct ohash_info str_info;
1081 struct mpage *mpage, *mpage_dest;
1082 struct mlink *mlink, *mlink_dest;
1083 struct mdoc *mdoc;
1084 struct man *man;
1085 char *sodest;
1086 char *cp;
1087 int fd;
1088 unsigned int pslot;
1089 enum mandoclevel lvl;
1090
1091 str_info.alloc = hash_alloc;
1092 str_info.calloc = hash_calloc;
1093 str_info.free = hash_free;
1094 str_info.key_offset = offsetof(struct str, key);
1095
1096 if ( ! nodb)
1097 SQL_EXEC("BEGIN TRANSACTION");
1098
1099 mpage = ohash_first(&mpages, &pslot);
1100 while (mpage != NULL) {
1101 mlinks_undupe(mpage);
1102 if (mpage->mlinks == NULL) {
1103 mpage = ohash_next(&mpages, &pslot);
1104 continue;
1105 }
1106
1107 name_mask = NAME_MASK;
1108 ohash_init(&names, 4, &str_info);
1109 ohash_init(&strings, 6, &str_info);
1110 mparse_reset(mp);
1111 mdoc = NULL;
1112 man = NULL;
1113 sodest = NULL;
1114
1115 mparse_open(mp, &fd, mpage->mlinks->file);
1116 if (fd == -1) {
1117 say(mpage->mlinks->file, "&open");
1118 goto nextpage;
1119 }
1120
1121 /*
1122 * Try interpreting the file as mdoc(7) or man(7)
1123 * source code, unless it is already known to be
1124 * formatted. Fall back to formatted mode.
1125 */
1126 if (mpage->mlinks->dform != FORM_CAT ||
1127 mpage->mlinks->fform != FORM_CAT) {
1128 lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1129 if (lvl < MANDOCLEVEL_FATAL)
1130 mparse_result(mp, &mdoc, &man, &sodest);
1131 }
1132
1133 if (sodest != NULL) {
1134 mlink_dest = ohash_find(&mlinks,
1135 ohash_qlookup(&mlinks, sodest));
1136 if (mlink_dest == NULL) {
1137 mandoc_asprintf(&cp, "%s.gz", sodest);
1138 mlink_dest = ohash_find(&mlinks,
1139 ohash_qlookup(&mlinks, cp));
1140 free(cp);
1141 }
1142 if (mlink_dest != NULL) {
1143
1144 /* The .so target exists. */
1145
1146 mpage_dest = mlink_dest->mpage;
1147 mlink = mpage->mlinks;
1148 while (1) {
1149 mlink->mpage = mpage_dest;
1150
1151 /*
1152 * If the target was already
1153 * processed, add the links
1154 * to the database now.
1155 * Otherwise, this will
1156 * happen when we come
1157 * to the target.
1158 */
1159
1160 if (mpage_dest->pageid)
1161 dbadd_mlink_name(mlink);
1162
1163 if (mlink->next == NULL)
1164 break;
1165 mlink = mlink->next;
1166 }
1167
1168 /* Move all links to the target. */
1169
1170 mlink->next = mlink_dest->next;
1171 mlink_dest->next = mpage->mlinks;
1172 mpage->mlinks = NULL;
1173 }
1174 goto nextpage;
1175 } else if (mdoc != NULL) {
1176 mpage->form = FORM_SRC;
1177 mpage->sec = mdoc_meta(mdoc)->msec;
1178 mpage->sec = mandoc_strdup(
1179 mpage->sec == NULL ? "" : mpage->sec);
1180 mpage->arch = mdoc_meta(mdoc)->arch;
1181 mpage->arch = mandoc_strdup(
1182 mpage->arch == NULL ? "" : mpage->arch);
1183 mpage->title =
1184 mandoc_strdup(mdoc_meta(mdoc)->title);
1185 } else if (man != NULL) {
1186 mpage->form = FORM_SRC;
1187 mpage->sec =
1188 mandoc_strdup(man_meta(man)->msec);
1189 mpage->arch =
1190 mandoc_strdup(mpage->mlinks->arch);
1191 mpage->title =
1192 mandoc_strdup(man_meta(man)->title);
1193 } else {
1194 mpage->form = FORM_CAT;
1195 mpage->sec =
1196 mandoc_strdup(mpage->mlinks->dsec);
1197 mpage->arch =
1198 mandoc_strdup(mpage->mlinks->arch);
1199 mpage->title =
1200 mandoc_strdup(mpage->mlinks->name);
1201 }
1202 putkey(mpage, mpage->sec, TYPE_sec);
1203 if (*mpage->arch != '\0')
1204 putkey(mpage, mpage->arch, TYPE_arch);
1205
1206 for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1207 if ('\0' != *mlink->dsec)
1208 putkey(mpage, mlink->dsec, TYPE_sec);
1209 if ('\0' != *mlink->fsec)
1210 putkey(mpage, mlink->fsec, TYPE_sec);
1211 putkey(mpage, '\0' == *mlink->arch ?
1212 any : mlink->arch, TYPE_arch);
1213 putkey(mpage, mlink->name, NAME_FILE);
1214 }
1215
1216 assert(NULL == mpage->desc);
1217 if (NULL != mdoc) {
1218 if (NULL != (cp = mdoc_meta(mdoc)->name))
1219 putkey(mpage, cp, NAME_HEAD);
1220 parse_mdoc(mpage, mdoc_node(mdoc));
1221 } else if (NULL != man)
1222 parse_man(mpage, man_node(man));
1223 else
1224 parse_cat(mpage, fd);
1225 if (NULL == mpage->desc)
1226 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1227
1228 if (warnings && !use_all)
1229 for (mlink = mpage->mlinks; mlink;
1230 mlink = mlink->next)
1231 mlink_check(mpage, mlink);
1232
1233 dbadd(mpage, mc);
1234
1235 nextpage:
1236 if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1237 exitcode = (int)MANDOCLEVEL_SYSERR;
1238 say(mpage->mlinks->file, "&wait gunzip");
1239 }
1240 ohash_delete(&strings);
1241 ohash_delete(&names);
1242 mpage = ohash_next(&mpages, &pslot);
1243 }
1244
1245 if (0 == nodb)
1246 SQL_EXEC("END TRANSACTION");
1247 }
1248
1249 static void
1250 names_check(void)
1251 {
1252 sqlite3_stmt *stmt;
1253 const char *name, *sec, *arch, *key;
1254 int irc;
1255
1256 sqlite3_prepare_v2(db,
1257 "SELECT name, sec, arch, key FROM ("
1258 "SELECT name AS key, pageid FROM names "
1259 "WHERE bits & ? AND NOT EXISTS ("
1260 "SELECT pageid FROM mlinks "
1261 "WHERE mlinks.pageid == names.pageid "
1262 "AND mlinks.name == names.name"
1263 ")"
1264 ") JOIN ("
1265 "SELECT sec, arch, name, pageid FROM mlinks "
1266 "GROUP BY pageid"
1267 ") USING (pageid);",
1268 -1, &stmt, NULL);
1269
1270 if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1271 say("", "%s", sqlite3_errmsg(db));
1272
1273 while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1274 name = (const char *)sqlite3_column_text(stmt, 0);
1275 sec = (const char *)sqlite3_column_text(stmt, 1);
1276 arch = (const char *)sqlite3_column_text(stmt, 2);
1277 key = (const char *)sqlite3_column_text(stmt, 3);
1278 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1279 '\0' == *arch ? "" : "/",
1280 '\0' == *arch ? "" : arch, key);
1281 }
1282 sqlite3_finalize(stmt);
1283 }
1284
1285 static void
1286 parse_cat(struct mpage *mpage, int fd)
1287 {
1288 FILE *stream;
1289 char *line, *p, *title;
1290 size_t len, plen, titlesz;
1291
1292 stream = (-1 == fd) ?
1293 fopen(mpage->mlinks->file, "r") :
1294 fdopen(fd, "r");
1295 if (NULL == stream) {
1296 if (-1 != fd)
1297 close(fd);
1298 if (warnings)
1299 say(mpage->mlinks->file, "&fopen");
1300 return;
1301 }
1302
1303 /* Skip to first blank line. */
1304
1305 while (NULL != (line = fgetln(stream, &len)))
1306 if ('\n' == *line)
1307 break;
1308
1309 /*
1310 * Assume the first line that is not indented
1311 * is the first section header. Skip to it.
1312 */
1313
1314 while (NULL != (line = fgetln(stream, &len)))
1315 if ('\n' != *line && ' ' != *line)
1316 break;
1317
1318 /*
1319 * Read up until the next section into a buffer.
1320 * Strip the leading and trailing newline from each read line,
1321 * appending a trailing space.
1322 * Ignore empty (whitespace-only) lines.
1323 */
1324
1325 titlesz = 0;
1326 title = NULL;
1327
1328 while (NULL != (line = fgetln(stream, &len))) {
1329 if (' ' != *line || '\n' != line[len - 1])
1330 break;
1331 while (len > 0 && isspace((unsigned char)*line)) {
1332 line++;
1333 len--;
1334 }
1335 if (1 == len)
1336 continue;
1337 title = mandoc_realloc(title, titlesz + len);
1338 memcpy(title + titlesz, line, len);
1339 titlesz += len;
1340 title[titlesz - 1] = ' ';
1341 }
1342
1343 /*
1344 * If no page content can be found, or the input line
1345 * is already the next section header, or there is no
1346 * trailing newline, reuse the page title as the page
1347 * description.
1348 */
1349
1350 if (NULL == title || '\0' == *title) {
1351 if (warnings)
1352 say(mpage->mlinks->file,
1353 "Cannot find NAME section");
1354 fclose(stream);
1355 free(title);
1356 return;
1357 }
1358
1359 title = mandoc_realloc(title, titlesz + 1);
1360 title[titlesz] = '\0';
1361
1362 /*
1363 * Skip to the first dash.
1364 * Use the remaining line as the description (no more than 70
1365 * bytes).
1366 */
1367
1368 if (NULL != (p = strstr(title, "- "))) {
1369 for (p += 2; ' ' == *p || '\b' == *p; p++)
1370 /* Skip to next word. */ ;
1371 } else {
1372 if (warnings)
1373 say(mpage->mlinks->file,
1374 "No dash in title line");
1375 p = title;
1376 }
1377
1378 plen = strlen(p);
1379
1380 /* Strip backspace-encoding from line. */
1381
1382 while (NULL != (line = memchr(p, '\b', plen))) {
1383 len = line - p;
1384 if (0 == len) {
1385 memmove(line, line + 1, plen--);
1386 continue;
1387 }
1388 memmove(line - 1, line + 1, plen - len);
1389 plen -= 2;
1390 }
1391
1392 mpage->desc = mandoc_strdup(p);
1393 fclose(stream);
1394 free(title);
1395 }
1396
1397 /*
1398 * Put a type/word pair into the word database for this particular file.
1399 */
1400 static void
1401 putkey(const struct mpage *mpage, char *value, uint64_t type)
1402 {
1403 char *cp;
1404
1405 assert(NULL != value);
1406 if (TYPE_arch == type)
1407 for (cp = value; *cp; cp++)
1408 if (isupper((unsigned char)*cp))
1409 *cp = _tolower((unsigned char)*cp);
1410 putkeys(mpage, value, strlen(value), type);
1411 }
1412
1413 /*
1414 * Grok all nodes at or below a certain mdoc node into putkey().
1415 */
1416 static void
1417 putmdockey(const struct mpage *mpage,
1418 const struct mdoc_node *n, uint64_t m)
1419 {
1420
1421 for ( ; NULL != n; n = n->next) {
1422 if (NULL != n->child)
1423 putmdockey(mpage, n->child, m);
1424 if (MDOC_TEXT == n->type)
1425 putkey(mpage, n->string, m);
1426 }
1427 }
1428
1429 static void
1430 parse_man(struct mpage *mpage, const struct man_node *n)
1431 {
1432 const struct man_node *head, *body;
1433 char *start, *title;
1434 char byte;
1435 size_t sz;
1436
1437 if (NULL == n)
1438 return;
1439
1440 /*
1441 * We're only searching for one thing: the first text child in
1442 * the BODY of a NAME section. Since we don't keep track of
1443 * sections in -man, run some hoops to find out whether we're in
1444 * the correct section or not.
1445 */
1446
1447 if (MAN_BODY == n->type && MAN_SH == n->tok) {
1448 body = n;
1449 assert(body->parent);
1450 if (NULL != (head = body->parent->head) &&
1451 1 == head->nchild &&
1452 NULL != (head = (head->child)) &&
1453 MAN_TEXT == head->type &&
1454 0 == strcmp(head->string, "NAME") &&
1455 NULL != body->child) {
1456
1457 /*
1458 * Suck the entire NAME section into memory.
1459 * Yes, we might run away.
1460 * But too many manuals have big, spread-out
1461 * NAME sections over many lines.
1462 */
1463
1464 title = NULL;
1465 man_deroff(&title, body);
1466 if (NULL == title)
1467 return;
1468
1469 /*
1470 * Go through a special heuristic dance here.
1471 * Conventionally, one or more manual names are
1472 * comma-specified prior to a whitespace, then a
1473 * dash, then a description. Try to puzzle out
1474 * the name parts here.
1475 */
1476
1477 start = title;
1478 for ( ;; ) {
1479 sz = strcspn(start, " ,");
1480 if ('\0' == start[sz])
1481 break;
1482
1483 byte = start[sz];
1484 start[sz] = '\0';
1485
1486 /*
1487 * Assume a stray trailing comma in the
1488 * name list if a name begins with a dash.
1489 */
1490
1491 if ('-' == start[0] ||
1492 ('\\' == start[0] && '-' == start[1]))
1493 break;
1494
1495 putkey(mpage, start, NAME_TITLE);
1496
1497 if (' ' == byte) {
1498 start += sz + 1;
1499 break;
1500 }
1501
1502 assert(',' == byte);
1503 start += sz + 1;
1504 while (' ' == *start)
1505 start++;
1506 }
1507
1508 if (start == title) {
1509 putkey(mpage, start, NAME_TITLE);
1510 free(title);
1511 return;
1512 }
1513
1514 while (isspace((unsigned char)*start))
1515 start++;
1516
1517 if (0 == strncmp(start, "-", 1))
1518 start += 1;
1519 else if (0 == strncmp(start, "\\-\\-", 4))
1520 start += 4;
1521 else if (0 == strncmp(start, "\\-", 2))
1522 start += 2;
1523 else if (0 == strncmp(start, "\\(en", 4))
1524 start += 4;
1525 else if (0 == strncmp(start, "\\(em", 4))
1526 start += 4;
1527
1528 while (' ' == *start)
1529 start++;
1530
1531 mpage->desc = mandoc_strdup(start);
1532 free(title);
1533 return;
1534 }
1535 }
1536
1537 for (n = n->child; n; n = n->next) {
1538 if (NULL != mpage->desc)
1539 break;
1540 parse_man(mpage, n);
1541 }
1542 }
1543
1544 static void
1545 parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1546 {
1547
1548 assert(NULL != n);
1549 for (n = n->child; NULL != n; n = n->next) {
1550 switch (n->type) {
1551 case MDOC_ELEM:
1552 /* FALLTHROUGH */
1553 case MDOC_BLOCK:
1554 /* FALLTHROUGH */
1555 case MDOC_HEAD:
1556 /* FALLTHROUGH */
1557 case MDOC_BODY:
1558 /* FALLTHROUGH */
1559 case MDOC_TAIL:
1560 if (NULL != mdocs[n->tok].fp)
1561 if (0 == (*mdocs[n->tok].fp)(mpage, n))
1562 break;
1563 if (mdocs[n->tok].mask)
1564 putmdockey(mpage, n->child,
1565 mdocs[n->tok].mask);
1566 break;
1567 default:
1568 assert(MDOC_ROOT != n->type);
1569 continue;
1570 }
1571 if (NULL != n->child)
1572 parse_mdoc(mpage, n);
1573 }
1574 }
1575
1576 static int
1577 parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1578 {
1579 const char *start, *end;
1580 size_t sz;
1581
1582 if (SEC_SYNOPSIS != n->sec ||
1583 NULL == (n = n->child) ||
1584 MDOC_TEXT != n->type)
1585 return(0);
1586
1587 /*
1588 * Only consider those `Fd' macro fields that begin with an
1589 * "inclusion" token (versus, e.g., #define).
1590 */
1591
1592 if (strcmp("#include", n->string))
1593 return(0);
1594
1595 if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1596 return(0);
1597
1598 /*
1599 * Strip away the enclosing angle brackets and make sure we're
1600 * not zero-length.
1601 */
1602
1603 start = n->string;
1604 if ('<' == *start || '"' == *start)
1605 start++;
1606
1607 if (0 == (sz = strlen(start)))
1608 return(0);
1609
1610 end = &start[(int)sz - 1];
1611 if ('>' == *end || '"' == *end)
1612 end--;
1613
1614 if (end > start)
1615 putkeys(mpage, start, end - start + 1, TYPE_In);
1616 return(0);
1617 }
1618
1619 static int
1620 parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1621 {
1622 char *cp;
1623
1624 if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1625 return(0);
1626
1627 /*
1628 * Parse: .Fn "struct type *name" "char *arg".
1629 * First strip away pointer symbol.
1630 * Then store the function name, then type.
1631 * Finally, store the arguments.
1632 */
1633
1634 if (NULL == (cp = strrchr(n->string, ' ')))
1635 cp = n->string;
1636
1637 while ('*' == *cp)
1638 cp++;
1639
1640 putkey(mpage, cp, TYPE_Fn);
1641
1642 if (n->string < cp)
1643 putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1644
1645 for (n = n->next; NULL != n; n = n->next)
1646 if (MDOC_TEXT == n->type)
1647 putkey(mpage, n->string, TYPE_Fa);
1648
1649 return(0);
1650 }
1651
1652 static int
1653 parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1654 {
1655 char *cp;
1656
1657 if (NULL == (n = n->child))
1658 return(0);
1659
1660 if (NULL == n->next) {
1661 putkey(mpage, n->string, TYPE_Xr);
1662 return(0);
1663 }
1664
1665 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1666 putkey(mpage, cp, TYPE_Xr);
1667 free(cp);
1668 return(0);
1669 }
1670
1671 static int
1672 parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1673 {
1674
1675 if (MDOC_BODY == n->type)
1676 mdoc_deroff(&mpage->desc, n);
1677 return(0);
1678 }
1679
1680 static int
1681 parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1682 {
1683
1684 if (SEC_NAME == n->sec)
1685 putmdockey(mpage, n->child, NAME_TITLE);
1686 else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1687 putmdockey(mpage, n->child, NAME_SYN);
1688 return(0);
1689 }
1690
1691 static int
1692 parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1693 {
1694
1695 return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1696 }
1697
1698 static int
1699 parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1700 {
1701
1702 return(MDOC_HEAD == n->type);
1703 }
1704
1705 static int
1706 parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1707 {
1708
1709 return(MDOC_BODY == n->type);
1710 }
1711
1712 /*
1713 * Add a string to the hash table for the current manual.
1714 * Each string has a bitmask telling which macros it belongs to.
1715 * When we finish the manual, we'll dump the table.
1716 */
1717 static void
1718 putkeys(const struct mpage *mpage,
1719 const char *cp, size_t sz, uint64_t v)
1720 {
1721 struct ohash *htab;
1722 struct str *s;
1723 const char *end;
1724 unsigned int slot;
1725 int i;
1726
1727 if (0 == sz)
1728 return;
1729
1730 if (TYPE_Nm & v) {
1731 htab = &names;
1732 v &= name_mask;
1733 if (v & NAME_FIRST)
1734 name_mask &= ~NAME_FIRST;
1735 if (debug > 1)
1736 say(mpage->mlinks->file,
1737 "Adding name %*s", sz, cp);
1738 } else {
1739 htab = &strings;
1740 if (debug > 1)
1741 for (i = 0; i < mansearch_keymax; i++)
1742 if ((uint64_t)1 << i & v)
1743 say(mpage->mlinks->file,
1744 "Adding key %s=%*s",
1745 mansearch_keynames[i], sz, cp);
1746 }
1747
1748 end = cp + sz;
1749 slot = ohash_qlookupi(htab, cp, &end);
1750 s = ohash_find(htab, slot);
1751
1752 if (NULL != s && mpage == s->mpage) {
1753 s->mask |= v;
1754 return;
1755 } else if (NULL == s) {
1756 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1757 memcpy(s->key, cp, sz);
1758 ohash_insert(htab, slot, s);
1759 }
1760 s->mpage = mpage;
1761 s->mask = v;
1762 }
1763
1764 /*
1765 * Take a Unicode codepoint and produce its UTF-8 encoding.
1766 * This isn't the best way to do this, but it works.
1767 * The magic numbers are from the UTF-8 packaging.
1768 * They're not as scary as they seem: read the UTF-8 spec for details.
1769 */
1770 static size_t
1771 utf8(unsigned int cp, char out[7])
1772 {
1773 size_t rc;
1774
1775 rc = 0;
1776 if (cp <= 0x0000007F) {
1777 rc = 1;
1778 out[0] = (char)cp;
1779 } else if (cp <= 0x000007FF) {
1780 rc = 2;
1781 out[0] = (cp >> 6 & 31) | 192;
1782 out[1] = (cp & 63) | 128;
1783 } else if (cp <= 0x0000FFFF) {
1784 rc = 3;
1785 out[0] = (cp >> 12 & 15) | 224;
1786 out[1] = (cp >> 6 & 63) | 128;
1787 out[2] = (cp & 63) | 128;
1788 } else if (cp <= 0x001FFFFF) {
1789 rc = 4;
1790 out[0] = (cp >> 18 & 7) | 240;
1791 out[1] = (cp >> 12 & 63) | 128;
1792 out[2] = (cp >> 6 & 63) | 128;
1793 out[3] = (cp & 63) | 128;
1794 } else if (cp <= 0x03FFFFFF) {
1795 rc = 5;
1796 out[0] = (cp >> 24 & 3) | 248;
1797 out[1] = (cp >> 18 & 63) | 128;
1798 out[2] = (cp >> 12 & 63) | 128;
1799 out[3] = (cp >> 6 & 63) | 128;
1800 out[4] = (cp & 63) | 128;
1801 } else if (cp <= 0x7FFFFFFF) {
1802 rc = 6;
1803 out[0] = (cp >> 30 & 1) | 252;
1804 out[1] = (cp >> 24 & 63) | 128;
1805 out[2] = (cp >> 18 & 63) | 128;
1806 out[3] = (cp >> 12 & 63) | 128;
1807 out[4] = (cp >> 6 & 63) | 128;
1808 out[5] = (cp & 63) | 128;
1809 } else
1810 return(0);
1811
1812 out[rc] = '\0';
1813 return(rc);
1814 }
1815
1816 /*
1817 * Store the rendered version of a key, or alias the pointer
1818 * if the key contains no escape sequences.
1819 */
1820 static void
1821 render_key(struct mchars *mc, struct str *key)
1822 {
1823 size_t sz, bsz, pos;
1824 char utfbuf[7], res[6];
1825 char *buf;
1826 const char *seq, *cpp, *val;
1827 int len, u;
1828 enum mandoc_esc esc;
1829
1830 assert(NULL == key->rendered);
1831
1832 res[0] = '\\';
1833 res[1] = '\t';
1834 res[2] = ASCII_NBRSP;
1835 res[3] = ASCII_HYPH;
1836 res[4] = ASCII_BREAK;
1837 res[5] = '\0';
1838
1839 val = key->key;
1840 bsz = strlen(val);
1841
1842 /*
1843 * Pre-check: if we have no stop-characters, then set the
1844 * pointer as ourselvse and get out of here.
1845 */
1846 if (strcspn(val, res) == bsz) {
1847 key->rendered = key->key;
1848 return;
1849 }
1850
1851 /* Pre-allocate by the length of the input */
1852
1853 buf = mandoc_malloc(++bsz);
1854 pos = 0;
1855
1856 while ('\0' != *val) {
1857 /*
1858 * Halt on the first escape sequence.
1859 * This also halts on the end of string, in which case
1860 * we just copy, fallthrough, and exit the loop.
1861 */
1862 if ((sz = strcspn(val, res)) > 0) {
1863 memcpy(&buf[pos], val, sz);
1864 pos += sz;
1865 val += sz;
1866 }
1867
1868 switch (*val) {
1869 case ASCII_HYPH:
1870 buf[pos++] = '-';
1871 val++;
1872 continue;
1873 case '\t':
1874 /* FALLTHROUGH */
1875 case ASCII_NBRSP:
1876 buf[pos++] = ' ';
1877 val++;
1878 /* FALLTHROUGH */
1879 case ASCII_BREAK:
1880 continue;
1881 default:
1882 break;
1883 }
1884 if ('\\' != *val)
1885 break;
1886
1887 /* Read past the slash. */
1888
1889 val++;
1890
1891 /*
1892 * Parse the escape sequence and see if it's a
1893 * predefined character or special character.
1894 */
1895
1896 esc = mandoc_escape((const char **)&val,
1897 &seq, &len);
1898 if (ESCAPE_ERROR == esc)
1899 break;
1900 if (ESCAPE_SPECIAL != esc)
1901 continue;
1902
1903 /*
1904 * Render the special character
1905 * as either UTF-8 or ASCII.
1906 */
1907
1908 if (write_utf8) {
1909 if ((u = mchars_spec2cp(mc, seq, len)) <= 0)
1910 continue;
1911 cpp = utfbuf;
1912 if (0 == (sz = utf8(u, utfbuf)))
1913 continue;
1914 sz = strlen(cpp);
1915 } else {
1916 cpp = mchars_spec2str(mc, seq, len, &sz);
1917 if (NULL == cpp)
1918 continue;
1919 if (ASCII_NBRSP == *cpp) {
1920 cpp = " ";
1921 sz = 1;
1922 }
1923 }
1924
1925 /* Copy the rendered glyph into the stream. */
1926
1927 bsz += sz;
1928 buf = mandoc_realloc(buf, bsz);
1929 memcpy(&buf[pos], cpp, sz);
1930 pos += sz;
1931 }
1932
1933 buf[pos] = '\0';
1934 key->rendered = buf;
1935 }
1936
1937 static void
1938 dbadd_mlink(const struct mlink *mlink)
1939 {
1940 size_t i;
1941
1942 i = 1;
1943 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1944 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1945 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1946 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1947 SQL_STEP(stmts[STMT_INSERT_LINK]);
1948 sqlite3_reset(stmts[STMT_INSERT_LINK]);
1949 }
1950
1951 static void
1952 dbadd_mlink_name(const struct mlink *mlink)
1953 {
1954 size_t i;
1955
1956 dbadd_mlink(mlink);
1957
1958 i = 1;
1959 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, NAME_FILE & NAME_MASK);
1960 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
1961 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
1962 SQL_STEP(stmts[STMT_INSERT_NAME]);
1963 sqlite3_reset(stmts[STMT_INSERT_NAME]);
1964 }
1965
1966 /*
1967 * Flush the current page's terms (and their bits) into the database.
1968 * Wrap the entire set of additions in a transaction to make sqlite be a
1969 * little faster.
1970 * Also, handle escape sequences at the last possible moment.
1971 */
1972 static void
1973 dbadd(struct mpage *mpage, struct mchars *mc)
1974 {
1975 struct mlink *mlink;
1976 struct str *key;
1977 size_t i;
1978 unsigned int slot;
1979
1980 mlink = mpage->mlinks;
1981
1982 if (nodb) {
1983 for (key = ohash_first(&names, &slot); NULL != key;
1984 key = ohash_next(&names, &slot)) {
1985 if (key->rendered != key->key)
1986 free(key->rendered);
1987 free(key);
1988 }
1989 for (key = ohash_first(&strings, &slot); NULL != key;
1990 key = ohash_next(&strings, &slot)) {
1991 if (key->rendered != key->key)
1992 free(key->rendered);
1993 free(key);
1994 }
1995 if (0 == debug)
1996 return;
1997 while (NULL != mlink) {
1998 fputs(mlink->name, stdout);
1999 if (NULL == mlink->next ||
2000 strcmp(mlink->dsec, mlink->next->dsec) ||
2001 strcmp(mlink->fsec, mlink->next->fsec) ||
2002 strcmp(mlink->arch, mlink->next->arch)) {
2003 putchar('(');
2004 if ('\0' == *mlink->dsec)
2005 fputs(mlink->fsec, stdout);
2006 else
2007 fputs(mlink->dsec, stdout);
2008 if ('\0' != *mlink->arch)
2009 printf("/%s", mlink->arch);
2010 putchar(')');
2011 }
2012 mlink = mlink->next;
2013 if (NULL != mlink)
2014 fputs(", ", stdout);
2015 }
2016 printf(" - %s\n", mpage->desc);
2017 return;
2018 }
2019
2020 if (debug)
2021 say(mlink->file, "Adding to database");
2022
2023 i = strlen(mpage->desc) + 1;
2024 key = mandoc_calloc(1, sizeof(struct str) + i);
2025 memcpy(key->key, mpage->desc, i);
2026 render_key(mc, key);
2027
2028 i = 1;
2029 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
2030 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
2031 SQL_STEP(stmts[STMT_INSERT_PAGE]);
2032 mpage->pageid = sqlite3_last_insert_rowid(db);
2033 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2034
2035 if (key->rendered != key->key)
2036 free(key->rendered);
2037 free(key);
2038
2039 while (NULL != mlink) {
2040 dbadd_mlink(mlink);
2041 mlink = mlink->next;
2042 }
2043 mlink = mpage->mlinks;
2044
2045 for (key = ohash_first(&names, &slot); NULL != key;
2046 key = ohash_next(&names, &slot)) {
2047 assert(key->mpage == mpage);
2048 if (NULL == key->rendered)
2049 render_key(mc, key);
2050 i = 1;
2051 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2052 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
2053 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
2054 SQL_STEP(stmts[STMT_INSERT_NAME]);
2055 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2056 if (key->rendered != key->key)
2057 free(key->rendered);
2058 free(key);
2059 }
2060 for (key = ohash_first(&strings, &slot); NULL != key;
2061 key = ohash_next(&strings, &slot)) {
2062 assert(key->mpage == mpage);
2063 if (NULL == key->rendered)
2064 render_key(mc, key);
2065 i = 1;
2066 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
2067 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
2068 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
2069 SQL_STEP(stmts[STMT_INSERT_KEY]);
2070 sqlite3_reset(stmts[STMT_INSERT_KEY]);
2071 if (key->rendered != key->key)
2072 free(key->rendered);
2073 free(key);
2074 }
2075 }
2076
2077 static void
2078 dbprune(void)
2079 {
2080 struct mpage *mpage;
2081 struct mlink *mlink;
2082 size_t i;
2083 unsigned int slot;
2084
2085 if (0 == nodb)
2086 SQL_EXEC("BEGIN TRANSACTION");
2087
2088 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2089 mpage = ohash_next(&mpages, &slot)) {
2090 mlink = mpage->mlinks;
2091 if (debug)
2092 say(mlink->file, "Deleting from database");
2093 if (nodb)
2094 continue;
2095 for ( ; NULL != mlink; mlink = mlink->next) {
2096 i = 1;
2097 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2098 i, mlink->dsec);
2099 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2100 i, mlink->arch);
2101 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2102 i, mlink->name);
2103 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2104 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2105 }
2106 }
2107
2108 if (0 == nodb)
2109 SQL_EXEC("END TRANSACTION");
2110 }
2111
2112 /*
2113 * Close an existing database and its prepared statements.
2114 * If "real" is not set, rename the temporary file into the real one.
2115 */
2116 static void
2117 dbclose(int real)
2118 {
2119 size_t i;
2120 int status;
2121 pid_t child;
2122
2123 if (nodb)
2124 return;
2125
2126 for (i = 0; i < STMT__MAX; i++) {
2127 sqlite3_finalize(stmts[i]);
2128 stmts[i] = NULL;
2129 }
2130
2131 sqlite3_close(db);
2132 db = NULL;
2133
2134 if (real)
2135 return;
2136
2137 if ('\0' == *tempfilename) {
2138 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2139 exitcode = (int)MANDOCLEVEL_SYSERR;
2140 say(MANDOC_DB, "&rename");
2141 }
2142 return;
2143 }
2144
2145 switch (child = fork()) {
2146 case -1:
2147 exitcode = (int)MANDOCLEVEL_SYSERR;
2148 say("", "&fork cmp");
2149 return;
2150 case 0:
2151 execlp("cmp", "cmp", "-s",
2152 tempfilename, MANDOC_DB, NULL);
2153 say("", "&exec cmp");
2154 exit(0);
2155 default:
2156 break;
2157 }
2158 if (-1 == waitpid(child, &status, 0)) {
2159 exitcode = (int)MANDOCLEVEL_SYSERR;
2160 say("", "&wait cmp");
2161 } else if (WIFSIGNALED(status)) {
2162 exitcode = (int)MANDOCLEVEL_SYSERR;
2163 say("", "cmp died from signal %d", WTERMSIG(status));
2164 } else if (WEXITSTATUS(status)) {
2165 exitcode = (int)MANDOCLEVEL_SYSERR;
2166 say(MANDOC_DB,
2167 "Data changed, but cannot replace database");
2168 }
2169
2170 *strrchr(tempfilename, '/') = '\0';
2171 switch (child = fork()) {
2172 case -1:
2173 exitcode = (int)MANDOCLEVEL_SYSERR;
2174 say("", "&fork rm");
2175 return;
2176 case 0:
2177 execlp("rm", "rm", "-rf", tempfilename, NULL);
2178 say("", "&exec rm");
2179 exit((int)MANDOCLEVEL_SYSERR);
2180 default:
2181 break;
2182 }
2183 if (-1 == waitpid(child, &status, 0)) {
2184 exitcode = (int)MANDOCLEVEL_SYSERR;
2185 say("", "&wait rm");
2186 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2187 exitcode = (int)MANDOCLEVEL_SYSERR;
2188 say("", "%s: Cannot remove temporary directory",
2189 tempfilename);
2190 }
2191 }
2192
2193 /*
2194 * This is straightforward stuff.
2195 * Open a database connection to a "temporary" database, then open a set
2196 * of prepared statements we'll use over and over again.
2197 * If "real" is set, we use the existing database; if not, we truncate a
2198 * temporary one.
2199 * Must be matched by dbclose().
2200 */
2201 static int
2202 dbopen(int real)
2203 {
2204 const char *sql;
2205 int rc, ofl;
2206
2207 if (nodb)
2208 return(1);
2209
2210 *tempfilename = '\0';
2211 ofl = SQLITE_OPEN_READWRITE;
2212
2213 if (real) {
2214 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2215 if (SQLITE_OK != rc) {
2216 exitcode = (int)MANDOCLEVEL_SYSERR;
2217 if (SQLITE_CANTOPEN != rc)
2218 say(MANDOC_DB, "%s", sqlite3_errstr(rc));
2219 return(0);
2220 }
2221 goto prepare_statements;
2222 }
2223
2224 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2225
2226 remove(MANDOC_DB "~");
2227 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2228 if (SQLITE_OK == rc)
2229 goto create_tables;
2230 if (MPARSE_QUICK & mparse_options) {
2231 exitcode = (int)MANDOCLEVEL_SYSERR;
2232 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
2233 return(0);
2234 }
2235
2236 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2237 sizeof(tempfilename));
2238 if (NULL == mkdtemp(tempfilename)) {
2239 exitcode = (int)MANDOCLEVEL_SYSERR;
2240 say("", "&%s", tempfilename);
2241 return(0);
2242 }
2243 (void)strlcat(tempfilename, "/" MANDOC_DB,
2244 sizeof(tempfilename));
2245 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2246 if (SQLITE_OK != rc) {
2247 exitcode = (int)MANDOCLEVEL_SYSERR;
2248 say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
2249 return(0);
2250 }
2251
2252 create_tables:
2253 sql = "CREATE TABLE \"mpages\" (\n"
2254 " \"desc\" TEXT NOT NULL,\n"
2255 " \"form\" INTEGER NOT NULL,\n"
2256 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2257 ");\n"
2258 "\n"
2259 "CREATE TABLE \"mlinks\" (\n"
2260 " \"sec\" TEXT NOT NULL,\n"
2261 " \"arch\" TEXT NOT NULL,\n"
2262 " \"name\" TEXT NOT NULL,\n"
2263 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2264 "ON DELETE CASCADE\n"
2265 ");\n"
2266 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2267 "\n"
2268 "CREATE TABLE \"names\" (\n"
2269 " \"bits\" INTEGER NOT NULL,\n"
2270 " \"name\" TEXT NOT NULL,\n"
2271 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2272 "ON DELETE CASCADE\n"
2273 ");\n"
2274 "\n"
2275 "CREATE TABLE \"keys\" (\n"
2276 " \"bits\" INTEGER NOT NULL,\n"
2277 " \"key\" TEXT NOT NULL,\n"
2278 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2279 "ON DELETE CASCADE\n"
2280 ");\n"
2281 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2282
2283 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2284 exitcode = (int)MANDOCLEVEL_SYSERR;
2285 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2286 sqlite3_close(db);
2287 return(0);
2288 }
2289
2290 prepare_statements:
2291 if (SQLITE_OK != sqlite3_exec(db,
2292 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2293 exitcode = (int)MANDOCLEVEL_SYSERR;
2294 say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2295 sqlite3_errmsg(db));
2296 sqlite3_close(db);
2297 return(0);
2298 }
2299
2300 sql = "DELETE FROM mpages WHERE pageid IN "
2301 "(SELECT pageid FROM mlinks WHERE "
2302 "sec=? AND arch=? AND name=?)";
2303 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2304 sql = "INSERT INTO mpages "
2305 "(desc,form) VALUES (?,?)";
2306 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2307 sql = "INSERT INTO mlinks "
2308 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2309 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2310 sql = "INSERT INTO names "
2311 "(bits,name,pageid) VALUES (?,?,?)";
2312 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2313 sql = "INSERT INTO keys "
2314 "(bits,key,pageid) VALUES (?,?,?)";
2315 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2316
2317 #ifndef __APPLE__
2318 /*
2319 * When opening a new database, we can turn off
2320 * synchronous mode for much better performance.
2321 */
2322
2323 if (real && SQLITE_OK != sqlite3_exec(db,
2324 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2325 exitcode = (int)MANDOCLEVEL_SYSERR;
2326 say(MANDOC_DB, "PRAGMA synchronous: %s",
2327 sqlite3_errmsg(db));
2328 sqlite3_close(db);
2329 return(0);
2330 }
2331 #endif
2332
2333 return(1);
2334 }
2335
2336 static void *
2337 hash_calloc(size_t n, size_t sz, void *arg)
2338 {
2339
2340 return(mandoc_calloc(n, sz));
2341 }
2342
2343 static void *
2344 hash_alloc(size_t sz, void *arg)
2345 {
2346
2347 return(mandoc_malloc(sz));
2348 }
2349
2350 static void
2351 hash_free(void *p, void *arg)
2352 {
2353
2354 free(p);
2355 }
2356
2357 static int
2358 set_basedir(const char *targetdir, int report_baddir)
2359 {
2360 static char startdir[PATH_MAX];
2361 static int getcwd_status; /* 1 = ok, 2 = failure */
2362 static int chdir_status; /* 1 = changed directory */
2363 char *cp;
2364
2365 /*
2366 * Remember the original working directory, if possible.
2367 * This will be needed if the second or a later directory
2368 * on the command line is given as a relative path.
2369 * Do not error out if the current directory is not
2370 * searchable: Maybe it won't be needed after all.
2371 */
2372 if (0 == getcwd_status) {
2373 if (NULL == getcwd(startdir, sizeof(startdir))) {
2374 getcwd_status = 2;
2375 (void)strlcpy(startdir, strerror(errno),
2376 sizeof(startdir));
2377 } else
2378 getcwd_status = 1;
2379 }
2380
2381 /*
2382 * We are leaving the old base directory.
2383 * Do not use it any longer, not even for messages.
2384 */
2385 *basedir = '\0';
2386
2387 /*
2388 * If and only if the directory was changed earlier and
2389 * the next directory to process is given as a relative path,
2390 * first go back, or bail out if that is impossible.
2391 */
2392 if (chdir_status && '/' != *targetdir) {
2393 if (2 == getcwd_status) {
2394 exitcode = (int)MANDOCLEVEL_SYSERR;
2395 say("", "getcwd: %s", startdir);
2396 return(0);
2397 }
2398 if (-1 == chdir(startdir)) {
2399 exitcode = (int)MANDOCLEVEL_SYSERR;
2400 say("", "&chdir %s", startdir);
2401 return(0);
2402 }
2403 }
2404
2405 /*
2406 * Always resolve basedir to the canonicalized absolute
2407 * pathname and append a trailing slash, such that
2408 * we can reliably check whether files are inside.
2409 */
2410 if (NULL == realpath(targetdir, basedir)) {
2411 if (report_baddir || errno != ENOENT) {
2412 exitcode = (int)MANDOCLEVEL_BADARG;
2413 say("", "&%s: realpath", targetdir);
2414 }
2415 return(0);
2416 } else if (-1 == chdir(basedir)) {
2417 if (report_baddir || errno != ENOENT) {
2418 exitcode = (int)MANDOCLEVEL_BADARG;
2419 say("", "&chdir");
2420 }
2421 return(0);
2422 }
2423 chdir_status = 1;
2424 cp = strchr(basedir, '\0');
2425 if ('/' != cp[-1]) {
2426 if (cp - basedir >= PATH_MAX - 1) {
2427 exitcode = (int)MANDOCLEVEL_SYSERR;
2428 say("", "Filename too long");
2429 return(0);
2430 }
2431 *cp++ = '/';
2432 *cp = '\0';
2433 }
2434 return(1);
2435 }
2436
2437 static void
2438 say(const char *file, const char *format, ...)
2439 {
2440 va_list ap;
2441 int use_errno;
2442
2443 if ('\0' != *basedir)
2444 fprintf(stderr, "%s", basedir);
2445 if ('\0' != *basedir && '\0' != *file)
2446 fputc('/', stderr);
2447 if ('\0' != *file)
2448 fprintf(stderr, "%s", file);
2449
2450 use_errno = 1;
2451 if (NULL != format) {
2452 switch (*format) {
2453 case '&':
2454 format++;
2455 break;
2456 case '\0':
2457 format = NULL;
2458 break;
2459 default:
2460 use_errno = 0;
2461 break;
2462 }
2463 }
2464 if (NULL != format) {
2465 if ('\0' != *basedir || '\0' != *file)
2466 fputs(": ", stderr);
2467 va_start(ap, format);
2468 vfprintf(stderr, format, ap);
2469 va_end(ap);
2470 }
2471 if (use_errno) {
2472 if ('\0' != *basedir || '\0' != *file || NULL != format)
2473 fputs(": ", stderr);
2474 perror(NULL);
2475 } else
2476 fputc('\n', stderr);
2477 }
A git conversion mirror of mandoc.bsd.lv