]> 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
Let -h imply -c (that is, not use the pager).
[mandoc.git] / mandocdb.c
1 /* $Id: mandocdb.c,v 1.168 2014/10/28 17:36:19 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 int dbopen(int);
142 static void dbprune(void);
143 static void filescan(const char *);
144 static void *hash_alloc(size_t, void *);
145 static void hash_free(void *, void *);
146 static void *hash_calloc(size_t, size_t, void *);
147 static void mlink_add(struct mlink *, const struct stat *);
148 static void mlink_check(struct mpage *, struct mlink *);
149 static void mlink_free(struct mlink *);
150 static void mlinks_undupe(struct mpage *);
151 static void mpages_free(void);
152 static void mpages_merge(struct mchars *, struct mparse *);
153 static void names_check(void);
154 static void parse_cat(struct mpage *, int);
155 static void parse_man(struct mpage *, const struct man_node *);
156 static void parse_mdoc(struct mpage *, const struct mdoc_node *);
157 static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
158 static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
159 static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
160 static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
161 static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
162 static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
163 static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
164 static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
165 static void putkey(const struct mpage *, char *, uint64_t);
166 static void putkeys(const struct mpage *,
167 const char *, size_t, uint64_t);
168 static void putmdockey(const struct mpage *,
169 const struct mdoc_node *, uint64_t);
170 static void render_key(struct mchars *, struct str *);
171 static void say(const char *, const char *, ...);
172 static int set_basedir(const char *, int);
173 static int treescan(void);
174 static size_t utf8(unsigned int, char [7]);
175
176 static char tempfilename[32];
177 static char *progname;
178 static int nodb; /* no database changes */
179 static int mparse_options; /* abort the parse early */
180 static int use_all; /* use all found files */
181 static int debug; /* print what we're doing */
182 static int warnings; /* warn about crap */
183 static int write_utf8; /* write UTF-8 output; else ASCII */
184 static int exitcode; /* to be returned by main */
185 static enum op op; /* operational mode */
186 static char basedir[PATH_MAX]; /* current base directory */
187 static struct ohash mpages; /* table of distinct manual pages */
188 static struct ohash mlinks; /* table of directory entries */
189 static struct ohash names; /* table of all names */
190 static struct ohash strings; /* table of all strings */
191 static sqlite3 *db = NULL; /* current database */
192 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
193 static uint64_t name_mask;
194
195 static const struct mdoc_handler mdocs[MDOC_MAX] = {
196 { NULL, 0 }, /* Ap */
197 { NULL, 0 }, /* Dd */
198 { NULL, 0 }, /* Dt */
199 { NULL, 0 }, /* Os */
200 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
201 { parse_mdoc_head, TYPE_Ss }, /* Ss */
202 { NULL, 0 }, /* Pp */
203 { NULL, 0 }, /* D1 */
204 { NULL, 0 }, /* Dl */
205 { NULL, 0 }, /* Bd */
206 { NULL, 0 }, /* Ed */
207 { NULL, 0 }, /* Bl */
208 { NULL, 0 }, /* El */
209 { NULL, 0 }, /* It */
210 { NULL, 0 }, /* Ad */
211 { NULL, TYPE_An }, /* An */
212 { NULL, TYPE_Ar }, /* Ar */
213 { NULL, TYPE_Cd }, /* Cd */
214 { NULL, TYPE_Cm }, /* Cm */
215 { NULL, TYPE_Dv }, /* Dv */
216 { NULL, TYPE_Er }, /* Er */
217 { NULL, TYPE_Ev }, /* Ev */
218 { NULL, 0 }, /* Ex */
219 { NULL, TYPE_Fa }, /* Fa */
220 { parse_mdoc_Fd, 0 }, /* Fd */
221 { NULL, TYPE_Fl }, /* Fl */
222 { parse_mdoc_Fn, 0 }, /* Fn */
223 { NULL, TYPE_Ft }, /* Ft */
224 { NULL, TYPE_Ic }, /* Ic */
225 { NULL, TYPE_In }, /* In */
226 { NULL, TYPE_Li }, /* Li */
227 { parse_mdoc_Nd, 0 }, /* Nd */
228 { parse_mdoc_Nm, 0 }, /* Nm */
229 { NULL, 0 }, /* Op */
230 { NULL, 0 }, /* Ot */
231 { NULL, TYPE_Pa }, /* Pa */
232 { NULL, 0 }, /* Rv */
233 { NULL, TYPE_St }, /* St */
234 { NULL, TYPE_Va }, /* Va */
235 { parse_mdoc_body, TYPE_Va }, /* Vt */
236 { parse_mdoc_Xr, 0 }, /* Xr */
237 { NULL, 0 }, /* %A */
238 { NULL, 0 }, /* %B */
239 { NULL, 0 }, /* %D */
240 { NULL, 0 }, /* %I */
241 { NULL, 0 }, /* %J */
242 { NULL, 0 }, /* %N */
243 { NULL, 0 }, /* %O */
244 { NULL, 0 }, /* %P */
245 { NULL, 0 }, /* %R */
246 { NULL, 0 }, /* %T */
247 { NULL, 0 }, /* %V */
248 { NULL, 0 }, /* Ac */
249 { NULL, 0 }, /* Ao */
250 { NULL, 0 }, /* Aq */
251 { NULL, TYPE_At }, /* At */
252 { NULL, 0 }, /* Bc */
253 { NULL, 0 }, /* Bf */
254 { NULL, 0 }, /* Bo */
255 { NULL, 0 }, /* Bq */
256 { NULL, TYPE_Bsx }, /* Bsx */
257 { NULL, TYPE_Bx }, /* Bx */
258 { NULL, 0 }, /* Db */
259 { NULL, 0 }, /* Dc */
260 { NULL, 0 }, /* Do */
261 { NULL, 0 }, /* Dq */
262 { NULL, 0 }, /* Ec */
263 { NULL, 0 }, /* Ef */
264 { NULL, TYPE_Em }, /* Em */
265 { NULL, 0 }, /* Eo */
266 { NULL, TYPE_Fx }, /* Fx */
267 { NULL, TYPE_Ms }, /* Ms */
268 { NULL, 0 }, /* No */
269 { NULL, 0 }, /* Ns */
270 { NULL, TYPE_Nx }, /* Nx */
271 { NULL, TYPE_Ox }, /* Ox */
272 { NULL, 0 }, /* Pc */
273 { NULL, 0 }, /* Pf */
274 { NULL, 0 }, /* Po */
275 { NULL, 0 }, /* Pq */
276 { NULL, 0 }, /* Qc */
277 { NULL, 0 }, /* Ql */
278 { NULL, 0 }, /* Qo */
279 { NULL, 0 }, /* Qq */
280 { NULL, 0 }, /* Re */
281 { NULL, 0 }, /* Rs */
282 { NULL, 0 }, /* Sc */
283 { NULL, 0 }, /* So */
284 { NULL, 0 }, /* Sq */
285 { NULL, 0 }, /* Sm */
286 { NULL, 0 }, /* Sx */
287 { NULL, TYPE_Sy }, /* Sy */
288 { NULL, TYPE_Tn }, /* Tn */
289 { NULL, 0 }, /* Ux */
290 { NULL, 0 }, /* Xc */
291 { NULL, 0 }, /* Xo */
292 { parse_mdoc_head, 0 }, /* Fo */
293 { NULL, 0 }, /* Fc */
294 { NULL, 0 }, /* Oo */
295 { NULL, 0 }, /* Oc */
296 { NULL, 0 }, /* Bk */
297 { NULL, 0 }, /* Ek */
298 { NULL, 0 }, /* Bt */
299 { NULL, 0 }, /* Hf */
300 { NULL, 0 }, /* Fr */
301 { NULL, 0 }, /* Ud */
302 { NULL, TYPE_Lb }, /* Lb */
303 { NULL, 0 }, /* Lp */
304 { NULL, TYPE_Lk }, /* Lk */
305 { NULL, TYPE_Mt }, /* Mt */
306 { NULL, 0 }, /* Brq */
307 { NULL, 0 }, /* Bro */
308 { NULL, 0 }, /* Brc */
309 { NULL, 0 }, /* %C */
310 { NULL, 0 }, /* Es */
311 { NULL, 0 }, /* En */
312 { NULL, TYPE_Dx }, /* Dx */
313 { NULL, 0 }, /* %Q */
314 { NULL, 0 }, /* br */
315 { NULL, 0 }, /* sp */
316 { NULL, 0 }, /* %U */
317 { NULL, 0 }, /* Ta */
318 { NULL, 0 }, /* ll */
319 };
320
321
322 int
323 main(int argc, char *argv[])
324 {
325 int ch, i;
326 size_t j, sz;
327 const char *path_arg;
328 struct mchars *mc;
329 struct manpaths dirs;
330 struct mparse *mp;
331 struct ohash_info mpages_info, mlinks_info;
332
333 memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
334 memset(&dirs, 0, sizeof(struct manpaths));
335
336 mpages_info.alloc = mlinks_info.alloc = hash_alloc;
337 mpages_info.calloc = mlinks_info.calloc = hash_calloc;
338 mpages_info.free = mlinks_info.free = hash_free;
339
340 mpages_info.key_offset = offsetof(struct mpage, inodev);
341 mlinks_info.key_offset = offsetof(struct mlink, file);
342
343 progname = strrchr(argv[0], '/');
344 if (progname == NULL)
345 progname = argv[0];
346 else
347 ++progname;
348
349 /*
350 * We accept a few different invocations.
351 * The CHECKOP macro makes sure that invocation styles don't
352 * clobber each other.
353 */
354 #define CHECKOP(_op, _ch) do \
355 if (OP_DEFAULT != (_op)) { \
356 fprintf(stderr, "%s: -%c: Conflicting option\n", \
357 progname, (_ch)); \
358 goto usage; \
359 } while (/*CONSTCOND*/0)
360
361 path_arg = NULL;
362 op = OP_DEFAULT;
363
364 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
365 switch (ch) {
366 case 'a':
367 use_all = 1;
368 break;
369 case 'C':
370 CHECKOP(op, ch);
371 path_arg = optarg;
372 op = OP_CONFFILE;
373 break;
374 case 'D':
375 debug++;
376 break;
377 case 'd':
378 CHECKOP(op, ch);
379 path_arg = optarg;
380 op = OP_UPDATE;
381 break;
382 case 'n':
383 nodb = 1;
384 break;
385 case 'p':
386 warnings = 1;
387 break;
388 case 'Q':
389 mparse_options |= MPARSE_QUICK;
390 break;
391 case 'T':
392 if (strcmp(optarg, "utf8")) {
393 fprintf(stderr, "%s: -T%s: "
394 "Unsupported output format\n",
395 progname, optarg);
396 goto usage;
397 }
398 write_utf8 = 1;
399 break;
400 case 't':
401 CHECKOP(op, ch);
402 dup2(STDOUT_FILENO, STDERR_FILENO);
403 op = OP_TEST;
404 nodb = warnings = 1;
405 break;
406 case 'u':
407 CHECKOP(op, ch);
408 path_arg = optarg;
409 op = OP_DELETE;
410 break;
411 case 'v':
412 /* Compatibility with espie@'s makewhatis. */
413 break;
414 default:
415 goto usage;
416 }
417
418 argc -= optind;
419 argv += optind;
420
421 if (OP_CONFFILE == op && argc > 0) {
422 fprintf(stderr, "%s: -C: Too many arguments\n",
423 progname);
424 goto usage;
425 }
426
427 exitcode = (int)MANDOCLEVEL_OK;
428 mc = mchars_alloc();
429 mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
430 mc, NULL);
431 ohash_init(&mpages, 6, &mpages_info);
432 ohash_init(&mlinks, 6, &mlinks_info);
433
434 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
435
436 /*
437 * Most of these deal with a specific directory.
438 * Jump into that directory first.
439 */
440 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
441 goto out;
442
443 if (dbopen(1)) {
444 /*
445 * The existing database is usable. Process
446 * all files specified on the command-line.
447 */
448 use_all = 1;
449 for (i = 0; i < argc; i++)
450 filescan(argv[i]);
451 if (OP_TEST != op)
452 dbprune();
453 } else {
454 /*
455 * Database missing or corrupt.
456 * Recreate from scratch.
457 */
458 exitcode = (int)MANDOCLEVEL_OK;
459 op = OP_DEFAULT;
460 if (0 == treescan())
461 goto out;
462 if (0 == dbopen(0))
463 goto out;
464 }
465 if (OP_DELETE != op)
466 mpages_merge(mc, mp);
467 dbclose(OP_DEFAULT == op ? 0 : 1);
468 } else {
469 /*
470 * If we have arguments, use them as our manpaths.
471 * If we don't, grok from manpath(1) or however else
472 * manpath_parse() wants to do it.
473 */
474 if (argc > 0) {
475 dirs.paths = mandoc_reallocarray(NULL,
476 argc, sizeof(char *));
477 dirs.sz = (size_t)argc;
478 for (i = 0; i < argc; i++)
479 dirs.paths[i] = mandoc_strdup(argv[i]);
480 } else
481 manpath_parse(&dirs, path_arg, NULL, NULL);
482
483 if (0 == dirs.sz) {
484 exitcode = (int)MANDOCLEVEL_BADARG;
485 say("", "Empty manpath");
486 }
487
488 /*
489 * First scan the tree rooted at a base directory, then
490 * build a new database and finally move it into place.
491 * Ignore zero-length directories and strip trailing
492 * slashes.
493 */
494 for (j = 0; j < dirs.sz; j++) {
495 sz = strlen(dirs.paths[j]);
496 if (sz && '/' == dirs.paths[j][sz - 1])
497 dirs.paths[j][--sz] = '\0';
498 if (0 == sz)
499 continue;
500
501 if (j) {
502 ohash_init(&mpages, 6, &mpages_info);
503 ohash_init(&mlinks, 6, &mlinks_info);
504 }
505
506 if (0 == set_basedir(dirs.paths[j], argc > 0))
507 continue;
508 if (0 == treescan())
509 continue;
510 if (0 == dbopen(0))
511 continue;
512
513 mpages_merge(mc, mp);
514 if (warnings && !nodb &&
515 ! (MPARSE_QUICK & mparse_options))
516 names_check();
517 dbclose(0);
518
519 if (j + 1 < dirs.sz) {
520 mpages_free();
521 ohash_delete(&mpages);
522 ohash_delete(&mlinks);
523 }
524 }
525 }
526 out:
527 manpath_free(&dirs);
528 mparse_free(mp);
529 mchars_free(mc);
530 mpages_free();
531 ohash_delete(&mpages);
532 ohash_delete(&mlinks);
533 return(exitcode);
534 usage:
535 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
536 " %s [-aDnpQ] [-Tutf8] dir ...\n"
537 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
538 " %s [-Dnp] -u dir [file ...]\n"
539 " %s [-Q] -t file ...\n",
540 progname, progname, progname,
541 progname, progname);
542
543 return((int)MANDOCLEVEL_BADARG);
544 }
545
546 /*
547 * Scan a directory tree rooted at "basedir" for manpages.
548 * We use fts(), scanning directory parts along the way for clues to our
549 * section and architecture.
550 *
551 * If use_all has been specified, grok all files.
552 * If not, sanitise paths to the following:
553 *
554 * [./]man*[/<arch>]/<name>.<section>
555 * or
556 * [./]cat<section>[/<arch>]/<name>.0
557 *
558 * TODO: accomodate for multi-language directories.
559 */
560 static int
561 treescan(void)
562 {
563 char buf[PATH_MAX];
564 FTS *f;
565 FTSENT *ff;
566 struct mlink *mlink;
567 int dform, gzip;
568 char *dsec, *arch, *fsec, *cp;
569 const char *path;
570 const char *argv[2];
571
572 argv[0] = ".";
573 argv[1] = (char *)NULL;
574
575 f = fts_open((char * const *)argv,
576 FTS_PHYSICAL | FTS_NOCHDIR, NULL);
577 if (NULL == f) {
578 exitcode = (int)MANDOCLEVEL_SYSERR;
579 say("", "&fts_open");
580 return(0);
581 }
582
583 dsec = arch = NULL;
584 dform = FORM_NONE;
585
586 while (NULL != (ff = fts_read(f))) {
587 path = ff->fts_path + 2;
588 switch (ff->fts_info) {
589
590 /*
591 * Symbolic links require various sanity checks,
592 * then get handled just like regular files.
593 */
594 case FTS_SL:
595 if (NULL == realpath(path, buf)) {
596 if (warnings)
597 say(path, "&realpath");
598 continue;
599 }
600 if (strstr(buf, basedir) != buf) {
601 if (warnings) say("",
602 "%s: outside base directory", buf);
603 continue;
604 }
605 /* Use logical inode to avoid mpages dupe. */
606 if (-1 == stat(path, ff->fts_statp)) {
607 if (warnings)
608 say(path, "&stat");
609 continue;
610 }
611 /* FALLTHROUGH */
612
613 /*
614 * If we're a regular file, add an mlink by using the
615 * stored directory data and handling the filename.
616 */
617 case FTS_F:
618 if (0 == strcmp(path, MANDOC_DB))
619 continue;
620 if ( ! use_all && ff->fts_level < 2) {
621 if (warnings)
622 say(path, "Extraneous file");
623 continue;
624 }
625 gzip = 0;
626 fsec = NULL;
627 while (NULL == fsec) {
628 fsec = strrchr(ff->fts_name, '.');
629 if (NULL == fsec || strcmp(fsec+1, "gz"))
630 break;
631 gzip = 1;
632 *fsec = '\0';
633 fsec = NULL;
634 }
635 if (NULL == fsec) {
636 if ( ! use_all) {
637 if (warnings)
638 say(path,
639 "No filename suffix");
640 continue;
641 }
642 } else if (0 == strcmp(++fsec, "html")) {
643 if (warnings)
644 say(path, "Skip html");
645 continue;
646 } else if (0 == strcmp(fsec, "ps")) {
647 if (warnings)
648 say(path, "Skip ps");
649 continue;
650 } else if (0 == strcmp(fsec, "pdf")) {
651 if (warnings)
652 say(path, "Skip pdf");
653 continue;
654 } else if ( ! use_all &&
655 ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
656 (FORM_CAT == dform && strcmp(fsec, "0")))) {
657 if (warnings)
658 say(path, "Wrong filename suffix");
659 continue;
660 } else
661 fsec[-1] = '\0';
662
663 mlink = mandoc_calloc(1, sizeof(struct mlink));
664 if (strlcpy(mlink->file, path,
665 sizeof(mlink->file)) >=
666 sizeof(mlink->file)) {
667 say(path, "Filename too long");
668 free(mlink);
669 continue;
670 }
671 mlink->dform = dform;
672 mlink->dsec = dsec;
673 mlink->arch = arch;
674 mlink->name = ff->fts_name;
675 mlink->fsec = fsec;
676 mlink->gzip = gzip;
677 mlink_add(mlink, ff->fts_statp);
678 continue;
679
680 case FTS_D:
681 /* FALLTHROUGH */
682 case FTS_DP:
683 break;
684
685 default:
686 if (warnings)
687 say(path, "Not a regular file");
688 continue;
689 }
690
691 switch (ff->fts_level) {
692 case 0:
693 /* Ignore the root directory. */
694 break;
695 case 1:
696 /*
697 * This might contain manX/ or catX/.
698 * Try to infer this from the name.
699 * If we're not in use_all, enforce it.
700 */
701 cp = ff->fts_name;
702 if (FTS_DP == ff->fts_info)
703 break;
704
705 if (0 == strncmp(cp, "man", 3)) {
706 dform = FORM_SRC;
707 dsec = cp + 3;
708 } else if (0 == strncmp(cp, "cat", 3)) {
709 dform = FORM_CAT;
710 dsec = cp + 3;
711 } else {
712 dform = FORM_NONE;
713 dsec = NULL;
714 }
715
716 if (NULL != dsec || use_all)
717 break;
718
719 if (warnings)
720 say(path, "Unknown directory part");
721 fts_set(f, ff, FTS_SKIP);
722 break;
723 case 2:
724 /*
725 * Possibly our architecture.
726 * If we're descending, keep tabs on it.
727 */
728 if (FTS_DP != ff->fts_info && NULL != dsec)
729 arch = ff->fts_name;
730 else
731 arch = NULL;
732 break;
733 default:
734 if (FTS_DP == ff->fts_info || use_all)
735 break;
736 if (warnings)
737 say(path, "Extraneous directory part");
738 fts_set(f, ff, FTS_SKIP);
739 break;
740 }
741 }
742
743 fts_close(f);
744 return(1);
745 }
746
747 /*
748 * Add a file to the mlinks table.
749 * Do not verify that it's a "valid" looking manpage (we'll do that
750 * later).
751 *
752 * Try to infer the manual section, architecture, and page name from the
753 * path, assuming it looks like
754 *
755 * [./]man*[/<arch>]/<name>.<section>
756 * or
757 * [./]cat<section>[/<arch>]/<name>.0
758 *
759 * See treescan() for the fts(3) version of this.
760 */
761 static void
762 filescan(const char *file)
763 {
764 char buf[PATH_MAX];
765 struct stat st;
766 struct mlink *mlink;
767 char *p, *start;
768
769 assert(use_all);
770
771 if (0 == strncmp(file, "./", 2))
772 file += 2;
773
774 /*
775 * We have to do lstat(2) before realpath(3) loses
776 * the information whether this is a symbolic link.
777 * We need to know that because for symbolic links,
778 * we want to use the orginal file name, while for
779 * regular files, we want to use the real path.
780 */
781 if (-1 == lstat(file, &st)) {
782 exitcode = (int)MANDOCLEVEL_BADARG;
783 say(file, "&lstat");
784 return;
785 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
786 exitcode = (int)MANDOCLEVEL_BADARG;
787 say(file, "Not a regular file");
788 return;
789 }
790
791 /*
792 * We have to resolve the file name to the real path
793 * in any case for the base directory check.
794 */
795 if (NULL == realpath(file, buf)) {
796 exitcode = (int)MANDOCLEVEL_BADARG;
797 say(file, "&realpath");
798 return;
799 }
800
801 if (OP_TEST == op)
802 start = buf;
803 else if (strstr(buf, basedir) == buf)
804 start = buf + strlen(basedir);
805 else {
806 exitcode = (int)MANDOCLEVEL_BADARG;
807 say("", "%s: outside base directory", buf);
808 return;
809 }
810
811 /*
812 * Now we are sure the file is inside our tree.
813 * If it is a symbolic link, ignore the real path
814 * and use the original name.
815 * This implies passing stuff like "cat1/../man1/foo.1"
816 * on the command line won't work. So don't do that.
817 * Note the stat(2) can still fail if the link target
818 * doesn't exist.
819 */
820 if (S_IFLNK & st.st_mode) {
821 if (-1 == stat(buf, &st)) {
822 exitcode = (int)MANDOCLEVEL_BADARG;
823 say(file, "&stat");
824 return;
825 }
826 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
827 say(file, "Filename too long");
828 return;
829 }
830 start = buf;
831 if (OP_TEST != op && strstr(buf, basedir) == buf)
832 start += strlen(basedir);
833 }
834
835 mlink = mandoc_calloc(1, sizeof(struct mlink));
836 mlink->dform = FORM_NONE;
837 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
838 sizeof(mlink->file)) {
839 say(start, "Filename too long");
840 return;
841 }
842
843 /*
844 * First try to guess our directory structure.
845 * If we find a separator, try to look for man* or cat*.
846 * If we find one of these and what's underneath is a directory,
847 * assume it's an architecture.
848 */
849 if (NULL != (p = strchr(start, '/'))) {
850 *p++ = '\0';
851 if (0 == strncmp(start, "man", 3)) {
852 mlink->dform = FORM_SRC;
853 mlink->dsec = start + 3;
854 } else if (0 == strncmp(start, "cat", 3)) {
855 mlink->dform = FORM_CAT;
856 mlink->dsec = start + 3;
857 }
858
859 start = p;
860 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
861 *p++ = '\0';
862 mlink->arch = start;
863 start = p;
864 }
865 }
866
867 /*
868 * Now check the file suffix.
869 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
870 */
871 p = strrchr(start, '\0');
872 while (p-- > start && '/' != *p && '.' != *p)
873 /* Loop. */ ;
874
875 if ('.' == *p) {
876 *p++ = '\0';
877 mlink->fsec = p;
878 }
879
880 /*
881 * Now try to parse the name.
882 * Use the filename portion of the path.
883 */
884 mlink->name = start;
885 if (NULL != (p = strrchr(start, '/'))) {
886 mlink->name = p + 1;
887 *p = '\0';
888 }
889 mlink_add(mlink, &st);
890 }
891
892 static void
893 mlink_add(struct mlink *mlink, const struct stat *st)
894 {
895 struct inodev inodev;
896 struct mpage *mpage;
897 unsigned int slot;
898
899 assert(NULL != mlink->file);
900
901 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
902 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
903 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
904 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
905
906 if ('0' == *mlink->fsec) {
907 free(mlink->fsec);
908 mlink->fsec = mandoc_strdup(mlink->dsec);
909 mlink->fform = FORM_CAT;
910 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
911 mlink->fform = FORM_SRC;
912 else
913 mlink->fform = FORM_NONE;
914
915 slot = ohash_qlookup(&mlinks, mlink->file);
916 assert(NULL == ohash_find(&mlinks, slot));
917 ohash_insert(&mlinks, slot, mlink);
918
919 inodev.st_ino = st->st_ino;
920 inodev.st_dev = st->st_dev;
921 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
922 sizeof(struct inodev), inodev.st_ino);
923 mpage = ohash_find(&mpages, slot);
924 if (NULL == mpage) {
925 mpage = mandoc_calloc(1, sizeof(struct mpage));
926 mpage->inodev.st_ino = inodev.st_ino;
927 mpage->inodev.st_dev = inodev.st_dev;
928 ohash_insert(&mpages, slot, mpage);
929 } else
930 mlink->next = mpage->mlinks;
931 mpage->mlinks = mlink;
932 mlink->mpage = mpage;
933 }
934
935 static void
936 mlink_free(struct mlink *mlink)
937 {
938
939 free(mlink->dsec);
940 free(mlink->arch);
941 free(mlink->name);
942 free(mlink->fsec);
943 free(mlink);
944 }
945
946 static void
947 mpages_free(void)
948 {
949 struct mpage *mpage;
950 struct mlink *mlink;
951 unsigned int slot;
952
953 mpage = ohash_first(&mpages, &slot);
954 while (NULL != mpage) {
955 while (NULL != (mlink = mpage->mlinks)) {
956 mpage->mlinks = mlink->next;
957 mlink_free(mlink);
958 }
959 free(mpage->sec);
960 free(mpage->arch);
961 free(mpage->title);
962 free(mpage->desc);
963 free(mpage);
964 mpage = ohash_next(&mpages, &slot);
965 }
966 }
967
968 /*
969 * For each mlink to the mpage, check whether the path looks like
970 * it is formatted, and if it does, check whether a source manual
971 * exists by the same name, ignoring the suffix.
972 * If both conditions hold, drop the mlink.
973 */
974 static void
975 mlinks_undupe(struct mpage *mpage)
976 {
977 char buf[PATH_MAX];
978 struct mlink **prev;
979 struct mlink *mlink;
980 char *bufp;
981
982 mpage->form = FORM_CAT;
983 prev = &mpage->mlinks;
984 while (NULL != (mlink = *prev)) {
985 if (FORM_CAT != mlink->dform) {
986 mpage->form = FORM_NONE;
987 goto nextlink;
988 }
989 (void)strlcpy(buf, mlink->file, sizeof(buf));
990 bufp = strstr(buf, "cat");
991 assert(NULL != bufp);
992 memcpy(bufp, "man", 3);
993 if (NULL != (bufp = strrchr(buf, '.')))
994 *++bufp = '\0';
995 (void)strlcat(buf, mlink->dsec, sizeof(buf));
996 if (NULL == ohash_find(&mlinks,
997 ohash_qlookup(&mlinks, buf)))
998 goto nextlink;
999 if (warnings)
1000 say(mlink->file, "Man source exists: %s", buf);
1001 if (use_all)
1002 goto nextlink;
1003 *prev = mlink->next;
1004 mlink_free(mlink);
1005 continue;
1006 nextlink:
1007 prev = &(*prev)->next;
1008 }
1009 }
1010
1011 static void
1012 mlink_check(struct mpage *mpage, struct mlink *mlink)
1013 {
1014 struct str *str;
1015 unsigned int slot;
1016
1017 /*
1018 * Check whether the manual section given in a file
1019 * agrees with the directory where the file is located.
1020 * Some manuals have suffixes like (3p) on their
1021 * section number either inside the file or in the
1022 * directory name, some are linked into more than one
1023 * section, like encrypt(1) = makekey(8).
1024 */
1025
1026 if (FORM_SRC == mpage->form &&
1027 strcasecmp(mpage->sec, mlink->dsec))
1028 say(mlink->file, "Section \"%s\" manual in %s directory",
1029 mpage->sec, mlink->dsec);
1030
1031 /*
1032 * Manual page directories exist for each kernel
1033 * architecture as returned by machine(1).
1034 * However, many manuals only depend on the
1035 * application architecture as returned by arch(1).
1036 * For example, some (2/ARM) manuals are shared
1037 * across the "armish" and "zaurus" kernel
1038 * architectures.
1039 * A few manuals are even shared across completely
1040 * different architectures, for example fdformat(1)
1041 * on amd64, i386, sparc, and sparc64.
1042 */
1043
1044 if (strcasecmp(mpage->arch, mlink->arch))
1045 say(mlink->file, "Architecture \"%s\" manual in "
1046 "\"%s\" directory", mpage->arch, mlink->arch);
1047
1048 /*
1049 * XXX
1050 * parse_cat() doesn't set NAME_TITLE yet.
1051 */
1052
1053 if (FORM_CAT == mpage->form)
1054 return;
1055
1056 /*
1057 * Check whether this mlink
1058 * appears as a name in the NAME section.
1059 */
1060
1061 slot = ohash_qlookup(&names, mlink->name);
1062 str = ohash_find(&names, slot);
1063 assert(NULL != str);
1064 if ( ! (NAME_TITLE & str->mask))
1065 say(mlink->file, "Name missing in NAME section");
1066 }
1067
1068 /*
1069 * Run through the files in the global vector "mpages"
1070 * and add them to the database specified in "basedir".
1071 *
1072 * This handles the parsing scheme itself, using the cues of directory
1073 * and filename to determine whether the file is parsable or not.
1074 */
1075 static void
1076 mpages_merge(struct mchars *mc, struct mparse *mp)
1077 {
1078 char any[] = "any";
1079 struct ohash_info str_info;
1080 struct mpage *mpage, *mpage_dest;
1081 struct mlink *mlink, *mlink_dest;
1082 struct mdoc *mdoc;
1083 struct man *man;
1084 char *sodest;
1085 char *cp;
1086 pid_t child_pid;
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 (0 == nodb)
1097 SQL_EXEC("BEGIN TRANSACTION");
1098
1099 mpage = ohash_first(&mpages, &pslot);
1100 while (NULL != mpage) {
1101 mlinks_undupe(mpage);
1102 if (NULL == mpage->mlinks) {
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 child_pid = 0;
1115
1116 mparse_open(mp, &fd, mpage->mlinks->file, &child_pid);
1117 if (fd == -1) {
1118 say(mpage->mlinks->file, "&open");
1119 goto nextpage;
1120 }
1121
1122 /*
1123 * Try interpreting the file as mdoc(7) or man(7)
1124 * source code, unless it is already known to be
1125 * formatted. Fall back to formatted mode.
1126 */
1127 if (FORM_CAT != mpage->mlinks->dform ||
1128 FORM_CAT != mpage->mlinks->fform) {
1129 lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1130 if (lvl < MANDOCLEVEL_FATAL)
1131 mparse_result(mp, &mdoc, &man, &sodest);
1132 }
1133
1134 if (NULL != sodest) {
1135 mlink_dest = ohash_find(&mlinks,
1136 ohash_qlookup(&mlinks, sodest));
1137 if (NULL != mlink_dest) {
1138
1139 /* The .so target exists. */
1140
1141 mpage_dest = mlink_dest->mpage;
1142 mlink = mpage->mlinks;
1143 while (1) {
1144 mlink->mpage = mpage_dest;
1145
1146 /*
1147 * If the target was already
1148 * processed, add the links
1149 * to the database now.
1150 * Otherwise, this will
1151 * happen when we come
1152 * to the target.
1153 */
1154
1155 if (mpage_dest->pageid)
1156 dbadd_mlink(mlink);
1157
1158 if (NULL == mlink->next)
1159 break;
1160 mlink = mlink->next;
1161 }
1162
1163 /* Move all links to the target. */
1164
1165 mlink->next = mlink_dest->next;
1166 mlink_dest->next = mpage->mlinks;
1167 mpage->mlinks = NULL;
1168 }
1169 goto nextpage;
1170 } else if (NULL != mdoc) {
1171 mpage->form = FORM_SRC;
1172 mpage->sec = mdoc_meta(mdoc)->msec;
1173 mpage->sec = mandoc_strdup(
1174 NULL == mpage->sec ? "" : mpage->sec);
1175 mpage->arch = mdoc_meta(mdoc)->arch;
1176 mpage->arch = mandoc_strdup(
1177 NULL == mpage->arch ? "" : mpage->arch);
1178 mpage->title =
1179 mandoc_strdup(mdoc_meta(mdoc)->title);
1180 } else if (NULL != man) {
1181 mpage->form = FORM_SRC;
1182 mpage->sec =
1183 mandoc_strdup(man_meta(man)->msec);
1184 mpage->arch =
1185 mandoc_strdup(mpage->mlinks->arch);
1186 mpage->title =
1187 mandoc_strdup(man_meta(man)->title);
1188 } else {
1189 mpage->form = FORM_CAT;
1190 mpage->sec =
1191 mandoc_strdup(mpage->mlinks->dsec);
1192 mpage->arch =
1193 mandoc_strdup(mpage->mlinks->arch);
1194 mpage->title =
1195 mandoc_strdup(mpage->mlinks->name);
1196 }
1197 if (mpage->mlinks->gzip)
1198 mpage->form |= FORM_GZ;
1199 putkey(mpage, mpage->sec, TYPE_sec);
1200 if (*mpage->arch != '\0')
1201 putkey(mpage, mpage->arch, TYPE_arch);
1202
1203 for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1204 if ('\0' != *mlink->dsec)
1205 putkey(mpage, mlink->dsec, TYPE_sec);
1206 if ('\0' != *mlink->fsec)
1207 putkey(mpage, mlink->fsec, TYPE_sec);
1208 putkey(mpage, '\0' == *mlink->arch ?
1209 any : mlink->arch, TYPE_arch);
1210 putkey(mpage, mlink->name, NAME_FILE);
1211 }
1212
1213 assert(NULL == mpage->desc);
1214 if (NULL != mdoc) {
1215 if (NULL != (cp = mdoc_meta(mdoc)->name))
1216 putkey(mpage, cp, NAME_HEAD);
1217 parse_mdoc(mpage, mdoc_node(mdoc));
1218 } else if (NULL != man)
1219 parse_man(mpage, man_node(man));
1220 else
1221 parse_cat(mpage, fd);
1222 if (NULL == mpage->desc)
1223 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1224
1225 if (warnings && !use_all)
1226 for (mlink = mpage->mlinks; mlink;
1227 mlink = mlink->next)
1228 mlink_check(mpage, mlink);
1229
1230 dbadd(mpage, mc);
1231
1232 nextpage:
1233 if (child_pid &&
1234 mparse_wait(mp, child_pid) != MANDOCLEVEL_OK) {
1235 exitcode = (int)MANDOCLEVEL_SYSERR;
1236 say(mpage->mlinks->file, "&wait gunzip");
1237 }
1238 ohash_delete(&strings);
1239 ohash_delete(&names);
1240 mpage = ohash_next(&mpages, &pslot);
1241 }
1242
1243 if (0 == nodb)
1244 SQL_EXEC("END TRANSACTION");
1245 }
1246
1247 static void
1248 names_check(void)
1249 {
1250 sqlite3_stmt *stmt;
1251 const char *name, *sec, *arch, *key;
1252 int irc;
1253
1254 sqlite3_prepare_v2(db,
1255 "SELECT name, sec, arch, key FROM ("
1256 "SELECT name AS key, pageid FROM names "
1257 "WHERE bits & ? AND NOT EXISTS ("
1258 "SELECT pageid FROM mlinks "
1259 "WHERE mlinks.pageid == names.pageid "
1260 "AND mlinks.name == names.name"
1261 ")"
1262 ") JOIN ("
1263 "SELECT sec, arch, name, pageid FROM mlinks "
1264 "GROUP BY pageid"
1265 ") USING (pageid);",
1266 -1, &stmt, NULL);
1267
1268 if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1269 say("", "%s", sqlite3_errmsg(db));
1270
1271 while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1272 name = (const char *)sqlite3_column_text(stmt, 0);
1273 sec = (const char *)sqlite3_column_text(stmt, 1);
1274 arch = (const char *)sqlite3_column_text(stmt, 2);
1275 key = (const char *)sqlite3_column_text(stmt, 3);
1276 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1277 '\0' == *arch ? "" : "/",
1278 '\0' == *arch ? "" : arch, key);
1279 }
1280 sqlite3_finalize(stmt);
1281 }
1282
1283 static void
1284 parse_cat(struct mpage *mpage, int fd)
1285 {
1286 FILE *stream;
1287 char *line, *p, *title;
1288 size_t len, plen, titlesz;
1289
1290 stream = (-1 == fd) ?
1291 fopen(mpage->mlinks->file, "r") :
1292 fdopen(fd, "r");
1293 if (NULL == stream) {
1294 if (-1 != fd)
1295 close(fd);
1296 if (warnings)
1297 say(mpage->mlinks->file, "&fopen");
1298 return;
1299 }
1300
1301 /* Skip to first blank line. */
1302
1303 while (NULL != (line = fgetln(stream, &len)))
1304 if ('\n' == *line)
1305 break;
1306
1307 /*
1308 * Assume the first line that is not indented
1309 * is the first section header. Skip to it.
1310 */
1311
1312 while (NULL != (line = fgetln(stream, &len)))
1313 if ('\n' != *line && ' ' != *line)
1314 break;
1315
1316 /*
1317 * Read up until the next section into a buffer.
1318 * Strip the leading and trailing newline from each read line,
1319 * appending a trailing space.
1320 * Ignore empty (whitespace-only) lines.
1321 */
1322
1323 titlesz = 0;
1324 title = NULL;
1325
1326 while (NULL != (line = fgetln(stream, &len))) {
1327 if (' ' != *line || '\n' != line[len - 1])
1328 break;
1329 while (len > 0 && isspace((unsigned char)*line)) {
1330 line++;
1331 len--;
1332 }
1333 if (1 == len)
1334 continue;
1335 title = mandoc_realloc(title, titlesz + len);
1336 memcpy(title + titlesz, line, len);
1337 titlesz += len;
1338 title[titlesz - 1] = ' ';
1339 }
1340
1341 /*
1342 * If no page content can be found, or the input line
1343 * is already the next section header, or there is no
1344 * trailing newline, reuse the page title as the page
1345 * description.
1346 */
1347
1348 if (NULL == title || '\0' == *title) {
1349 if (warnings)
1350 say(mpage->mlinks->file,
1351 "Cannot find NAME section");
1352 fclose(stream);
1353 free(title);
1354 return;
1355 }
1356
1357 title = mandoc_realloc(title, titlesz + 1);
1358 title[titlesz] = '\0';
1359
1360 /*
1361 * Skip to the first dash.
1362 * Use the remaining line as the description (no more than 70
1363 * bytes).
1364 */
1365
1366 if (NULL != (p = strstr(title, "- "))) {
1367 for (p += 2; ' ' == *p || '\b' == *p; p++)
1368 /* Skip to next word. */ ;
1369 } else {
1370 if (warnings)
1371 say(mpage->mlinks->file,
1372 "No dash in title line");
1373 p = title;
1374 }
1375
1376 plen = strlen(p);
1377
1378 /* Strip backspace-encoding from line. */
1379
1380 while (NULL != (line = memchr(p, '\b', plen))) {
1381 len = line - p;
1382 if (0 == len) {
1383 memmove(line, line + 1, plen--);
1384 continue;
1385 }
1386 memmove(line - 1, line + 1, plen - len);
1387 plen -= 2;
1388 }
1389
1390 mpage->desc = mandoc_strdup(p);
1391 fclose(stream);
1392 free(title);
1393 }
1394
1395 /*
1396 * Put a type/word pair into the word database for this particular file.
1397 */
1398 static void
1399 putkey(const struct mpage *mpage, char *value, uint64_t type)
1400 {
1401 char *cp;
1402
1403 assert(NULL != value);
1404 if (TYPE_arch == type)
1405 for (cp = value; *cp; cp++)
1406 if (isupper((unsigned char)*cp))
1407 *cp = _tolower((unsigned char)*cp);
1408 putkeys(mpage, value, strlen(value), type);
1409 }
1410
1411 /*
1412 * Grok all nodes at or below a certain mdoc node into putkey().
1413 */
1414 static void
1415 putmdockey(const struct mpage *mpage,
1416 const struct mdoc_node *n, uint64_t m)
1417 {
1418
1419 for ( ; NULL != n; n = n->next) {
1420 if (NULL != n->child)
1421 putmdockey(mpage, n->child, m);
1422 if (MDOC_TEXT == n->type)
1423 putkey(mpage, n->string, m);
1424 }
1425 }
1426
1427 static void
1428 parse_man(struct mpage *mpage, const struct man_node *n)
1429 {
1430 const struct man_node *head, *body;
1431 char *start, *title;
1432 char byte;
1433 size_t sz;
1434
1435 if (NULL == n)
1436 return;
1437
1438 /*
1439 * We're only searching for one thing: the first text child in
1440 * the BODY of a NAME section. Since we don't keep track of
1441 * sections in -man, run some hoops to find out whether we're in
1442 * the correct section or not.
1443 */
1444
1445 if (MAN_BODY == n->type && MAN_SH == n->tok) {
1446 body = n;
1447 assert(body->parent);
1448 if (NULL != (head = body->parent->head) &&
1449 1 == head->nchild &&
1450 NULL != (head = (head->child)) &&
1451 MAN_TEXT == head->type &&
1452 0 == strcmp(head->string, "NAME") &&
1453 NULL != body->child) {
1454
1455 /*
1456 * Suck the entire NAME section into memory.
1457 * Yes, we might run away.
1458 * But too many manuals have big, spread-out
1459 * NAME sections over many lines.
1460 */
1461
1462 title = NULL;
1463 man_deroff(&title, body);
1464 if (NULL == title)
1465 return;
1466
1467 /*
1468 * Go through a special heuristic dance here.
1469 * Conventionally, one or more manual names are
1470 * comma-specified prior to a whitespace, then a
1471 * dash, then a description. Try to puzzle out
1472 * the name parts here.
1473 */
1474
1475 start = title;
1476 for ( ;; ) {
1477 sz = strcspn(start, " ,");
1478 if ('\0' == start[sz])
1479 break;
1480
1481 byte = start[sz];
1482 start[sz] = '\0';
1483
1484 /*
1485 * Assume a stray trailing comma in the
1486 * name list if a name begins with a dash.
1487 */
1488
1489 if ('-' == start[0] ||
1490 ('\\' == start[0] && '-' == start[1]))
1491 break;
1492
1493 putkey(mpage, start, NAME_TITLE);
1494
1495 if (' ' == byte) {
1496 start += sz + 1;
1497 break;
1498 }
1499
1500 assert(',' == byte);
1501 start += sz + 1;
1502 while (' ' == *start)
1503 start++;
1504 }
1505
1506 if (start == title) {
1507 putkey(mpage, start, NAME_TITLE);
1508 free(title);
1509 return;
1510 }
1511
1512 while (isspace((unsigned char)*start))
1513 start++;
1514
1515 if (0 == strncmp(start, "-", 1))
1516 start += 1;
1517 else if (0 == strncmp(start, "\\-\\-", 4))
1518 start += 4;
1519 else if (0 == strncmp(start, "\\-", 2))
1520 start += 2;
1521 else if (0 == strncmp(start, "\\(en", 4))
1522 start += 4;
1523 else if (0 == strncmp(start, "\\(em", 4))
1524 start += 4;
1525
1526 while (' ' == *start)
1527 start++;
1528
1529 mpage->desc = mandoc_strdup(start);
1530 free(title);
1531 return;
1532 }
1533 }
1534
1535 for (n = n->child; n; n = n->next) {
1536 if (NULL != mpage->desc)
1537 break;
1538 parse_man(mpage, n);
1539 }
1540 }
1541
1542 static void
1543 parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1544 {
1545
1546 assert(NULL != n);
1547 for (n = n->child; NULL != n; n = n->next) {
1548 switch (n->type) {
1549 case MDOC_ELEM:
1550 /* FALLTHROUGH */
1551 case MDOC_BLOCK:
1552 /* FALLTHROUGH */
1553 case MDOC_HEAD:
1554 /* FALLTHROUGH */
1555 case MDOC_BODY:
1556 /* FALLTHROUGH */
1557 case MDOC_TAIL:
1558 if (NULL != mdocs[n->tok].fp)
1559 if (0 == (*mdocs[n->tok].fp)(mpage, n))
1560 break;
1561 if (mdocs[n->tok].mask)
1562 putmdockey(mpage, n->child,
1563 mdocs[n->tok].mask);
1564 break;
1565 default:
1566 assert(MDOC_ROOT != n->type);
1567 continue;
1568 }
1569 if (NULL != n->child)
1570 parse_mdoc(mpage, n);
1571 }
1572 }
1573
1574 static int
1575 parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1576 {
1577 const char *start, *end;
1578 size_t sz;
1579
1580 if (SEC_SYNOPSIS != n->sec ||
1581 NULL == (n = n->child) ||
1582 MDOC_TEXT != n->type)
1583 return(0);
1584
1585 /*
1586 * Only consider those `Fd' macro fields that begin with an
1587 * "inclusion" token (versus, e.g., #define).
1588 */
1589
1590 if (strcmp("#include", n->string))
1591 return(0);
1592
1593 if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1594 return(0);
1595
1596 /*
1597 * Strip away the enclosing angle brackets and make sure we're
1598 * not zero-length.
1599 */
1600
1601 start = n->string;
1602 if ('<' == *start || '"' == *start)
1603 start++;
1604
1605 if (0 == (sz = strlen(start)))
1606 return(0);
1607
1608 end = &start[(int)sz - 1];
1609 if ('>' == *end || '"' == *end)
1610 end--;
1611
1612 if (end > start)
1613 putkeys(mpage, start, end - start + 1, TYPE_In);
1614 return(0);
1615 }
1616
1617 static int
1618 parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1619 {
1620 char *cp;
1621
1622 if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1623 return(0);
1624
1625 /*
1626 * Parse: .Fn "struct type *name" "char *arg".
1627 * First strip away pointer symbol.
1628 * Then store the function name, then type.
1629 * Finally, store the arguments.
1630 */
1631
1632 if (NULL == (cp = strrchr(n->string, ' ')))
1633 cp = n->string;
1634
1635 while ('*' == *cp)
1636 cp++;
1637
1638 putkey(mpage, cp, TYPE_Fn);
1639
1640 if (n->string < cp)
1641 putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1642
1643 for (n = n->next; NULL != n; n = n->next)
1644 if (MDOC_TEXT == n->type)
1645 putkey(mpage, n->string, TYPE_Fa);
1646
1647 return(0);
1648 }
1649
1650 static int
1651 parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1652 {
1653 char *cp;
1654
1655 if (NULL == (n = n->child))
1656 return(0);
1657
1658 if (NULL == n->next) {
1659 putkey(mpage, n->string, TYPE_Xr);
1660 return(0);
1661 }
1662
1663 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1664 putkey(mpage, cp, TYPE_Xr);
1665 free(cp);
1666 return(0);
1667 }
1668
1669 static int
1670 parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1671 {
1672
1673 if (MDOC_BODY == n->type)
1674 mdoc_deroff(&mpage->desc, n);
1675 return(0);
1676 }
1677
1678 static int
1679 parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1680 {
1681
1682 if (SEC_NAME == n->sec)
1683 putmdockey(mpage, n->child, NAME_TITLE);
1684 else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1685 putmdockey(mpage, n->child, NAME_SYN);
1686 return(0);
1687 }
1688
1689 static int
1690 parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1691 {
1692
1693 return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1694 }
1695
1696 static int
1697 parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1698 {
1699
1700 return(MDOC_HEAD == n->type);
1701 }
1702
1703 static int
1704 parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1705 {
1706
1707 return(MDOC_BODY == n->type);
1708 }
1709
1710 /*
1711 * Add a string to the hash table for the current manual.
1712 * Each string has a bitmask telling which macros it belongs to.
1713 * When we finish the manual, we'll dump the table.
1714 */
1715 static void
1716 putkeys(const struct mpage *mpage,
1717 const char *cp, size_t sz, uint64_t v)
1718 {
1719 struct ohash *htab;
1720 struct str *s;
1721 const char *end;
1722 unsigned int slot;
1723 int i;
1724
1725 if (0 == sz)
1726 return;
1727
1728 if (TYPE_Nm & v) {
1729 htab = &names;
1730 v &= name_mask;
1731 name_mask &= ~NAME_FIRST;
1732 if (debug > 1)
1733 say(mpage->mlinks->file,
1734 "Adding name %*s", sz, cp);
1735 } else {
1736 htab = &strings;
1737 if (debug > 1)
1738 for (i = 0; i < mansearch_keymax; i++)
1739 if ((uint64_t)1 << i & v)
1740 say(mpage->mlinks->file,
1741 "Adding key %s=%*s",
1742 mansearch_keynames[i], sz, cp);
1743 }
1744
1745 end = cp + sz;
1746 slot = ohash_qlookupi(htab, cp, &end);
1747 s = ohash_find(htab, slot);
1748
1749 if (NULL != s && mpage == s->mpage) {
1750 s->mask |= v;
1751 return;
1752 } else if (NULL == s) {
1753 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1754 memcpy(s->key, cp, sz);
1755 ohash_insert(htab, slot, s);
1756 }
1757 s->mpage = mpage;
1758 s->mask = v;
1759 }
1760
1761 /*
1762 * Take a Unicode codepoint and produce its UTF-8 encoding.
1763 * This isn't the best way to do this, but it works.
1764 * The magic numbers are from the UTF-8 packaging.
1765 * They're not as scary as they seem: read the UTF-8 spec for details.
1766 */
1767 static size_t
1768 utf8(unsigned int cp, char out[7])
1769 {
1770 size_t rc;
1771
1772 rc = 0;
1773 if (cp <= 0x0000007F) {
1774 rc = 1;
1775 out[0] = (char)cp;
1776 } else if (cp <= 0x000007FF) {
1777 rc = 2;
1778 out[0] = (cp >> 6 & 31) | 192;
1779 out[1] = (cp & 63) | 128;
1780 } else if (cp <= 0x0000FFFF) {
1781 rc = 3;
1782 out[0] = (cp >> 12 & 15) | 224;
1783 out[1] = (cp >> 6 & 63) | 128;
1784 out[2] = (cp & 63) | 128;
1785 } else if (cp <= 0x001FFFFF) {
1786 rc = 4;
1787 out[0] = (cp >> 18 & 7) | 240;
1788 out[1] = (cp >> 12 & 63) | 128;
1789 out[2] = (cp >> 6 & 63) | 128;
1790 out[3] = (cp & 63) | 128;
1791 } else if (cp <= 0x03FFFFFF) {
1792 rc = 5;
1793 out[0] = (cp >> 24 & 3) | 248;
1794 out[1] = (cp >> 18 & 63) | 128;
1795 out[2] = (cp >> 12 & 63) | 128;
1796 out[3] = (cp >> 6 & 63) | 128;
1797 out[4] = (cp & 63) | 128;
1798 } else if (cp <= 0x7FFFFFFF) {
1799 rc = 6;
1800 out[0] = (cp >> 30 & 1) | 252;
1801 out[1] = (cp >> 24 & 63) | 128;
1802 out[2] = (cp >> 18 & 63) | 128;
1803 out[3] = (cp >> 12 & 63) | 128;
1804 out[4] = (cp >> 6 & 63) | 128;
1805 out[5] = (cp & 63) | 128;
1806 } else
1807 return(0);
1808
1809 out[rc] = '\0';
1810 return(rc);
1811 }
1812
1813 /*
1814 * Store the rendered version of a key, or alias the pointer
1815 * if the key contains no escape sequences.
1816 */
1817 static void
1818 render_key(struct mchars *mc, struct str *key)
1819 {
1820 size_t sz, bsz, pos;
1821 char utfbuf[7], res[6];
1822 char *buf;
1823 const char *seq, *cpp, *val;
1824 int len, u;
1825 enum mandoc_esc esc;
1826
1827 assert(NULL == key->rendered);
1828
1829 res[0] = '\\';
1830 res[1] = '\t';
1831 res[2] = ASCII_NBRSP;
1832 res[3] = ASCII_HYPH;
1833 res[4] = ASCII_BREAK;
1834 res[5] = '\0';
1835
1836 val = key->key;
1837 bsz = strlen(val);
1838
1839 /*
1840 * Pre-check: if we have no stop-characters, then set the
1841 * pointer as ourselvse and get out of here.
1842 */
1843 if (strcspn(val, res) == bsz) {
1844 key->rendered = key->key;
1845 return;
1846 }
1847
1848 /* Pre-allocate by the length of the input */
1849
1850 buf = mandoc_malloc(++bsz);
1851 pos = 0;
1852
1853 while ('\0' != *val) {
1854 /*
1855 * Halt on the first escape sequence.
1856 * This also halts on the end of string, in which case
1857 * we just copy, fallthrough, and exit the loop.
1858 */
1859 if ((sz = strcspn(val, res)) > 0) {
1860 memcpy(&buf[pos], val, sz);
1861 pos += sz;
1862 val += sz;
1863 }
1864
1865 switch (*val) {
1866 case ASCII_HYPH:
1867 buf[pos++] = '-';
1868 val++;
1869 continue;
1870 case '\t':
1871 /* FALLTHROUGH */
1872 case ASCII_NBRSP:
1873 buf[pos++] = ' ';
1874 val++;
1875 /* FALLTHROUGH */
1876 case ASCII_BREAK:
1877 continue;
1878 default:
1879 break;
1880 }
1881 if ('\\' != *val)
1882 break;
1883
1884 /* Read past the slash. */
1885
1886 val++;
1887
1888 /*
1889 * Parse the escape sequence and see if it's a
1890 * predefined character or special character.
1891 */
1892
1893 esc = mandoc_escape((const char **)&val,
1894 &seq, &len);
1895 if (ESCAPE_ERROR == esc)
1896 break;
1897 if (ESCAPE_SPECIAL != esc)
1898 continue;
1899
1900 /*
1901 * Render the special character
1902 * as either UTF-8 or ASCII.
1903 */
1904
1905 if (write_utf8) {
1906 if ((u = mchars_spec2cp(mc, seq, len)) <= 0)
1907 continue;
1908 cpp = utfbuf;
1909 if (0 == (sz = utf8(u, utfbuf)))
1910 continue;
1911 sz = strlen(cpp);
1912 } else {
1913 cpp = mchars_spec2str(mc, seq, len, &sz);
1914 if (NULL == cpp)
1915 continue;
1916 if (ASCII_NBRSP == *cpp) {
1917 cpp = " ";
1918 sz = 1;
1919 }
1920 }
1921
1922 /* Copy the rendered glyph into the stream. */
1923
1924 bsz += sz;
1925 buf = mandoc_realloc(buf, bsz);
1926 memcpy(&buf[pos], cpp, sz);
1927 pos += sz;
1928 }
1929
1930 buf[pos] = '\0';
1931 key->rendered = buf;
1932 }
1933
1934 static void
1935 dbadd_mlink(const struct mlink *mlink)
1936 {
1937 size_t i;
1938
1939 i = 1;
1940 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1941 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1942 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1943 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
1944 SQL_STEP(stmts[STMT_INSERT_LINK]);
1945 sqlite3_reset(stmts[STMT_INSERT_LINK]);
1946
1947 i = 1;
1948 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, NAME_FILE);
1949 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
1950 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
1951 SQL_STEP(stmts[STMT_INSERT_NAME]);
1952 sqlite3_reset(stmts[STMT_INSERT_NAME]);
1953 }
1954
1955 /*
1956 * Flush the current page's terms (and their bits) into the database.
1957 * Wrap the entire set of additions in a transaction to make sqlite be a
1958 * little faster.
1959 * Also, handle escape sequences at the last possible moment.
1960 */
1961 static void
1962 dbadd(struct mpage *mpage, struct mchars *mc)
1963 {
1964 struct mlink *mlink;
1965 struct str *key;
1966 size_t i;
1967 unsigned int slot;
1968
1969 mlink = mpage->mlinks;
1970
1971 if (nodb) {
1972 for (key = ohash_first(&names, &slot); NULL != key;
1973 key = ohash_next(&names, &slot)) {
1974 if (key->rendered != key->key)
1975 free(key->rendered);
1976 free(key);
1977 }
1978 for (key = ohash_first(&strings, &slot); NULL != key;
1979 key = ohash_next(&strings, &slot)) {
1980 if (key->rendered != key->key)
1981 free(key->rendered);
1982 free(key);
1983 }
1984 if (0 == debug)
1985 return;
1986 while (NULL != mlink) {
1987 fputs(mlink->name, stdout);
1988 if (NULL == mlink->next ||
1989 strcmp(mlink->dsec, mlink->next->dsec) ||
1990 strcmp(mlink->fsec, mlink->next->fsec) ||
1991 strcmp(mlink->arch, mlink->next->arch)) {
1992 putchar('(');
1993 if ('\0' == *mlink->dsec)
1994 fputs(mlink->fsec, stdout);
1995 else
1996 fputs(mlink->dsec, stdout);
1997 if ('\0' != *mlink->arch)
1998 printf("/%s", mlink->arch);
1999 putchar(')');
2000 }
2001 mlink = mlink->next;
2002 if (NULL != mlink)
2003 fputs(", ", stdout);
2004 }
2005 printf(" - %s\n", mpage->desc);
2006 return;
2007 }
2008
2009 if (debug)
2010 say(mlink->file, "Adding to database");
2011
2012 i = strlen(mpage->desc) + 1;
2013 key = mandoc_calloc(1, sizeof(struct str) + i);
2014 memcpy(key->key, mpage->desc, i);
2015 render_key(mc, key);
2016
2017 i = 1;
2018 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, key->rendered);
2019 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
2020 SQL_STEP(stmts[STMT_INSERT_PAGE]);
2021 mpage->pageid = sqlite3_last_insert_rowid(db);
2022 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2023
2024 if (key->rendered != key->key)
2025 free(key->rendered);
2026 free(key);
2027
2028 while (NULL != mlink) {
2029 dbadd_mlink(mlink);
2030 mlink = mlink->next;
2031 }
2032 mlink = mpage->mlinks;
2033
2034 for (key = ohash_first(&names, &slot); NULL != key;
2035 key = ohash_next(&names, &slot)) {
2036 assert(key->mpage == mpage);
2037 if (NULL == key->rendered)
2038 render_key(mc, key);
2039 i = 1;
2040 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2041 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
2042 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
2043 SQL_STEP(stmts[STMT_INSERT_NAME]);
2044 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2045 if (key->rendered != key->key)
2046 free(key->rendered);
2047 free(key);
2048 }
2049 for (key = ohash_first(&strings, &slot); NULL != key;
2050 key = ohash_next(&strings, &slot)) {
2051 assert(key->mpage == mpage);
2052 if (NULL == key->rendered)
2053 render_key(mc, key);
2054 i = 1;
2055 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
2056 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
2057 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
2058 SQL_STEP(stmts[STMT_INSERT_KEY]);
2059 sqlite3_reset(stmts[STMT_INSERT_KEY]);
2060 if (key->rendered != key->key)
2061 free(key->rendered);
2062 free(key);
2063 }
2064 }
2065
2066 static void
2067 dbprune(void)
2068 {
2069 struct mpage *mpage;
2070 struct mlink *mlink;
2071 size_t i;
2072 unsigned int slot;
2073
2074 if (0 == nodb)
2075 SQL_EXEC("BEGIN TRANSACTION");
2076
2077 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2078 mpage = ohash_next(&mpages, &slot)) {
2079 mlink = mpage->mlinks;
2080 if (debug)
2081 say(mlink->file, "Deleting from database");
2082 if (nodb)
2083 continue;
2084 for ( ; NULL != mlink; mlink = mlink->next) {
2085 i = 1;
2086 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2087 i, mlink->dsec);
2088 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2089 i, mlink->arch);
2090 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2091 i, mlink->name);
2092 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2093 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2094 }
2095 }
2096
2097 if (0 == nodb)
2098 SQL_EXEC("END TRANSACTION");
2099 }
2100
2101 /*
2102 * Close an existing database and its prepared statements.
2103 * If "real" is not set, rename the temporary file into the real one.
2104 */
2105 static void
2106 dbclose(int real)
2107 {
2108 size_t i;
2109 int status;
2110 pid_t child;
2111
2112 if (nodb)
2113 return;
2114
2115 for (i = 0; i < STMT__MAX; i++) {
2116 sqlite3_finalize(stmts[i]);
2117 stmts[i] = NULL;
2118 }
2119
2120 sqlite3_close(db);
2121 db = NULL;
2122
2123 if (real)
2124 return;
2125
2126 if ('\0' == *tempfilename) {
2127 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2128 exitcode = (int)MANDOCLEVEL_SYSERR;
2129 say(MANDOC_DB, "&rename");
2130 }
2131 return;
2132 }
2133
2134 switch (child = fork()) {
2135 case -1:
2136 exitcode = (int)MANDOCLEVEL_SYSERR;
2137 say("", "&fork cmp");
2138 return;
2139 case 0:
2140 execlp("cmp", "cmp", "-s",
2141 tempfilename, MANDOC_DB, NULL);
2142 say("", "&exec cmp");
2143 exit(0);
2144 default:
2145 break;
2146 }
2147 if (-1 == waitpid(child, &status, 0)) {
2148 exitcode = (int)MANDOCLEVEL_SYSERR;
2149 say("", "&wait cmp");
2150 } else if (WIFSIGNALED(status)) {
2151 exitcode = (int)MANDOCLEVEL_SYSERR;
2152 say("", "cmp died from signal %d", WTERMSIG(status));
2153 } else if (WEXITSTATUS(status)) {
2154 exitcode = (int)MANDOCLEVEL_SYSERR;
2155 say(MANDOC_DB,
2156 "Data changed, but cannot replace database");
2157 }
2158
2159 *strrchr(tempfilename, '/') = '\0';
2160 switch (child = fork()) {
2161 case -1:
2162 exitcode = (int)MANDOCLEVEL_SYSERR;
2163 say("", "&fork rm");
2164 return;
2165 case 0:
2166 execlp("rm", "rm", "-rf", tempfilename, NULL);
2167 say("", "&exec rm");
2168 exit((int)MANDOCLEVEL_SYSERR);
2169 default:
2170 break;
2171 }
2172 if (-1 == waitpid(child, &status, 0)) {
2173 exitcode = (int)MANDOCLEVEL_SYSERR;
2174 say("", "&wait rm");
2175 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2176 exitcode = (int)MANDOCLEVEL_SYSERR;
2177 say("", "%s: Cannot remove temporary directory",
2178 tempfilename);
2179 }
2180 }
2181
2182 /*
2183 * This is straightforward stuff.
2184 * Open a database connection to a "temporary" database, then open a set
2185 * of prepared statements we'll use over and over again.
2186 * If "real" is set, we use the existing database; if not, we truncate a
2187 * temporary one.
2188 * Must be matched by dbclose().
2189 */
2190 static int
2191 dbopen(int real)
2192 {
2193 const char *sql;
2194 int rc, ofl;
2195
2196 if (nodb)
2197 return(1);
2198
2199 *tempfilename = '\0';
2200 ofl = SQLITE_OPEN_READWRITE;
2201
2202 if (real) {
2203 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2204 if (SQLITE_OK != rc) {
2205 exitcode = (int)MANDOCLEVEL_SYSERR;
2206 if (SQLITE_CANTOPEN != rc)
2207 say(MANDOC_DB, "%s", sqlite3_errstr(rc));
2208 return(0);
2209 }
2210 goto prepare_statements;
2211 }
2212
2213 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2214
2215 remove(MANDOC_DB "~");
2216 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2217 if (SQLITE_OK == rc)
2218 goto create_tables;
2219 if (MPARSE_QUICK & mparse_options) {
2220 exitcode = (int)MANDOCLEVEL_SYSERR;
2221 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
2222 return(0);
2223 }
2224
2225 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2226 sizeof(tempfilename));
2227 if (NULL == mkdtemp(tempfilename)) {
2228 exitcode = (int)MANDOCLEVEL_SYSERR;
2229 say("", "&%s", tempfilename);
2230 return(0);
2231 }
2232 (void)strlcat(tempfilename, "/" MANDOC_DB,
2233 sizeof(tempfilename));
2234 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2235 if (SQLITE_OK != rc) {
2236 exitcode = (int)MANDOCLEVEL_SYSERR;
2237 say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
2238 return(0);
2239 }
2240
2241 create_tables:
2242 sql = "CREATE TABLE \"mpages\" (\n"
2243 " \"desc\" TEXT NOT NULL,\n"
2244 " \"form\" INTEGER NOT NULL,\n"
2245 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2246 ");\n"
2247 "\n"
2248 "CREATE TABLE \"mlinks\" (\n"
2249 " \"sec\" TEXT NOT NULL,\n"
2250 " \"arch\" TEXT NOT NULL,\n"
2251 " \"name\" TEXT NOT NULL,\n"
2252 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2253 "ON DELETE CASCADE\n"
2254 ");\n"
2255 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2256 "\n"
2257 "CREATE TABLE \"names\" (\n"
2258 " \"bits\" INTEGER NOT NULL,\n"
2259 " \"name\" TEXT NOT NULL,\n"
2260 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2261 "ON DELETE CASCADE\n"
2262 ");\n"
2263 "\n"
2264 "CREATE TABLE \"keys\" (\n"
2265 " \"bits\" INTEGER NOT NULL,\n"
2266 " \"key\" TEXT NOT NULL,\n"
2267 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2268 "ON DELETE CASCADE\n"
2269 ");\n"
2270 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2271
2272 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2273 exitcode = (int)MANDOCLEVEL_SYSERR;
2274 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2275 sqlite3_close(db);
2276 return(0);
2277 }
2278
2279 prepare_statements:
2280 if (SQLITE_OK != sqlite3_exec(db,
2281 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2282 exitcode = (int)MANDOCLEVEL_SYSERR;
2283 say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2284 sqlite3_errmsg(db));
2285 sqlite3_close(db);
2286 return(0);
2287 }
2288
2289 sql = "DELETE FROM mpages WHERE pageid IN "
2290 "(SELECT pageid FROM mlinks WHERE "
2291 "sec=? AND arch=? AND name=?)";
2292 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2293 sql = "INSERT INTO mpages "
2294 "(desc,form) VALUES (?,?)";
2295 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2296 sql = "INSERT INTO mlinks "
2297 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2298 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2299 sql = "INSERT INTO names "
2300 "(bits,name,pageid) VALUES (?,?,?)";
2301 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2302 sql = "INSERT INTO keys "
2303 "(bits,key,pageid) VALUES (?,?,?)";
2304 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2305
2306 #ifndef __APPLE__
2307 /*
2308 * When opening a new database, we can turn off
2309 * synchronous mode for much better performance.
2310 */
2311
2312 if (real && SQLITE_OK != sqlite3_exec(db,
2313 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2314 exitcode = (int)MANDOCLEVEL_SYSERR;
2315 say(MANDOC_DB, "PRAGMA synchronous: %s",
2316 sqlite3_errmsg(db));
2317 sqlite3_close(db);
2318 return(0);
2319 }
2320 #endif
2321
2322 return(1);
2323 }
2324
2325 static void *
2326 hash_calloc(size_t n, size_t sz, void *arg)
2327 {
2328
2329 return(mandoc_calloc(n, sz));
2330 }
2331
2332 static void *
2333 hash_alloc(size_t sz, void *arg)
2334 {
2335
2336 return(mandoc_malloc(sz));
2337 }
2338
2339 static void
2340 hash_free(void *p, void *arg)
2341 {
2342
2343 free(p);
2344 }
2345
2346 static int
2347 set_basedir(const char *targetdir, int report_baddir)
2348 {
2349 static char startdir[PATH_MAX];
2350 static int getcwd_status; /* 1 = ok, 2 = failure */
2351 static int chdir_status; /* 1 = changed directory */
2352 char *cp;
2353
2354 /*
2355 * Remember the original working directory, if possible.
2356 * This will be needed if the second or a later directory
2357 * on the command line is given as a relative path.
2358 * Do not error out if the current directory is not
2359 * searchable: Maybe it won't be needed after all.
2360 */
2361 if (0 == getcwd_status) {
2362 if (NULL == getcwd(startdir, sizeof(startdir))) {
2363 getcwd_status = 2;
2364 (void)strlcpy(startdir, strerror(errno),
2365 sizeof(startdir));
2366 } else
2367 getcwd_status = 1;
2368 }
2369
2370 /*
2371 * We are leaving the old base directory.
2372 * Do not use it any longer, not even for messages.
2373 */
2374 *basedir = '\0';
2375
2376 /*
2377 * If and only if the directory was changed earlier and
2378 * the next directory to process is given as a relative path,
2379 * first go back, or bail out if that is impossible.
2380 */
2381 if (chdir_status && '/' != *targetdir) {
2382 if (2 == getcwd_status) {
2383 exitcode = (int)MANDOCLEVEL_SYSERR;
2384 say("", "getcwd: %s", startdir);
2385 return(0);
2386 }
2387 if (-1 == chdir(startdir)) {
2388 exitcode = (int)MANDOCLEVEL_SYSERR;
2389 say("", "&chdir %s", startdir);
2390 return(0);
2391 }
2392 }
2393
2394 /*
2395 * Always resolve basedir to the canonicalized absolute
2396 * pathname and append a trailing slash, such that
2397 * we can reliably check whether files are inside.
2398 */
2399 if (NULL == realpath(targetdir, basedir)) {
2400 if (report_baddir || errno != ENOENT) {
2401 exitcode = (int)MANDOCLEVEL_BADARG;
2402 say("", "&%s: realpath", targetdir);
2403 }
2404 return(0);
2405 } else if (-1 == chdir(basedir)) {
2406 if (report_baddir || errno != ENOENT) {
2407 exitcode = (int)MANDOCLEVEL_BADARG;
2408 say("", "&chdir");
2409 }
2410 return(0);
2411 }
2412 chdir_status = 1;
2413 cp = strchr(basedir, '\0');
2414 if ('/' != cp[-1]) {
2415 if (cp - basedir >= PATH_MAX - 1) {
2416 exitcode = (int)MANDOCLEVEL_SYSERR;
2417 say("", "Filename too long");
2418 return(0);
2419 }
2420 *cp++ = '/';
2421 *cp = '\0';
2422 }
2423 return(1);
2424 }
2425
2426 static void
2427 say(const char *file, const char *format, ...)
2428 {
2429 va_list ap;
2430 int use_errno;
2431
2432 if ('\0' != *basedir)
2433 fprintf(stderr, "%s", basedir);
2434 if ('\0' != *basedir && '\0' != *file)
2435 fputc('/', stderr);
2436 if ('\0' != *file)
2437 fprintf(stderr, "%s", file);
2438
2439 use_errno = 1;
2440 if (NULL != format) {
2441 switch (*format) {
2442 case '&':
2443 format++;
2444 break;
2445 case '\0':
2446 format = NULL;
2447 break;
2448 default:
2449 use_errno = 0;
2450 break;
2451 }
2452 }
2453 if (NULL != format) {
2454 if ('\0' != *basedir || '\0' != *file)
2455 fputs(": ", stderr);
2456 va_start(ap, format);
2457 vfprintf(stderr, format, ap);
2458 va_end(ap);
2459 }
2460 if (use_errno) {
2461 if ('\0' != *basedir || '\0' != *file || NULL != format)
2462 fputs(": ", stderr);
2463 perror(NULL);
2464 } else
2465 fputc('\n', stderr);
2466 }
A git conversion mirror of mandoc.bsd.lv