]> 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
Next speed optimization step for the new apropos(1).
[mandoc.git] / mandocdb.c
1 /* $Id: mandocdb.c,v 1.133 2014/04/10 02:46:21 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 #ifdef HAVE_CONFIG_H
19 #include "config.h"
20 #endif
21
22 #include <sys/stat.h>
23 #include <sys/wait.h>
24
25 #include <assert.h>
26 #include <ctype.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <fts.h>
30 #include <getopt.h>
31 #include <limits.h>
32 #include <stddef.h>
33 #include <stdio.h>
34 #include <stdint.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <unistd.h>
38
39 #ifdef HAVE_OHASH
40 #include <ohash.h>
41 #else
42 #include "compat_ohash.h"
43 #endif
44 #include <sqlite3.h>
45
46 #include "mdoc.h"
47 #include "man.h"
48 #include "mandoc.h"
49 #include "mandoc_aux.h"
50 #include "manpath.h"
51 #include "mansearch.h"
52
53 extern int mansearch_keymax;
54 extern const char *const mansearch_keynames[];
55
56 #define SQL_EXEC(_v) \
57 if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
58 fprintf(stderr, "%s\n", sqlite3_errmsg(db))
59 #define SQL_BIND_TEXT(_s, _i, _v) \
60 if (SQLITE_OK != sqlite3_bind_text \
61 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
62 fprintf(stderr, "%s\n", sqlite3_errmsg(db))
63 #define SQL_BIND_INT(_s, _i, _v) \
64 if (SQLITE_OK != sqlite3_bind_int \
65 ((_s), (_i)++, (_v))) \
66 fprintf(stderr, "%s\n", sqlite3_errmsg(db))
67 #define SQL_BIND_INT64(_s, _i, _v) \
68 if (SQLITE_OK != sqlite3_bind_int64 \
69 ((_s), (_i)++, (_v))) \
70 fprintf(stderr, "%s\n", sqlite3_errmsg(db))
71 #define SQL_STEP(_s) \
72 if (SQLITE_DONE != sqlite3_step((_s))) \
73 fprintf(stderr, "%s\n", sqlite3_errmsg(db))
74
75 enum op {
76 OP_DEFAULT = 0, /* new dbs from dir list or default config */
77 OP_CONFFILE, /* new databases from custom config file */
78 OP_UPDATE, /* delete/add entries in existing database */
79 OP_DELETE, /* delete entries from existing database */
80 OP_TEST /* change no databases, report potential problems */
81 };
82
83 enum form {
84 FORM_NONE, /* format is unknown */
85 FORM_SRC, /* format is -man or -mdoc */
86 FORM_CAT /* format is cat */
87 };
88
89 struct str {
90 char *rendered; /* key in UTF-8 or ASCII form */
91 const struct mpage *mpage; /* if set, the owning parse */
92 uint64_t mask; /* bitmask in sequence */
93 char key[]; /* may contain escape sequences */
94 };
95
96 struct inodev {
97 ino_t st_ino;
98 dev_t st_dev;
99 };
100
101 struct mpage {
102 struct inodev inodev; /* used for hashing routine */
103 int64_t recno; /* id in mpages SQL table */
104 enum form form; /* format from file content */
105 char *sec; /* section from file content */
106 char *arch; /* architecture from file content */
107 char *title; /* title from file content */
108 char *desc; /* description from file content */
109 struct mlink *mlinks; /* singly linked list */
110 };
111
112 struct mlink {
113 char file[PATH_MAX]; /* filename rel. to manpath */
114 enum form dform; /* format from directory */
115 enum form fform; /* format from file name suffix */
116 char *dsec; /* section from directory */
117 char *arch; /* architecture from directory */
118 char *name; /* name from file name (not empty) */
119 char *fsec; /* section from file name suffix */
120 struct mlink *next; /* singly linked list */
121 struct mpage *mpage; /* parent */
122 int gzip; /* filename has a .gz suffix */
123 };
124
125 enum stmt {
126 STMT_DELETE_PAGE = 0, /* delete mpage */
127 STMT_INSERT_PAGE, /* insert mpage */
128 STMT_INSERT_LINK, /* insert mlink */
129 STMT_INSERT_NAME, /* insert name */
130 STMT_INSERT_KEY, /* insert parsed key */
131 STMT__MAX
132 };
133
134 typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
135
136 struct mdoc_handler {
137 mdoc_fp fp; /* optional handler */
138 uint64_t mask; /* set unless handler returns 0 */
139 };
140
141 static void dbclose(int);
142 static void dbadd(struct mpage *, struct mchars *);
143 static void dbadd_mlink(const struct mlink *mlink);
144 static int dbopen(int);
145 static void dbprune(void);
146 static void filescan(const char *);
147 static void *hash_alloc(size_t, void *);
148 static void hash_free(void *, size_t, void *);
149 static void *hash_halloc(size_t, void *);
150 static void mlink_add(struct mlink *, const struct stat *);
151 static void mlink_check(struct mpage *, struct mlink *);
152 static void mlink_free(struct mlink *);
153 static void mlinks_undupe(struct mpage *);
154 static void mpages_free(void);
155 static void mpages_merge(struct mchars *, struct mparse *);
156 static void names_check(void);
157 static void parse_cat(struct mpage *, int);
158 static void parse_man(struct mpage *, const struct man_node *);
159 static void parse_mdoc(struct mpage *, const struct mdoc_node *);
160 static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
161 static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
162 static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
163 static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
164 static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
165 static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
166 static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
167 static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
168 static void putkey(const struct mpage *, char *, uint64_t);
169 static void putkeys(const struct mpage *,
170 const char *, size_t, uint64_t);
171 static void putmdockey(const struct mpage *,
172 const struct mdoc_node *, uint64_t);
173 static void render_key(struct mchars *, struct str *);
174 static void say(const char *, const char *, ...);
175 static int set_basedir(const char *);
176 static int treescan(void);
177 static size_t utf8(unsigned int, char [7]);
178
179 static char tempfilename[32];
180 static char *progname;
181 static int nodb; /* no database changes */
182 static int mparse_options; /* abort the parse early */
183 static int use_all; /* use all found files */
184 static int debug; /* print what we're doing */
185 static int warnings; /* warn about crap */
186 static int write_utf8; /* write UTF-8 output; else ASCII */
187 static int exitcode; /* to be returned by main */
188 static enum op op; /* operational mode */
189 static char basedir[PATH_MAX]; /* current base directory */
190 static struct ohash mpages; /* table of distinct manual pages */
191 static struct ohash mlinks; /* table of directory entries */
192 static struct ohash names; /* table of all names */
193 static struct ohash strings; /* table of all strings */
194 static sqlite3 *db = NULL; /* current database */
195 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
196 static uint64_t name_mask;
197
198 static const struct mdoc_handler mdocs[MDOC_MAX] = {
199 { NULL, 0 }, /* Ap */
200 { NULL, 0 }, /* Dd */
201 { NULL, 0 }, /* Dt */
202 { NULL, 0 }, /* Os */
203 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
204 { parse_mdoc_head, TYPE_Ss }, /* Ss */
205 { NULL, 0 }, /* Pp */
206 { NULL, 0 }, /* D1 */
207 { NULL, 0 }, /* Dl */
208 { NULL, 0 }, /* Bd */
209 { NULL, 0 }, /* Ed */
210 { NULL, 0 }, /* Bl */
211 { NULL, 0 }, /* El */
212 { NULL, 0 }, /* It */
213 { NULL, 0 }, /* Ad */
214 { NULL, TYPE_An }, /* An */
215 { NULL, TYPE_Ar }, /* Ar */
216 { NULL, TYPE_Cd }, /* Cd */
217 { NULL, TYPE_Cm }, /* Cm */
218 { NULL, TYPE_Dv }, /* Dv */
219 { NULL, TYPE_Er }, /* Er */
220 { NULL, TYPE_Ev }, /* Ev */
221 { NULL, 0 }, /* Ex */
222 { NULL, TYPE_Fa }, /* Fa */
223 { parse_mdoc_Fd, 0 }, /* Fd */
224 { NULL, TYPE_Fl }, /* Fl */
225 { parse_mdoc_Fn, 0 }, /* Fn */
226 { NULL, TYPE_Ft }, /* Ft */
227 { NULL, TYPE_Ic }, /* Ic */
228 { NULL, TYPE_In }, /* In */
229 { NULL, TYPE_Li }, /* Li */
230 { parse_mdoc_Nd, 0 }, /* Nd */
231 { parse_mdoc_Nm, 0 }, /* Nm */
232 { NULL, 0 }, /* Op */
233 { NULL, 0 }, /* Ot */
234 { NULL, TYPE_Pa }, /* Pa */
235 { NULL, 0 }, /* Rv */
236 { NULL, TYPE_St }, /* St */
237 { NULL, TYPE_Va }, /* Va */
238 { parse_mdoc_body, TYPE_Va }, /* Vt */
239 { parse_mdoc_Xr, 0 }, /* Xr */
240 { NULL, 0 }, /* %A */
241 { NULL, 0 }, /* %B */
242 { NULL, 0 }, /* %D */
243 { NULL, 0 }, /* %I */
244 { NULL, 0 }, /* %J */
245 { NULL, 0 }, /* %N */
246 { NULL, 0 }, /* %O */
247 { NULL, 0 }, /* %P */
248 { NULL, 0 }, /* %R */
249 { NULL, 0 }, /* %T */
250 { NULL, 0 }, /* %V */
251 { NULL, 0 }, /* Ac */
252 { NULL, 0 }, /* Ao */
253 { NULL, 0 }, /* Aq */
254 { NULL, TYPE_At }, /* At */
255 { NULL, 0 }, /* Bc */
256 { NULL, 0 }, /* Bf */
257 { NULL, 0 }, /* Bo */
258 { NULL, 0 }, /* Bq */
259 { NULL, TYPE_Bsx }, /* Bsx */
260 { NULL, TYPE_Bx }, /* Bx */
261 { NULL, 0 }, /* Db */
262 { NULL, 0 }, /* Dc */
263 { NULL, 0 }, /* Do */
264 { NULL, 0 }, /* Dq */
265 { NULL, 0 }, /* Ec */
266 { NULL, 0 }, /* Ef */
267 { NULL, TYPE_Em }, /* Em */
268 { NULL, 0 }, /* Eo */
269 { NULL, TYPE_Fx }, /* Fx */
270 { NULL, TYPE_Ms }, /* Ms */
271 { NULL, 0 }, /* No */
272 { NULL, 0 }, /* Ns */
273 { NULL, TYPE_Nx }, /* Nx */
274 { NULL, TYPE_Ox }, /* Ox */
275 { NULL, 0 }, /* Pc */
276 { NULL, 0 }, /* Pf */
277 { NULL, 0 }, /* Po */
278 { NULL, 0 }, /* Pq */
279 { NULL, 0 }, /* Qc */
280 { NULL, 0 }, /* Ql */
281 { NULL, 0 }, /* Qo */
282 { NULL, 0 }, /* Qq */
283 { NULL, 0 }, /* Re */
284 { NULL, 0 }, /* Rs */
285 { NULL, 0 }, /* Sc */
286 { NULL, 0 }, /* So */
287 { NULL, 0 }, /* Sq */
288 { NULL, 0 }, /* Sm */
289 { NULL, 0 }, /* Sx */
290 { NULL, TYPE_Sy }, /* Sy */
291 { NULL, TYPE_Tn }, /* Tn */
292 { NULL, 0 }, /* Ux */
293 { NULL, 0 }, /* Xc */
294 { NULL, 0 }, /* Xo */
295 { parse_mdoc_head, 0 }, /* Fo */
296 { NULL, 0 }, /* Fc */
297 { NULL, 0 }, /* Oo */
298 { NULL, 0 }, /* Oc */
299 { NULL, 0 }, /* Bk */
300 { NULL, 0 }, /* Ek */
301 { NULL, 0 }, /* Bt */
302 { NULL, 0 }, /* Hf */
303 { NULL, 0 }, /* Fr */
304 { NULL, 0 }, /* Ud */
305 { NULL, TYPE_Lb }, /* Lb */
306 { NULL, 0 }, /* Lp */
307 { NULL, TYPE_Lk }, /* Lk */
308 { NULL, TYPE_Mt }, /* Mt */
309 { NULL, 0 }, /* Brq */
310 { NULL, 0 }, /* Bro */
311 { NULL, 0 }, /* Brc */
312 { NULL, 0 }, /* %C */
313 { NULL, 0 }, /* Es */
314 { NULL, 0 }, /* En */
315 { NULL, TYPE_Dx }, /* Dx */
316 { NULL, 0 }, /* %Q */
317 { NULL, 0 }, /* br */
318 { NULL, 0 }, /* sp */
319 { NULL, 0 }, /* %U */
320 { NULL, 0 }, /* Ta */
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.halloc = mlinks_info.halloc = hash_halloc;
339 mpages_info.hfree = mlinks_info.hfree = 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, "-%c: Conflicting option\n", (_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, "-T%s: Unsupported "
394 "output format\n", optarg);
395 goto usage;
396 }
397 write_utf8 = 1;
398 break;
399 case ('t'):
400 CHECKOP(op, ch);
401 dup2(STDOUT_FILENO, STDERR_FILENO);
402 op = OP_TEST;
403 nodb = warnings = 1;
404 break;
405 case ('u'):
406 CHECKOP(op, ch);
407 path_arg = optarg;
408 op = OP_DELETE;
409 break;
410 case ('v'):
411 /* Compatibility with espie@'s makewhatis. */
412 break;
413 default:
414 goto usage;
415 }
416
417 argc -= optind;
418 argv += optind;
419
420 if (OP_CONFFILE == op && argc > 0) {
421 fprintf(stderr, "-C: Too many arguments\n");
422 goto usage;
423 }
424
425 exitcode = (int)MANDOCLEVEL_OK;
426 mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
427 mc = mchars_alloc();
428
429 ohash_init(&mpages, 6, &mpages_info);
430 ohash_init(&mlinks, 6, &mlinks_info);
431
432 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
433 /*
434 * Force processing all files.
435 */
436 use_all = 1;
437
438 /*
439 * All of these deal with a specific directory.
440 * Jump into that directory then collect files specified
441 * on the command-line.
442 */
443 if (0 == set_basedir(path_arg))
444 goto out;
445 for (i = 0; i < argc; i++)
446 filescan(argv[i]);
447 if (0 == dbopen(1))
448 goto out;
449 if (OP_TEST != op)
450 dbprune();
451 if (OP_DELETE != op)
452 mpages_merge(mc, mp);
453 dbclose(1);
454 } else {
455 /*
456 * If we have arguments, use them as our manpaths.
457 * If we don't, grok from manpath(1) or however else
458 * manpath_parse() wants to do it.
459 */
460 if (argc > 0) {
461 dirs.paths = mandoc_calloc
462 (argc, sizeof(char *));
463 dirs.sz = (size_t)argc;
464 for (i = 0; i < argc; i++)
465 dirs.paths[i] = mandoc_strdup(argv[i]);
466 } else
467 manpath_parse(&dirs, path_arg, NULL, NULL);
468
469 if (0 == dirs.sz) {
470 exitcode = (int)MANDOCLEVEL_BADARG;
471 say("", "Empty manpath");
472 }
473
474 /*
475 * First scan the tree rooted at a base directory, then
476 * build a new database and finally move it into place.
477 * Ignore zero-length directories and strip trailing
478 * slashes.
479 */
480 for (j = 0; j < dirs.sz; j++) {
481 sz = strlen(dirs.paths[j]);
482 if (sz && '/' == dirs.paths[j][sz - 1])
483 dirs.paths[j][--sz] = '\0';
484 if (0 == sz)
485 continue;
486
487 if (j) {
488 ohash_init(&mpages, 6, &mpages_info);
489 ohash_init(&mlinks, 6, &mlinks_info);
490 }
491
492 if (0 == set_basedir(dirs.paths[j]))
493 goto out;
494 if (0 == treescan())
495 goto out;
496 if (0 == set_basedir(dirs.paths[j]))
497 goto out;
498 if (0 == dbopen(0))
499 goto out;
500
501 mpages_merge(mc, mp);
502 if (warnings &&
503 ! (MPARSE_QUICK & mparse_options))
504 names_check();
505 dbclose(0);
506
507 if (j + 1 < dirs.sz) {
508 mpages_free();
509 ohash_delete(&mpages);
510 ohash_delete(&mlinks);
511 }
512 }
513 }
514 out:
515 set_basedir(NULL);
516 manpath_free(&dirs);
517 mchars_free(mc);
518 mparse_free(mp);
519 mpages_free();
520 ohash_delete(&mpages);
521 ohash_delete(&mlinks);
522 return(exitcode);
523 usage:
524 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
525 " %s [-aDnpQ] [-Tutf8] dir ...\n"
526 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
527 " %s [-Dnp] -u dir [file ...]\n"
528 " %s [-Q] -t file ...\n",
529 progname, progname, progname,
530 progname, progname);
531
532 return((int)MANDOCLEVEL_BADARG);
533 }
534
535 /*
536 * Scan a directory tree rooted at "basedir" for manpages.
537 * We use fts(), scanning directory parts along the way for clues to our
538 * section and architecture.
539 *
540 * If use_all has been specified, grok all files.
541 * If not, sanitise paths to the following:
542 *
543 * [./]man*[/<arch>]/<name>.<section>
544 * or
545 * [./]cat<section>[/<arch>]/<name>.0
546 *
547 * TODO: accomodate for multi-language directories.
548 */
549 static int
550 treescan(void)
551 {
552 FTS *f;
553 FTSENT *ff;
554 struct mlink *mlink;
555 int dform, gzip;
556 char *dsec, *arch, *fsec, *cp;
557 const char *path;
558 const char *argv[2];
559
560 argv[0] = ".";
561 argv[1] = (char *)NULL;
562
563 /*
564 * Walk through all components under the directory, using the
565 * logical descent of files.
566 */
567 f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
568 if (NULL == f) {
569 exitcode = (int)MANDOCLEVEL_SYSERR;
570 say("", "&fts_open");
571 return(0);
572 }
573
574 dsec = arch = NULL;
575 dform = FORM_NONE;
576
577 while (NULL != (ff = fts_read(f))) {
578 path = ff->fts_path + 2;
579 /*
580 * If we're a regular file, add an mlink by using the
581 * stored directory data and handling the filename.
582 */
583 if (FTS_F == ff->fts_info) {
584 if (0 == strcmp(path, MANDOC_DB))
585 continue;
586 if ( ! use_all && ff->fts_level < 2) {
587 if (warnings)
588 say(path, "Extraneous file");
589 continue;
590 }
591 gzip = 0;
592 fsec = NULL;
593 while (NULL == fsec) {
594 fsec = strrchr(ff->fts_name, '.');
595 if (NULL == fsec || strcmp(fsec+1, "gz"))
596 break;
597 gzip = 1;
598 *fsec = '\0';
599 fsec = NULL;
600 }
601 if (NULL == fsec) {
602 if ( ! use_all) {
603 if (warnings)
604 say(path,
605 "No filename suffix");
606 continue;
607 }
608 } else if (0 == strcmp(++fsec, "html")) {
609 if (warnings)
610 say(path, "Skip html");
611 continue;
612 } else if (0 == strcmp(fsec, "ps")) {
613 if (warnings)
614 say(path, "Skip ps");
615 continue;
616 } else if (0 == strcmp(fsec, "pdf")) {
617 if (warnings)
618 say(path, "Skip pdf");
619 continue;
620 } else if ( ! use_all &&
621 ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
622 (FORM_CAT == dform && strcmp(fsec, "0")))) {
623 if (warnings)
624 say(path, "Wrong filename suffix");
625 continue;
626 } else
627 fsec[-1] = '\0';
628
629 mlink = mandoc_calloc(1, sizeof(struct mlink));
630 strlcpy(mlink->file, path, sizeof(mlink->file));
631 mlink->dform = dform;
632 mlink->dsec = dsec;
633 mlink->arch = arch;
634 mlink->name = ff->fts_name;
635 mlink->fsec = fsec;
636 mlink->gzip = gzip;
637 mlink_add(mlink, ff->fts_statp);
638 continue;
639 } else if (FTS_D != ff->fts_info &&
640 FTS_DP != ff->fts_info) {
641 if (warnings)
642 say(path, "Not a regular file");
643 continue;
644 }
645
646 switch (ff->fts_level) {
647 case (0):
648 /* Ignore the root directory. */
649 break;
650 case (1):
651 /*
652 * This might contain manX/ or catX/.
653 * Try to infer this from the name.
654 * If we're not in use_all, enforce it.
655 */
656 cp = ff->fts_name;
657 if (FTS_DP == ff->fts_info)
658 break;
659
660 if (0 == strncmp(cp, "man", 3)) {
661 dform = FORM_SRC;
662 dsec = cp + 3;
663 } else if (0 == strncmp(cp, "cat", 3)) {
664 dform = FORM_CAT;
665 dsec = cp + 3;
666 } else {
667 dform = FORM_NONE;
668 dsec = NULL;
669 }
670
671 if (NULL != dsec || use_all)
672 break;
673
674 if (warnings)
675 say(path, "Unknown directory part");
676 fts_set(f, ff, FTS_SKIP);
677 break;
678 case (2):
679 /*
680 * Possibly our architecture.
681 * If we're descending, keep tabs on it.
682 */
683 if (FTS_DP != ff->fts_info && NULL != dsec)
684 arch = ff->fts_name;
685 else
686 arch = NULL;
687 break;
688 default:
689 if (FTS_DP == ff->fts_info || use_all)
690 break;
691 if (warnings)
692 say(path, "Extraneous directory part");
693 fts_set(f, ff, FTS_SKIP);
694 break;
695 }
696 }
697
698 fts_close(f);
699 return(1);
700 }
701
702 /*
703 * Add a file to the mlinks table.
704 * Do not verify that it's a "valid" looking manpage (we'll do that
705 * later).
706 *
707 * Try to infer the manual section, architecture, and page name from the
708 * path, assuming it looks like
709 *
710 * [./]man*[/<arch>]/<name>.<section>
711 * or
712 * [./]cat<section>[/<arch>]/<name>.0
713 *
714 * See treescan() for the fts(3) version of this.
715 */
716 static void
717 filescan(const char *file)
718 {
719 char buf[PATH_MAX];
720 struct stat st;
721 struct mlink *mlink;
722 char *p, *start;
723
724 assert(use_all);
725
726 if (0 == strncmp(file, "./", 2))
727 file += 2;
728
729 if (NULL == realpath(file, buf)) {
730 exitcode = (int)MANDOCLEVEL_BADARG;
731 say(file, "&realpath");
732 return;
733 }
734
735 if (strstr(buf, basedir) == buf)
736 start = buf + strlen(basedir) + 1;
737 else if (OP_TEST == op)
738 start = buf;
739 else {
740 exitcode = (int)MANDOCLEVEL_BADARG;
741 say("", "%s: outside base directory", buf);
742 return;
743 }
744
745 if (-1 == stat(buf, &st)) {
746 exitcode = (int)MANDOCLEVEL_BADARG;
747 say(file, "&stat");
748 return;
749 } else if ( ! (S_IFREG & st.st_mode)) {
750 exitcode = (int)MANDOCLEVEL_BADARG;
751 say(file, "Not a regular file");
752 return;
753 }
754
755 mlink = mandoc_calloc(1, sizeof(struct mlink));
756 strlcpy(mlink->file, start, sizeof(mlink->file));
757
758 /*
759 * First try to guess our directory structure.
760 * If we find a separator, try to look for man* or cat*.
761 * If we find one of these and what's underneath is a directory,
762 * assume it's an architecture.
763 */
764 if (NULL != (p = strchr(start, '/'))) {
765 *p++ = '\0';
766 if (0 == strncmp(start, "man", 3)) {
767 mlink->dform = FORM_SRC;
768 mlink->dsec = start + 3;
769 } else if (0 == strncmp(start, "cat", 3)) {
770 mlink->dform = FORM_CAT;
771 mlink->dsec = start + 3;
772 }
773
774 start = p;
775 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
776 *p++ = '\0';
777 mlink->arch = start;
778 start = p;
779 }
780 }
781
782 /*
783 * Now check the file suffix.
784 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
785 */
786 p = strrchr(start, '\0');
787 while (p-- > start && '/' != *p && '.' != *p)
788 /* Loop. */ ;
789
790 if ('.' == *p) {
791 *p++ = '\0';
792 mlink->fsec = p;
793 }
794
795 /*
796 * Now try to parse the name.
797 * Use the filename portion of the path.
798 */
799 mlink->name = start;
800 if (NULL != (p = strrchr(start, '/'))) {
801 mlink->name = p + 1;
802 *p = '\0';
803 }
804 mlink_add(mlink, &st);
805 }
806
807 static void
808 mlink_add(struct mlink *mlink, const struct stat *st)
809 {
810 struct inodev inodev;
811 struct mpage *mpage;
812 unsigned int slot;
813
814 assert(NULL != mlink->file);
815
816 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
817 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
818 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
819 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
820
821 if ('0' == *mlink->fsec) {
822 free(mlink->fsec);
823 mlink->fsec = mandoc_strdup(mlink->dsec);
824 mlink->fform = FORM_CAT;
825 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
826 mlink->fform = FORM_SRC;
827 else
828 mlink->fform = FORM_NONE;
829
830 slot = ohash_qlookup(&mlinks, mlink->file);
831 assert(NULL == ohash_find(&mlinks, slot));
832 ohash_insert(&mlinks, slot, mlink);
833
834 inodev.st_ino = st->st_ino;
835 inodev.st_dev = st->st_dev;
836 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
837 sizeof(struct inodev), inodev.st_ino);
838 mpage = ohash_find(&mpages, slot);
839 if (NULL == mpage) {
840 mpage = mandoc_calloc(1, sizeof(struct mpage));
841 mpage->inodev.st_ino = inodev.st_ino;
842 mpage->inodev.st_dev = inodev.st_dev;
843 ohash_insert(&mpages, slot, mpage);
844 } else
845 mlink->next = mpage->mlinks;
846 mpage->mlinks = mlink;
847 mlink->mpage = mpage;
848 }
849
850 static void
851 mlink_free(struct mlink *mlink)
852 {
853
854 free(mlink->dsec);
855 free(mlink->arch);
856 free(mlink->name);
857 free(mlink->fsec);
858 free(mlink);
859 }
860
861 static void
862 mpages_free(void)
863 {
864 struct mpage *mpage;
865 struct mlink *mlink;
866 unsigned int slot;
867
868 mpage = ohash_first(&mpages, &slot);
869 while (NULL != mpage) {
870 while (NULL != (mlink = mpage->mlinks)) {
871 mpage->mlinks = mlink->next;
872 mlink_free(mlink);
873 }
874 free(mpage->sec);
875 free(mpage->arch);
876 free(mpage->title);
877 free(mpage->desc);
878 free(mpage);
879 mpage = ohash_next(&mpages, &slot);
880 }
881 }
882
883 /*
884 * For each mlink to the mpage, check whether the path looks like
885 * it is formatted, and if it does, check whether a source manual
886 * exists by the same name, ignoring the suffix.
887 * If both conditions hold, drop the mlink.
888 */
889 static void
890 mlinks_undupe(struct mpage *mpage)
891 {
892 char buf[PATH_MAX];
893 struct mlink **prev;
894 struct mlink *mlink;
895 char *bufp;
896
897 mpage->form = FORM_CAT;
898 prev = &mpage->mlinks;
899 while (NULL != (mlink = *prev)) {
900 if (FORM_CAT != mlink->dform) {
901 mpage->form = FORM_NONE;
902 goto nextlink;
903 }
904 if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
905 if (warnings)
906 say(mlink->file, "Filename too long");
907 goto nextlink;
908 }
909 bufp = strstr(buf, "cat");
910 assert(NULL != bufp);
911 memcpy(bufp, "man", 3);
912 if (NULL != (bufp = strrchr(buf, '.')))
913 *++bufp = '\0';
914 strlcat(buf, mlink->dsec, PATH_MAX);
915 if (NULL == ohash_find(&mlinks,
916 ohash_qlookup(&mlinks, buf)))
917 goto nextlink;
918 if (warnings)
919 say(mlink->file, "Man source exists: %s", buf);
920 if (use_all)
921 goto nextlink;
922 *prev = mlink->next;
923 mlink_free(mlink);
924 continue;
925 nextlink:
926 prev = &(*prev)->next;
927 }
928 }
929
930 static void
931 mlink_check(struct mpage *mpage, struct mlink *mlink)
932 {
933 struct str *str;
934 unsigned int slot;
935
936 /*
937 * Check whether the manual section given in a file
938 * agrees with the directory where the file is located.
939 * Some manuals have suffixes like (3p) on their
940 * section number either inside the file or in the
941 * directory name, some are linked into more than one
942 * section, like encrypt(1) = makekey(8).
943 */
944
945 if (FORM_SRC == mpage->form &&
946 strcasecmp(mpage->sec, mlink->dsec))
947 say(mlink->file, "Section \"%s\" manual in %s directory",
948 mpage->sec, mlink->dsec);
949
950 /*
951 * Manual page directories exist for each kernel
952 * architecture as returned by machine(1).
953 * However, many manuals only depend on the
954 * application architecture as returned by arch(1).
955 * For example, some (2/ARM) manuals are shared
956 * across the "armish" and "zaurus" kernel
957 * architectures.
958 * A few manuals are even shared across completely
959 * different architectures, for example fdformat(1)
960 * on amd64, i386, sparc, and sparc64.
961 */
962
963 if (strcasecmp(mpage->arch, mlink->arch))
964 say(mlink->file, "Architecture \"%s\" manual in "
965 "\"%s\" directory", mpage->arch, mlink->arch);
966
967 /*
968 * XXX
969 * parse_cat() doesn't set NAME_TITLE yet.
970 */
971
972 if (FORM_CAT == mpage->form)
973 return;
974
975 /*
976 * Check whether this mlink
977 * appears as a name in the NAME section.
978 */
979
980 slot = ohash_qlookup(&names, mlink->name);
981 str = ohash_find(&names, slot);
982 assert(NULL != str);
983 if ( ! (NAME_TITLE & str->mask))
984 say(mlink->file, "Name missing in NAME section");
985 }
986
987 /*
988 * Run through the files in the global vector "mpages"
989 * and add them to the database specified in "basedir".
990 *
991 * This handles the parsing scheme itself, using the cues of directory
992 * and filename to determine whether the file is parsable or not.
993 */
994 static void
995 mpages_merge(struct mchars *mc, struct mparse *mp)
996 {
997 char any[] = "any";
998 struct ohash_info str_info;
999 int fd[2];
1000 struct mpage *mpage, *mpage_dest;
1001 struct mlink *mlink, *mlink_dest;
1002 struct mdoc *mdoc;
1003 struct man *man;
1004 char *sodest;
1005 char *cp;
1006 pid_t child_pid;
1007 int status;
1008 unsigned int pslot;
1009 enum mandoclevel lvl;
1010
1011 str_info.alloc = hash_alloc;
1012 str_info.halloc = hash_halloc;
1013 str_info.hfree = hash_free;
1014 str_info.key_offset = offsetof(struct str, key);
1015
1016 if (0 == nodb)
1017 SQL_EXEC("BEGIN TRANSACTION");
1018
1019 mpage = ohash_first(&mpages, &pslot);
1020 while (NULL != mpage) {
1021 mlinks_undupe(mpage);
1022 if (NULL == mpage->mlinks) {
1023 mpage = ohash_next(&mpages, &pslot);
1024 continue;
1025 }
1026
1027 name_mask = NAME_MASK;
1028 ohash_init(&names, 4, &str_info);
1029 ohash_init(&strings, 6, &str_info);
1030 mparse_reset(mp);
1031 mdoc = NULL;
1032 man = NULL;
1033 sodest = NULL;
1034 child_pid = 0;
1035 fd[0] = -1;
1036 fd[1] = -1;
1037
1038 if (mpage->mlinks->gzip) {
1039 if (-1 == pipe(fd)) {
1040 exitcode = (int)MANDOCLEVEL_SYSERR;
1041 say(mpage->mlinks->file, "&pipe gunzip");
1042 goto nextpage;
1043 }
1044 switch (child_pid = fork()) {
1045 case (-1):
1046 exitcode = (int)MANDOCLEVEL_SYSERR;
1047 say(mpage->mlinks->file, "&fork gunzip");
1048 child_pid = 0;
1049 close(fd[1]);
1050 close(fd[0]);
1051 goto nextpage;
1052 case (0):
1053 close(fd[0]);
1054 if (-1 == dup2(fd[1], STDOUT_FILENO)) {
1055 say(mpage->mlinks->file,
1056 "&dup gunzip");
1057 exit(1);
1058 }
1059 execlp("gunzip", "gunzip", "-c",
1060 mpage->mlinks->file, NULL);
1061 say(mpage->mlinks->file, "&exec gunzip");
1062 exit(1);
1063 default:
1064 close(fd[1]);
1065 break;
1066 }
1067 }
1068
1069 /*
1070 * Try interpreting the file as mdoc(7) or man(7)
1071 * source code, unless it is already known to be
1072 * formatted. Fall back to formatted mode.
1073 */
1074 if (FORM_CAT != mpage->mlinks->dform ||
1075 FORM_CAT != mpage->mlinks->fform) {
1076 lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
1077 if (lvl < MANDOCLEVEL_FATAL)
1078 mparse_result(mp, &mdoc, &man, &sodest);
1079 }
1080
1081 if (NULL != sodest) {
1082 mlink_dest = ohash_find(&mlinks,
1083 ohash_qlookup(&mlinks, sodest));
1084 if (NULL != mlink_dest) {
1085
1086 /* The .so target exists. */
1087
1088 mpage_dest = mlink_dest->mpage;
1089 mlink = mpage->mlinks;
1090 while (1) {
1091 mlink->mpage = mpage_dest;
1092
1093 /*
1094 * If the target was already
1095 * processed, add the links
1096 * to the database now.
1097 * Otherwise, this will
1098 * happen when we come
1099 * to the target.
1100 */
1101
1102 if (mpage_dest->recno)
1103 dbadd_mlink(mlink);
1104
1105 if (NULL == mlink->next)
1106 break;
1107 mlink = mlink->next;
1108 }
1109
1110 /* Move all links to the target. */
1111
1112 mlink->next = mlink_dest->next;
1113 mlink_dest->next = mpage->mlinks;
1114 mpage->mlinks = NULL;
1115 }
1116 goto nextpage;
1117 } else if (NULL != mdoc) {
1118 mpage->form = FORM_SRC;
1119 mpage->sec =
1120 mandoc_strdup(mdoc_meta(mdoc)->msec);
1121 mpage->arch = mdoc_meta(mdoc)->arch;
1122 mpage->arch = mandoc_strdup(
1123 NULL == mpage->arch ? "" : mpage->arch);
1124 mpage->title =
1125 mandoc_strdup(mdoc_meta(mdoc)->title);
1126 } else if (NULL != man) {
1127 mpage->form = FORM_SRC;
1128 mpage->sec =
1129 mandoc_strdup(man_meta(man)->msec);
1130 mpage->arch =
1131 mandoc_strdup(mpage->mlinks->arch);
1132 mpage->title =
1133 mandoc_strdup(man_meta(man)->title);
1134 } else {
1135 mpage->form = FORM_CAT;
1136 mpage->sec =
1137 mandoc_strdup(mpage->mlinks->dsec);
1138 mpage->arch =
1139 mandoc_strdup(mpage->mlinks->arch);
1140 mpage->title =
1141 mandoc_strdup(mpage->mlinks->name);
1142 }
1143 putkey(mpage, mpage->sec, TYPE_sec);
1144 putkey(mpage, '\0' == *mpage->arch ?
1145 any : mpage->arch, TYPE_arch);
1146
1147 for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1148 if ('\0' != *mlink->dsec)
1149 putkey(mpage, mlink->dsec, TYPE_sec);
1150 if ('\0' != *mlink->fsec)
1151 putkey(mpage, mlink->fsec, TYPE_sec);
1152 putkey(mpage, '\0' == *mlink->arch ?
1153 any : mlink->arch, TYPE_arch);
1154 putkey(mpage, mlink->name, NAME_FILE);
1155 }
1156
1157 if (NULL != mdoc) {
1158 if (NULL != (cp = mdoc_meta(mdoc)->name))
1159 putkey(mpage, cp, NAME_HEAD);
1160 assert(NULL == mpage->desc);
1161 parse_mdoc(mpage, mdoc_node(mdoc));
1162 if (NULL == mpage->desc)
1163 mpage->desc = mandoc_strdup(
1164 mpage->mlinks->name);
1165 } else if (NULL != man)
1166 parse_man(mpage, man_node(man));
1167 else
1168 parse_cat(mpage, fd[0]);
1169
1170 if (warnings && !use_all)
1171 for (mlink = mpage->mlinks; mlink;
1172 mlink = mlink->next)
1173 mlink_check(mpage, mlink);
1174
1175 dbadd(mpage, mc);
1176
1177 nextpage:
1178 if (child_pid) {
1179 if (-1 == waitpid(child_pid, &status, 0)) {
1180 exitcode = (int)MANDOCLEVEL_SYSERR;
1181 say(mpage->mlinks->file, "&wait gunzip");
1182 } else if (WIFSIGNALED(status)) {
1183 exitcode = (int)MANDOCLEVEL_SYSERR;
1184 say(mpage->mlinks->file,
1185 "gunzip died from signal %d",
1186 WTERMSIG(status));
1187 } else if (WEXITSTATUS(status)) {
1188 exitcode = (int)MANDOCLEVEL_SYSERR;
1189 say(mpage->mlinks->file,
1190 "gunzip failed with code %d",
1191 WEXITSTATUS(status));
1192 }
1193 }
1194 ohash_delete(&strings);
1195 ohash_delete(&names);
1196 mpage = ohash_next(&mpages, &pslot);
1197 }
1198
1199 if (0 == nodb)
1200 SQL_EXEC("END TRANSACTION");
1201 }
1202
1203 static void
1204 names_check(void)
1205 {
1206 sqlite3_stmt *stmt;
1207 const char *name, *sec, *arch, *key;
1208 size_t i;
1209 int irc;
1210
1211 sqlite3_prepare_v2(db,
1212 "SELECT name, sec, arch, key FROM ("
1213 "SELECT name AS key, pageid FROM names "
1214 "WHERE bits & ? AND NOT EXISTS ("
1215 "SELECT pageid FROM mlinks "
1216 "WHERE mlinks.pageid == names.pageid "
1217 "AND mlinks.name == names.name"
1218 ")"
1219 ") JOIN ("
1220 "SELECT * FROM mlinks GROUP BY pageid"
1221 ") USING (pageid);",
1222 -1, &stmt, NULL);
1223
1224 i = 1;
1225 SQL_BIND_INT64(stmt, i, NAME_TITLE);
1226
1227 while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1228 name = sqlite3_column_text(stmt, 0);
1229 sec = sqlite3_column_text(stmt, 1);
1230 arch = sqlite3_column_text(stmt, 2);
1231 key = sqlite3_column_text(stmt, 3);
1232 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1233 '\0' == *arch ? "" : "/",
1234 '\0' == *arch ? "" : arch, key);
1235 }
1236 sqlite3_finalize(stmt);
1237 }
1238
1239 static void
1240 parse_cat(struct mpage *mpage, int fd)
1241 {
1242 FILE *stream;
1243 char *line, *p, *title;
1244 size_t len, plen, titlesz;
1245
1246 stream = (-1 == fd) ?
1247 fopen(mpage->mlinks->file, "r") :
1248 fdopen(fd, "r");
1249 if (NULL == stream) {
1250 if (warnings)
1251 say(mpage->mlinks->file, "&fopen");
1252 return;
1253 }
1254
1255 /* Skip to first blank line. */
1256
1257 while (NULL != (line = fgetln(stream, &len)))
1258 if ('\n' == *line)
1259 break;
1260
1261 /*
1262 * Assume the first line that is not indented
1263 * is the first section header. Skip to it.
1264 */
1265
1266 while (NULL != (line = fgetln(stream, &len)))
1267 if ('\n' != *line && ' ' != *line)
1268 break;
1269
1270 /*
1271 * Read up until the next section into a buffer.
1272 * Strip the leading and trailing newline from each read line,
1273 * appending a trailing space.
1274 * Ignore empty (whitespace-only) lines.
1275 */
1276
1277 titlesz = 0;
1278 title = NULL;
1279
1280 while (NULL != (line = fgetln(stream, &len))) {
1281 if (' ' != *line || '\n' != line[len - 1])
1282 break;
1283 while (len > 0 && isspace((unsigned char)*line)) {
1284 line++;
1285 len--;
1286 }
1287 if (1 == len)
1288 continue;
1289 title = mandoc_realloc(title, titlesz + len);
1290 memcpy(title + titlesz, line, len);
1291 titlesz += len;
1292 title[titlesz - 1] = ' ';
1293 }
1294
1295 /*
1296 * If no page content can be found, or the input line
1297 * is already the next section header, or there is no
1298 * trailing newline, reuse the page title as the page
1299 * description.
1300 */
1301
1302 if (NULL == title || '\0' == *title) {
1303 if (warnings)
1304 say(mpage->mlinks->file,
1305 "Cannot find NAME section");
1306 assert(NULL == mpage->desc);
1307 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1308 fclose(stream);
1309 free(title);
1310 return;
1311 }
1312
1313 title = mandoc_realloc(title, titlesz + 1);
1314 title[titlesz] = '\0';
1315
1316 /*
1317 * Skip to the first dash.
1318 * Use the remaining line as the description (no more than 70
1319 * bytes).
1320 */
1321
1322 if (NULL != (p = strstr(title, "- "))) {
1323 for (p += 2; ' ' == *p || '\b' == *p; p++)
1324 /* Skip to next word. */ ;
1325 } else {
1326 if (warnings)
1327 say(mpage->mlinks->file,
1328 "No dash in title line");
1329 p = title;
1330 }
1331
1332 plen = strlen(p);
1333
1334 /* Strip backspace-encoding from line. */
1335
1336 while (NULL != (line = memchr(p, '\b', plen))) {
1337 len = line - p;
1338 if (0 == len) {
1339 memmove(line, line + 1, plen--);
1340 continue;
1341 }
1342 memmove(line - 1, line + 1, plen - len);
1343 plen -= 2;
1344 }
1345
1346 assert(NULL == mpage->desc);
1347 mpage->desc = mandoc_strdup(p);
1348 fclose(stream);
1349 free(title);
1350 }
1351
1352 /*
1353 * Put a type/word pair into the word database for this particular file.
1354 */
1355 static void
1356 putkey(const struct mpage *mpage, char *value, uint64_t type)
1357 {
1358 char *cp;
1359
1360 assert(NULL != value);
1361 if (TYPE_arch == type)
1362 for (cp = value; *cp; cp++)
1363 if (isupper((unsigned char)*cp))
1364 *cp = _tolower((unsigned char)*cp);
1365 putkeys(mpage, value, strlen(value), type);
1366 }
1367
1368 /*
1369 * Grok all nodes at or below a certain mdoc node into putkey().
1370 */
1371 static void
1372 putmdockey(const struct mpage *mpage,
1373 const struct mdoc_node *n, uint64_t m)
1374 {
1375
1376 for ( ; NULL != n; n = n->next) {
1377 if (NULL != n->child)
1378 putmdockey(mpage, n->child, m);
1379 if (MDOC_TEXT == n->type)
1380 putkey(mpage, n->string, m);
1381 }
1382 }
1383
1384 static void
1385 parse_man(struct mpage *mpage, const struct man_node *n)
1386 {
1387 const struct man_node *head, *body;
1388 char *start, *title;
1389 char byte;
1390 size_t sz;
1391
1392 if (NULL == n)
1393 return;
1394
1395 /*
1396 * We're only searching for one thing: the first text child in
1397 * the BODY of a NAME section. Since we don't keep track of
1398 * sections in -man, run some hoops to find out whether we're in
1399 * the correct section or not.
1400 */
1401
1402 if (MAN_BODY == n->type && MAN_SH == n->tok) {
1403 body = n;
1404 assert(body->parent);
1405 if (NULL != (head = body->parent->head) &&
1406 1 == head->nchild &&
1407 NULL != (head = (head->child)) &&
1408 MAN_TEXT == head->type &&
1409 0 == strcmp(head->string, "NAME") &&
1410 NULL != body->child) {
1411
1412 /*
1413 * Suck the entire NAME section into memory.
1414 * Yes, we might run away.
1415 * But too many manuals have big, spread-out
1416 * NAME sections over many lines.
1417 */
1418
1419 title = NULL;
1420 man_deroff(&title, body);
1421 if (NULL == title)
1422 return;
1423
1424 /*
1425 * Go through a special heuristic dance here.
1426 * Conventionally, one or more manual names are
1427 * comma-specified prior to a whitespace, then a
1428 * dash, then a description. Try to puzzle out
1429 * the name parts here.
1430 */
1431
1432 start = title;
1433 for ( ;; ) {
1434 sz = strcspn(start, " ,");
1435 if ('\0' == start[sz])
1436 break;
1437
1438 byte = start[sz];
1439 start[sz] = '\0';
1440
1441 /*
1442 * Assume a stray trailing comma in the
1443 * name list if a name begins with a dash.
1444 */
1445
1446 if ('-' == start[0] ||
1447 ('\\' == start[0] && '-' == start[1]))
1448 break;
1449
1450 putkey(mpage, start, NAME_TITLE);
1451
1452 if (' ' == byte) {
1453 start += sz + 1;
1454 break;
1455 }
1456
1457 assert(',' == byte);
1458 start += sz + 1;
1459 while (' ' == *start)
1460 start++;
1461 }
1462
1463 if (start == title) {
1464 putkey(mpage, start, NAME_TITLE);
1465 free(title);
1466 return;
1467 }
1468
1469 while (isspace((unsigned char)*start))
1470 start++;
1471
1472 if (0 == strncmp(start, "-", 1))
1473 start += 1;
1474 else if (0 == strncmp(start, "\\-\\-", 4))
1475 start += 4;
1476 else if (0 == strncmp(start, "\\-", 2))
1477 start += 2;
1478 else if (0 == strncmp(start, "\\(en", 4))
1479 start += 4;
1480 else if (0 == strncmp(start, "\\(em", 4))
1481 start += 4;
1482
1483 while (' ' == *start)
1484 start++;
1485
1486 assert(NULL == mpage->desc);
1487 mpage->desc = mandoc_strdup(start);
1488 free(title);
1489 return;
1490 }
1491 }
1492
1493 for (n = n->child; n; n = n->next) {
1494 if (NULL != mpage->desc)
1495 break;
1496 parse_man(mpage, n);
1497 }
1498 }
1499
1500 static void
1501 parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
1502 {
1503
1504 assert(NULL != n);
1505 for (n = n->child; NULL != n; n = n->next) {
1506 switch (n->type) {
1507 case (MDOC_ELEM):
1508 /* FALLTHROUGH */
1509 case (MDOC_BLOCK):
1510 /* FALLTHROUGH */
1511 case (MDOC_HEAD):
1512 /* FALLTHROUGH */
1513 case (MDOC_BODY):
1514 /* FALLTHROUGH */
1515 case (MDOC_TAIL):
1516 if (NULL != mdocs[n->tok].fp)
1517 if (0 == (*mdocs[n->tok].fp)(mpage, n))
1518 break;
1519 if (mdocs[n->tok].mask)
1520 putmdockey(mpage, n->child,
1521 mdocs[n->tok].mask);
1522 break;
1523 default:
1524 assert(MDOC_ROOT != n->type);
1525 continue;
1526 }
1527 if (NULL != n->child)
1528 parse_mdoc(mpage, n);
1529 }
1530 }
1531
1532 static int
1533 parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
1534 {
1535 const char *start, *end;
1536 size_t sz;
1537
1538 if (SEC_SYNOPSIS != n->sec ||
1539 NULL == (n = n->child) ||
1540 MDOC_TEXT != n->type)
1541 return(0);
1542
1543 /*
1544 * Only consider those `Fd' macro fields that begin with an
1545 * "inclusion" token (versus, e.g., #define).
1546 */
1547
1548 if (strcmp("#include", n->string))
1549 return(0);
1550
1551 if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1552 return(0);
1553
1554 /*
1555 * Strip away the enclosing angle brackets and make sure we're
1556 * not zero-length.
1557 */
1558
1559 start = n->string;
1560 if ('<' == *start || '"' == *start)
1561 start++;
1562
1563 if (0 == (sz = strlen(start)))
1564 return(0);
1565
1566 end = &start[(int)sz - 1];
1567 if ('>' == *end || '"' == *end)
1568 end--;
1569
1570 if (end > start)
1571 putkeys(mpage, start, end - start + 1, TYPE_In);
1572 return(0);
1573 }
1574
1575 static int
1576 parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
1577 {
1578 char *cp;
1579
1580 if (NULL == (n = n->child) || MDOC_TEXT != n->type)
1581 return(0);
1582
1583 /*
1584 * Parse: .Fn "struct type *name" "char *arg".
1585 * First strip away pointer symbol.
1586 * Then store the function name, then type.
1587 * Finally, store the arguments.
1588 */
1589
1590 if (NULL == (cp = strrchr(n->string, ' ')))
1591 cp = n->string;
1592
1593 while ('*' == *cp)
1594 cp++;
1595
1596 putkey(mpage, cp, TYPE_Fn);
1597
1598 if (n->string < cp)
1599 putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
1600
1601 for (n = n->next; NULL != n; n = n->next)
1602 if (MDOC_TEXT == n->type)
1603 putkey(mpage, n->string, TYPE_Fa);
1604
1605 return(0);
1606 }
1607
1608 static int
1609 parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
1610 {
1611 char *cp;
1612
1613 if (NULL == (n = n->child))
1614 return(0);
1615
1616 if (NULL == n->next) {
1617 putkey(mpage, n->string, TYPE_Xr);
1618 return(0);
1619 }
1620
1621 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1622 putkey(mpage, cp, TYPE_Xr);
1623 free(cp);
1624 return(0);
1625 }
1626
1627 static int
1628 parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
1629 {
1630
1631 if (MDOC_BODY == n->type)
1632 mdoc_deroff(&mpage->desc, n);
1633 return(0);
1634 }
1635
1636 static int
1637 parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
1638 {
1639
1640 if (SEC_NAME == n->sec)
1641 putmdockey(mpage, n->child, NAME_TITLE);
1642 else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
1643 putmdockey(mpage, n->child, NAME_SYN);
1644 return(0);
1645 }
1646
1647 static int
1648 parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
1649 {
1650
1651 return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1652 }
1653
1654 static int
1655 parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
1656 {
1657
1658 return(MDOC_HEAD == n->type);
1659 }
1660
1661 static int
1662 parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
1663 {
1664
1665 return(MDOC_BODY == n->type);
1666 }
1667
1668 /*
1669 * Add a string to the hash table for the current manual.
1670 * Each string has a bitmask telling which macros it belongs to.
1671 * When we finish the manual, we'll dump the table.
1672 */
1673 static void
1674 putkeys(const struct mpage *mpage,
1675 const char *cp, size_t sz, uint64_t v)
1676 {
1677 struct ohash *htab;
1678 struct str *s;
1679 const char *end;
1680 unsigned int slot;
1681 int i;
1682
1683 if (0 == sz)
1684 return;
1685
1686 if (TYPE_Nm & v) {
1687 htab = &names;
1688 v &= name_mask;
1689 name_mask &= ~NAME_FIRST;
1690 if (debug > 1)
1691 say(mpage->mlinks->file,
1692 "Adding name %*s", sz, cp);
1693 } else {
1694 htab = &strings;
1695 if (debug > 1)
1696 for (i = 0; i < mansearch_keymax; i++)
1697 if (1 << i & v)
1698 say(mpage->mlinks->file,
1699 "Adding key %s=%*s",
1700 mansearch_keynames[i], sz, cp);
1701 }
1702
1703 end = cp + sz;
1704 slot = ohash_qlookupi(htab, cp, &end);
1705 s = ohash_find(htab, slot);
1706
1707 if (NULL != s && mpage == s->mpage) {
1708 s->mask |= v;
1709 return;
1710 } else if (NULL == s) {
1711 s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
1712 memcpy(s->key, cp, sz);
1713 ohash_insert(htab, slot, s);
1714 }
1715 s->mpage = mpage;
1716 s->mask = v;
1717 }
1718
1719 /*
1720 * Take a Unicode codepoint and produce its UTF-8 encoding.
1721 * This isn't the best way to do this, but it works.
1722 * The magic numbers are from the UTF-8 packaging.
1723 * They're not as scary as they seem: read the UTF-8 spec for details.
1724 */
1725 static size_t
1726 utf8(unsigned int cp, char out[7])
1727 {
1728 size_t rc;
1729
1730 rc = 0;
1731 if (cp <= 0x0000007F) {
1732 rc = 1;
1733 out[0] = (char)cp;
1734 } else if (cp <= 0x000007FF) {
1735 rc = 2;
1736 out[0] = (cp >> 6 & 31) | 192;
1737 out[1] = (cp & 63) | 128;
1738 } else if (cp <= 0x0000FFFF) {
1739 rc = 3;
1740 out[0] = (cp >> 12 & 15) | 224;
1741 out[1] = (cp >> 6 & 63) | 128;
1742 out[2] = (cp & 63) | 128;
1743 } else if (cp <= 0x001FFFFF) {
1744 rc = 4;
1745 out[0] = (cp >> 18 & 7) | 240;
1746 out[1] = (cp >> 12 & 63) | 128;
1747 out[2] = (cp >> 6 & 63) | 128;
1748 out[3] = (cp & 63) | 128;
1749 } else if (cp <= 0x03FFFFFF) {
1750 rc = 5;
1751 out[0] = (cp >> 24 & 3) | 248;
1752 out[1] = (cp >> 18 & 63) | 128;
1753 out[2] = (cp >> 12 & 63) | 128;
1754 out[3] = (cp >> 6 & 63) | 128;
1755 out[4] = (cp & 63) | 128;
1756 } else if (cp <= 0x7FFFFFFF) {
1757 rc = 6;
1758 out[0] = (cp >> 30 & 1) | 252;
1759 out[1] = (cp >> 24 & 63) | 128;
1760 out[2] = (cp >> 18 & 63) | 128;
1761 out[3] = (cp >> 12 & 63) | 128;
1762 out[4] = (cp >> 6 & 63) | 128;
1763 out[5] = (cp & 63) | 128;
1764 } else
1765 return(0);
1766
1767 out[rc] = '\0';
1768 return(rc);
1769 }
1770
1771 /*
1772 * Store the rendered version of a key, or alias the pointer
1773 * if the key contains no escape sequences.
1774 */
1775 static void
1776 render_key(struct mchars *mc, struct str *key)
1777 {
1778 size_t sz, bsz, pos;
1779 char utfbuf[7], res[6];
1780 char *buf;
1781 const char *seq, *cpp, *val;
1782 int len, u;
1783 enum mandoc_esc esc;
1784
1785 assert(NULL == key->rendered);
1786
1787 res[0] = '\\';
1788 res[1] = '\t';
1789 res[2] = ASCII_NBRSP;
1790 res[3] = ASCII_HYPH;
1791 res[4] = ASCII_BREAK;
1792 res[5] = '\0';
1793
1794 val = key->key;
1795 bsz = strlen(val);
1796
1797 /*
1798 * Pre-check: if we have no stop-characters, then set the
1799 * pointer as ourselvse and get out of here.
1800 */
1801 if (strcspn(val, res) == bsz) {
1802 key->rendered = key->key;
1803 return;
1804 }
1805
1806 /* Pre-allocate by the length of the input */
1807
1808 buf = mandoc_malloc(++bsz);
1809 pos = 0;
1810
1811 while ('\0' != *val) {
1812 /*
1813 * Halt on the first escape sequence.
1814 * This also halts on the end of string, in which case
1815 * we just copy, fallthrough, and exit the loop.
1816 */
1817 if ((sz = strcspn(val, res)) > 0) {
1818 memcpy(&buf[pos], val, sz);
1819 pos += sz;
1820 val += sz;
1821 }
1822
1823 switch (*val) {
1824 case (ASCII_HYPH):
1825 buf[pos++] = '-';
1826 val++;
1827 continue;
1828 case ('\t'):
1829 /* FALLTHROUGH */
1830 case (ASCII_NBRSP):
1831 buf[pos++] = ' ';
1832 val++;
1833 /* FALLTHROUGH */
1834 case (ASCII_BREAK):
1835 continue;
1836 default:
1837 break;
1838 }
1839 if ('\\' != *val)
1840 break;
1841
1842 /* Read past the slash. */
1843
1844 val++;
1845
1846 /*
1847 * Parse the escape sequence and see if it's a
1848 * predefined character or special character.
1849 */
1850
1851 esc = mandoc_escape
1852 ((const char **)&val, &seq, &len);
1853 if (ESCAPE_ERROR == esc)
1854 break;
1855 if (ESCAPE_SPECIAL != esc)
1856 continue;
1857
1858 /*
1859 * Render the special character
1860 * as either UTF-8 or ASCII.
1861 */
1862
1863 if (write_utf8) {
1864 if (0 == (u = mchars_spec2cp(mc, seq, len)))
1865 continue;
1866 cpp = utfbuf;
1867 if (0 == (sz = utf8(u, utfbuf)))
1868 continue;
1869 sz = strlen(cpp);
1870 } else {
1871 cpp = mchars_spec2str(mc, seq, len, &sz);
1872 if (NULL == cpp)
1873 continue;
1874 if (ASCII_NBRSP == *cpp) {
1875 cpp = " ";
1876 sz = 1;
1877 }
1878 }
1879
1880 /* Copy the rendered glyph into the stream. */
1881
1882 bsz += sz;
1883 buf = mandoc_realloc(buf, bsz);
1884 memcpy(&buf[pos], cpp, sz);
1885 pos += sz;
1886 }
1887
1888 buf[pos] = '\0';
1889 key->rendered = buf;
1890 }
1891
1892 static void
1893 dbadd_mlink(const struct mlink *mlink)
1894 {
1895 size_t i;
1896
1897 i = 1;
1898 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
1899 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
1900 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
1901 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno);
1902 SQL_STEP(stmts[STMT_INSERT_LINK]);
1903 sqlite3_reset(stmts[STMT_INSERT_LINK]);
1904 }
1905
1906 /*
1907 * Flush the current page's terms (and their bits) into the database.
1908 * Wrap the entire set of additions in a transaction to make sqlite be a
1909 * little faster.
1910 * Also, handle escape sequences at the last possible moment.
1911 */
1912 static void
1913 dbadd(struct mpage *mpage, struct mchars *mc)
1914 {
1915 struct mlink *mlink;
1916 struct str *key;
1917 size_t i;
1918 unsigned int slot;
1919
1920 mlink = mpage->mlinks;
1921
1922 if (nodb) {
1923 while (NULL != mlink) {
1924 fputs(mlink->name, stdout);
1925 if (NULL == mlink->next ||
1926 strcmp(mlink->dsec, mlink->next->dsec) ||
1927 strcmp(mlink->fsec, mlink->next->fsec) ||
1928 strcmp(mlink->arch, mlink->next->arch)) {
1929 putchar('(');
1930 if ('\0' == *mlink->dsec)
1931 fputs(mlink->fsec, stdout);
1932 else
1933 fputs(mlink->dsec, stdout);
1934 if ('\0' != *mlink->arch)
1935 printf("/%s", mlink->arch);
1936 putchar(')');
1937 }
1938 mlink = mlink->next;
1939 if (NULL != mlink)
1940 fputs(", ", stdout);
1941 }
1942 printf(" - %s\n", mpage->desc);
1943 return;
1944 }
1945
1946 if (debug)
1947 say(mlink->file, "Adding to database");
1948
1949 i = 1;
1950 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, mpage->desc);
1951 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
1952 SQL_STEP(stmts[STMT_INSERT_PAGE]);
1953 mpage->recno = sqlite3_last_insert_rowid(db);
1954 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
1955
1956 while (NULL != mlink) {
1957 dbadd_mlink(mlink);
1958 mlink = mlink->next;
1959 }
1960
1961 for (key = ohash_first(&names, &slot); NULL != key;
1962 key = ohash_next(&names, &slot)) {
1963 assert(key->mpage == mpage);
1964 if (NULL == key->rendered)
1965 render_key(mc, key);
1966 i = 1;
1967 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
1968 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->rendered);
1969 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->recno);
1970 SQL_STEP(stmts[STMT_INSERT_NAME]);
1971 sqlite3_reset(stmts[STMT_INSERT_NAME]);
1972 if (key->rendered != key->key)
1973 free(key->rendered);
1974 free(key);
1975 }
1976 for (key = ohash_first(&strings, &slot); NULL != key;
1977 key = ohash_next(&strings, &slot)) {
1978 assert(key->mpage == mpage);
1979 if (NULL == key->rendered)
1980 render_key(mc, key);
1981 i = 1;
1982 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
1983 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
1984 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno);
1985 SQL_STEP(stmts[STMT_INSERT_KEY]);
1986 sqlite3_reset(stmts[STMT_INSERT_KEY]);
1987 if (key->rendered != key->key)
1988 free(key->rendered);
1989 free(key);
1990 }
1991 }
1992
1993 static void
1994 dbprune(void)
1995 {
1996 struct mpage *mpage;
1997 struct mlink *mlink;
1998 size_t i;
1999 unsigned int slot;
2000
2001 if (0 == nodb)
2002 SQL_EXEC("BEGIN TRANSACTION");
2003
2004 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2005 mpage = ohash_next(&mpages, &slot)) {
2006 mlink = mpage->mlinks;
2007 if (debug)
2008 say(mlink->file, "Deleting from database");
2009 if (nodb)
2010 continue;
2011 for ( ; NULL != mlink; mlink = mlink->next) {
2012 i = 1;
2013 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2014 i, mlink->dsec);
2015 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2016 i, mlink->arch);
2017 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2018 i, mlink->name);
2019 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2020 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2021 }
2022 }
2023
2024 if (0 == nodb)
2025 SQL_EXEC("END TRANSACTION");
2026 }
2027
2028 /*
2029 * Close an existing database and its prepared statements.
2030 * If "real" is not set, rename the temporary file into the real one.
2031 */
2032 static void
2033 dbclose(int real)
2034 {
2035 size_t i;
2036 int status;
2037 pid_t child;
2038
2039 if (nodb)
2040 return;
2041
2042 for (i = 0; i < STMT__MAX; i++) {
2043 sqlite3_finalize(stmts[i]);
2044 stmts[i] = NULL;
2045 }
2046
2047 sqlite3_close(db);
2048 db = NULL;
2049
2050 if (real)
2051 return;
2052
2053 if ('\0' == *tempfilename) {
2054 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2055 exitcode = (int)MANDOCLEVEL_SYSERR;
2056 say(MANDOC_DB, "&rename");
2057 }
2058 return;
2059 }
2060
2061 switch (child = fork()) {
2062 case (-1):
2063 exitcode = (int)MANDOCLEVEL_SYSERR;
2064 say("", "&fork cmp");
2065 return;
2066 case (0):
2067 execlp("cmp", "cmp", "-s",
2068 tempfilename, MANDOC_DB, NULL);
2069 say("", "&exec cmp");
2070 exit(0);
2071 default:
2072 break;
2073 }
2074 if (-1 == waitpid(child, &status, 0)) {
2075 exitcode = (int)MANDOCLEVEL_SYSERR;
2076 say("", "&wait cmp");
2077 } else if (WIFSIGNALED(status)) {
2078 exitcode = (int)MANDOCLEVEL_SYSERR;
2079 say("", "cmp died from signal %d", WTERMSIG(status));
2080 } else if (WEXITSTATUS(status)) {
2081 exitcode = (int)MANDOCLEVEL_SYSERR;
2082 say(MANDOC_DB,
2083 "Data changed, but cannot replace database");
2084 }
2085
2086 *strrchr(tempfilename, '/') = '\0';
2087 switch (child = fork()) {
2088 case (-1):
2089 exitcode = (int)MANDOCLEVEL_SYSERR;
2090 say("", "&fork rm");
2091 return;
2092 case (0):
2093 execlp("rm", "rm", "-rf", tempfilename, NULL);
2094 say("", "&exec rm");
2095 exit((int)MANDOCLEVEL_SYSERR);
2096 default:
2097 break;
2098 }
2099 if (-1 == waitpid(child, &status, 0)) {
2100 exitcode = (int)MANDOCLEVEL_SYSERR;
2101 say("", "&wait rm");
2102 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2103 exitcode = (int)MANDOCLEVEL_SYSERR;
2104 say("", "%s: Cannot remove temporary directory",
2105 tempfilename);
2106 }
2107 }
2108
2109 /*
2110 * This is straightforward stuff.
2111 * Open a database connection to a "temporary" database, then open a set
2112 * of prepared statements we'll use over and over again.
2113 * If "real" is set, we use the existing database; if not, we truncate a
2114 * temporary one.
2115 * Must be matched by dbclose().
2116 */
2117 static int
2118 dbopen(int real)
2119 {
2120 const char *sql;
2121 int rc, ofl;
2122
2123 if (nodb)
2124 return(1);
2125
2126 *tempfilename = '\0';
2127 ofl = SQLITE_OPEN_READWRITE;
2128
2129 if (real) {
2130 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2131 if (SQLITE_OK != rc) {
2132 exitcode = (int)MANDOCLEVEL_SYSERR;
2133 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2134 return(0);
2135 }
2136 goto prepare_statements;
2137 }
2138
2139 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2140
2141 remove(MANDOC_DB "~");
2142 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2143 if (SQLITE_OK == rc)
2144 goto create_tables;
2145 if (MPARSE_QUICK & mparse_options) {
2146 exitcode = (int)MANDOCLEVEL_SYSERR;
2147 say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
2148 return(0);
2149 }
2150
2151 if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2152 sizeof(tempfilename)) >= sizeof(tempfilename)) {
2153 exitcode = (int)MANDOCLEVEL_SYSERR;
2154 say("", "/tmp/mandocdb.XXXXXX: Filename too long");
2155 return(0);
2156 }
2157 if (NULL == mkdtemp(tempfilename)) {
2158 exitcode = (int)MANDOCLEVEL_SYSERR;
2159 say("", "&%s", tempfilename);
2160 return(0);
2161 }
2162 if (strlcat(tempfilename, "/" MANDOC_DB,
2163 sizeof(tempfilename)) >= sizeof(tempfilename)) {
2164 exitcode = (int)MANDOCLEVEL_SYSERR;
2165 say("", "%s/" MANDOC_DB ": Filename too long",
2166 tempfilename);
2167 return(0);
2168 }
2169 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2170 if (SQLITE_OK != rc) {
2171 exitcode = (int)MANDOCLEVEL_SYSERR;
2172 say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
2173 return(0);
2174 }
2175
2176 create_tables:
2177 sql = "CREATE TABLE \"mpages\" (\n"
2178 " \"desc\" TEXT NOT NULL,\n"
2179 " \"form\" INTEGER NOT NULL,\n"
2180 " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2181 ");\n"
2182 "\n"
2183 "CREATE TABLE \"mlinks\" (\n"
2184 " \"sec\" TEXT NOT NULL,\n"
2185 " \"arch\" TEXT NOT NULL,\n"
2186 " \"name\" TEXT NOT NULL,\n"
2187 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
2188 "ON DELETE CASCADE\n"
2189 ");\n"
2190 "\n"
2191 "CREATE TABLE \"names\" (\n"
2192 " \"bits\" INTEGER NOT NULL,\n"
2193 " \"name\" TEXT NOT NULL,\n"
2194 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
2195 "ON DELETE CASCADE\n"
2196 ");\n"
2197 "\n"
2198 "CREATE TABLE \"keys\" (\n"
2199 " \"bits\" INTEGER NOT NULL,\n"
2200 " \"key\" TEXT NOT NULL,\n"
2201 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
2202 "ON DELETE CASCADE\n"
2203 ");\n";
2204
2205 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2206 exitcode = (int)MANDOCLEVEL_SYSERR;
2207 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2208 return(0);
2209 }
2210
2211 prepare_statements:
2212 SQL_EXEC("PRAGMA foreign_keys = ON");
2213 sql = "DELETE FROM mpages WHERE id IN "
2214 "(SELECT pageid FROM mlinks WHERE "
2215 "sec=? AND arch=? AND name=?)";
2216 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2217 sql = "INSERT INTO mpages "
2218 "(desc,form) VALUES (?,?)";
2219 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2220 sql = "INSERT INTO mlinks "
2221 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2222 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2223 sql = "INSERT INTO names "
2224 "(bits,name,pageid) VALUES (?,?,?)";
2225 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2226 sql = "INSERT INTO keys "
2227 "(bits,key,pageid) VALUES (?,?,?)";
2228 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2229
2230 #ifndef __APPLE__
2231 /*
2232 * When opening a new database, we can turn off
2233 * synchronous mode for much better performance.
2234 */
2235
2236 if (real)
2237 SQL_EXEC("PRAGMA synchronous = OFF");
2238 #endif
2239
2240 return(1);
2241 }
2242
2243 static void *
2244 hash_halloc(size_t sz, void *arg)
2245 {
2246
2247 return(mandoc_calloc(sz, 1));
2248 }
2249
2250 static void *
2251 hash_alloc(size_t sz, void *arg)
2252 {
2253
2254 return(mandoc_malloc(sz));
2255 }
2256
2257 static void
2258 hash_free(void *p, size_t sz, void *arg)
2259 {
2260
2261 free(p);
2262 }
2263
2264 static int
2265 set_basedir(const char *targetdir)
2266 {
2267 static char startdir[PATH_MAX];
2268 static int fd;
2269
2270 /*
2271 * Remember where we started by keeping a fd open to the origin
2272 * path component: throughout this utility, we chdir() a lot to
2273 * handle relative paths, and by doing this, we can return to
2274 * the starting point.
2275 */
2276 if ('\0' == *startdir) {
2277 if (NULL == getcwd(startdir, PATH_MAX)) {
2278 exitcode = (int)MANDOCLEVEL_SYSERR;
2279 if (NULL != targetdir)
2280 say("", "&getcwd");
2281 return(0);
2282 }
2283 if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
2284 exitcode = (int)MANDOCLEVEL_SYSERR;
2285 say("", "&open %s", startdir);
2286 return(0);
2287 }
2288 if (NULL == targetdir)
2289 targetdir = startdir;
2290 } else {
2291 if (-1 == fd)
2292 return(0);
2293 if (-1 == fchdir(fd)) {
2294 close(fd);
2295 basedir[0] = '\0';
2296 exitcode = (int)MANDOCLEVEL_SYSERR;
2297 say("", "&chdir %s", startdir);
2298 return(0);
2299 }
2300 if (NULL == targetdir) {
2301 close(fd);
2302 return(1);
2303 }
2304 }
2305 if (NULL == realpath(targetdir, basedir)) {
2306 basedir[0] = '\0';
2307 exitcode = (int)MANDOCLEVEL_BADARG;
2308 say("", "&%s: realpath", targetdir);
2309 return(0);
2310 } else if (-1 == chdir(basedir)) {
2311 exitcode = (int)MANDOCLEVEL_BADARG;
2312 say("", "&chdir");
2313 return(0);
2314 }
2315 return(1);
2316 }
2317
2318 static void
2319 say(const char *file, const char *format, ...)
2320 {
2321 va_list ap;
2322 int use_errno;
2323
2324 if ('\0' != *basedir)
2325 fprintf(stderr, "%s", basedir);
2326 if ('\0' != *basedir && '\0' != *file)
2327 fputs("//", stderr);
2328 if ('\0' != *file)
2329 fprintf(stderr, "%s", file);
2330
2331 use_errno = 1;
2332 if (NULL != format) {
2333 switch (*format) {
2334 case ('&'):
2335 format++;
2336 break;
2337 case ('\0'):
2338 format = NULL;
2339 break;
2340 default:
2341 use_errno = 0;
2342 break;
2343 }
2344 }
2345 if (NULL != format) {
2346 if ('\0' != *basedir || '\0' != *file)
2347 fputs(": ", stderr);
2348 va_start(ap, format);
2349 vfprintf(stderr, format, ap);
2350 va_end(ap);
2351 }
2352 if (use_errno) {
2353 if ('\0' != *basedir || '\0' != *file || NULL != format)
2354 fputs(": ", stderr);
2355 perror(NULL);
2356 } else
2357 fputc('\n', stderr);
2358 }
A git conversion mirror of mandoc.bsd.lv