]>
git.cameronkatri.com Git - mandoc.git/blob - mandocdb.c
1 /* $Id: mandocdb.c,v 1.145 2014/04/25 12:13:15 schwarze Exp $ */
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
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.
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.
42 #include "compat_ohash.h"
49 #include "mandoc_aux.h"
51 #include "mansearch.h"
53 extern int mansearch_keymax
;
54 extern const char *const mansearch_keynames
[];
56 #define SQL_EXEC(_v) \
57 if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
58 say("", "%s: %s", (_v), 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 say(mlink->file, "%s", sqlite3_errmsg(db))
63 #define SQL_BIND_INT(_s, _i, _v) \
64 if (SQLITE_OK != sqlite3_bind_int \
65 ((_s), (_i)++, (_v))) \
66 say(mlink->file, "%s", sqlite3_errmsg(db))
67 #define SQL_BIND_INT64(_s, _i, _v) \
68 if (SQLITE_OK != sqlite3_bind_int64 \
69 ((_s), (_i)++, (_v))) \
70 say(mlink->file, "%s", sqlite3_errmsg(db))
71 #define SQL_STEP(_s) \
72 if (SQLITE_DONE != sqlite3_step((_s))) \
73 say(mlink->file, "%s", sqlite3_errmsg(db))
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 */
84 FORM_NONE
, /* format is unknown */
85 FORM_SRC
, /* format is -man or -mdoc */
86 FORM_CAT
/* format is cat */
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 */
102 struct inodev inodev
; /* used for hashing routine */
103 int64_t pageid
; /* pageid 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 */
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 */
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 */
134 typedef int (*mdoc_fp
)(struct mpage
*, const struct mdoc_node
*);
136 struct mdoc_handler
{
137 mdoc_fp fp
; /* optional handler */
138 uint64_t mask
; /* set unless handler returns 0 */
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]);
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
;
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 */
325 main(int argc
, char *argv
[])
329 const char *path_arg
;
331 struct manpaths dirs
;
333 struct ohash_info mpages_info
, mlinks_info
;
335 memset(stmts
, 0, STMT__MAX
* sizeof(sqlite3_stmt
*));
336 memset(&dirs
, 0, sizeof(struct manpaths
));
338 mpages_info
.alloc
= mlinks_info
.alloc
= hash_alloc
;
339 mpages_info
.halloc
= mlinks_info
.halloc
= hash_halloc
;
340 mpages_info
.hfree
= mlinks_info
.hfree
= hash_free
;
342 mpages_info
.key_offset
= offsetof(struct mpage
, inodev
);
343 mlinks_info
.key_offset
= offsetof(struct mlink
, file
);
345 progname
= strrchr(argv
[0], '/');
346 if (progname
== NULL
)
352 * We accept a few different invocations.
353 * The CHECKOP macro makes sure that invocation styles don't
354 * clobber each other.
356 #define CHECKOP(_op, _ch) do \
357 if (OP_DEFAULT != (_op)) { \
358 fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
360 } while (/*CONSTCOND*/0)
365 while (-1 != (ch
= getopt(argc
, argv
, "aC:Dd:npQT:tu:v")))
390 mparse_options
|= MPARSE_QUICK
;
393 if (strcmp(optarg
, "utf8")) {
394 fprintf(stderr
, "-T%s: Unsupported "
395 "output format\n", optarg
);
402 dup2(STDOUT_FILENO
, STDERR_FILENO
);
412 /* Compatibility with espie@'s makewhatis. */
421 if (OP_CONFFILE
== op
&& argc
> 0) {
422 fprintf(stderr
, "-C: Too many arguments\n");
426 exitcode
= (int)MANDOCLEVEL_OK
;
427 mp
= mparse_alloc(mparse_options
, MANDOCLEVEL_FATAL
, NULL
, NULL
);
430 ohash_init(&mpages
, 6, &mpages_info
);
431 ohash_init(&mlinks
, 6, &mlinks_info
);
433 if (OP_UPDATE
== op
|| OP_DELETE
== op
|| OP_TEST
== op
) {
436 * All of these deal with a specific directory.
437 * Jump into that directory first.
439 if (0 == set_basedir(path_arg
))
444 * The existing database is usable. Process
445 * all files specified on the command-line.
448 for (i
= 0; i
< argc
; i
++)
454 * Database missing or corrupt.
455 * Recreate from scratch.
457 exitcode
= (int)MANDOCLEVEL_OK
;
465 mpages_merge(mc
, mp
);
466 dbclose(OP_DEFAULT
== op
? 0 : 1);
469 * If we have arguments, use them as our manpaths.
470 * If we don't, grok from manpath(1) or however else
471 * manpath_parse() wants to do it.
474 dirs
.paths
= mandoc_reallocarray(NULL
,
475 argc
, sizeof(char *));
476 dirs
.sz
= (size_t)argc
;
477 for (i
= 0; i
< argc
; i
++)
478 dirs
.paths
[i
] = mandoc_strdup(argv
[i
]);
480 manpath_parse(&dirs
, path_arg
, NULL
, NULL
);
483 exitcode
= (int)MANDOCLEVEL_BADARG
;
484 say("", "Empty manpath");
488 * First scan the tree rooted at a base directory, then
489 * build a new database and finally move it into place.
490 * Ignore zero-length directories and strip trailing
493 for (j
= 0; j
< dirs
.sz
; j
++) {
494 sz
= strlen(dirs
.paths
[j
]);
495 if (sz
&& '/' == dirs
.paths
[j
][sz
- 1])
496 dirs
.paths
[j
][--sz
] = '\0';
501 ohash_init(&mpages
, 6, &mpages_info
);
502 ohash_init(&mlinks
, 6, &mlinks_info
);
505 if (0 == set_basedir(dirs
.paths
[j
]))
509 if (0 == set_basedir(dirs
.paths
[j
]))
514 mpages_merge(mc
, mp
);
516 ! (MPARSE_QUICK
& mparse_options
))
520 if (j
+ 1 < dirs
.sz
) {
522 ohash_delete(&mpages
);
523 ohash_delete(&mlinks
);
533 ohash_delete(&mpages
);
534 ohash_delete(&mlinks
);
537 fprintf(stderr
, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
538 " %s [-aDnpQ] [-Tutf8] dir ...\n"
539 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
540 " %s [-Dnp] -u dir [file ...]\n"
541 " %s [-Q] -t file ...\n",
542 progname
, progname
, progname
,
545 return((int)MANDOCLEVEL_BADARG
);
549 * Scan a directory tree rooted at "basedir" for manpages.
550 * We use fts(), scanning directory parts along the way for clues to our
551 * section and architecture.
553 * If use_all has been specified, grok all files.
554 * If not, sanitise paths to the following:
556 * [./]man*[/<arch>]/<name>.<section>
558 * [./]cat<section>[/<arch>]/<name>.0
560 * TODO: accomodate for multi-language directories.
570 char *dsec
, *arch
, *fsec
, *cp
;
575 argv
[1] = (char *)NULL
;
577 f
= fts_open((char * const *)argv
,
578 FTS_PHYSICAL
| FTS_NOCHDIR
, NULL
);
580 exitcode
= (int)MANDOCLEVEL_SYSERR
;
581 say("", "&fts_open");
588 while (NULL
!= (ff
= fts_read(f
))) {
589 path
= ff
->fts_path
+ 2;
590 switch (ff
->fts_info
) {
593 * Symbolic links require various sanity checks,
594 * then get handled just like regular files.
597 if (NULL
== realpath(path
, buf
)) {
599 say(path
, "&realpath");
602 if (strstr(buf
, basedir
) != buf
) {
603 if (warnings
) say("",
604 "%s: outside base directory", buf
);
607 /* Use logical inode to avoid mpages dupe. */
608 if (-1 == stat(path
, ff
->fts_statp
)) {
616 * If we're a regular file, add an mlink by using the
617 * stored directory data and handling the filename.
620 if (0 == strcmp(path
, MANDOC_DB
))
622 if ( ! use_all
&& ff
->fts_level
< 2) {
624 say(path
, "Extraneous file");
629 while (NULL
== fsec
) {
630 fsec
= strrchr(ff
->fts_name
, '.');
631 if (NULL
== fsec
|| strcmp(fsec
+1, "gz"))
641 "No filename suffix");
644 } else if (0 == strcmp(++fsec
, "html")) {
646 say(path
, "Skip html");
648 } else if (0 == strcmp(fsec
, "ps")) {
650 say(path
, "Skip ps");
652 } else if (0 == strcmp(fsec
, "pdf")) {
654 say(path
, "Skip pdf");
656 } else if ( ! use_all
&&
657 ((FORM_SRC
== dform
&& strcmp(fsec
, dsec
)) ||
658 (FORM_CAT
== dform
&& strcmp(fsec
, "0")))) {
660 say(path
, "Wrong filename suffix");
665 mlink
= mandoc_calloc(1, sizeof(struct mlink
));
666 if (strlcpy(mlink
->file
, path
,
667 sizeof(mlink
->file
)) >=
668 sizeof(mlink
->file
)) {
669 say(path
, "Filename too long");
673 mlink
->dform
= dform
;
676 mlink
->name
= ff
->fts_name
;
679 mlink_add(mlink
, ff
->fts_statp
);
689 say(path
, "Not a regular file");
693 switch (ff
->fts_level
) {
695 /* Ignore the root directory. */
699 * This might contain manX/ or catX/.
700 * Try to infer this from the name.
701 * If we're not in use_all, enforce it.
704 if (FTS_DP
== ff
->fts_info
)
707 if (0 == strncmp(cp
, "man", 3)) {
710 } else if (0 == strncmp(cp
, "cat", 3)) {
718 if (NULL
!= dsec
|| use_all
)
722 say(path
, "Unknown directory part");
723 fts_set(f
, ff
, FTS_SKIP
);
727 * Possibly our architecture.
728 * If we're descending, keep tabs on it.
730 if (FTS_DP
!= ff
->fts_info
&& NULL
!= dsec
)
736 if (FTS_DP
== ff
->fts_info
|| use_all
)
739 say(path
, "Extraneous directory part");
740 fts_set(f
, ff
, FTS_SKIP
);
750 * Add a file to the mlinks table.
751 * Do not verify that it's a "valid" looking manpage (we'll do that
754 * Try to infer the manual section, architecture, and page name from the
755 * path, assuming it looks like
757 * [./]man*[/<arch>]/<name>.<section>
759 * [./]cat<section>[/<arch>]/<name>.0
761 * See treescan() for the fts(3) version of this.
764 filescan(const char *file
)
773 if (0 == strncmp(file
, "./", 2))
777 * We have to do lstat(2) before realpath(3) loses
778 * the information whether this is a symbolic link.
779 * We need to know that because for symbolic links,
780 * we want to use the orginal file name, while for
781 * regular files, we want to use the real path.
783 if (-1 == lstat(file
, &st
)) {
784 exitcode
= (int)MANDOCLEVEL_BADARG
;
787 } else if (0 == ((S_IFREG
| S_IFLNK
) & st
.st_mode
)) {
788 exitcode
= (int)MANDOCLEVEL_BADARG
;
789 say(file
, "Not a regular file");
794 * We have to resolve the file name to the real path
795 * in any case for the base directory check.
797 if (NULL
== realpath(file
, buf
)) {
798 exitcode
= (int)MANDOCLEVEL_BADARG
;
799 say(file
, "&realpath");
803 if (strstr(buf
, basedir
) == buf
)
804 start
= buf
+ strlen(basedir
) + 1;
805 else if (OP_TEST
== op
)
808 exitcode
= (int)MANDOCLEVEL_BADARG
;
809 say("", "%s: outside base directory", buf
);
814 * Now we are sure the file is inside our tree.
815 * If it is a symbolic link, ignore the real path
816 * and use the original name.
817 * This implies passing stuff like "cat1/../man1/foo.1"
818 * on the command line won't work. So don't do that.
819 * Note the stat(2) can still fail if the link target
822 if (S_IFLNK
& st
.st_mode
) {
823 if (-1 == stat(buf
, &st
)) {
824 exitcode
= (int)MANDOCLEVEL_BADARG
;
828 if (strlcpy(buf
, file
, sizeof(buf
)) >= sizeof(buf
)) {
829 say(file
, "Filename too long");
832 start
= strstr(buf
, basedir
) == buf
?
833 buf
+ strlen(basedir
) + 1 : buf
;
836 mlink
= mandoc_calloc(1, sizeof(struct mlink
));
837 if (strlcpy(mlink
->file
, start
, sizeof(mlink
->file
)) >=
838 sizeof(mlink
->file
)) {
839 say(start
, "Filename too long");
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.
849 if (NULL
!= (p
= strchr(start
, '/'))) {
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;
860 if (NULL
!= mlink
->dsec
&& NULL
!= (p
= strchr(start
, '/'))) {
868 * Now check the file suffix.
869 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
871 p
= strrchr(start
, '\0');
872 while (p
-- > start
&& '/' != *p
&& '.' != *p
)
881 * Now try to parse the name.
882 * Use the filename portion of the path.
885 if (NULL
!= (p
= strrchr(start
, '/'))) {
889 mlink_add(mlink
, &st
);
893 mlink_add(struct mlink
*mlink
, const struct stat
*st
)
895 struct inodev inodev
;
899 assert(NULL
!= mlink
->file
);
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
: "");
906 if ('0' == *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
;
913 mlink
->fform
= FORM_NONE
;
915 slot
= ohash_qlookup(&mlinks
, mlink
->file
);
916 assert(NULL
== ohash_find(&mlinks
, slot
));
917 ohash_insert(&mlinks
, slot
, mlink
);
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
);
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
);
930 mlink
->next
= mpage
->mlinks
;
931 mpage
->mlinks
= mlink
;
932 mlink
->mpage
= mpage
;
936 mlink_free(struct mlink
*mlink
)
953 mpage
= ohash_first(&mpages
, &slot
);
954 while (NULL
!= mpage
) {
955 while (NULL
!= (mlink
= mpage
->mlinks
)) {
956 mpage
->mlinks
= mlink
->next
;
964 mpage
= ohash_next(&mpages
, &slot
);
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.
975 mlinks_undupe(struct mpage
*mpage
)
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
;
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
, '.')))
995 (void)strlcat(buf
, mlink
->dsec
, sizeof(buf
));
996 if (NULL
== ohash_find(&mlinks
,
997 ohash_qlookup(&mlinks
, buf
)))
1000 say(mlink
->file
, "Man source exists: %s", buf
);
1003 *prev
= mlink
->next
;
1007 prev
= &(*prev
)->next
;
1012 mlink_check(struct mpage
*mpage
, struct mlink
*mlink
)
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).
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
);
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
1039 * A few manuals are even shared across completely
1040 * different architectures, for example fdformat(1)
1041 * on amd64, i386, sparc, and sparc64.
1044 if (strcasecmp(mpage
->arch
, mlink
->arch
))
1045 say(mlink
->file
, "Architecture \"%s\" manual in "
1046 "\"%s\" directory", mpage
->arch
, mlink
->arch
);
1050 * parse_cat() doesn't set NAME_TITLE yet.
1053 if (FORM_CAT
== mpage
->form
)
1057 * Check whether this mlink
1058 * appears as a name in the NAME section.
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");
1069 * Run through the files in the global vector "mpages"
1070 * and add them to the database specified in "basedir".
1072 * This handles the parsing scheme itself, using the cues of directory
1073 * and filename to determine whether the file is parsable or not.
1076 mpages_merge(struct mchars
*mc
, struct mparse
*mp
)
1079 struct ohash_info str_info
;
1081 struct mpage
*mpage
, *mpage_dest
;
1082 struct mlink
*mlink
, *mlink_dest
;
1090 enum mandoclevel lvl
;
1092 str_info
.alloc
= hash_alloc
;
1093 str_info
.halloc
= hash_halloc
;
1094 str_info
.hfree
= hash_free
;
1095 str_info
.key_offset
= offsetof(struct str
, key
);
1098 SQL_EXEC("BEGIN TRANSACTION");
1100 mpage
= ohash_first(&mpages
, &pslot
);
1101 while (NULL
!= mpage
) {
1102 mlinks_undupe(mpage
);
1103 if (NULL
== mpage
->mlinks
) {
1104 mpage
= ohash_next(&mpages
, &pslot
);
1108 name_mask
= NAME_MASK
;
1109 ohash_init(&names
, 4, &str_info
);
1110 ohash_init(&strings
, 6, &str_info
);
1119 if (mpage
->mlinks
->gzip
) {
1120 if (-1 == pipe(fd
)) {
1121 exitcode
= (int)MANDOCLEVEL_SYSERR
;
1122 say(mpage
->mlinks
->file
, "&pipe gunzip");
1125 switch (child_pid
= fork()) {
1127 exitcode
= (int)MANDOCLEVEL_SYSERR
;
1128 say(mpage
->mlinks
->file
, "&fork gunzip");
1135 if (-1 == dup2(fd
[1], STDOUT_FILENO
)) {
1136 say(mpage
->mlinks
->file
,
1140 execlp("gunzip", "gunzip", "-c",
1141 mpage
->mlinks
->file
, NULL
);
1142 say(mpage
->mlinks
->file
, "&exec gunzip");
1151 * Try interpreting the file as mdoc(7) or man(7)
1152 * source code, unless it is already known to be
1153 * formatted. Fall back to formatted mode.
1155 if (FORM_CAT
!= mpage
->mlinks
->dform
||
1156 FORM_CAT
!= mpage
->mlinks
->fform
) {
1157 lvl
= mparse_readfd(mp
, fd
[0], mpage
->mlinks
->file
);
1158 if (lvl
< MANDOCLEVEL_FATAL
)
1159 mparse_result(mp
, &mdoc
, &man
, &sodest
);
1162 if (NULL
!= sodest
) {
1163 mlink_dest
= ohash_find(&mlinks
,
1164 ohash_qlookup(&mlinks
, sodest
));
1165 if (NULL
!= mlink_dest
) {
1167 /* The .so target exists. */
1169 mpage_dest
= mlink_dest
->mpage
;
1170 mlink
= mpage
->mlinks
;
1172 mlink
->mpage
= mpage_dest
;
1175 * If the target was already
1176 * processed, add the links
1177 * to the database now.
1178 * Otherwise, this will
1179 * happen when we come
1183 if (mpage_dest
->pageid
)
1186 if (NULL
== mlink
->next
)
1188 mlink
= mlink
->next
;
1191 /* Move all links to the target. */
1193 mlink
->next
= mlink_dest
->next
;
1194 mlink_dest
->next
= mpage
->mlinks
;
1195 mpage
->mlinks
= NULL
;
1198 } else if (NULL
!= mdoc
) {
1199 mpage
->form
= FORM_SRC
;
1201 mandoc_strdup(mdoc_meta(mdoc
)->msec
);
1202 mpage
->arch
= mdoc_meta(mdoc
)->arch
;
1203 mpage
->arch
= mandoc_strdup(
1204 NULL
== mpage
->arch
? "" : mpage
->arch
);
1206 mandoc_strdup(mdoc_meta(mdoc
)->title
);
1207 } else if (NULL
!= man
) {
1208 mpage
->form
= FORM_SRC
;
1210 mandoc_strdup(man_meta(man
)->msec
);
1212 mandoc_strdup(mpage
->mlinks
->arch
);
1214 mandoc_strdup(man_meta(man
)->title
);
1216 mpage
->form
= FORM_CAT
;
1218 mandoc_strdup(mpage
->mlinks
->dsec
);
1220 mandoc_strdup(mpage
->mlinks
->arch
);
1222 mandoc_strdup(mpage
->mlinks
->name
);
1224 putkey(mpage
, mpage
->sec
, TYPE_sec
);
1225 putkey(mpage
, '\0' == *mpage
->arch
?
1226 any
: mpage
->arch
, TYPE_arch
);
1228 for (mlink
= mpage
->mlinks
; mlink
; mlink
= mlink
->next
) {
1229 if ('\0' != *mlink
->dsec
)
1230 putkey(mpage
, mlink
->dsec
, TYPE_sec
);
1231 if ('\0' != *mlink
->fsec
)
1232 putkey(mpage
, mlink
->fsec
, TYPE_sec
);
1233 putkey(mpage
, '\0' == *mlink
->arch
?
1234 any
: mlink
->arch
, TYPE_arch
);
1235 putkey(mpage
, mlink
->name
, NAME_FILE
);
1238 assert(NULL
== mpage
->desc
);
1240 if (NULL
!= (cp
= mdoc_meta(mdoc
)->name
))
1241 putkey(mpage
, cp
, NAME_HEAD
);
1242 parse_mdoc(mpage
, mdoc_node(mdoc
));
1243 } else if (NULL
!= man
)
1244 parse_man(mpage
, man_node(man
));
1246 parse_cat(mpage
, fd
[0]);
1247 if (NULL
== mpage
->desc
)
1248 mpage
->desc
= mandoc_strdup(mpage
->mlinks
->name
);
1250 if (warnings
&& !use_all
)
1251 for (mlink
= mpage
->mlinks
; mlink
;
1252 mlink
= mlink
->next
)
1253 mlink_check(mpage
, mlink
);
1259 if (-1 == waitpid(child_pid
, &status
, 0)) {
1260 exitcode
= (int)MANDOCLEVEL_SYSERR
;
1261 say(mpage
->mlinks
->file
, "&wait gunzip");
1262 } else if (WIFSIGNALED(status
)) {
1263 exitcode
= (int)MANDOCLEVEL_SYSERR
;
1264 say(mpage
->mlinks
->file
,
1265 "gunzip died from signal %d",
1267 } else if (WEXITSTATUS(status
)) {
1268 exitcode
= (int)MANDOCLEVEL_SYSERR
;
1269 say(mpage
->mlinks
->file
,
1270 "gunzip failed with code %d",
1271 WEXITSTATUS(status
));
1274 ohash_delete(&strings
);
1275 ohash_delete(&names
);
1276 mpage
= ohash_next(&mpages
, &pslot
);
1280 SQL_EXEC("END TRANSACTION");
1287 const char *name
, *sec
, *arch
, *key
;
1290 sqlite3_prepare_v2(db
,
1291 "SELECT name, sec, arch, key FROM ("
1292 "SELECT name AS key, pageid FROM names "
1293 "WHERE bits & ? AND NOT EXISTS ("
1294 "SELECT pageid FROM mlinks "
1295 "WHERE mlinks.pageid == names.pageid "
1296 "AND mlinks.name == names.name"
1299 "SELECT sec, arch, name, pageid FROM mlinks "
1301 ") USING (pageid);",
1304 if (SQLITE_OK
!= sqlite3_bind_int64(stmt
, 1, NAME_TITLE
))
1305 say("", "%s", sqlite3_errmsg(db
));
1307 while (SQLITE_ROW
== (irc
= sqlite3_step(stmt
))) {
1308 name
= sqlite3_column_text(stmt
, 0);
1309 sec
= sqlite3_column_text(stmt
, 1);
1310 arch
= sqlite3_column_text(stmt
, 2);
1311 key
= sqlite3_column_text(stmt
, 3);
1312 say("", "%s(%s%s%s) lacks mlink \"%s\"", name
, sec
,
1313 '\0' == *arch
? "" : "/",
1314 '\0' == *arch
? "" : arch
, key
);
1316 sqlite3_finalize(stmt
);
1320 parse_cat(struct mpage
*mpage
, int fd
)
1323 char *line
, *p
, *title
;
1324 size_t len
, plen
, titlesz
;
1326 stream
= (-1 == fd
) ?
1327 fopen(mpage
->mlinks
->file
, "r") :
1329 if (NULL
== stream
) {
1331 say(mpage
->mlinks
->file
, "&fopen");
1335 /* Skip to first blank line. */
1337 while (NULL
!= (line
= fgetln(stream
, &len
)))
1342 * Assume the first line that is not indented
1343 * is the first section header. Skip to it.
1346 while (NULL
!= (line
= fgetln(stream
, &len
)))
1347 if ('\n' != *line
&& ' ' != *line
)
1351 * Read up until the next section into a buffer.
1352 * Strip the leading and trailing newline from each read line,
1353 * appending a trailing space.
1354 * Ignore empty (whitespace-only) lines.
1360 while (NULL
!= (line
= fgetln(stream
, &len
))) {
1361 if (' ' != *line
|| '\n' != line
[len
- 1])
1363 while (len
> 0 && isspace((unsigned char)*line
)) {
1369 title
= mandoc_realloc(title
, titlesz
+ len
);
1370 memcpy(title
+ titlesz
, line
, len
);
1372 title
[titlesz
- 1] = ' ';
1376 * If no page content can be found, or the input line
1377 * is already the next section header, or there is no
1378 * trailing newline, reuse the page title as the page
1382 if (NULL
== title
|| '\0' == *title
) {
1384 say(mpage
->mlinks
->file
,
1385 "Cannot find NAME section");
1391 title
= mandoc_realloc(title
, titlesz
+ 1);
1392 title
[titlesz
] = '\0';
1395 * Skip to the first dash.
1396 * Use the remaining line as the description (no more than 70
1400 if (NULL
!= (p
= strstr(title
, "- "))) {
1401 for (p
+= 2; ' ' == *p
|| '\b' == *p
; p
++)
1402 /* Skip to next word. */ ;
1405 say(mpage
->mlinks
->file
,
1406 "No dash in title line");
1412 /* Strip backspace-encoding from line. */
1414 while (NULL
!= (line
= memchr(p
, '\b', plen
))) {
1417 memmove(line
, line
+ 1, plen
--);
1420 memmove(line
- 1, line
+ 1, plen
- len
);
1424 mpage
->desc
= mandoc_strdup(p
);
1430 * Put a type/word pair into the word database for this particular file.
1433 putkey(const struct mpage
*mpage
, char *value
, uint64_t type
)
1437 assert(NULL
!= value
);
1438 if (TYPE_arch
== type
)
1439 for (cp
= value
; *cp
; cp
++)
1440 if (isupper((unsigned char)*cp
))
1441 *cp
= _tolower((unsigned char)*cp
);
1442 putkeys(mpage
, value
, strlen(value
), type
);
1446 * Grok all nodes at or below a certain mdoc node into putkey().
1449 putmdockey(const struct mpage
*mpage
,
1450 const struct mdoc_node
*n
, uint64_t m
)
1453 for ( ; NULL
!= n
; n
= n
->next
) {
1454 if (NULL
!= n
->child
)
1455 putmdockey(mpage
, n
->child
, m
);
1456 if (MDOC_TEXT
== n
->type
)
1457 putkey(mpage
, n
->string
, m
);
1462 parse_man(struct mpage
*mpage
, const struct man_node
*n
)
1464 const struct man_node
*head
, *body
;
1465 char *start
, *title
;
1473 * We're only searching for one thing: the first text child in
1474 * the BODY of a NAME section. Since we don't keep track of
1475 * sections in -man, run some hoops to find out whether we're in
1476 * the correct section or not.
1479 if (MAN_BODY
== n
->type
&& MAN_SH
== n
->tok
) {
1481 assert(body
->parent
);
1482 if (NULL
!= (head
= body
->parent
->head
) &&
1483 1 == head
->nchild
&&
1484 NULL
!= (head
= (head
->child
)) &&
1485 MAN_TEXT
== head
->type
&&
1486 0 == strcmp(head
->string
, "NAME") &&
1487 NULL
!= body
->child
) {
1490 * Suck the entire NAME section into memory.
1491 * Yes, we might run away.
1492 * But too many manuals have big, spread-out
1493 * NAME sections over many lines.
1497 man_deroff(&title
, body
);
1502 * Go through a special heuristic dance here.
1503 * Conventionally, one or more manual names are
1504 * comma-specified prior to a whitespace, then a
1505 * dash, then a description. Try to puzzle out
1506 * the name parts here.
1511 sz
= strcspn(start
, " ,");
1512 if ('\0' == start
[sz
])
1519 * Assume a stray trailing comma in the
1520 * name list if a name begins with a dash.
1523 if ('-' == start
[0] ||
1524 ('\\' == start
[0] && '-' == start
[1]))
1527 putkey(mpage
, start
, NAME_TITLE
);
1534 assert(',' == byte
);
1536 while (' ' == *start
)
1540 if (start
== title
) {
1541 putkey(mpage
, start
, NAME_TITLE
);
1546 while (isspace((unsigned char)*start
))
1549 if (0 == strncmp(start
, "-", 1))
1551 else if (0 == strncmp(start
, "\\-\\-", 4))
1553 else if (0 == strncmp(start
, "\\-", 2))
1555 else if (0 == strncmp(start
, "\\(en", 4))
1557 else if (0 == strncmp(start
, "\\(em", 4))
1560 while (' ' == *start
)
1563 mpage
->desc
= mandoc_strdup(start
);
1569 for (n
= n
->child
; n
; n
= n
->next
) {
1570 if (NULL
!= mpage
->desc
)
1572 parse_man(mpage
, n
);
1577 parse_mdoc(struct mpage
*mpage
, const struct mdoc_node
*n
)
1581 for (n
= n
->child
; NULL
!= n
; n
= n
->next
) {
1592 if (NULL
!= mdocs
[n
->tok
].fp
)
1593 if (0 == (*mdocs
[n
->tok
].fp
)(mpage
, n
))
1595 if (mdocs
[n
->tok
].mask
)
1596 putmdockey(mpage
, n
->child
,
1597 mdocs
[n
->tok
].mask
);
1600 assert(MDOC_ROOT
!= n
->type
);
1603 if (NULL
!= n
->child
)
1604 parse_mdoc(mpage
, n
);
1609 parse_mdoc_Fd(struct mpage
*mpage
, const struct mdoc_node
*n
)
1611 const char *start
, *end
;
1614 if (SEC_SYNOPSIS
!= n
->sec
||
1615 NULL
== (n
= n
->child
) ||
1616 MDOC_TEXT
!= n
->type
)
1620 * Only consider those `Fd' macro fields that begin with an
1621 * "inclusion" token (versus, e.g., #define).
1624 if (strcmp("#include", n
->string
))
1627 if (NULL
== (n
= n
->next
) || MDOC_TEXT
!= n
->type
)
1631 * Strip away the enclosing angle brackets and make sure we're
1636 if ('<' == *start
|| '"' == *start
)
1639 if (0 == (sz
= strlen(start
)))
1642 end
= &start
[(int)sz
- 1];
1643 if ('>' == *end
|| '"' == *end
)
1647 putkeys(mpage
, start
, end
- start
+ 1, TYPE_In
);
1652 parse_mdoc_Fn(struct mpage
*mpage
, const struct mdoc_node
*n
)
1656 if (NULL
== (n
= n
->child
) || MDOC_TEXT
!= n
->type
)
1660 * Parse: .Fn "struct type *name" "char *arg".
1661 * First strip away pointer symbol.
1662 * Then store the function name, then type.
1663 * Finally, store the arguments.
1666 if (NULL
== (cp
= strrchr(n
->string
, ' ')))
1672 putkey(mpage
, cp
, TYPE_Fn
);
1675 putkeys(mpage
, n
->string
, cp
- n
->string
, TYPE_Ft
);
1677 for (n
= n
->next
; NULL
!= n
; n
= n
->next
)
1678 if (MDOC_TEXT
== n
->type
)
1679 putkey(mpage
, n
->string
, TYPE_Fa
);
1685 parse_mdoc_Xr(struct mpage
*mpage
, const struct mdoc_node
*n
)
1689 if (NULL
== (n
= n
->child
))
1692 if (NULL
== n
->next
) {
1693 putkey(mpage
, n
->string
, TYPE_Xr
);
1697 mandoc_asprintf(&cp
, "%s(%s)", n
->string
, n
->next
->string
);
1698 putkey(mpage
, cp
, TYPE_Xr
);
1704 parse_mdoc_Nd(struct mpage
*mpage
, const struct mdoc_node
*n
)
1707 if (MDOC_BODY
== n
->type
)
1708 mdoc_deroff(&mpage
->desc
, n
);
1713 parse_mdoc_Nm(struct mpage
*mpage
, const struct mdoc_node
*n
)
1716 if (SEC_NAME
== n
->sec
)
1717 putmdockey(mpage
, n
->child
, NAME_TITLE
);
1718 else if (SEC_SYNOPSIS
== n
->sec
&& MDOC_HEAD
== n
->type
)
1719 putmdockey(mpage
, n
->child
, NAME_SYN
);
1724 parse_mdoc_Sh(struct mpage
*mpage
, const struct mdoc_node
*n
)
1727 return(SEC_CUSTOM
== n
->sec
&& MDOC_HEAD
== n
->type
);
1731 parse_mdoc_head(struct mpage
*mpage
, const struct mdoc_node
*n
)
1734 return(MDOC_HEAD
== n
->type
);
1738 parse_mdoc_body(struct mpage
*mpage
, const struct mdoc_node
*n
)
1741 return(MDOC_BODY
== n
->type
);
1745 * Add a string to the hash table for the current manual.
1746 * Each string has a bitmask telling which macros it belongs to.
1747 * When we finish the manual, we'll dump the table.
1750 putkeys(const struct mpage
*mpage
,
1751 const char *cp
, size_t sz
, uint64_t v
)
1765 name_mask
&= ~NAME_FIRST
;
1767 say(mpage
->mlinks
->file
,
1768 "Adding name %*s", sz
, cp
);
1772 for (i
= 0; i
< mansearch_keymax
; i
++)
1774 say(mpage
->mlinks
->file
,
1775 "Adding key %s=%*s",
1776 mansearch_keynames
[i
], sz
, cp
);
1780 slot
= ohash_qlookupi(htab
, cp
, &end
);
1781 s
= ohash_find(htab
, slot
);
1783 if (NULL
!= s
&& mpage
== s
->mpage
) {
1786 } else if (NULL
== s
) {
1787 s
= mandoc_calloc(1, sizeof(struct str
) + sz
+ 1);
1788 memcpy(s
->key
, cp
, sz
);
1789 ohash_insert(htab
, slot
, s
);
1796 * Take a Unicode codepoint and produce its UTF-8 encoding.
1797 * This isn't the best way to do this, but it works.
1798 * The magic numbers are from the UTF-8 packaging.
1799 * They're not as scary as they seem: read the UTF-8 spec for details.
1802 utf8(unsigned int cp
, char out
[7])
1807 if (cp
<= 0x0000007F) {
1810 } else if (cp
<= 0x000007FF) {
1812 out
[0] = (cp
>> 6 & 31) | 192;
1813 out
[1] = (cp
& 63) | 128;
1814 } else if (cp
<= 0x0000FFFF) {
1816 out
[0] = (cp
>> 12 & 15) | 224;
1817 out
[1] = (cp
>> 6 & 63) | 128;
1818 out
[2] = (cp
& 63) | 128;
1819 } else if (cp
<= 0x001FFFFF) {
1821 out
[0] = (cp
>> 18 & 7) | 240;
1822 out
[1] = (cp
>> 12 & 63) | 128;
1823 out
[2] = (cp
>> 6 & 63) | 128;
1824 out
[3] = (cp
& 63) | 128;
1825 } else if (cp
<= 0x03FFFFFF) {
1827 out
[0] = (cp
>> 24 & 3) | 248;
1828 out
[1] = (cp
>> 18 & 63) | 128;
1829 out
[2] = (cp
>> 12 & 63) | 128;
1830 out
[3] = (cp
>> 6 & 63) | 128;
1831 out
[4] = (cp
& 63) | 128;
1832 } else if (cp
<= 0x7FFFFFFF) {
1834 out
[0] = (cp
>> 30 & 1) | 252;
1835 out
[1] = (cp
>> 24 & 63) | 128;
1836 out
[2] = (cp
>> 18 & 63) | 128;
1837 out
[3] = (cp
>> 12 & 63) | 128;
1838 out
[4] = (cp
>> 6 & 63) | 128;
1839 out
[5] = (cp
& 63) | 128;
1848 * Store the rendered version of a key, or alias the pointer
1849 * if the key contains no escape sequences.
1852 render_key(struct mchars
*mc
, struct str
*key
)
1854 size_t sz
, bsz
, pos
;
1855 char utfbuf
[7], res
[6];
1857 const char *seq
, *cpp
, *val
;
1859 enum mandoc_esc esc
;
1861 assert(NULL
== key
->rendered
);
1865 res
[2] = ASCII_NBRSP
;
1866 res
[3] = ASCII_HYPH
;
1867 res
[4] = ASCII_BREAK
;
1874 * Pre-check: if we have no stop-characters, then set the
1875 * pointer as ourselvse and get out of here.
1877 if (strcspn(val
, res
) == bsz
) {
1878 key
->rendered
= key
->key
;
1882 /* Pre-allocate by the length of the input */
1884 buf
= mandoc_malloc(++bsz
);
1887 while ('\0' != *val
) {
1889 * Halt on the first escape sequence.
1890 * This also halts on the end of string, in which case
1891 * we just copy, fallthrough, and exit the loop.
1893 if ((sz
= strcspn(val
, res
)) > 0) {
1894 memcpy(&buf
[pos
], val
, sz
);
1918 /* Read past the slash. */
1923 * Parse the escape sequence and see if it's a
1924 * predefined character or special character.
1927 esc
= mandoc_escape((const char **)&val
,
1929 if (ESCAPE_ERROR
== esc
)
1931 if (ESCAPE_SPECIAL
!= esc
)
1935 * Render the special character
1936 * as either UTF-8 or ASCII.
1940 if (0 == (u
= mchars_spec2cp(mc
, seq
, len
)))
1943 if (0 == (sz
= utf8(u
, utfbuf
)))
1947 cpp
= mchars_spec2str(mc
, seq
, len
, &sz
);
1950 if (ASCII_NBRSP
== *cpp
) {
1956 /* Copy the rendered glyph into the stream. */
1959 buf
= mandoc_realloc(buf
, bsz
);
1960 memcpy(&buf
[pos
], cpp
, sz
);
1965 key
->rendered
= buf
;
1969 dbadd_mlink(const struct mlink
*mlink
)
1974 SQL_BIND_TEXT(stmts
[STMT_INSERT_LINK
], i
, mlink
->dsec
);
1975 SQL_BIND_TEXT(stmts
[STMT_INSERT_LINK
], i
, mlink
->arch
);
1976 SQL_BIND_TEXT(stmts
[STMT_INSERT_LINK
], i
, mlink
->name
);
1977 SQL_BIND_INT64(stmts
[STMT_INSERT_LINK
], i
, mlink
->mpage
->pageid
);
1978 SQL_STEP(stmts
[STMT_INSERT_LINK
]);
1979 sqlite3_reset(stmts
[STMT_INSERT_LINK
]);
1983 * Flush the current page's terms (and their bits) into the database.
1984 * Wrap the entire set of additions in a transaction to make sqlite be a
1986 * Also, handle escape sequences at the last possible moment.
1989 dbadd(struct mpage
*mpage
, struct mchars
*mc
)
1991 struct mlink
*mlink
;
1996 mlink
= mpage
->mlinks
;
2001 while (NULL
!= mlink
) {
2002 fputs(mlink
->name
, stdout
);
2003 if (NULL
== mlink
->next
||
2004 strcmp(mlink
->dsec
, mlink
->next
->dsec
) ||
2005 strcmp(mlink
->fsec
, mlink
->next
->fsec
) ||
2006 strcmp(mlink
->arch
, mlink
->next
->arch
)) {
2008 if ('\0' == *mlink
->dsec
)
2009 fputs(mlink
->fsec
, stdout
);
2011 fputs(mlink
->dsec
, stdout
);
2012 if ('\0' != *mlink
->arch
)
2013 printf("/%s", mlink
->arch
);
2016 mlink
= mlink
->next
;
2018 fputs(", ", stdout
);
2020 printf(" - %s\n", mpage
->desc
);
2025 say(mlink
->file
, "Adding to database");
2028 SQL_BIND_TEXT(stmts
[STMT_INSERT_PAGE
], i
, mpage
->desc
);
2029 SQL_BIND_INT(stmts
[STMT_INSERT_PAGE
], i
, FORM_SRC
== mpage
->form
);
2030 SQL_STEP(stmts
[STMT_INSERT_PAGE
]);
2031 mpage
->pageid
= sqlite3_last_insert_rowid(db
);
2032 sqlite3_reset(stmts
[STMT_INSERT_PAGE
]);
2034 while (NULL
!= mlink
) {
2036 mlink
= mlink
->next
;
2038 mlink
= mpage
->mlinks
;
2040 for (key
= ohash_first(&names
, &slot
); NULL
!= key
;
2041 key
= ohash_next(&names
, &slot
)) {
2042 assert(key
->mpage
== mpage
);
2043 if (NULL
== key
->rendered
)
2044 render_key(mc
, key
);
2046 SQL_BIND_INT64(stmts
[STMT_INSERT_NAME
], i
, key
->mask
);
2047 SQL_BIND_TEXT(stmts
[STMT_INSERT_NAME
], i
, key
->rendered
);
2048 SQL_BIND_INT64(stmts
[STMT_INSERT_NAME
], i
, mpage
->pageid
);
2049 SQL_STEP(stmts
[STMT_INSERT_NAME
]);
2050 sqlite3_reset(stmts
[STMT_INSERT_NAME
]);
2051 if (key
->rendered
!= key
->key
)
2052 free(key
->rendered
);
2055 for (key
= ohash_first(&strings
, &slot
); NULL
!= key
;
2056 key
= ohash_next(&strings
, &slot
)) {
2057 assert(key
->mpage
== mpage
);
2058 if (NULL
== key
->rendered
)
2059 render_key(mc
, key
);
2061 SQL_BIND_INT64(stmts
[STMT_INSERT_KEY
], i
, key
->mask
);
2062 SQL_BIND_TEXT(stmts
[STMT_INSERT_KEY
], i
, key
->rendered
);
2063 SQL_BIND_INT64(stmts
[STMT_INSERT_KEY
], i
, mpage
->pageid
);
2064 SQL_STEP(stmts
[STMT_INSERT_KEY
]);
2065 sqlite3_reset(stmts
[STMT_INSERT_KEY
]);
2066 if (key
->rendered
!= key
->key
)
2067 free(key
->rendered
);
2075 struct mpage
*mpage
;
2076 struct mlink
*mlink
;
2081 SQL_EXEC("BEGIN TRANSACTION");
2083 for (mpage
= ohash_first(&mpages
, &slot
); NULL
!= mpage
;
2084 mpage
= ohash_next(&mpages
, &slot
)) {
2085 mlink
= mpage
->mlinks
;
2087 say(mlink
->file
, "Deleting from database");
2090 for ( ; NULL
!= mlink
; mlink
= mlink
->next
) {
2092 SQL_BIND_TEXT(stmts
[STMT_DELETE_PAGE
],
2094 SQL_BIND_TEXT(stmts
[STMT_DELETE_PAGE
],
2096 SQL_BIND_TEXT(stmts
[STMT_DELETE_PAGE
],
2098 SQL_STEP(stmts
[STMT_DELETE_PAGE
]);
2099 sqlite3_reset(stmts
[STMT_DELETE_PAGE
]);
2104 SQL_EXEC("END TRANSACTION");
2108 * Close an existing database and its prepared statements.
2109 * If "real" is not set, rename the temporary file into the real one.
2121 for (i
= 0; i
< STMT__MAX
; i
++) {
2122 sqlite3_finalize(stmts
[i
]);
2132 if ('\0' == *tempfilename
) {
2133 if (-1 == rename(MANDOC_DB
"~", MANDOC_DB
)) {
2134 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2135 say(MANDOC_DB
, "&rename");
2140 switch (child
= fork()) {
2142 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2143 say("", "&fork cmp");
2146 execlp("cmp", "cmp", "-s",
2147 tempfilename
, MANDOC_DB
, NULL
);
2148 say("", "&exec cmp");
2153 if (-1 == waitpid(child
, &status
, 0)) {
2154 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2155 say("", "&wait cmp");
2156 } else if (WIFSIGNALED(status
)) {
2157 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2158 say("", "cmp died from signal %d", WTERMSIG(status
));
2159 } else if (WEXITSTATUS(status
)) {
2160 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2162 "Data changed, but cannot replace database");
2165 *strrchr(tempfilename
, '/') = '\0';
2166 switch (child
= fork()) {
2168 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2169 say("", "&fork rm");
2172 execlp("rm", "rm", "-rf", tempfilename
, NULL
);
2173 say("", "&exec rm");
2174 exit((int)MANDOCLEVEL_SYSERR
);
2178 if (-1 == waitpid(child
, &status
, 0)) {
2179 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2180 say("", "&wait rm");
2181 } else if (WIFSIGNALED(status
) || WEXITSTATUS(status
)) {
2182 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2183 say("", "%s: Cannot remove temporary directory",
2189 * This is straightforward stuff.
2190 * Open a database connection to a "temporary" database, then open a set
2191 * of prepared statements we'll use over and over again.
2192 * If "real" is set, we use the existing database; if not, we truncate a
2194 * Must be matched by dbclose().
2205 *tempfilename
= '\0';
2206 ofl
= SQLITE_OPEN_READWRITE
;
2209 rc
= sqlite3_open_v2(MANDOC_DB
, &db
, ofl
, NULL
);
2210 if (SQLITE_OK
!= rc
) {
2211 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2212 say(MANDOC_DB
, "%s", sqlite3_errmsg(db
));
2215 goto prepare_statements
;
2218 ofl
|= SQLITE_OPEN_CREATE
| SQLITE_OPEN_EXCLUSIVE
;
2220 remove(MANDOC_DB
"~");
2221 rc
= sqlite3_open_v2(MANDOC_DB
"~", &db
, ofl
, NULL
);
2222 if (SQLITE_OK
== rc
)
2224 if (MPARSE_QUICK
& mparse_options
) {
2225 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2226 say(MANDOC_DB
"~", "%s", sqlite3_errmsg(db
));
2230 (void)strlcpy(tempfilename
, "/tmp/mandocdb.XXXXXX",
2231 sizeof(tempfilename
));
2232 if (NULL
== mkdtemp(tempfilename
)) {
2233 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2234 say("", "&%s", tempfilename
);
2237 (void)strlcat(tempfilename
, "/" MANDOC_DB
,
2238 sizeof(tempfilename
));
2239 rc
= sqlite3_open_v2(tempfilename
, &db
, ofl
, NULL
);
2240 if (SQLITE_OK
!= rc
) {
2241 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2242 say("", "%s: %s", tempfilename
, sqlite3_errmsg(db
));
2247 sql
= "CREATE TABLE \"mpages\" (\n"
2248 " \"desc\" TEXT NOT NULL,\n"
2249 " \"form\" INTEGER NOT NULL,\n"
2250 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2253 "CREATE TABLE \"mlinks\" (\n"
2254 " \"sec\" TEXT NOT NULL,\n"
2255 " \"arch\" TEXT NOT NULL,\n"
2256 " \"name\" TEXT NOT NULL,\n"
2257 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2258 "ON DELETE CASCADE\n"
2260 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2262 "CREATE TABLE \"names\" (\n"
2263 " \"bits\" INTEGER NOT NULL,\n"
2264 " \"name\" TEXT NOT NULL,\n"
2265 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2266 "ON DELETE CASCADE\n"
2269 "CREATE TABLE \"keys\" (\n"
2270 " \"bits\" INTEGER NOT NULL,\n"
2271 " \"key\" TEXT NOT NULL,\n"
2272 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2273 "ON DELETE CASCADE\n"
2275 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2277 if (SQLITE_OK
!= sqlite3_exec(db
, sql
, NULL
, NULL
, NULL
)) {
2278 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2279 say(MANDOC_DB
, "%s", sqlite3_errmsg(db
));
2284 SQL_EXEC("PRAGMA foreign_keys = ON");
2285 sql
= "DELETE FROM mpages WHERE pageid IN "
2286 "(SELECT pageid FROM mlinks WHERE "
2287 "sec=? AND arch=? AND name=?)";
2288 sqlite3_prepare_v2(db
, sql
, -1, &stmts
[STMT_DELETE_PAGE
], NULL
);
2289 sql
= "INSERT INTO mpages "
2290 "(desc,form) VALUES (?,?)";
2291 sqlite3_prepare_v2(db
, sql
, -1, &stmts
[STMT_INSERT_PAGE
], NULL
);
2292 sql
= "INSERT INTO mlinks "
2293 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2294 sqlite3_prepare_v2(db
, sql
, -1, &stmts
[STMT_INSERT_LINK
], NULL
);
2295 sql
= "INSERT INTO names "
2296 "(bits,name,pageid) VALUES (?,?,?)";
2297 sqlite3_prepare_v2(db
, sql
, -1, &stmts
[STMT_INSERT_NAME
], NULL
);
2298 sql
= "INSERT INTO keys "
2299 "(bits,key,pageid) VALUES (?,?,?)";
2300 sqlite3_prepare_v2(db
, sql
, -1, &stmts
[STMT_INSERT_KEY
], NULL
);
2304 * When opening a new database, we can turn off
2305 * synchronous mode for much better performance.
2309 SQL_EXEC("PRAGMA synchronous = OFF");
2316 hash_halloc(size_t sz
, void *arg
)
2319 return(mandoc_calloc(1, sz
));
2323 hash_alloc(size_t sz
, void *arg
)
2326 return(mandoc_malloc(sz
));
2330 hash_free(void *p
, size_t sz
, void *arg
)
2337 set_basedir(const char *targetdir
)
2339 static char startdir
[PATH_MAX
];
2343 * Remember where we started by keeping a fd open to the origin
2344 * path component: throughout this utility, we chdir() a lot to
2345 * handle relative paths, and by doing this, we can return to
2346 * the starting point.
2348 if ('\0' == *startdir
) {
2349 if (NULL
== getcwd(startdir
, PATH_MAX
)) {
2350 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2351 if (NULL
!= targetdir
)
2355 if (-1 == (fd
= open(startdir
, O_RDONLY
, 0))) {
2356 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2357 say("", "&open %s", startdir
);
2360 if (NULL
== targetdir
)
2361 targetdir
= startdir
;
2365 if (-1 == fchdir(fd
)) {
2368 exitcode
= (int)MANDOCLEVEL_SYSERR
;
2369 say("", "&chdir %s", startdir
);
2372 if (NULL
== targetdir
) {
2377 if (NULL
== realpath(targetdir
, basedir
)) {
2379 exitcode
= (int)MANDOCLEVEL_BADARG
;
2380 say("", "&%s: realpath", targetdir
);
2382 } else if (-1 == chdir(basedir
)) {
2383 exitcode
= (int)MANDOCLEVEL_BADARG
;
2391 say(const char *file
, const char *format
, ...)
2396 if ('\0' != *basedir
)
2397 fprintf(stderr
, "%s", basedir
);
2398 if ('\0' != *basedir
&& '\0' != *file
)
2399 fputs("//", stderr
);
2401 fprintf(stderr
, "%s", file
);
2404 if (NULL
!= format
) {
2417 if (NULL
!= format
) {
2418 if ('\0' != *basedir
|| '\0' != *file
)
2419 fputs(": ", stderr
);
2420 va_start(ap
, format
);
2421 vfprintf(stderr
, format
, ap
);
2425 if ('\0' != *basedir
|| '\0' != *file
|| NULL
!= format
)
2426 fputs(": ", stderr
);
2429 fputc('\n', stderr
);