1 /* $Id: mandocdb.c,v 1.215 2016/01/08 17:48:09 schwarze Exp $ */
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2016 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 AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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.
20 #include <sys/types.h>
34 #include "compat_fts.h"
47 #include "mandoc_aux.h"
48 #include "mandoc_ohash.h"
54 #include "mansearch.h"
56 extern int mansearch_keymax;
57 extern const char *const mansearch_keynames[];
59 #define SQL_EXEC(_v) \
60 if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
61 say("", "%s: %s", (_v), sqlite3_errmsg(db))
62 #define SQL_BIND_TEXT(_s, _i, _v) \
63 if (SQLITE_OK != sqlite3_bind_text \
64 ((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
65 say(mlink->file, "%s", sqlite3_errmsg(db))
66 #define SQL_BIND_INT(_s, _i, _v) \
67 if (SQLITE_OK != sqlite3_bind_int \
68 ((_s), (_i)++, (_v))) \
69 say(mlink->file, "%s", sqlite3_errmsg(db))
70 #define SQL_BIND_INT64(_s, _i, _v) \
71 if (SQLITE_OK != sqlite3_bind_int64 \
72 ((_s), (_i)++, (_v))) \
73 say(mlink->file, "%s", sqlite3_errmsg(db))
74 #define SQL_STEP(_s) \
75 if (SQLITE_DONE != sqlite3_step((_s))) \
76 say(mlink->file, "%s", sqlite3_errmsg(db))
79 OP_DEFAULT = 0, /* new dbs from dir list or default config */
80 OP_CONFFILE, /* new databases from custom config file */
81 OP_UPDATE, /* delete/add entries in existing database */
82 OP_DELETE, /* delete entries from existing database */
83 OP_TEST /* change no databases, report potential problems */
87 const struct mpage *mpage; /* if set, the owning parse */
88 uint64_t mask; /* bitmask in sequence */
89 char key[]; /* rendered text */
98 struct inodev inodev; /* used for hashing routine */
99 int64_t pageid; /* pageid in mpages SQL table */
100 char *sec; /* section from file content */
101 char *arch; /* architecture from file content */
102 char *title; /* title from file content */
103 char *desc; /* description from file content */
104 struct mlink *mlinks; /* singly linked list */
105 int form; /* format from file content */
110 char file[PATH_MAX]; /* filename rel. to manpath */
111 char *dsec; /* section from directory */
112 char *arch; /* architecture from directory */
113 char *name; /* name from file name (not empty) */
114 char *fsec; /* section from file name suffix */
115 struct mlink *next; /* singly linked list */
116 struct mpage *mpage; /* parent */
117 int dform; /* format from directory */
118 int fform; /* format from file name suffix */
119 int gzip; /* filename has a .gz suffix */
123 STMT_DELETE_PAGE = 0, /* delete mpage */
124 STMT_INSERT_PAGE, /* insert mpage */
125 STMT_INSERT_LINK, /* insert mlink */
126 STMT_INSERT_NAME, /* insert name */
127 STMT_SELECT_NAME, /* retrieve existing name flags */
128 STMT_INSERT_KEY, /* insert parsed key */
132 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
133 const struct roff_node *);
135 struct mdoc_handler {
136 mdoc_fp fp; /* optional handler */
137 uint64_t mask; /* set unless handler returns 0 */
140 static void dbclose(int);
141 static void dbadd(struct mpage *);
142 static void dbadd_mlink(const struct mlink *mlink);
143 static void dbadd_mlink_name(const struct mlink *mlink);
144 static int dbopen(int);
145 static void dbprune(void);
146 static void filescan(const char *);
147 static void mlink_add(struct mlink *, const struct stat *);
148 static void mlink_check(struct mpage *, struct mlink *);
149 static void mlink_free(struct mlink *);
150 static void mlinks_undupe(struct mpage *);
151 static void mpages_free(void);
152 static void mpages_merge(struct mparse *);
153 static void names_check(void);
154 static void parse_cat(struct mpage *, int);
155 static void parse_man(struct mpage *, const struct roff_meta *,
156 const struct roff_node *);
157 static void parse_mdoc(struct mpage *, const struct roff_meta *,
158 const struct roff_node *);
159 static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
160 const struct roff_node *);
161 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
162 const struct roff_node *);
163 static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
164 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
165 const struct roff_node *);
166 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
167 const struct roff_node *);
168 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
169 const struct roff_node *);
170 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
171 const struct roff_node *);
172 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
173 const struct roff_node *);
174 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
175 const struct roff_node *);
176 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
177 const struct roff_node *);
178 static void putkey(const struct mpage *, char *, uint64_t);
179 static void putkeys(const struct mpage *, char *, size_t, uint64_t);
180 static void putmdockey(const struct mpage *,
181 const struct roff_node *, uint64_t);
182 static int render_string(char **, size_t *);
183 static void say(const char *, const char *, ...);
184 static int set_basedir(const char *, int);
185 static int treescan(void);
186 static size_t utf8(unsigned int, char [7]);
188 static char tempfilename[32];
189 static int nodb; /* no database changes */
190 static int mparse_options; /* abort the parse early */
191 static int use_all; /* use all found files */
192 static int debug; /* print what we're doing */
193 static int warnings; /* warn about crap */
194 static int write_utf8; /* write UTF-8 output; else ASCII */
195 static int exitcode; /* to be returned by main */
196 static enum op op; /* operational mode */
197 static char basedir[PATH_MAX]; /* current base directory */
198 static struct ohash mpages; /* table of distinct manual pages */
199 static struct ohash mlinks; /* table of directory entries */
200 static struct ohash names; /* table of all names */
201 static struct ohash strings; /* table of all strings */
202 static sqlite3 *db = NULL; /* current database */
203 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
204 static uint64_t name_mask;
206 static const struct mdoc_handler mdocs[MDOC_MAX] = {
207 { NULL, 0 }, /* Ap */
208 { NULL, 0 }, /* Dd */
209 { NULL, 0 }, /* Dt */
210 { NULL, 0 }, /* Os */
211 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
212 { parse_mdoc_head, TYPE_Ss }, /* Ss */
213 { NULL, 0 }, /* Pp */
214 { NULL, 0 }, /* D1 */
215 { NULL, 0 }, /* Dl */
216 { NULL, 0 }, /* Bd */
217 { NULL, 0 }, /* Ed */
218 { NULL, 0 }, /* Bl */
219 { NULL, 0 }, /* El */
220 { NULL, 0 }, /* It */
221 { NULL, 0 }, /* Ad */
222 { NULL, TYPE_An }, /* An */
223 { NULL, TYPE_Ar }, /* Ar */
224 { NULL, TYPE_Cd }, /* Cd */
225 { NULL, TYPE_Cm }, /* Cm */
226 { NULL, TYPE_Dv }, /* Dv */
227 { NULL, TYPE_Er }, /* Er */
228 { NULL, TYPE_Ev }, /* Ev */
229 { NULL, 0 }, /* Ex */
230 { NULL, TYPE_Fa }, /* Fa */
231 { parse_mdoc_Fd, 0 }, /* Fd */
232 { NULL, TYPE_Fl }, /* Fl */
233 { parse_mdoc_Fn, 0 }, /* Fn */
234 { NULL, TYPE_Ft }, /* Ft */
235 { NULL, TYPE_Ic }, /* Ic */
236 { NULL, TYPE_In }, /* In */
237 { NULL, TYPE_Li }, /* Li */
238 { parse_mdoc_Nd, 0 }, /* Nd */
239 { parse_mdoc_Nm, 0 }, /* Nm */
240 { NULL, 0 }, /* Op */
241 { NULL, 0 }, /* Ot */
242 { NULL, TYPE_Pa }, /* Pa */
243 { NULL, 0 }, /* Rv */
244 { NULL, TYPE_St }, /* St */
245 { parse_mdoc_Va, TYPE_Va }, /* Va */
246 { parse_mdoc_Va, TYPE_Vt }, /* Vt */
247 { parse_mdoc_Xr, 0 }, /* Xr */
248 { NULL, 0 }, /* %A */
249 { NULL, 0 }, /* %B */
250 { NULL, 0 }, /* %D */
251 { NULL, 0 }, /* %I */
252 { NULL, 0 }, /* %J */
253 { NULL, 0 }, /* %N */
254 { NULL, 0 }, /* %O */
255 { NULL, 0 }, /* %P */
256 { NULL, 0 }, /* %R */
257 { NULL, 0 }, /* %T */
258 { NULL, 0 }, /* %V */
259 { NULL, 0 }, /* Ac */
260 { NULL, 0 }, /* Ao */
261 { NULL, 0 }, /* Aq */
262 { NULL, TYPE_At }, /* At */
263 { NULL, 0 }, /* Bc */
264 { NULL, 0 }, /* Bf */
265 { NULL, 0 }, /* Bo */
266 { NULL, 0 }, /* Bq */
267 { NULL, TYPE_Bsx }, /* Bsx */
268 { NULL, TYPE_Bx }, /* Bx */
269 { NULL, 0 }, /* Db */
270 { NULL, 0 }, /* Dc */
271 { NULL, 0 }, /* Do */
272 { NULL, 0 }, /* Dq */
273 { NULL, 0 }, /* Ec */
274 { NULL, 0 }, /* Ef */
275 { NULL, TYPE_Em }, /* Em */
276 { NULL, 0 }, /* Eo */
277 { NULL, TYPE_Fx }, /* Fx */
278 { NULL, TYPE_Ms }, /* Ms */
279 { NULL, 0 }, /* No */
280 { NULL, 0 }, /* Ns */
281 { NULL, TYPE_Nx }, /* Nx */
282 { NULL, TYPE_Ox }, /* Ox */
283 { NULL, 0 }, /* Pc */
284 { NULL, 0 }, /* Pf */
285 { NULL, 0 }, /* Po */
286 { NULL, 0 }, /* Pq */
287 { NULL, 0 }, /* Qc */
288 { NULL, 0 }, /* Ql */
289 { NULL, 0 }, /* Qo */
290 { NULL, 0 }, /* Qq */
291 { NULL, 0 }, /* Re */
292 { NULL, 0 }, /* Rs */
293 { NULL, 0 }, /* Sc */
294 { NULL, 0 }, /* So */
295 { NULL, 0 }, /* Sq */
296 { NULL, 0 }, /* Sm */
297 { NULL, 0 }, /* Sx */
298 { NULL, TYPE_Sy }, /* Sy */
299 { NULL, TYPE_Tn }, /* Tn */
300 { NULL, 0 }, /* Ux */
301 { NULL, 0 }, /* Xc */
302 { NULL, 0 }, /* Xo */
303 { parse_mdoc_Fo, 0 }, /* Fo */
304 { NULL, 0 }, /* Fc */
305 { NULL, 0 }, /* Oo */
306 { NULL, 0 }, /* Oc */
307 { NULL, 0 }, /* Bk */
308 { NULL, 0 }, /* Ek */
309 { NULL, 0 }, /* Bt */
310 { NULL, 0 }, /* Hf */
311 { NULL, 0 }, /* Fr */
312 { NULL, 0 }, /* Ud */
313 { NULL, TYPE_Lb }, /* Lb */
314 { NULL, 0 }, /* Lp */
315 { NULL, TYPE_Lk }, /* Lk */
316 { NULL, TYPE_Mt }, /* Mt */
317 { NULL, 0 }, /* Brq */
318 { NULL, 0 }, /* Bro */
319 { NULL, 0 }, /* Brc */
320 { NULL, 0 }, /* %C */
321 { NULL, 0 }, /* Es */
322 { NULL, 0 }, /* En */
323 { NULL, TYPE_Dx }, /* Dx */
324 { NULL, 0 }, /* %Q */
325 { NULL, 0 }, /* br */
326 { NULL, 0 }, /* sp */
327 { NULL, 0 }, /* %U */
328 { NULL, 0 }, /* Ta */
329 { NULL, 0 }, /* ll */
334 mandocdb(int argc, char *argv[])
338 const char *path_arg, *progname;
343 if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
345 return (int)MANDOCLEVEL_SYSERR;
349 memset(&conf, 0, sizeof(conf));
350 memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
353 * We accept a few different invocations.
354 * The CHECKOP macro makes sure that invocation styles don't
355 * clobber each other.
357 #define CHECKOP(_op, _ch) do \
358 if (OP_DEFAULT != (_op)) { \
359 warnx("-%c: Conflicting option", (_ch)); \
361 } while (/*CONSTCOND*/0)
366 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
391 mparse_options |= MPARSE_QUICK;
394 if (strcmp(optarg, "utf8")) {
395 warnx("-T%s: Unsupported output format",
403 dup2(STDOUT_FILENO, STDERR_FILENO);
413 /* Compatibility with espie@'s makewhatis. */
424 if (pledge("stdio rpath", NULL) == -1) {
426 return (int)MANDOCLEVEL_SYSERR;
431 if (OP_CONFFILE == op && argc > 0) {
432 warnx("-C: Too many arguments");
436 exitcode = (int)MANDOCLEVEL_OK;
438 mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
439 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
440 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
442 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
445 * Most of these deal with a specific directory.
446 * Jump into that directory first.
448 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
453 * The existing database is usable. Process
454 * all files specified on the command-line.
458 if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
460 exitcode = (int)MANDOCLEVEL_SYSERR;
466 for (i = 0; i < argc; i++)
472 * Database missing or corrupt.
473 * Recreate from scratch.
475 exitcode = (int)MANDOCLEVEL_OK;
484 dbclose(OP_DEFAULT == op ? 0 : 1);
487 * If we have arguments, use them as our manpaths.
488 * If we don't, grok from manpath(1) or however else
489 * manconf_parse() wants to do it.
492 conf.manpath.paths = mandoc_reallocarray(NULL,
493 argc, sizeof(char *));
494 conf.manpath.sz = (size_t)argc;
495 for (i = 0; i < argc; i++)
496 conf.manpath.paths[i] = mandoc_strdup(argv[i]);
498 manconf_parse(&conf, path_arg, NULL, NULL);
500 if (conf.manpath.sz == 0) {
501 exitcode = (int)MANDOCLEVEL_BADARG;
502 say("", "Empty manpath");
506 * First scan the tree rooted at a base directory, then
507 * build a new database and finally move it into place.
508 * Ignore zero-length directories and strip trailing
511 for (j = 0; j < conf.manpath.sz; j++) {
512 sz = strlen(conf.manpath.paths[j]);
513 if (sz && conf.manpath.paths[j][sz - 1] == '/')
514 conf.manpath.paths[j][--sz] = '\0';
519 mandoc_ohash_init(&mpages, 6,
520 offsetof(struct mpage, inodev));
521 mandoc_ohash_init(&mlinks, 6,
522 offsetof(struct mlink, file));
525 if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
533 if (warnings && !nodb &&
534 ! (MPARSE_QUICK & mparse_options))
538 if (j + 1 < conf.manpath.sz) {
540 ohash_delete(&mpages);
541 ohash_delete(&mlinks);
550 ohash_delete(&mpages);
551 ohash_delete(&mlinks);
554 progname = getprogname();
555 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
556 " %s [-aDnpQ] [-Tutf8] dir ...\n"
557 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
558 " %s [-Dnp] -u dir [file ...]\n"
559 " %s [-Q] -t file ...\n",
560 progname, progname, progname, progname, progname);
562 return (int)MANDOCLEVEL_BADARG;
566 * Scan a directory tree rooted at "basedir" for manpages.
567 * We use fts(), scanning directory parts along the way for clues to our
568 * section and architecture.
570 * If use_all has been specified, grok all files.
571 * If not, sanitise paths to the following:
573 * [./]man*[/<arch>]/<name>.<section>
575 * [./]cat<section>[/<arch>]/<name>.0
577 * TODO: accomodate for multi-language directories.
587 char *dsec, *arch, *fsec, *cp;
592 argv[1] = (char *)NULL;
594 f = fts_open((char * const *)argv,
595 FTS_PHYSICAL | FTS_NOCHDIR, NULL);
597 exitcode = (int)MANDOCLEVEL_SYSERR;
598 say("", "&fts_open");
605 while ((ff = fts_read(f)) != NULL) {
606 path = ff->fts_path + 2;
607 switch (ff->fts_info) {
610 * Symbolic links require various sanity checks,
611 * then get handled just like regular files.
614 if (realpath(path, buf) == NULL) {
616 say(path, "&realpath");
619 if (strstr(buf, basedir) != buf
621 && strstr(buf, HOMEBREWDIR) != buf
624 if (warnings) say("",
625 "%s: outside base directory", buf);
628 /* Use logical inode to avoid mpages dupe. */
629 if (stat(path, ff->fts_statp) == -1) {
637 * If we're a regular file, add an mlink by using the
638 * stored directory data and handling the filename.
641 if ( ! strcmp(path, MANDOC_DB))
643 if ( ! use_all && ff->fts_level < 2) {
645 say(path, "Extraneous file");
650 while (fsec == NULL) {
651 fsec = strrchr(ff->fts_name, '.');
652 if (fsec == NULL || strcmp(fsec+1, "gz"))
662 "No filename suffix");
665 } else if ( ! strcmp(++fsec, "html")) {
667 say(path, "Skip html");
669 } else if ( ! strcmp(fsec, "ps")) {
671 say(path, "Skip ps");
673 } else if ( ! strcmp(fsec, "pdf")) {
675 say(path, "Skip pdf");
677 } else if ( ! use_all &&
678 ((dform == FORM_SRC &&
679 strncmp(fsec, dsec, strlen(dsec))) ||
680 (dform == FORM_CAT && strcmp(fsec, "0")))) {
682 say(path, "Wrong filename suffix");
687 mlink = mandoc_calloc(1, sizeof(struct mlink));
688 if (strlcpy(mlink->file, path,
689 sizeof(mlink->file)) >=
690 sizeof(mlink->file)) {
691 say(path, "Filename too long");
695 mlink->dform = dform;
698 mlink->name = ff->fts_name;
701 mlink_add(mlink, ff->fts_statp);
710 say(path, "Not a regular file");
714 switch (ff->fts_level) {
716 /* Ignore the root directory. */
720 * This might contain manX/ or catX/.
721 * Try to infer this from the name.
722 * If we're not in use_all, enforce it.
725 if (ff->fts_info == FTS_DP) {
731 if ( ! strncmp(cp, "man", 3)) {
734 } else if ( ! strncmp(cp, "cat", 3)) {
742 if (dsec != NULL || use_all)
746 say(path, "Unknown directory part");
747 fts_set(f, ff, FTS_SKIP);
751 * Possibly our architecture.
752 * If we're descending, keep tabs on it.
754 if (ff->fts_info != FTS_DP && dsec != NULL)
760 if (ff->fts_info == FTS_DP || use_all)
763 say(path, "Extraneous directory part");
764 fts_set(f, ff, FTS_SKIP);
774 * Add a file to the mlinks table.
775 * Do not verify that it's a "valid" looking manpage (we'll do that
778 * Try to infer the manual section, architecture, and page name from the
779 * path, assuming it looks like
781 * [./]man*[/<arch>]/<name>.<section>
783 * [./]cat<section>[/<arch>]/<name>.0
785 * See treescan() for the fts(3) version of this.
788 filescan(const char *file)
797 if (0 == strncmp(file, "./", 2))
801 * We have to do lstat(2) before realpath(3) loses
802 * the information whether this is a symbolic link.
803 * We need to know that because for symbolic links,
804 * we want to use the orginal file name, while for
805 * regular files, we want to use the real path.
807 if (-1 == lstat(file, &st)) {
808 exitcode = (int)MANDOCLEVEL_BADARG;
811 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
812 exitcode = (int)MANDOCLEVEL_BADARG;
813 say(file, "Not a regular file");
818 * We have to resolve the file name to the real path
819 * in any case for the base directory check.
821 if (NULL == realpath(file, buf)) {
822 exitcode = (int)MANDOCLEVEL_BADARG;
823 say(file, "&realpath");
829 else if (strstr(buf, basedir) == buf)
830 start = buf + strlen(basedir);
832 else if (strstr(buf, HOMEBREWDIR) == buf)
836 exitcode = (int)MANDOCLEVEL_BADARG;
837 say("", "%s: outside base directory", buf);
842 * Now we are sure the file is inside our tree.
843 * If it is a symbolic link, ignore the real path
844 * and use the original name.
845 * This implies passing stuff like "cat1/../man1/foo.1"
846 * on the command line won't work. So don't do that.
847 * Note the stat(2) can still fail if the link target
850 if (S_IFLNK & st.st_mode) {
851 if (-1 == stat(buf, &st)) {
852 exitcode = (int)MANDOCLEVEL_BADARG;
856 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
857 say(file, "Filename too long");
861 if (OP_TEST != op && strstr(buf, basedir) == buf)
862 start += strlen(basedir);
865 mlink = mandoc_calloc(1, sizeof(struct mlink));
866 mlink->dform = FORM_NONE;
867 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
868 sizeof(mlink->file)) {
869 say(start, "Filename too long");
875 * First try to guess our directory structure.
876 * If we find a separator, try to look for man* or cat*.
877 * If we find one of these and what's underneath is a directory,
878 * assume it's an architecture.
880 if (NULL != (p = strchr(start, '/'))) {
882 if (0 == strncmp(start, "man", 3)) {
883 mlink->dform = FORM_SRC;
884 mlink->dsec = start + 3;
885 } else if (0 == strncmp(start, "cat", 3)) {
886 mlink->dform = FORM_CAT;
887 mlink->dsec = start + 3;
891 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
899 * Now check the file suffix.
900 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
902 p = strrchr(start, '\0');
903 while (p-- > start && '/' != *p && '.' != *p)
912 * Now try to parse the name.
913 * Use the filename portion of the path.
916 if (NULL != (p = strrchr(start, '/'))) {
920 mlink_add(mlink, &st);
924 mlink_add(struct mlink *mlink, const struct stat *st)
926 struct inodev inodev;
930 assert(NULL != mlink->file);
932 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
933 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
934 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
935 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
937 if ('0' == *mlink->fsec) {
939 mlink->fsec = mandoc_strdup(mlink->dsec);
940 mlink->fform = FORM_CAT;
941 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
942 mlink->fform = FORM_SRC;
944 mlink->fform = FORM_NONE;
946 slot = ohash_qlookup(&mlinks, mlink->file);
947 assert(NULL == ohash_find(&mlinks, slot));
948 ohash_insert(&mlinks, slot, mlink);
950 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
951 inodev.st_ino = st->st_ino;
952 inodev.st_dev = st->st_dev;
953 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
954 sizeof(struct inodev), inodev.st_ino);
955 mpage = ohash_find(&mpages, slot);
957 mpage = mandoc_calloc(1, sizeof(struct mpage));
958 mpage->inodev.st_ino = inodev.st_ino;
959 mpage->inodev.st_dev = inodev.st_dev;
960 ohash_insert(&mpages, slot, mpage);
962 mlink->next = mpage->mlinks;
963 mpage->mlinks = mlink;
964 mlink->mpage = mpage;
968 mlink_free(struct mlink *mlink)
985 mpage = ohash_first(&mpages, &slot);
986 while (NULL != mpage) {
987 while (NULL != (mlink = mpage->mlinks)) {
988 mpage->mlinks = mlink->next;
996 mpage = ohash_next(&mpages, &slot);
1001 * For each mlink to the mpage, check whether the path looks like
1002 * it is formatted, and if it does, check whether a source manual
1003 * exists by the same name, ignoring the suffix.
1004 * If both conditions hold, drop the mlink.
1007 mlinks_undupe(struct mpage *mpage)
1010 struct mlink **prev;
1011 struct mlink *mlink;
1014 mpage->form = FORM_CAT;
1015 prev = &mpage->mlinks;
1016 while (NULL != (mlink = *prev)) {
1017 if (FORM_CAT != mlink->dform) {
1018 mpage->form = FORM_NONE;
1021 (void)strlcpy(buf, mlink->file, sizeof(buf));
1022 bufp = strstr(buf, "cat");
1023 assert(NULL != bufp);
1024 memcpy(bufp, "man", 3);
1025 if (NULL != (bufp = strrchr(buf, '.')))
1027 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1028 if (NULL == ohash_find(&mlinks,
1029 ohash_qlookup(&mlinks, buf)))
1032 say(mlink->file, "Man source exists: %s", buf);
1035 *prev = mlink->next;
1039 prev = &(*prev)->next;
1044 mlink_check(struct mpage *mpage, struct mlink *mlink)
1050 * Check whether the manual section given in a file
1051 * agrees with the directory where the file is located.
1052 * Some manuals have suffixes like (3p) on their
1053 * section number either inside the file or in the
1054 * directory name, some are linked into more than one
1055 * section, like encrypt(1) = makekey(8).
1058 if (FORM_SRC == mpage->form &&
1059 strcasecmp(mpage->sec, mlink->dsec))
1060 say(mlink->file, "Section \"%s\" manual in %s directory",
1061 mpage->sec, mlink->dsec);
1064 * Manual page directories exist for each kernel
1065 * architecture as returned by machine(1).
1066 * However, many manuals only depend on the
1067 * application architecture as returned by arch(1).
1068 * For example, some (2/ARM) manuals are shared
1069 * across the "armish" and "zaurus" kernel
1071 * A few manuals are even shared across completely
1072 * different architectures, for example fdformat(1)
1073 * on amd64, i386, sparc, and sparc64.
1076 if (strcasecmp(mpage->arch, mlink->arch))
1077 say(mlink->file, "Architecture \"%s\" manual in "
1078 "\"%s\" directory", mpage->arch, mlink->arch);
1082 * parse_cat() doesn't set NAME_TITLE yet.
1085 if (FORM_CAT == mpage->form)
1089 * Check whether this mlink
1090 * appears as a name in the NAME section.
1093 slot = ohash_qlookup(&names, mlink->name);
1094 str = ohash_find(&names, slot);
1095 assert(NULL != str);
1096 if ( ! (NAME_TITLE & str->mask))
1097 say(mlink->file, "Name missing in NAME section");
1101 * Run through the files in the global vector "mpages"
1102 * and add them to the database specified in "basedir".
1104 * This handles the parsing scheme itself, using the cues of directory
1105 * and filename to determine whether the file is parsable or not.
1108 mpages_merge(struct mparse *mp)
1111 struct mpage *mpage, *mpage_dest;
1112 struct mlink *mlink, *mlink_dest;
1113 struct roff_man *man;
1120 SQL_EXEC("BEGIN TRANSACTION");
1122 mpage = ohash_first(&mpages, &pslot);
1123 while (mpage != NULL) {
1124 mlinks_undupe(mpage);
1125 if ((mlink = mpage->mlinks) == NULL) {
1126 mpage = ohash_next(&mpages, &pslot);
1130 name_mask = NAME_MASK;
1131 mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1132 mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1137 if ((fd = mparse_open(mp, mlink->file)) == -1) {
1138 say(mlink->file, "&open");
1143 * Interpret the file as mdoc(7) or man(7) source
1144 * code, unless it is known to be formatted.
1146 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1147 mparse_readfd(mp, fd, mlink->file);
1149 mparse_result(mp, &man, &sodest);
1152 if (sodest != NULL) {
1153 mlink_dest = ohash_find(&mlinks,
1154 ohash_qlookup(&mlinks, sodest));
1155 if (mlink_dest == NULL) {
1156 mandoc_asprintf(&cp, "%s.gz", sodest);
1157 mlink_dest = ohash_find(&mlinks,
1158 ohash_qlookup(&mlinks, cp));
1161 if (mlink_dest != NULL) {
1163 /* The .so target exists. */
1165 mpage_dest = mlink_dest->mpage;
1167 mlink->mpage = mpage_dest;
1170 * If the target was already
1171 * processed, add the links
1172 * to the database now.
1173 * Otherwise, this will
1174 * happen when we come
1178 if (mpage_dest->pageid)
1179 dbadd_mlink_name(mlink);
1181 if (mlink->next == NULL)
1183 mlink = mlink->next;
1186 /* Move all links to the target. */
1188 mlink->next = mlink_dest->next;
1189 mlink_dest->next = mpage->mlinks;
1190 mpage->mlinks = NULL;
1193 } else if (man != NULL && man->macroset == MACROSET_MDOC) {
1195 mpage->form = FORM_SRC;
1196 mpage->sec = man->meta.msec;
1197 mpage->sec = mandoc_strdup(
1198 mpage->sec == NULL ? "" : mpage->sec);
1199 mpage->arch = man->meta.arch;
1200 mpage->arch = mandoc_strdup(
1201 mpage->arch == NULL ? "" : mpage->arch);
1202 mpage->title = mandoc_strdup(man->meta.title);
1203 } else if (man != NULL && man->macroset == MACROSET_MAN) {
1205 mpage->form = FORM_SRC;
1206 mpage->sec = mandoc_strdup(man->meta.msec);
1207 mpage->arch = mandoc_strdup(mlink->arch);
1208 mpage->title = mandoc_strdup(man->meta.title);
1210 mpage->form = FORM_CAT;
1211 mpage->sec = mandoc_strdup(mlink->dsec);
1212 mpage->arch = mandoc_strdup(mlink->arch);
1213 mpage->title = mandoc_strdup(mlink->name);
1215 putkey(mpage, mpage->sec, TYPE_sec);
1216 if (*mpage->arch != '\0')
1217 putkey(mpage, mpage->arch, TYPE_arch);
1219 for ( ; mlink != NULL; mlink = mlink->next) {
1220 if ('\0' != *mlink->dsec)
1221 putkey(mpage, mlink->dsec, TYPE_sec);
1222 if ('\0' != *mlink->fsec)
1223 putkey(mpage, mlink->fsec, TYPE_sec);
1224 putkey(mpage, '\0' == *mlink->arch ?
1225 any : mlink->arch, TYPE_arch);
1226 putkey(mpage, mlink->name, NAME_FILE);
1229 assert(mpage->desc == NULL);
1230 if (man != NULL && man->macroset == MACROSET_MDOC)
1231 parse_mdoc(mpage, &man->meta, man->first);
1232 else if (man != NULL)
1233 parse_man(mpage, &man->meta, man->first);
1235 parse_cat(mpage, fd);
1236 if (mpage->desc == NULL)
1237 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1239 if (warnings && !use_all)
1240 for (mlink = mpage->mlinks; mlink;
1241 mlink = mlink->next)
1242 mlink_check(mpage, mlink);
1245 mlink = mpage->mlinks;
1248 ohash_delete(&strings);
1249 ohash_delete(&names);
1250 mpage = ohash_next(&mpages, &pslot);
1254 SQL_EXEC("END TRANSACTION");
1261 const char *name, *sec, *arch, *key;
1263 sqlite3_prepare_v2(db,
1264 "SELECT name, sec, arch, key FROM ("
1265 "SELECT name AS key, pageid FROM names "
1266 "WHERE bits & ? AND NOT EXISTS ("
1267 "SELECT pageid FROM mlinks "
1268 "WHERE mlinks.pageid == names.pageid "
1269 "AND mlinks.name == names.name"
1272 "SELECT sec, arch, name, pageid FROM mlinks "
1274 ") USING (pageid);",
1277 if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
1278 say("", "%s", sqlite3_errmsg(db));
1280 while (sqlite3_step(stmt) == SQLITE_ROW) {
1281 name = (const char *)sqlite3_column_text(stmt, 0);
1282 sec = (const char *)sqlite3_column_text(stmt, 1);
1283 arch = (const char *)sqlite3_column_text(stmt, 2);
1284 key = (const char *)sqlite3_column_text(stmt, 3);
1285 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1286 '\0' == *arch ? "" : "/",
1287 '\0' == *arch ? "" : arch, key);
1289 sqlite3_finalize(stmt);
1293 parse_cat(struct mpage *mpage, int fd)
1296 char *line, *p, *title;
1297 size_t linesz, plen, titlesz;
1301 stream = (-1 == fd) ?
1302 fopen(mpage->mlinks->file, "r") :
1304 if (NULL == stream) {
1308 say(mpage->mlinks->file, "&fopen");
1315 /* Skip to first blank line. */
1317 while (getline(&line, &linesz, stream) != -1)
1322 * Assume the first line that is not indented
1323 * is the first section header. Skip to it.
1326 while (getline(&line, &linesz, stream) != -1)
1327 if (*line != '\n' && *line != ' ')
1331 * Read up until the next section into a buffer.
1332 * Strip the leading and trailing newline from each read line,
1333 * appending a trailing space.
1334 * Ignore empty (whitespace-only) lines.
1340 while ((len = getline(&line, &linesz, stream)) != -1) {
1344 while (isspace((unsigned char)line[offs]))
1346 if (line[offs] == '\0')
1348 title = mandoc_realloc(title, titlesz + len - offs);
1349 memcpy(title + titlesz, line + offs, len - offs);
1350 titlesz += len - offs;
1351 title[titlesz - 1] = ' ';
1356 * If no page content can be found, or the input line
1357 * is already the next section header, or there is no
1358 * trailing newline, reuse the page title as the page
1362 if (NULL == title || '\0' == *title) {
1364 say(mpage->mlinks->file,
1365 "Cannot find NAME section");
1371 title[titlesz - 1] = '\0';
1374 * Skip to the first dash.
1375 * Use the remaining line as the description (no more than 70
1379 if (NULL != (p = strstr(title, "- "))) {
1380 for (p += 2; ' ' == *p || '\b' == *p; p++)
1381 /* Skip to next word. */ ;
1384 say(mpage->mlinks->file,
1385 "No dash in title line");
1391 /* Strip backspace-encoding from line. */
1393 while (NULL != (line = memchr(p, '\b', plen))) {
1396 memmove(line, line + 1, plen--);
1399 memmove(line - 1, line + 1, plen - len);
1403 mpage->desc = mandoc_strdup(p);
1409 * Put a type/word pair into the word database for this particular file.
1412 putkey(const struct mpage *mpage, char *value, uint64_t type)
1416 assert(NULL != value);
1417 if (TYPE_arch == type)
1418 for (cp = value; *cp; cp++)
1419 if (isupper((unsigned char)*cp))
1420 *cp = _tolower((unsigned char)*cp);
1421 putkeys(mpage, value, strlen(value), type);
1425 * Grok all nodes at or below a certain mdoc node into putkey().
1428 putmdockey(const struct mpage *mpage,
1429 const struct roff_node *n, uint64_t m)
1432 for ( ; NULL != n; n = n->next) {
1433 if (NULL != n->child)
1434 putmdockey(mpage, n->child, m);
1435 if (n->type == ROFFT_TEXT)
1436 putkey(mpage, n->string, m);
1441 parse_man(struct mpage *mpage, const struct roff_meta *meta,
1442 const struct roff_node *n)
1444 const struct roff_node *head, *body;
1445 char *start, *title;
1453 * We're only searching for one thing: the first text child in
1454 * the BODY of a NAME section. Since we don't keep track of
1455 * sections in -man, run some hoops to find out whether we're in
1456 * the correct section or not.
1459 if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1461 if ((head = body->parent->head) != NULL &&
1462 (head = head->child) != NULL &&
1463 head->next == NULL &&
1464 head->type == ROFFT_TEXT &&
1465 strcmp(head->string, "NAME") == 0 &&
1466 body->child != NULL) {
1469 * Suck the entire NAME section into memory.
1470 * Yes, we might run away.
1471 * But too many manuals have big, spread-out
1472 * NAME sections over many lines.
1476 deroff(&title, body);
1481 * Go through a special heuristic dance here.
1482 * Conventionally, one or more manual names are
1483 * comma-specified prior to a whitespace, then a
1484 * dash, then a description. Try to puzzle out
1485 * the name parts here.
1490 sz = strcspn(start, " ,");
1491 if ('\0' == start[sz])
1498 * Assume a stray trailing comma in the
1499 * name list if a name begins with a dash.
1502 if ('-' == start[0] ||
1503 ('\\' == start[0] && '-' == start[1]))
1506 putkey(mpage, start, NAME_TITLE);
1507 if ( ! (mpage->name_head_done ||
1508 strcasecmp(start, meta->title))) {
1509 putkey(mpage, start, NAME_HEAD);
1510 mpage->name_head_done = 1;
1518 assert(',' == byte);
1520 while (' ' == *start)
1524 if (start == title) {
1525 putkey(mpage, start, NAME_TITLE);
1526 if ( ! (mpage->name_head_done ||
1527 strcasecmp(start, meta->title))) {
1528 putkey(mpage, start, NAME_HEAD);
1529 mpage->name_head_done = 1;
1535 while (isspace((unsigned char)*start))
1538 if (0 == strncmp(start, "-", 1))
1540 else if (0 == strncmp(start, "\\-\\-", 4))
1542 else if (0 == strncmp(start, "\\-", 2))
1544 else if (0 == strncmp(start, "\\(en", 4))
1546 else if (0 == strncmp(start, "\\(em", 4))
1549 while (' ' == *start)
1552 mpage->desc = mandoc_strdup(start);
1558 for (n = n->child; n; n = n->next) {
1559 if (NULL != mpage->desc)
1561 parse_man(mpage, meta, n);
1566 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1567 const struct roff_node *n)
1571 for (n = n->child; NULL != n; n = n->next) {
1578 if (NULL != mdocs[n->tok].fp)
1579 if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1581 if (mdocs[n->tok].mask)
1582 putmdockey(mpage, n->child,
1583 mdocs[n->tok].mask);
1586 assert(n->type != ROFFT_ROOT);
1589 if (NULL != n->child)
1590 parse_mdoc(mpage, meta, n);
1595 parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
1596 const struct roff_node *n)
1601 if (SEC_SYNOPSIS != n->sec ||
1602 NULL == (n = n->child) ||
1603 n->type != ROFFT_TEXT)
1607 * Only consider those `Fd' macro fields that begin with an
1608 * "inclusion" token (versus, e.g., #define).
1611 if (strcmp("#include", n->string))
1614 if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
1618 * Strip away the enclosing angle brackets and make sure we're
1623 if ('<' == *start || '"' == *start)
1626 if (0 == (sz = strlen(start)))
1629 end = &start[(int)sz - 1];
1630 if ('>' == *end || '"' == *end)
1634 putkeys(mpage, start, end - start + 1, TYPE_In);
1639 parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
1644 if (n->type != ROFFT_TEXT)
1647 /* Skip function pointer punctuation. */
1650 while (*cp == '(' || *cp == '*')
1652 sz = strcspn(cp, "()");
1654 putkeys(mpage, cp, sz, TYPE_Fn);
1655 if (n->sec == SEC_SYNOPSIS)
1656 putkeys(mpage, cp, sz, NAME_SYN);
1660 parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
1661 const struct roff_node *n)
1664 if (n->child == NULL)
1667 parse_mdoc_fname(mpage, n->child);
1669 for (n = n->child->next; n != NULL; n = n->next)
1670 if (n->type == ROFFT_TEXT)
1671 putkey(mpage, n->string, TYPE_Fa);
1677 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1678 const struct roff_node *n)
1681 if (n->type != ROFFT_HEAD)
1684 if (n->child != NULL)
1685 parse_mdoc_fname(mpage, n->child);
1691 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1692 const struct roff_node *n)
1696 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1699 if (n->child != NULL &&
1700 n->child->next == NULL &&
1701 n->child->type == ROFFT_TEXT)
1707 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1708 n->type == ROFFT_BODY ? TYPE_Va : 0));
1716 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1717 const struct roff_node *n)
1721 if (NULL == (n = n->child))
1724 if (NULL == n->next) {
1725 putkey(mpage, n->string, TYPE_Xr);
1729 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1730 putkey(mpage, cp, TYPE_Xr);
1736 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1737 const struct roff_node *n)
1740 if (n->type == ROFFT_BODY)
1741 deroff(&mpage->desc, n);
1746 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1747 const struct roff_node *n)
1750 if (SEC_NAME == n->sec)
1751 putmdockey(mpage, n->child, NAME_TITLE);
1752 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1753 if (n->child == NULL)
1754 putkey(mpage, meta->name, NAME_SYN);
1756 putmdockey(mpage, n->child, NAME_SYN);
1758 if ( ! (mpage->name_head_done ||
1759 n->child == NULL || n->child->string == NULL ||
1760 strcasecmp(n->child->string, meta->title))) {
1761 putkey(mpage, n->child->string, ROFFT_HEAD);
1762 mpage->name_head_done = 1;
1768 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1769 const struct roff_node *n)
1772 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1776 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1777 const struct roff_node *n)
1780 return n->type == ROFFT_HEAD;
1784 * Add a string to the hash table for the current manual.
1785 * Each string has a bitmask telling which macros it belongs to.
1786 * When we finish the manual, we'll dump the table.
1789 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1800 mustfree = render_string(&cp, &sz);
1806 name_mask &= ~NAME_FIRST;
1808 say(mpage->mlinks->file,
1809 "Adding name %*s, bits=%d", sz, cp, v);
1813 for (i = 0; i < mansearch_keymax; i++)
1814 if ((uint64_t)1 << i & v)
1815 say(mpage->mlinks->file,
1816 "Adding key %s=%*s",
1817 mansearch_keynames[i], sz, cp);
1821 slot = ohash_qlookupi(htab, cp, &end);
1822 s = ohash_find(htab, slot);
1824 if (NULL != s && mpage == s->mpage) {
1827 } else if (NULL == s) {
1828 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1829 memcpy(s->key, cp, sz);
1830 ohash_insert(htab, slot, s);
1840 * Take a Unicode codepoint and produce its UTF-8 encoding.
1841 * This isn't the best way to do this, but it works.
1842 * The magic numbers are from the UTF-8 packaging.
1843 * They're not as scary as they seem: read the UTF-8 spec for details.
1846 utf8(unsigned int cp, char out[7])
1851 if (cp <= 0x0000007F) {
1854 } else if (cp <= 0x000007FF) {
1856 out[0] = (cp >> 6 & 31) | 192;
1857 out[1] = (cp & 63) | 128;
1858 } else if (cp <= 0x0000FFFF) {
1860 out[0] = (cp >> 12 & 15) | 224;
1861 out[1] = (cp >> 6 & 63) | 128;
1862 out[2] = (cp & 63) | 128;
1863 } else if (cp <= 0x001FFFFF) {
1865 out[0] = (cp >> 18 & 7) | 240;
1866 out[1] = (cp >> 12 & 63) | 128;
1867 out[2] = (cp >> 6 & 63) | 128;
1868 out[3] = (cp & 63) | 128;
1869 } else if (cp <= 0x03FFFFFF) {
1871 out[0] = (cp >> 24 & 3) | 248;
1872 out[1] = (cp >> 18 & 63) | 128;
1873 out[2] = (cp >> 12 & 63) | 128;
1874 out[3] = (cp >> 6 & 63) | 128;
1875 out[4] = (cp & 63) | 128;
1876 } else if (cp <= 0x7FFFFFFF) {
1878 out[0] = (cp >> 30 & 1) | 252;
1879 out[1] = (cp >> 24 & 63) | 128;
1880 out[2] = (cp >> 18 & 63) | 128;
1881 out[3] = (cp >> 12 & 63) | 128;
1882 out[4] = (cp >> 6 & 63) | 128;
1883 out[5] = (cp & 63) | 128;
1892 * If the string contains escape sequences,
1893 * replace it with an allocated rendering and return 1,
1894 * such that the caller can free it after use.
1895 * Otherwise, do nothing and return 0.
1898 render_string(char **public, size_t *psz)
1900 const char *src, *scp, *addcp, *seq;
1902 size_t ssz, dsz, addsz;
1903 char utfbuf[7], res[6];
1904 int seqlen, unicode;
1908 res[2] = ASCII_NBRSP;
1909 res[3] = ASCII_HYPH;
1910 res[4] = ASCII_BREAK;
1913 src = scp = *public;
1918 while (scp < src + *psz) {
1920 /* Leave normal characters unchanged. */
1922 if (strchr(res, *scp) == NULL) {
1930 * Found something that requires replacing,
1931 * make sure we have a destination buffer.
1935 dst = mandoc_malloc(ssz + 1);
1937 memcpy(dst, src, dsz);
1940 /* Handle single-char special characters. */
1961 * Found an escape sequence.
1962 * Read past the slash, then parse it.
1963 * Ignore everything except characters.
1967 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1971 * Render the special character
1972 * as either UTF-8 or ASCII.
1976 unicode = mchars_spec2cp(seq, seqlen);
1979 addsz = utf8(unicode, utfbuf);
1984 addcp = mchars_spec2str(seq, seqlen, &addsz);
1987 if (*addcp == ASCII_NBRSP) {
1993 /* Copy the rendered glyph into the stream. */
1996 dst = mandoc_realloc(dst, ssz + 1);
1997 memcpy(dst + dsz, addcp, addsz);
2005 /* Trim trailing whitespace and NUL-terminate. */
2007 while (*psz > 0 && (*public)[*psz - 1] == ' ')
2010 (*public)[*psz] = '\0';
2017 dbadd_mlink(const struct mlink *mlink)
2022 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2023 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2024 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
2025 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
2026 SQL_STEP(stmts[STMT_INSERT_LINK]);
2027 sqlite3_reset(stmts[STMT_INSERT_LINK]);
2031 dbadd_mlink_name(const struct mlink *mlink)
2039 SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2040 bits = NAME_FILE & NAME_MASK;
2041 if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2042 bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2043 sqlite3_reset(stmts[STMT_SELECT_NAME]);
2047 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
2048 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2049 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2050 SQL_STEP(stmts[STMT_INSERT_NAME]);
2051 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2055 * Flush the current page's terms (and their bits) into the database.
2056 * Wrap the entire set of additions in a transaction to make sqlite be a
2058 * Also, handle escape sequences at the last possible moment.
2061 dbadd(struct mpage *mpage)
2063 struct mlink *mlink;
2070 mlink = mpage->mlinks;
2073 for (key = ohash_first(&names, &slot); NULL != key;
2074 key = ohash_next(&names, &slot))
2076 for (key = ohash_first(&strings, &slot); NULL != key;
2077 key = ohash_next(&strings, &slot))
2081 while (NULL != mlink) {
2082 fputs(mlink->name, stdout);
2083 if (NULL == mlink->next ||
2084 strcmp(mlink->dsec, mlink->next->dsec) ||
2085 strcmp(mlink->fsec, mlink->next->fsec) ||
2086 strcmp(mlink->arch, mlink->next->arch)) {
2088 if ('\0' == *mlink->dsec)
2089 fputs(mlink->fsec, stdout);
2091 fputs(mlink->dsec, stdout);
2092 if ('\0' != *mlink->arch)
2093 printf("/%s", mlink->arch);
2096 mlink = mlink->next;
2098 fputs(", ", stdout);
2100 printf(" - %s\n", mpage->desc);
2105 say(mlink->file, "Adding to database");
2109 mustfree = render_string(&cp, &i);
2111 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
2112 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
2113 SQL_STEP(stmts[STMT_INSERT_PAGE]);
2114 mpage->pageid = sqlite3_last_insert_rowid(db);
2115 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2119 while (NULL != mlink) {
2121 mlink = mlink->next;
2123 mlink = mpage->mlinks;
2125 for (key = ohash_first(&names, &slot); NULL != key;
2126 key = ohash_next(&names, &slot)) {
2127 assert(key->mpage == mpage);
2129 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2130 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
2131 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
2132 SQL_STEP(stmts[STMT_INSERT_NAME]);
2133 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2136 for (key = ohash_first(&strings, &slot); NULL != key;
2137 key = ohash_next(&strings, &slot)) {
2138 assert(key->mpage == mpage);
2140 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
2141 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
2142 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
2143 SQL_STEP(stmts[STMT_INSERT_KEY]);
2144 sqlite3_reset(stmts[STMT_INSERT_KEY]);
2152 struct mpage *mpage;
2153 struct mlink *mlink;
2158 SQL_EXEC("BEGIN TRANSACTION");
2160 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2161 mpage = ohash_next(&mpages, &slot)) {
2162 mlink = mpage->mlinks;
2164 say(mlink->file, "Deleting from database");
2167 for ( ; NULL != mlink; mlink = mlink->next) {
2169 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2171 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2173 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2175 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2176 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2181 SQL_EXEC("END TRANSACTION");
2185 * Close an existing database and its prepared statements.
2186 * If "real" is not set, rename the temporary file into the real one.
2198 for (i = 0; i < STMT__MAX; i++) {
2199 sqlite3_finalize(stmts[i]);
2209 if ('\0' == *tempfilename) {
2210 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2211 exitcode = (int)MANDOCLEVEL_SYSERR;
2212 say(MANDOC_DB, "&rename");
2217 switch (child = fork()) {
2219 exitcode = (int)MANDOCLEVEL_SYSERR;
2220 say("", "&fork cmp");
2223 execlp("cmp", "cmp", "-s",
2224 tempfilename, MANDOC_DB, (char *)NULL);
2225 say("", "&exec cmp");
2230 if (-1 == waitpid(child, &status, 0)) {
2231 exitcode = (int)MANDOCLEVEL_SYSERR;
2232 say("", "&wait cmp");
2233 } else if (WIFSIGNALED(status)) {
2234 exitcode = (int)MANDOCLEVEL_SYSERR;
2235 say("", "cmp died from signal %d", WTERMSIG(status));
2236 } else if (WEXITSTATUS(status)) {
2237 exitcode = (int)MANDOCLEVEL_SYSERR;
2239 "Data changed, but cannot replace database");
2242 *strrchr(tempfilename, '/') = '\0';
2243 switch (child = fork()) {
2245 exitcode = (int)MANDOCLEVEL_SYSERR;
2246 say("", "&fork rm");
2249 execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
2250 say("", "&exec rm");
2251 exit((int)MANDOCLEVEL_SYSERR);
2255 if (-1 == waitpid(child, &status, 0)) {
2256 exitcode = (int)MANDOCLEVEL_SYSERR;
2257 say("", "&wait rm");
2258 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2259 exitcode = (int)MANDOCLEVEL_SYSERR;
2260 say("", "%s: Cannot remove temporary directory",
2266 * This is straightforward stuff.
2267 * Open a database connection to a "temporary" database, then open a set
2268 * of prepared statements we'll use over and over again.
2269 * If "real" is set, we use the existing database; if not, we truncate a
2271 * Must be matched by dbclose().
2282 *tempfilename = '\0';
2283 ofl = SQLITE_OPEN_READWRITE;
2286 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2287 if (SQLITE_OK != rc) {
2288 exitcode = (int)MANDOCLEVEL_SYSERR;
2289 if (SQLITE_CANTOPEN != rc)
2290 say(MANDOC_DB, "%s", sqlite3_errstr(rc));
2293 goto prepare_statements;
2296 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2298 remove(MANDOC_DB "~");
2299 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2300 if (SQLITE_OK == rc)
2302 if (MPARSE_QUICK & mparse_options) {
2303 exitcode = (int)MANDOCLEVEL_SYSERR;
2304 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
2308 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2309 sizeof(tempfilename));
2310 if (NULL == mkdtemp(tempfilename)) {
2311 exitcode = (int)MANDOCLEVEL_SYSERR;
2312 say("", "&%s", tempfilename);
2315 (void)strlcat(tempfilename, "/" MANDOC_DB,
2316 sizeof(tempfilename));
2317 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2318 if (SQLITE_OK != rc) {
2319 exitcode = (int)MANDOCLEVEL_SYSERR;
2320 say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
2325 sql = "CREATE TABLE \"mpages\" (\n"
2326 " \"desc\" TEXT NOT NULL,\n"
2327 " \"form\" INTEGER NOT NULL,\n"
2328 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2331 "CREATE TABLE \"mlinks\" (\n"
2332 " \"sec\" TEXT NOT NULL,\n"
2333 " \"arch\" TEXT NOT NULL,\n"
2334 " \"name\" TEXT NOT NULL,\n"
2335 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2336 "ON DELETE CASCADE\n"
2338 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2340 "CREATE TABLE \"names\" (\n"
2341 " \"bits\" INTEGER NOT NULL,\n"
2342 " \"name\" TEXT NOT NULL,\n"
2343 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2344 "ON DELETE CASCADE,\n"
2345 " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
2348 "CREATE TABLE \"keys\" (\n"
2349 " \"bits\" INTEGER NOT NULL,\n"
2350 " \"key\" TEXT NOT NULL,\n"
2351 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2352 "ON DELETE CASCADE\n"
2354 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2356 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2357 exitcode = (int)MANDOCLEVEL_SYSERR;
2358 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2364 if (SQLITE_OK != sqlite3_exec(db,
2365 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2366 exitcode = (int)MANDOCLEVEL_SYSERR;
2367 say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2368 sqlite3_errmsg(db));
2373 sql = "DELETE FROM mpages WHERE pageid IN "
2374 "(SELECT pageid FROM mlinks WHERE "
2375 "sec=? AND arch=? AND name=?)";
2376 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2377 sql = "INSERT INTO mpages "
2378 "(desc,form) VALUES (?,?)";
2379 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2380 sql = "INSERT INTO mlinks "
2381 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2382 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2383 sql = "SELECT bits FROM names where pageid = ?";
2384 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
2385 sql = "INSERT INTO names "
2386 "(bits,name,pageid) VALUES (?,?,?)";
2387 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2388 sql = "INSERT INTO keys "
2389 "(bits,key,pageid) VALUES (?,?,?)";
2390 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2394 * When opening a new database, we can turn off
2395 * synchronous mode for much better performance.
2398 if (real && SQLITE_OK != sqlite3_exec(db,
2399 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2400 exitcode = (int)MANDOCLEVEL_SYSERR;
2401 say(MANDOC_DB, "PRAGMA synchronous: %s",
2402 sqlite3_errmsg(db));
2412 set_basedir(const char *targetdir, int report_baddir)
2414 static char startdir[PATH_MAX];
2415 static int getcwd_status; /* 1 = ok, 2 = failure */
2416 static int chdir_status; /* 1 = changed directory */
2420 * Remember the original working directory, if possible.
2421 * This will be needed if the second or a later directory
2422 * on the command line is given as a relative path.
2423 * Do not error out if the current directory is not
2424 * searchable: Maybe it won't be needed after all.
2426 if (0 == getcwd_status) {
2427 if (NULL == getcwd(startdir, sizeof(startdir))) {
2429 (void)strlcpy(startdir, strerror(errno),
2436 * We are leaving the old base directory.
2437 * Do not use it any longer, not even for messages.
2442 * If and only if the directory was changed earlier and
2443 * the next directory to process is given as a relative path,
2444 * first go back, or bail out if that is impossible.
2446 if (chdir_status && '/' != *targetdir) {
2447 if (2 == getcwd_status) {
2448 exitcode = (int)MANDOCLEVEL_SYSERR;
2449 say("", "getcwd: %s", startdir);
2452 if (-1 == chdir(startdir)) {
2453 exitcode = (int)MANDOCLEVEL_SYSERR;
2454 say("", "&chdir %s", startdir);
2460 * Always resolve basedir to the canonicalized absolute
2461 * pathname and append a trailing slash, such that
2462 * we can reliably check whether files are inside.
2464 if (NULL == realpath(targetdir, basedir)) {
2465 if (report_baddir || errno != ENOENT) {
2466 exitcode = (int)MANDOCLEVEL_BADARG;
2467 say("", "&%s: realpath", targetdir);
2470 } else if (-1 == chdir(basedir)) {
2471 if (report_baddir || errno != ENOENT) {
2472 exitcode = (int)MANDOCLEVEL_BADARG;
2478 cp = strchr(basedir, '\0');
2479 if ('/' != cp[-1]) {
2480 if (cp - basedir >= PATH_MAX - 1) {
2481 exitcode = (int)MANDOCLEVEL_SYSERR;
2482 say("", "Filename too long");
2492 say(const char *file, const char *format, ...)
2497 if ('\0' != *basedir)
2498 fprintf(stderr, "%s", basedir);
2499 if ('\0' != *basedir && '\0' != *file)
2502 fprintf(stderr, "%s", file);
2505 if (NULL != format) {
2518 if (NULL != format) {
2519 if ('\0' != *basedir || '\0' != *file)
2520 fputs(": ", stderr);
2521 va_start(ap, format);
2522 vfprintf(stderr, format, ap);
2526 if ('\0' != *basedir || '\0' != *file || NULL != format)
2527 fputs(": ", stderr);
2530 fputc('\n', stderr);