1 /* $Id: mandocdb.c,v 1.263 2019/05/03 18:17:12 schwarze Exp $ */
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2019 Ingo Schwarze <schwarze@openbsd.org>
5 * Copyright (c) 2016 Ed Maste <emaste@freebsd.org>
7 * Permission to use, copy, modify, and distribute this software for any
8 * purpose with or without fee is hereby granted, provided that the above
9 * copyright notice and this permission notice appear in all copies.
11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
21 #include <sys/types.h>
35 #include "compat_fts.h"
49 #include "mandoc_aux.h"
50 #include "mandoc_ohash.h"
55 #include "mandoc_parse.h"
57 #include "mansearch.h"
58 #include "dba_array.h"
61 extern const char *const mansearch_keynames[];
64 OP_DEFAULT = 0, /* new dbs from dir list or default config */
65 OP_CONFFILE, /* new databases from custom config file */
66 OP_UPDATE, /* delete/add entries in existing database */
67 OP_DELETE, /* delete entries from existing database */
68 OP_TEST /* change no databases, report potential problems */
72 const struct mpage *mpage; /* if set, the owning parse */
73 uint64_t mask; /* bitmask in sequence */
74 char key[]; /* rendered text */
83 struct inodev inodev; /* used for hashing routine */
84 struct dba_array *dba;
85 char *sec; /* section from file content */
86 char *arch; /* architecture from file content */
87 char *title; /* title from file content */
88 char *desc; /* description from file content */
89 struct mpage *next; /* singly linked list */
90 struct mlink *mlinks; /* singly linked list */
92 enum form form; /* format from file content */
96 char file[PATH_MAX]; /* filename rel. to manpath */
97 char *dsec; /* section from directory */
98 char *arch; /* architecture from directory */
99 char *name; /* name from file name (not empty) */
100 char *fsec; /* section from file name suffix */
101 struct mlink *next; /* singly linked list */
102 struct mpage *mpage; /* parent */
103 int gzip; /* filename has a .gz suffix */
104 enum form dform; /* format from directory */
105 enum form fform; /* format from file name suffix */
108 typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
109 const struct roff_node *);
111 struct mdoc_handler {
112 mdoc_fp fp; /* optional handler */
113 uint64_t mask; /* set unless handler returns 0 */
114 int taboo; /* node flags that must not be set */
118 int mandocdb(int, char *[]);
120 static void dbadd(struct dba *, struct mpage *);
121 static void dbadd_mlink(const struct mlink *mlink);
122 static void dbprune(struct dba *);
123 static void dbwrite(struct dba *);
124 static void filescan(const char *);
125 #if HAVE_FTS_COMPARE_CONST
126 static int fts_compare(const FTSENT *const *, const FTSENT *const *);
128 static int fts_compare(const FTSENT **, const FTSENT **);
130 static void mlink_add(struct mlink *, const struct stat *);
131 static void mlink_check(struct mpage *, struct mlink *);
132 static void mlink_free(struct mlink *);
133 static void mlinks_undupe(struct mpage *);
134 static void mpages_free(void);
135 static void mpages_merge(struct dba *, struct mparse *);
136 static void parse_cat(struct mpage *, int);
137 static void parse_man(struct mpage *, const struct roff_meta *,
138 const struct roff_node *);
139 static void parse_mdoc(struct mpage *, const struct roff_meta *,
140 const struct roff_node *);
141 static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
142 const struct roff_node *);
143 static int parse_mdoc_Fa(struct mpage *, const struct roff_meta *,
144 const struct roff_node *);
145 static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
146 const struct roff_node *);
147 static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
148 static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
149 const struct roff_node *);
150 static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
151 const struct roff_node *);
152 static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
153 const struct roff_node *);
154 static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
155 const struct roff_node *);
156 static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
157 const struct roff_node *);
158 static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
159 const struct roff_node *);
160 static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
161 const struct roff_node *);
162 static void putkey(const struct mpage *, char *, uint64_t);
163 static void putkeys(const struct mpage *, char *, size_t, uint64_t);
164 static void putmdockey(const struct mpage *,
165 const struct roff_node *, uint64_t, int);
166 static int render_string(char **, size_t *);
167 static void say(const char *, const char *, ...)
168 __attribute__((__format__ (__printf__, 2, 3)));
169 static int set_basedir(const char *, int);
170 static int treescan(void);
171 static size_t utf8(unsigned int, char [7]);
173 static int nodb; /* no database changes */
174 static int mparse_options; /* abort the parse early */
175 static int use_all; /* use all found files */
176 static int debug; /* print what we're doing */
177 static int warnings; /* warn about crap */
178 static int write_utf8; /* write UTF-8 output; else ASCII */
179 static int exitcode; /* to be returned by main */
180 static enum op op; /* operational mode */
181 static char basedir[PATH_MAX]; /* current base directory */
182 static struct mpage *mpage_head; /* list of distinct manual pages */
183 static struct ohash mpages; /* table of distinct manual pages */
184 static struct ohash mlinks; /* table of directory entries */
185 static struct ohash names; /* table of all names */
186 static struct ohash strings; /* table of all strings */
187 static uint64_t name_mask;
189 static const struct mdoc_handler mdoc_handlers[MDOC_MAX - MDOC_Dd] = {
190 { NULL, 0, NODE_NOPRT }, /* Dd */
191 { NULL, 0, NODE_NOPRT }, /* Dt */
192 { NULL, 0, NODE_NOPRT }, /* Os */
193 { parse_mdoc_Sh, TYPE_Sh, 0 }, /* Sh */
194 { parse_mdoc_head, TYPE_Ss, 0 }, /* Ss */
195 { NULL, 0, 0 }, /* Pp */
196 { NULL, 0, 0 }, /* D1 */
197 { NULL, 0, 0 }, /* Dl */
198 { NULL, 0, 0 }, /* Bd */
199 { NULL, 0, 0 }, /* Ed */
200 { NULL, 0, 0 }, /* Bl */
201 { NULL, 0, 0 }, /* El */
202 { NULL, 0, 0 }, /* It */
203 { NULL, 0, 0 }, /* Ad */
204 { NULL, TYPE_An, 0 }, /* An */
205 { NULL, 0, 0 }, /* Ap */
206 { NULL, TYPE_Ar, 0 }, /* Ar */
207 { NULL, TYPE_Cd, 0 }, /* Cd */
208 { NULL, TYPE_Cm, 0 }, /* Cm */
209 { NULL, TYPE_Dv, 0 }, /* Dv */
210 { NULL, TYPE_Er, 0 }, /* Er */
211 { NULL, TYPE_Ev, 0 }, /* Ev */
212 { NULL, 0, 0 }, /* Ex */
213 { parse_mdoc_Fa, 0, 0 }, /* Fa */
214 { parse_mdoc_Fd, 0, 0 }, /* Fd */
215 { NULL, TYPE_Fl, 0 }, /* Fl */
216 { parse_mdoc_Fn, 0, 0 }, /* Fn */
217 { NULL, TYPE_Ft | TYPE_Vt, 0 }, /* Ft */
218 { NULL, TYPE_Ic, 0 }, /* Ic */
219 { NULL, TYPE_In, 0 }, /* In */
220 { NULL, TYPE_Li, 0 }, /* Li */
221 { parse_mdoc_Nd, 0, 0 }, /* Nd */
222 { parse_mdoc_Nm, 0, 0 }, /* Nm */
223 { NULL, 0, 0 }, /* Op */
224 { NULL, 0, 0 }, /* Ot */
225 { NULL, TYPE_Pa, NODE_NOSRC }, /* Pa */
226 { NULL, 0, 0 }, /* Rv */
227 { NULL, TYPE_St, 0 }, /* St */
228 { parse_mdoc_Va, TYPE_Va, 0 }, /* Va */
229 { parse_mdoc_Va, TYPE_Vt, 0 }, /* Vt */
230 { parse_mdoc_Xr, 0, 0 }, /* Xr */
231 { NULL, 0, 0 }, /* %A */
232 { NULL, 0, 0 }, /* %B */
233 { NULL, 0, 0 }, /* %D */
234 { NULL, 0, 0 }, /* %I */
235 { NULL, 0, 0 }, /* %J */
236 { NULL, 0, 0 }, /* %N */
237 { NULL, 0, 0 }, /* %O */
238 { NULL, 0, 0 }, /* %P */
239 { NULL, 0, 0 }, /* %R */
240 { NULL, 0, 0 }, /* %T */
241 { NULL, 0, 0 }, /* %V */
242 { NULL, 0, 0 }, /* Ac */
243 { NULL, 0, 0 }, /* Ao */
244 { NULL, 0, 0 }, /* Aq */
245 { NULL, TYPE_At, 0 }, /* At */
246 { NULL, 0, 0 }, /* Bc */
247 { NULL, 0, 0 }, /* Bf */
248 { NULL, 0, 0 }, /* Bo */
249 { NULL, 0, 0 }, /* Bq */
250 { NULL, TYPE_Bsx, NODE_NOSRC }, /* Bsx */
251 { NULL, TYPE_Bx, NODE_NOSRC }, /* Bx */
252 { NULL, 0, 0 }, /* Db */
253 { NULL, 0, 0 }, /* Dc */
254 { NULL, 0, 0 }, /* Do */
255 { NULL, 0, 0 }, /* Dq */
256 { NULL, 0, 0 }, /* Ec */
257 { NULL, 0, 0 }, /* Ef */
258 { NULL, TYPE_Em, 0 }, /* Em */
259 { NULL, 0, 0 }, /* Eo */
260 { NULL, TYPE_Fx, NODE_NOSRC }, /* Fx */
261 { NULL, TYPE_Ms, 0 }, /* Ms */
262 { NULL, 0, 0 }, /* No */
263 { NULL, 0, 0 }, /* Ns */
264 { NULL, TYPE_Nx, NODE_NOSRC }, /* Nx */
265 { NULL, TYPE_Ox, NODE_NOSRC }, /* Ox */
266 { NULL, 0, 0 }, /* Pc */
267 { NULL, 0, 0 }, /* Pf */
268 { NULL, 0, 0 }, /* Po */
269 { NULL, 0, 0 }, /* Pq */
270 { NULL, 0, 0 }, /* Qc */
271 { NULL, 0, 0 }, /* Ql */
272 { NULL, 0, 0 }, /* Qo */
273 { NULL, 0, 0 }, /* Qq */
274 { NULL, 0, 0 }, /* Re */
275 { NULL, 0, 0 }, /* Rs */
276 { NULL, 0, 0 }, /* Sc */
277 { NULL, 0, 0 }, /* So */
278 { NULL, 0, 0 }, /* Sq */
279 { NULL, 0, 0 }, /* Sm */
280 { NULL, 0, 0 }, /* Sx */
281 { NULL, TYPE_Sy, 0 }, /* Sy */
282 { NULL, TYPE_Tn, 0 }, /* Tn */
283 { NULL, 0, NODE_NOSRC }, /* Ux */
284 { NULL, 0, 0 }, /* Xc */
285 { NULL, 0, 0 }, /* Xo */
286 { parse_mdoc_Fo, 0, 0 }, /* Fo */
287 { NULL, 0, 0 }, /* Fc */
288 { NULL, 0, 0 }, /* Oo */
289 { NULL, 0, 0 }, /* Oc */
290 { NULL, 0, 0 }, /* Bk */
291 { NULL, 0, 0 }, /* Ek */
292 { NULL, 0, 0 }, /* Bt */
293 { NULL, 0, 0 }, /* Hf */
294 { NULL, 0, 0 }, /* Fr */
295 { NULL, 0, 0 }, /* Ud */
296 { NULL, TYPE_Lb, NODE_NOSRC }, /* Lb */
297 { NULL, 0, 0 }, /* Lp */
298 { NULL, TYPE_Lk, 0 }, /* Lk */
299 { NULL, TYPE_Mt, NODE_NOSRC }, /* Mt */
300 { NULL, 0, 0 }, /* Brq */
301 { NULL, 0, 0 }, /* Bro */
302 { NULL, 0, 0 }, /* Brc */
303 { NULL, 0, 0 }, /* %C */
304 { NULL, 0, 0 }, /* Es */
305 { NULL, 0, 0 }, /* En */
306 { NULL, TYPE_Dx, NODE_NOSRC }, /* Dx */
307 { NULL, 0, 0 }, /* %Q */
308 { NULL, 0, 0 }, /* %U */
309 { NULL, 0, 0 }, /* Ta */
314 mandocdb(int argc, char *argv[])
319 const char *path_arg, *progname;
324 if (pledge("stdio rpath wpath cpath", NULL) == -1) {
326 return (int)MANDOCLEVEL_SYSERR;
330 #if HAVE_SANDBOX_INIT
331 if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
332 warnx("sandbox_init");
333 return (int)MANDOCLEVEL_SYSERR;
337 memset(&conf, 0, sizeof(conf));
340 * We accept a few different invocations.
341 * The CHECKOP macro makes sure that invocation styles don't
342 * clobber each other.
344 #define CHECKOP(_op, _ch) do \
345 if (OP_DEFAULT != (_op)) { \
346 warnx("-%c: Conflicting option", (_ch)); \
348 } while (/*CONSTCOND*/0)
350 mparse_options = MPARSE_VALIDATE;
354 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
379 mparse_options |= MPARSE_QUICK;
382 if (strcmp(optarg, "utf8")) {
383 warnx("-T%s: Unsupported output format",
391 dup2(STDOUT_FILENO, STDERR_FILENO);
401 /* Compatibility with espie@'s makewhatis. */
412 if (pledge("stdio rpath", NULL) == -1) {
414 return (int)MANDOCLEVEL_SYSERR;
419 if (OP_CONFFILE == op && argc > 0) {
420 warnx("-C: Too many arguments");
424 exitcode = (int)MANDOCLEVEL_OK;
426 mp = mparse_alloc(mparse_options, MANDOC_OS_OTHER, NULL);
427 mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
428 mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
430 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
433 * Most of these deal with a specific directory.
434 * Jump into that directory first.
436 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
439 dba = nodb ? dba_new(128) : dba_read(MANDOC_DB);
442 * The existing database is usable. Process
443 * all files specified on the command-line.
446 for (i = 0; i < argc; i++)
451 /* Database missing or corrupt. */
452 if (op != OP_UPDATE || errno != ENOENT)
453 say(MANDOC_DB, "%s: Automatically recreating"
454 " from scratch", strerror(errno));
455 exitcode = (int)MANDOCLEVEL_OK;
462 mpages_merge(dba, mp);
468 * If we have arguments, use them as our manpaths.
469 * If we don't, use man.conf(5).
472 conf.manpath.paths = mandoc_reallocarray(NULL,
473 argc, sizeof(char *));
474 conf.manpath.sz = (size_t)argc;
475 for (i = 0; i < argc; i++)
476 conf.manpath.paths[i] = mandoc_strdup(argv[i]);
478 manconf_parse(&conf, path_arg, NULL, NULL);
480 if (conf.manpath.sz == 0) {
481 exitcode = (int)MANDOCLEVEL_BADARG;
482 say("", "Empty manpath");
486 * First scan the tree rooted at a base directory, then
487 * build a new database and finally move it into place.
488 * Ignore zero-length directories and strip trailing
491 for (j = 0; j < conf.manpath.sz; j++) {
492 sz = strlen(conf.manpath.paths[j]);
493 if (sz && conf.manpath.paths[j][sz - 1] == '/')
494 conf.manpath.paths[j][--sz] = '\0';
499 mandoc_ohash_init(&mpages, 6,
500 offsetof(struct mpage, inodev));
501 mandoc_ohash_init(&mlinks, 6,
502 offsetof(struct mlink, file));
505 if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
510 mpages_merge(dba, mp);
515 if (j + 1 < conf.manpath.sz) {
517 ohash_delete(&mpages);
518 ohash_delete(&mlinks);
527 ohash_delete(&mpages);
528 ohash_delete(&mlinks);
531 progname = getprogname();
532 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
533 " %s [-aDnpQ] [-Tutf8] dir ...\n"
534 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
535 " %s [-Dnp] -u dir [file ...]\n"
536 " %s [-Q] -t file ...\n",
537 progname, progname, progname, progname, progname);
539 return (int)MANDOCLEVEL_BADARG;
543 * To get a singly linked list in alpha order while inserting entries
544 * at the beginning, process directory entries in reverse alpha order.
547 #if HAVE_FTS_COMPARE_CONST
548 fts_compare(const FTSENT *const *a, const FTSENT *const *b)
550 fts_compare(const FTSENT **a, const FTSENT **b)
553 return -strcmp((*a)->fts_name, (*b)->fts_name);
557 * Scan a directory tree rooted at "basedir" for manpages.
558 * We use fts(), scanning directory parts along the way for clues to our
559 * section and architecture.
561 * If use_all has been specified, grok all files.
562 * If not, sanitise paths to the following:
564 * [./]man*[/<arch>]/<name>.<section>
566 * [./]cat<section>[/<arch>]/<name>.0
568 * TODO: accommodate for multi-language directories.
579 char *dsec, *arch, *fsec, *cp;
586 f = fts_open((char * const *)argv, FTS_PHYSICAL | FTS_NOCHDIR,
589 exitcode = (int)MANDOCLEVEL_SYSERR;
590 say("", "&fts_open");
597 while ((ff = fts_read(f)) != NULL) {
598 path = ff->fts_path + 2;
599 switch (ff->fts_info) {
602 * Symbolic links require various sanity checks,
603 * then get handled just like regular files.
606 if (realpath(path, buf) == NULL) {
608 say(path, "&realpath");
611 if (strstr(buf, basedir) != buf
613 && strstr(buf, HOMEBREWDIR) != buf
616 if (warnings) say("",
617 "%s: outside base directory", buf);
620 /* Use logical inode to avoid mpages dupe. */
621 if (stat(path, ff->fts_statp) == -1) {
629 * If we're a regular file, add an mlink by using the
630 * stored directory data and handling the filename.
633 if ( ! strcmp(path, MANDOC_DB))
635 if ( ! use_all && ff->fts_level < 2) {
637 say(path, "Extraneous file");
642 while (fsec == NULL) {
643 fsec = strrchr(ff->fts_name, '.');
644 if (fsec == NULL || strcmp(fsec+1, "gz"))
654 "No filename suffix");
657 } else if ( ! strcmp(++fsec, "html")) {
659 say(path, "Skip html");
661 } else if ( ! strcmp(fsec, "ps")) {
663 say(path, "Skip ps");
665 } else if ( ! strcmp(fsec, "pdf")) {
667 say(path, "Skip pdf");
669 } else if ( ! use_all &&
670 ((dform == FORM_SRC &&
671 strncmp(fsec, dsec, strlen(dsec))) ||
672 (dform == FORM_CAT && strcmp(fsec, "0")))) {
674 say(path, "Wrong filename suffix");
679 mlink = mandoc_calloc(1, sizeof(struct mlink));
680 if (strlcpy(mlink->file, path,
681 sizeof(mlink->file)) >=
682 sizeof(mlink->file)) {
683 say(path, "Filename too long");
687 mlink->dform = dform;
690 mlink->name = ff->fts_name;
693 mlink_add(mlink, ff->fts_statp);
702 say(path, "Not a regular file");
706 switch (ff->fts_level) {
708 /* Ignore the root directory. */
712 * This might contain manX/ or catX/.
713 * Try to infer this from the name.
714 * If we're not in use_all, enforce it.
717 if (ff->fts_info == FTS_DP) {
723 if ( ! strncmp(cp, "man", 3)) {
726 } else if ( ! strncmp(cp, "cat", 3)) {
734 if (dsec != NULL || use_all)
738 say(path, "Unknown directory part");
739 fts_set(f, ff, FTS_SKIP);
743 * Possibly our architecture.
744 * If we're descending, keep tabs on it.
746 if (ff->fts_info != FTS_DP && dsec != NULL)
752 if (ff->fts_info == FTS_DP || use_all)
755 say(path, "Extraneous directory part");
756 fts_set(f, ff, FTS_SKIP);
766 * Add a file to the mlinks table.
767 * Do not verify that it's a "valid" looking manpage (we'll do that
770 * Try to infer the manual section, architecture, and page name from the
771 * path, assuming it looks like
773 * [./]man*[/<arch>]/<name>.<section>
775 * [./]cat<section>[/<arch>]/<name>.0
777 * See treescan() for the fts(3) version of this.
780 filescan(const char *file)
789 if (0 == strncmp(file, "./", 2))
793 * We have to do lstat(2) before realpath(3) loses
794 * the information whether this is a symbolic link.
795 * We need to know that because for symbolic links,
796 * we want to use the orginal file name, while for
797 * regular files, we want to use the real path.
799 if (-1 == lstat(file, &st)) {
800 exitcode = (int)MANDOCLEVEL_BADARG;
803 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
804 exitcode = (int)MANDOCLEVEL_BADARG;
805 say(file, "Not a regular file");
810 * We have to resolve the file name to the real path
811 * in any case for the base directory check.
813 if (NULL == realpath(file, buf)) {
814 exitcode = (int)MANDOCLEVEL_BADARG;
815 say(file, "&realpath");
821 else if (strstr(buf, basedir) == buf)
822 start = buf + strlen(basedir);
824 else if (strstr(buf, HOMEBREWDIR) == buf)
828 exitcode = (int)MANDOCLEVEL_BADARG;
829 say("", "%s: outside base directory", buf);
834 * Now we are sure the file is inside our tree.
835 * If it is a symbolic link, ignore the real path
836 * and use the original name.
837 * This implies passing stuff like "cat1/../man1/foo.1"
838 * on the command line won't work. So don't do that.
839 * Note the stat(2) can still fail if the link target
842 if (S_IFLNK & st.st_mode) {
843 if (-1 == stat(buf, &st)) {
844 exitcode = (int)MANDOCLEVEL_BADARG;
848 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
849 say(file, "Filename too long");
853 if (OP_TEST != op && strstr(buf, basedir) == buf)
854 start += strlen(basedir);
857 mlink = mandoc_calloc(1, sizeof(struct mlink));
858 mlink->dform = FORM_NONE;
859 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
860 sizeof(mlink->file)) {
861 say(start, "Filename too long");
867 * In test mode or when the original name is absolute
868 * but outside our tree, guess the base directory.
871 if (op == OP_TEST || (start == buf && *start == '/')) {
872 if (strncmp(buf, "man/", 4) == 0)
874 else if ((start = strstr(buf, "/man/")) != NULL)
881 * First try to guess our directory structure.
882 * If we find a separator, try to look for man* or cat*.
883 * If we find one of these and what's underneath is a directory,
884 * assume it's an architecture.
886 if (NULL != (p = strchr(start, '/'))) {
888 if (0 == strncmp(start, "man", 3)) {
889 mlink->dform = FORM_SRC;
890 mlink->dsec = start + 3;
891 } else if (0 == strncmp(start, "cat", 3)) {
892 mlink->dform = FORM_CAT;
893 mlink->dsec = start + 3;
897 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
905 * Now check the file suffix.
906 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
908 p = strrchr(start, '\0');
909 while (p-- > start && '/' != *p && '.' != *p)
918 * Now try to parse the name.
919 * Use the filename portion of the path.
922 if (NULL != (p = strrchr(start, '/'))) {
926 mlink_add(mlink, &st);
930 mlink_add(struct mlink *mlink, const struct stat *st)
932 struct inodev inodev;
936 assert(NULL != mlink->file);
938 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
939 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
940 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
941 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
943 if ('0' == *mlink->fsec) {
945 mlink->fsec = mandoc_strdup(mlink->dsec);
946 mlink->fform = FORM_CAT;
947 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
948 mlink->fform = FORM_SRC;
950 mlink->fform = FORM_NONE;
952 slot = ohash_qlookup(&mlinks, mlink->file);
953 assert(NULL == ohash_find(&mlinks, slot));
954 ohash_insert(&mlinks, slot, mlink);
956 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
957 inodev.st_ino = st->st_ino;
958 inodev.st_dev = st->st_dev;
959 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
960 sizeof(struct inodev), inodev.st_ino);
961 mpage = ohash_find(&mpages, slot);
963 mpage = mandoc_calloc(1, sizeof(struct mpage));
964 mpage->inodev.st_ino = inodev.st_ino;
965 mpage->inodev.st_dev = inodev.st_dev;
966 mpage->form = FORM_NONE;
967 mpage->next = mpage_head;
969 ohash_insert(&mpages, slot, mpage);
971 mlink->next = mpage->mlinks;
972 mpage->mlinks = mlink;
973 mlink->mpage = mpage;
977 mlink_free(struct mlink *mlink)
993 while ((mpage = mpage_head) != NULL) {
994 while ((mlink = mpage->mlinks) != NULL) {
995 mpage->mlinks = mlink->next;
998 mpage_head = mpage->next;
1008 * For each mlink to the mpage, check whether the path looks like
1009 * it is formatted, and if it does, check whether a source manual
1010 * exists by the same name, ignoring the suffix.
1011 * If both conditions hold, drop the mlink.
1014 mlinks_undupe(struct mpage *mpage)
1017 struct mlink **prev;
1018 struct mlink *mlink;
1021 mpage->form = FORM_CAT;
1022 prev = &mpage->mlinks;
1023 while (NULL != (mlink = *prev)) {
1024 if (FORM_CAT != mlink->dform) {
1025 mpage->form = FORM_NONE;
1028 (void)strlcpy(buf, mlink->file, sizeof(buf));
1029 bufp = strstr(buf, "cat");
1030 assert(NULL != bufp);
1031 memcpy(bufp, "man", 3);
1032 if (NULL != (bufp = strrchr(buf, '.')))
1034 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1035 if (NULL == ohash_find(&mlinks,
1036 ohash_qlookup(&mlinks, buf)))
1039 say(mlink->file, "Man source exists: %s", buf);
1042 *prev = mlink->next;
1046 prev = &(*prev)->next;
1051 mlink_check(struct mpage *mpage, struct mlink *mlink)
1057 * Check whether the manual section given in a file
1058 * agrees with the directory where the file is located.
1059 * Some manuals have suffixes like (3p) on their
1060 * section number either inside the file or in the
1061 * directory name, some are linked into more than one
1062 * section, like encrypt(1) = makekey(8).
1065 if (FORM_SRC == mpage->form &&
1066 strcasecmp(mpage->sec, mlink->dsec))
1067 say(mlink->file, "Section \"%s\" manual in %s directory",
1068 mpage->sec, mlink->dsec);
1071 * Manual page directories exist for each kernel
1072 * architecture as returned by machine(1).
1073 * However, many manuals only depend on the
1074 * application architecture as returned by arch(1).
1075 * For example, some (2/ARM) manuals are shared
1076 * across the "armish" and "zaurus" kernel
1078 * A few manuals are even shared across completely
1079 * different architectures, for example fdformat(1)
1080 * on amd64, i386, and sparc64.
1083 if (strcasecmp(mpage->arch, mlink->arch))
1084 say(mlink->file, "Architecture \"%s\" manual in "
1085 "\"%s\" directory", mpage->arch, mlink->arch);
1089 * parse_cat() doesn't set NAME_TITLE yet.
1092 if (FORM_CAT == mpage->form)
1096 * Check whether this mlink
1097 * appears as a name in the NAME section.
1100 slot = ohash_qlookup(&names, mlink->name);
1101 str = ohash_find(&names, slot);
1102 assert(NULL != str);
1103 if ( ! (NAME_TITLE & str->mask))
1104 say(mlink->file, "Name missing in NAME section");
1108 * Run through the files in the global vector "mpages"
1109 * and add them to the database specified in "basedir".
1111 * This handles the parsing scheme itself, using the cues of directory
1112 * and filename to determine whether the file is parsable or not.
1115 mpages_merge(struct dba *dba, struct mparse *mp)
1117 struct mpage *mpage, *mpage_dest;
1118 struct mlink *mlink, *mlink_dest;
1119 struct roff_meta *meta;
1123 for (mpage = mpage_head; mpage != NULL; mpage = mpage->next) {
1124 mlinks_undupe(mpage);
1125 if ((mlink = mpage->mlinks) == NULL)
1128 name_mask = NAME_MASK;
1129 mandoc_ohash_init(&names, 4, offsetof(struct str, key));
1130 mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
1134 if ((fd = mparse_open(mp, mlink->file)) == -1) {
1135 say(mlink->file, "&open");
1140 * Interpret the file as mdoc(7) or man(7) source
1141 * code, unless it is known to be formatted.
1143 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1144 mparse_readfd(mp, fd, mlink->file);
1147 meta = mparse_result(mp);
1150 if (meta != NULL && meta->sodest != NULL) {
1151 mlink_dest = ohash_find(&mlinks,
1152 ohash_qlookup(&mlinks, meta->sodest));
1153 if (mlink_dest == NULL) {
1154 mandoc_asprintf(&cp, "%s.gz", meta->sodest);
1155 mlink_dest = ohash_find(&mlinks,
1156 ohash_qlookup(&mlinks, cp));
1159 if (mlink_dest != NULL) {
1161 /* The .so target exists. */
1163 mpage_dest = mlink_dest->mpage;
1165 mlink->mpage = mpage_dest;
1168 * If the target was already
1169 * processed, add the links
1170 * to the database now.
1171 * Otherwise, this will
1172 * happen when we come
1176 if (mpage_dest->dba != NULL)
1179 if (mlink->next == NULL)
1181 mlink = mlink->next;
1184 /* Move all links to the target. */
1186 mlink->next = mlink_dest->next;
1187 mlink_dest->next = mpage->mlinks;
1188 mpage->mlinks = NULL;
1191 meta->macroset = MACROSET_NONE;
1193 if (meta != NULL && meta->macroset == MACROSET_MDOC) {
1194 mpage->form = FORM_SRC;
1195 mpage->sec = meta->msec;
1196 mpage->sec = mandoc_strdup(
1197 mpage->sec == NULL ? "" : mpage->sec);
1198 mpage->arch = meta->arch;
1199 mpage->arch = mandoc_strdup(
1200 mpage->arch == NULL ? "" : mpage->arch);
1201 mpage->title = mandoc_strdup(meta->title);
1202 } else if (meta != NULL && meta->macroset == MACROSET_MAN) {
1203 if (*meta->msec != '\0' || *meta->title != '\0') {
1204 mpage->form = FORM_SRC;
1205 mpage->sec = mandoc_strdup(meta->msec);
1206 mpage->arch = mandoc_strdup(mlink->arch);
1207 mpage->title = mandoc_strdup(meta->title);
1212 assert(mpage->desc == NULL);
1213 if (meta == NULL || meta->sodest != NULL) {
1214 mpage->sec = mandoc_strdup(mlink->dsec);
1215 mpage->arch = mandoc_strdup(mlink->arch);
1216 mpage->title = mandoc_strdup(mlink->name);
1218 mpage->form = FORM_CAT;
1219 parse_cat(mpage, fd);
1221 mpage->form = FORM_SRC;
1222 } else if (meta->macroset == MACROSET_MDOC)
1223 parse_mdoc(mpage, meta, meta->first);
1225 parse_man(mpage, meta, meta->first);
1226 if (mpage->desc == NULL) {
1227 mpage->desc = mandoc_strdup(mlink->name);
1229 say(mlink->file, "No one-line description, "
1230 "using filename \"%s\"", mlink->name);
1233 for (mlink = mpage->mlinks;
1235 mlink = mlink->next) {
1236 putkey(mpage, mlink->name, NAME_FILE);
1237 if (warnings && !use_all)
1238 mlink_check(mpage, mlink);
1244 ohash_delete(&strings);
1245 ohash_delete(&names);
1250 parse_cat(struct mpage *mpage, int fd)
1253 struct mlink *mlink;
1254 char *line, *p, *title, *sec;
1255 size_t linesz, plen, titlesz;
1259 mlink = mpage->mlinks;
1260 stream = fd == -1 ? fopen(mlink->file, "r") : fdopen(fd, "r");
1261 if (stream == NULL) {
1265 say(mlink->file, "&fopen");
1272 /* Parse the section number from the header line. */
1274 while (getline(&line, &linesz, stream) != -1) {
1277 if ((sec = strchr(line, '(')) == NULL)
1279 if ((p = strchr(++sec, ')')) == NULL)
1282 mpage->sec = mandoc_strndup(sec, p - sec);
1283 if (warnings && *mlink->dsec != '\0' &&
1284 strcasecmp(mpage->sec, mlink->dsec))
1286 "Section \"%s\" manual in %s directory",
1287 mpage->sec, mlink->dsec);
1291 /* Skip to first blank line. */
1293 while (line == NULL || *line != '\n')
1294 if (getline(&line, &linesz, stream) == -1)
1298 * Assume the first line that is not indented
1299 * is the first section header. Skip to it.
1302 while (getline(&line, &linesz, stream) != -1)
1303 if (*line != '\n' && *line != ' ')
1307 * Read up until the next section into a buffer.
1308 * Strip the leading and trailing newline from each read line,
1309 * appending a trailing space.
1310 * Ignore empty (whitespace-only) lines.
1316 while ((len = getline(&line, &linesz, stream)) != -1) {
1320 while (isspace((unsigned char)line[offs]))
1322 if (line[offs] == '\0')
1324 title = mandoc_realloc(title, titlesz + len - offs);
1325 memcpy(title + titlesz, line + offs, len - offs);
1326 titlesz += len - offs;
1327 title[titlesz - 1] = ' ';
1332 * If no page content can be found, or the input line
1333 * is already the next section header, or there is no
1334 * trailing newline, reuse the page title as the page
1338 if (NULL == title || '\0' == *title) {
1340 say(mlink->file, "Cannot find NAME section");
1346 title[titlesz - 1] = '\0';
1349 * Skip to the first dash.
1350 * Use the remaining line as the description (no more than 70
1354 if (NULL != (p = strstr(title, "- "))) {
1355 for (p += 2; ' ' == *p || '\b' == *p; p++)
1356 /* Skip to next word. */ ;
1359 say(mlink->file, "No dash in title line, "
1360 "reusing \"%s\" as one-line description", title);
1366 /* Strip backspace-encoding from line. */
1368 while (NULL != (line = memchr(p, '\b', plen))) {
1371 memmove(line, line + 1, plen--);
1374 memmove(line - 1, line + 1, plen - len);
1379 * Cut off excessive one-line descriptions.
1380 * Bad pages are not worth better heuristics.
1383 mpage->desc = mandoc_strndup(p, 150);
1389 * Put a type/word pair into the word database for this particular file.
1392 putkey(const struct mpage *mpage, char *value, uint64_t type)
1394 putkeys(mpage, value, strlen(value), type);
1398 * Grok all nodes at or below a certain mdoc node into putkey().
1401 putmdockey(const struct mpage *mpage,
1402 const struct roff_node *n, uint64_t m, int taboo)
1405 for ( ; NULL != n; n = n->next) {
1406 if (n->flags & taboo)
1408 if (NULL != n->child)
1409 putmdockey(mpage, n->child, m, taboo);
1410 if (n->type == ROFFT_TEXT)
1411 putkey(mpage, n->string, m);
1416 parse_man(struct mpage *mpage, const struct roff_meta *meta,
1417 const struct roff_node *n)
1419 const struct roff_node *head, *body;
1420 char *start, *title;
1428 * We're only searching for one thing: the first text child in
1429 * the BODY of a NAME section. Since we don't keep track of
1430 * sections in -man, run some hoops to find out whether we're in
1431 * the correct section or not.
1434 if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
1436 if ((head = body->parent->head) != NULL &&
1437 (head = head->child) != NULL &&
1438 head->next == NULL &&
1439 head->type == ROFFT_TEXT &&
1440 strcmp(head->string, "NAME") == 0 &&
1441 body->child != NULL) {
1444 * Suck the entire NAME section into memory.
1445 * Yes, we might run away.
1446 * But too many manuals have big, spread-out
1447 * NAME sections over many lines.
1451 deroff(&title, body);
1456 * Go through a special heuristic dance here.
1457 * Conventionally, one or more manual names are
1458 * comma-specified prior to a whitespace, then a
1459 * dash, then a description. Try to puzzle out
1460 * the name parts here.
1465 sz = strcspn(start, " ,");
1466 if ('\0' == start[sz])
1473 * Assume a stray trailing comma in the
1474 * name list if a name begins with a dash.
1477 if ('-' == start[0] ||
1478 ('\\' == start[0] && '-' == start[1]))
1481 putkey(mpage, start, NAME_TITLE);
1482 if ( ! (mpage->name_head_done ||
1483 strcasecmp(start, meta->title))) {
1484 putkey(mpage, start, NAME_HEAD);
1485 mpage->name_head_done = 1;
1493 assert(',' == byte);
1495 while (' ' == *start)
1499 if (start == title) {
1500 putkey(mpage, start, NAME_TITLE);
1501 if ( ! (mpage->name_head_done ||
1502 strcasecmp(start, meta->title))) {
1503 putkey(mpage, start, NAME_HEAD);
1504 mpage->name_head_done = 1;
1510 while (isspace((unsigned char)*start))
1513 if (0 == strncmp(start, "-", 1))
1515 else if (0 == strncmp(start, "\\-\\-", 4))
1517 else if (0 == strncmp(start, "\\-", 2))
1519 else if (0 == strncmp(start, "\\(en", 4))
1521 else if (0 == strncmp(start, "\\(em", 4))
1524 while (' ' == *start)
1528 * Cut off excessive one-line descriptions.
1529 * Bad pages are not worth better heuristics.
1532 mpage->desc = mandoc_strndup(start, 150);
1538 for (n = n->child; n; n = n->next) {
1539 if (NULL != mpage->desc)
1541 parse_man(mpage, meta, n);
1546 parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
1547 const struct roff_node *n)
1549 const struct mdoc_handler *handler;
1551 for (n = n->child; n != NULL; n = n->next) {
1552 if (n->tok == TOKEN_NONE || n->tok < ROFF_MAX)
1554 assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX);
1555 handler = mdoc_handlers + (n->tok - MDOC_Dd);
1556 if (n->flags & handler->taboo)
1565 if (handler->fp != NULL &&
1566 (*handler->fp)(mpage, meta, n) == 0)
1569 putmdockey(mpage, n->child,
1570 handler->mask, handler->taboo);
1575 if (NULL != n->child)
1576 parse_mdoc(mpage, meta, n);
1581 parse_mdoc_Fa(struct mpage *mpage, const struct roff_meta *meta,
1582 const struct roff_node *n)
1587 if (n->sec == SEC_SYNOPSIS)
1590 putmdockey(mpage, n->child, mask, 0);
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)
1665 if (n->child == NULL)
1668 parse_mdoc_fname(mpage, n->child);
1671 if (n != NULL && n->type == ROFFT_TEXT) {
1673 if (n->sec == SEC_SYNOPSIS)
1675 putmdockey(mpage, n, mask, 0);
1682 parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
1683 const struct roff_node *n)
1686 if (n->type != ROFFT_HEAD)
1689 if (n->child != NULL)
1690 parse_mdoc_fname(mpage, n->child);
1696 parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
1697 const struct roff_node *n)
1701 if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
1704 if (n->child != NULL &&
1705 n->child->next == NULL &&
1706 n->child->type == ROFFT_TEXT)
1712 putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
1713 n->type == ROFFT_BODY ? TYPE_Va : 0));
1721 parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
1722 const struct roff_node *n)
1726 if (NULL == (n = n->child))
1729 if (NULL == n->next) {
1730 putkey(mpage, n->string, TYPE_Xr);
1734 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1735 putkey(mpage, cp, TYPE_Xr);
1741 parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
1742 const struct roff_node *n)
1745 if (n->type == ROFFT_BODY)
1746 deroff(&mpage->desc, n);
1751 parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
1752 const struct roff_node *n)
1755 if (SEC_NAME == n->sec)
1756 putmdockey(mpage, n->child, NAME_TITLE, 0);
1757 else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
1758 if (n->child == NULL)
1759 putkey(mpage, meta->name, NAME_SYN);
1761 putmdockey(mpage, n->child, NAME_SYN, 0);
1763 if ( ! (mpage->name_head_done ||
1764 n->child == NULL || n->child->string == NULL ||
1765 strcasecmp(n->child->string, meta->title))) {
1766 putkey(mpage, n->child->string, NAME_HEAD);
1767 mpage->name_head_done = 1;
1773 parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
1774 const struct roff_node *n)
1777 return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
1781 parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
1782 const struct roff_node *n)
1785 return n->type == ROFFT_HEAD;
1789 * Add a string to the hash table for the current manual.
1790 * Each string has a bitmask telling which macros it belongs to.
1791 * When we finish the manual, we'll dump the table.
1794 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1805 mustfree = render_string(&cp, &sz);
1811 name_mask &= ~NAME_FIRST;
1813 say(mpage->mlinks->file,
1814 "Adding name %*s, bits=0x%llx", (int)sz, cp,
1815 (unsigned long long)v);
1819 for (i = 0; i < KEY_MAX; i++)
1820 if ((uint64_t)1 << i & v)
1821 say(mpage->mlinks->file,
1822 "Adding key %s=%*s",
1823 mansearch_keynames[i], (int)sz, cp);
1827 slot = ohash_qlookupi(htab, cp, &end);
1828 s = ohash_find(htab, slot);
1830 if (NULL != s && mpage == s->mpage) {
1833 } else if (NULL == s) {
1834 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1835 memcpy(s->key, cp, sz);
1836 ohash_insert(htab, slot, s);
1846 * Take a Unicode codepoint and produce its UTF-8 encoding.
1847 * This isn't the best way to do this, but it works.
1848 * The magic numbers are from the UTF-8 packaging.
1849 * They're not as scary as they seem: read the UTF-8 spec for details.
1852 utf8(unsigned int cp, char out[7])
1857 if (cp <= 0x0000007F) {
1860 } else if (cp <= 0x000007FF) {
1862 out[0] = (cp >> 6 & 31) | 192;
1863 out[1] = (cp & 63) | 128;
1864 } else if (cp <= 0x0000FFFF) {
1866 out[0] = (cp >> 12 & 15) | 224;
1867 out[1] = (cp >> 6 & 63) | 128;
1868 out[2] = (cp & 63) | 128;
1869 } else if (cp <= 0x001FFFFF) {
1871 out[0] = (cp >> 18 & 7) | 240;
1872 out[1] = (cp >> 12 & 63) | 128;
1873 out[2] = (cp >> 6 & 63) | 128;
1874 out[3] = (cp & 63) | 128;
1875 } else if (cp <= 0x03FFFFFF) {
1877 out[0] = (cp >> 24 & 3) | 248;
1878 out[1] = (cp >> 18 & 63) | 128;
1879 out[2] = (cp >> 12 & 63) | 128;
1880 out[3] = (cp >> 6 & 63) | 128;
1881 out[4] = (cp & 63) | 128;
1882 } else if (cp <= 0x7FFFFFFF) {
1884 out[0] = (cp >> 30 & 1) | 252;
1885 out[1] = (cp >> 24 & 63) | 128;
1886 out[2] = (cp >> 18 & 63) | 128;
1887 out[3] = (cp >> 12 & 63) | 128;
1888 out[4] = (cp >> 6 & 63) | 128;
1889 out[5] = (cp & 63) | 128;
1898 * If the string contains escape sequences,
1899 * replace it with an allocated rendering and return 1,
1900 * such that the caller can free it after use.
1901 * Otherwise, do nothing and return 0.
1904 render_string(char **public, size_t *psz)
1906 const char *src, *scp, *addcp, *seq;
1908 size_t ssz, dsz, addsz;
1909 char utfbuf[7], res[6];
1910 int seqlen, unicode;
1914 res[2] = ASCII_NBRSP;
1915 res[3] = ASCII_HYPH;
1916 res[4] = ASCII_BREAK;
1919 src = scp = *public;
1924 while (scp < src + *psz) {
1926 /* Leave normal characters unchanged. */
1928 if (strchr(res, *scp) == NULL) {
1936 * Found something that requires replacing,
1937 * make sure we have a destination buffer.
1941 dst = mandoc_malloc(ssz + 1);
1943 memcpy(dst, src, dsz);
1946 /* Handle single-char special characters. */
1967 * Found an escape sequence.
1968 * Read past the slash, then parse it.
1969 * Ignore everything except characters.
1973 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1977 * Render the special character
1978 * as either UTF-8 or ASCII.
1982 unicode = mchars_spec2cp(seq, seqlen);
1985 addsz = utf8(unicode, utfbuf);
1990 addcp = mchars_spec2str(seq, seqlen, &addsz);
1993 if (*addcp == ASCII_NBRSP) {
1999 /* Copy the rendered glyph into the stream. */
2002 dst = mandoc_realloc(dst, ssz + 1);
2003 memcpy(dst + dsz, addcp, addsz);
2011 /* Trim trailing whitespace and NUL-terminate. */
2013 while (*psz > 0 && (*public)[*psz - 1] == ' ')
2016 (*public)[*psz] = '\0';
2023 dbadd_mlink(const struct mlink *mlink)
2025 dba_page_alias(mlink->mpage->dba, mlink->name, NAME_FILE);
2026 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->dsec);
2027 dba_page_add(mlink->mpage->dba, DBP_SECT, mlink->fsec);
2028 dba_page_add(mlink->mpage->dba, DBP_ARCH, mlink->arch);
2029 dba_page_add(mlink->mpage->dba, DBP_FILE, mlink->file);
2033 * Flush the current page's terms (and their bits) into the database.
2034 * Also, handle escape sequences at the last possible moment.
2037 dbadd(struct dba *dba, struct mpage *mpage)
2039 struct mlink *mlink;
2047 mlink = mpage->mlinks;
2050 for (key = ohash_first(&names, &slot); NULL != key;
2051 key = ohash_next(&names, &slot))
2053 for (key = ohash_first(&strings, &slot); NULL != key;
2054 key = ohash_next(&strings, &slot))
2058 while (NULL != mlink) {
2059 fputs(mlink->name, stdout);
2060 if (NULL == mlink->next ||
2061 strcmp(mlink->dsec, mlink->next->dsec) ||
2062 strcmp(mlink->fsec, mlink->next->fsec) ||
2063 strcmp(mlink->arch, mlink->next->arch)) {
2065 if ('\0' == *mlink->dsec)
2066 fputs(mlink->fsec, stdout);
2068 fputs(mlink->dsec, stdout);
2069 if ('\0' != *mlink->arch)
2070 printf("/%s", mlink->arch);
2073 mlink = mlink->next;
2075 fputs(", ", stdout);
2077 printf(" - %s\n", mpage->desc);
2082 say(mlink->file, "Adding to database");
2086 mustfree = render_string(&cp, &i);
2087 mpage->dba = dba_page_new(dba->pages,
2088 *mpage->arch == '\0' ? mlink->arch : mpage->arch,
2089 cp, mlink->file, mpage->form);
2092 dba_page_add(mpage->dba, DBP_SECT, mpage->sec);
2094 while (mlink != NULL) {
2096 mlink = mlink->next;
2099 for (key = ohash_first(&names, &slot); NULL != key;
2100 key = ohash_next(&names, &slot)) {
2101 assert(key->mpage == mpage);
2102 dba_page_alias(mpage->dba, key->key, key->mask);
2105 for (key = ohash_first(&strings, &slot); NULL != key;
2106 key = ohash_next(&strings, &slot)) {
2107 assert(key->mpage == mpage);
2109 for (mask = TYPE_Xr; mask <= TYPE_Lb; mask *= 2) {
2110 if (key->mask & mask)
2111 dba_macro_add(dba->macros, i,
2112 key->key, mpage->dba);
2120 dbprune(struct dba *dba)
2122 struct dba_array *page, *files;
2125 dba_array_FOREACH(dba->pages, page) {
2126 files = dba_array_get(page, DBP_FILE);
2127 dba_array_FOREACH(files, file) {
2130 if (ohash_find(&mlinks, ohash_qlookup(&mlinks,
2133 say(file, "Deleting from database");
2134 dba_array_del(dba->pages);
2142 * Write the database from memory to disk.
2145 dbwrite(struct dba *dba)
2147 struct stat sb1, sb2;
2148 char tfn[33], *cp1, *cp2;
2153 * Do not write empty databases, and delete existing ones
2154 * when makewhatis -u causes them to become empty.
2157 dba_array_start(dba->pages);
2158 if (dba_array_next(dba->pages) == NULL) {
2159 if (unlink(MANDOC_DB) == -1 && errno != ENOENT)
2160 say(MANDOC_DB, "&unlink");
2165 * Build the database in a temporary file,
2166 * then atomically move it into place.
2169 if (dba_write(MANDOC_DB "~", dba) != -1) {
2170 if (rename(MANDOC_DB "~", MANDOC_DB) == -1) {
2171 exitcode = (int)MANDOCLEVEL_SYSERR;
2172 say(MANDOC_DB, "&rename");
2173 unlink(MANDOC_DB "~");
2179 * We lack write permission and cannot replace the database
2180 * file, but let's at least check whether the data changed.
2183 (void)strlcpy(tfn, "/tmp/mandocdb.XXXXXXXX", sizeof(tfn));
2184 if (mkdtemp(tfn) == NULL) {
2185 exitcode = (int)MANDOCLEVEL_SYSERR;
2186 say("", "&%s", tfn);
2189 cp1 = cp2 = MAP_FAILED;
2191 (void)strlcat(tfn, "/" MANDOC_DB, sizeof(tfn));
2192 if (dba_write(tfn, dba) == -1) {
2193 say(tfn, "&dba_write");
2196 if ((fd1 = open(MANDOC_DB, O_RDONLY, 0)) == -1) {
2197 say(MANDOC_DB, "&open");
2200 if ((fd2 = open(tfn, O_RDONLY, 0)) == -1) {
2204 if (fstat(fd1, &sb1) == -1) {
2205 say(MANDOC_DB, "&fstat");
2208 if (fstat(fd2, &sb2) == -1) {
2212 if (sb1.st_size != sb2.st_size)
2214 if ((cp1 = mmap(NULL, sb1.st_size, PROT_READ, MAP_PRIVATE,
2215 fd1, 0)) == MAP_FAILED) {
2216 say(MANDOC_DB, "&mmap");
2219 if ((cp2 = mmap(NULL, sb2.st_size, PROT_READ, MAP_PRIVATE,
2220 fd2, 0)) == MAP_FAILED) {
2224 for (i = 0; i < sb1.st_size; i++)
2225 if (cp1[i] != cp2[i])
2230 exitcode = (int)MANDOCLEVEL_SYSERR;
2231 say(MANDOC_DB, "Data changed, but cannot replace database");
2234 if (cp1 != MAP_FAILED)
2235 munmap(cp1, sb1.st_size);
2236 if (cp2 != MAP_FAILED)
2237 munmap(cp2, sb2.st_size);
2243 *strrchr(tfn, '/') = '\0';
2248 set_basedir(const char *targetdir, int report_baddir)
2250 static char startdir[PATH_MAX];
2251 static int getcwd_status; /* 1 = ok, 2 = failure */
2252 static int chdir_status; /* 1 = changed directory */
2256 * Remember the original working directory, if possible.
2257 * This will be needed if the second or a later directory
2258 * on the command line is given as a relative path.
2259 * Do not error out if the current directory is not
2260 * searchable: Maybe it won't be needed after all.
2262 if (0 == getcwd_status) {
2263 if (NULL == getcwd(startdir, sizeof(startdir))) {
2265 (void)strlcpy(startdir, strerror(errno),
2272 * We are leaving the old base directory.
2273 * Do not use it any longer, not even for messages.
2278 * If and only if the directory was changed earlier and
2279 * the next directory to process is given as a relative path,
2280 * first go back, or bail out if that is impossible.
2282 if (chdir_status && '/' != *targetdir) {
2283 if (2 == getcwd_status) {
2284 exitcode = (int)MANDOCLEVEL_SYSERR;
2285 say("", "getcwd: %s", startdir);
2288 if (-1 == chdir(startdir)) {
2289 exitcode = (int)MANDOCLEVEL_SYSERR;
2290 say("", "&chdir %s", startdir);
2296 * Always resolve basedir to the canonicalized absolute
2297 * pathname and append a trailing slash, such that
2298 * we can reliably check whether files are inside.
2300 if (NULL == realpath(targetdir, basedir)) {
2301 if (report_baddir || errno != ENOENT) {
2302 exitcode = (int)MANDOCLEVEL_BADARG;
2303 say("", "&%s: realpath", targetdir);
2306 } else if (-1 == chdir(basedir)) {
2307 if (report_baddir || errno != ENOENT) {
2308 exitcode = (int)MANDOCLEVEL_BADARG;
2314 cp = strchr(basedir, '\0');
2315 if ('/' != cp[-1]) {
2316 if (cp - basedir >= PATH_MAX - 1) {
2317 exitcode = (int)MANDOCLEVEL_SYSERR;
2318 say("", "Filename too long");
2328 say(const char *file, const char *format, ...)
2333 if ('\0' != *basedir)
2334 fprintf(stderr, "%s", basedir);
2335 if ('\0' != *basedir && '\0' != *file)
2338 fprintf(stderr, "%s", file);
2341 if (NULL != format) {
2354 if (NULL != format) {
2355 if ('\0' != *basedir || '\0' != *file)
2356 fputs(": ", stderr);
2357 va_start(ap, format);
2358 vfprintf(stderr, format, ap);
2362 if ('\0' != *basedir || '\0' != *file || NULL != format)
2363 fputs(": ", stderr);
2366 fputc('\n', stderr);