1 /* $Id: mandocdb.c,v 1.179 2014/12/09 07:29:42 schwarze Exp $ */
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
20 #include <sys/types.h>
31 #include "compat_fts.h"
45 #include "compat_ohash.h"
52 #include "mandoc_aux.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 mdoc_meta *,
133 const struct mdoc_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 *hash_alloc(size_t, void *);
148 static void hash_free(void *, void *);
149 static void *hash_calloc(size_t, size_t, void *);
150 static void mlink_add(struct mlink *, const struct stat *);
151 static void mlink_check(struct mpage *, struct mlink *);
152 static void mlink_free(struct mlink *);
153 static void mlinks_undupe(struct mpage *);
154 static void mpages_free(void);
155 static void mpages_merge(struct mparse *);
156 static void names_check(void);
157 static void parse_cat(struct mpage *, int);
158 static void parse_man(struct mpage *, const struct man_meta *,
159 const struct man_node *);
160 static void parse_mdoc(struct mpage *, const struct mdoc_meta *,
161 const struct mdoc_node *);
162 static int parse_mdoc_body(struct mpage *, const struct mdoc_meta *,
163 const struct mdoc_node *);
164 static int parse_mdoc_head(struct mpage *, const struct mdoc_meta *,
165 const struct mdoc_node *);
166 static int parse_mdoc_Fd(struct mpage *, const struct mdoc_meta *,
167 const struct mdoc_node *);
168 static void parse_mdoc_fname(struct mpage *, const struct mdoc_node *);
169 static int parse_mdoc_Fn(struct mpage *, const struct mdoc_meta *,
170 const struct mdoc_node *);
171 static int parse_mdoc_Fo(struct mpage *, const struct mdoc_meta *,
172 const struct mdoc_node *);
173 static int parse_mdoc_Nd(struct mpage *, const struct mdoc_meta *,
174 const struct mdoc_node *);
175 static int parse_mdoc_Nm(struct mpage *, const struct mdoc_meta *,
176 const struct mdoc_node *);
177 static int parse_mdoc_Sh(struct mpage *, const struct mdoc_meta *,
178 const struct mdoc_node *);
179 static int parse_mdoc_Xr(struct mpage *, const struct mdoc_meta *,
180 const struct mdoc_node *);
181 static void putkey(const struct mpage *, char *, uint64_t);
182 static void putkeys(const struct mpage *, char *, size_t, uint64_t);
183 static void putmdockey(const struct mpage *,
184 const struct mdoc_node *, uint64_t);
185 static int render_string(char **, size_t *);
186 static void say(const char *, const char *, ...);
187 static int set_basedir(const char *, int);
188 static int treescan(void);
189 static size_t utf8(unsigned int, char [7]);
191 static char tempfilename[32];
192 static char *progname;
193 static int nodb; /* no database changes */
194 static int mparse_options; /* abort the parse early */
195 static int use_all; /* use all found files */
196 static int debug; /* print what we're doing */
197 static int warnings; /* warn about crap */
198 static int write_utf8; /* write UTF-8 output; else ASCII */
199 static int exitcode; /* to be returned by main */
200 static enum op op; /* operational mode */
201 static char basedir[PATH_MAX]; /* current base directory */
202 static struct mchars *mchars; /* table of named characters */
203 static struct ohash mpages; /* table of distinct manual pages */
204 static struct ohash mlinks; /* table of directory entries */
205 static struct ohash names; /* table of all names */
206 static struct ohash strings; /* table of all strings */
207 static sqlite3 *db = NULL; /* current database */
208 static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
209 static uint64_t name_mask;
211 static const struct mdoc_handler mdocs[MDOC_MAX] = {
212 { NULL, 0 }, /* Ap */
213 { NULL, 0 }, /* Dd */
214 { NULL, 0 }, /* Dt */
215 { NULL, 0 }, /* Os */
216 { parse_mdoc_Sh, TYPE_Sh }, /* Sh */
217 { parse_mdoc_head, TYPE_Ss }, /* Ss */
218 { NULL, 0 }, /* Pp */
219 { NULL, 0 }, /* D1 */
220 { NULL, 0 }, /* Dl */
221 { NULL, 0 }, /* Bd */
222 { NULL, 0 }, /* Ed */
223 { NULL, 0 }, /* Bl */
224 { NULL, 0 }, /* El */
225 { NULL, 0 }, /* It */
226 { NULL, 0 }, /* Ad */
227 { NULL, TYPE_An }, /* An */
228 { NULL, TYPE_Ar }, /* Ar */
229 { NULL, TYPE_Cd }, /* Cd */
230 { NULL, TYPE_Cm }, /* Cm */
231 { NULL, TYPE_Dv }, /* Dv */
232 { NULL, TYPE_Er }, /* Er */
233 { NULL, TYPE_Ev }, /* Ev */
234 { NULL, 0 }, /* Ex */
235 { NULL, TYPE_Fa }, /* Fa */
236 { parse_mdoc_Fd, 0 }, /* Fd */
237 { NULL, TYPE_Fl }, /* Fl */
238 { parse_mdoc_Fn, 0 }, /* Fn */
239 { NULL, TYPE_Ft }, /* Ft */
240 { NULL, TYPE_Ic }, /* Ic */
241 { NULL, TYPE_In }, /* In */
242 { NULL, TYPE_Li }, /* Li */
243 { parse_mdoc_Nd, 0 }, /* Nd */
244 { parse_mdoc_Nm, 0 }, /* Nm */
245 { NULL, 0 }, /* Op */
246 { NULL, 0 }, /* Ot */
247 { NULL, TYPE_Pa }, /* Pa */
248 { NULL, 0 }, /* Rv */
249 { NULL, TYPE_St }, /* St */
250 { NULL, TYPE_Va }, /* Va */
251 { parse_mdoc_body, TYPE_Va }, /* Vt */
252 { parse_mdoc_Xr, 0 }, /* Xr */
253 { NULL, 0 }, /* %A */
254 { NULL, 0 }, /* %B */
255 { NULL, 0 }, /* %D */
256 { NULL, 0 }, /* %I */
257 { NULL, 0 }, /* %J */
258 { NULL, 0 }, /* %N */
259 { NULL, 0 }, /* %O */
260 { NULL, 0 }, /* %P */
261 { NULL, 0 }, /* %R */
262 { NULL, 0 }, /* %T */
263 { NULL, 0 }, /* %V */
264 { NULL, 0 }, /* Ac */
265 { NULL, 0 }, /* Ao */
266 { NULL, 0 }, /* Aq */
267 { NULL, TYPE_At }, /* At */
268 { NULL, 0 }, /* Bc */
269 { NULL, 0 }, /* Bf */
270 { NULL, 0 }, /* Bo */
271 { NULL, 0 }, /* Bq */
272 { NULL, TYPE_Bsx }, /* Bsx */
273 { NULL, TYPE_Bx }, /* Bx */
274 { NULL, 0 }, /* Db */
275 { NULL, 0 }, /* Dc */
276 { NULL, 0 }, /* Do */
277 { NULL, 0 }, /* Dq */
278 { NULL, 0 }, /* Ec */
279 { NULL, 0 }, /* Ef */
280 { NULL, TYPE_Em }, /* Em */
281 { NULL, 0 }, /* Eo */
282 { NULL, TYPE_Fx }, /* Fx */
283 { NULL, TYPE_Ms }, /* Ms */
284 { NULL, 0 }, /* No */
285 { NULL, 0 }, /* Ns */
286 { NULL, TYPE_Nx }, /* Nx */
287 { NULL, TYPE_Ox }, /* Ox */
288 { NULL, 0 }, /* Pc */
289 { NULL, 0 }, /* Pf */
290 { NULL, 0 }, /* Po */
291 { NULL, 0 }, /* Pq */
292 { NULL, 0 }, /* Qc */
293 { NULL, 0 }, /* Ql */
294 { NULL, 0 }, /* Qo */
295 { NULL, 0 }, /* Qq */
296 { NULL, 0 }, /* Re */
297 { NULL, 0 }, /* Rs */
298 { NULL, 0 }, /* Sc */
299 { NULL, 0 }, /* So */
300 { NULL, 0 }, /* Sq */
301 { NULL, 0 }, /* Sm */
302 { NULL, 0 }, /* Sx */
303 { NULL, TYPE_Sy }, /* Sy */
304 { NULL, TYPE_Tn }, /* Tn */
305 { NULL, 0 }, /* Ux */
306 { NULL, 0 }, /* Xc */
307 { NULL, 0 }, /* Xo */
308 { parse_mdoc_Fo, 0 }, /* Fo */
309 { NULL, 0 }, /* Fc */
310 { NULL, 0 }, /* Oo */
311 { NULL, 0 }, /* Oc */
312 { NULL, 0 }, /* Bk */
313 { NULL, 0 }, /* Ek */
314 { NULL, 0 }, /* Bt */
315 { NULL, 0 }, /* Hf */
316 { NULL, 0 }, /* Fr */
317 { NULL, 0 }, /* Ud */
318 { NULL, TYPE_Lb }, /* Lb */
319 { NULL, 0 }, /* Lp */
320 { NULL, TYPE_Lk }, /* Lk */
321 { NULL, TYPE_Mt }, /* Mt */
322 { NULL, 0 }, /* Brq */
323 { NULL, 0 }, /* Bro */
324 { NULL, 0 }, /* Brc */
325 { NULL, 0 }, /* %C */
326 { NULL, 0 }, /* Es */
327 { NULL, 0 }, /* En */
328 { NULL, TYPE_Dx }, /* Dx */
329 { NULL, 0 }, /* %Q */
330 { NULL, 0 }, /* br */
331 { NULL, 0 }, /* sp */
332 { NULL, 0 }, /* %U */
333 { NULL, 0 }, /* Ta */
334 { NULL, 0 }, /* ll */
339 mandocdb(int argc, char *argv[])
343 const char *path_arg;
344 struct manpaths dirs;
346 struct ohash_info mpages_info, mlinks_info;
348 memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
349 memset(&dirs, 0, sizeof(struct manpaths));
351 mpages_info.alloc = mlinks_info.alloc = hash_alloc;
352 mpages_info.calloc = mlinks_info.calloc = hash_calloc;
353 mpages_info.free = mlinks_info.free = hash_free;
355 mpages_info.key_offset = offsetof(struct mpage, inodev);
356 mlinks_info.key_offset = offsetof(struct mlink, file);
358 progname = strrchr(argv[0], '/');
359 if (progname == NULL)
365 * We accept a few different invocations.
366 * The CHECKOP macro makes sure that invocation styles don't
367 * clobber each other.
369 #define CHECKOP(_op, _ch) do \
370 if (OP_DEFAULT != (_op)) { \
371 fprintf(stderr, "%s: -%c: Conflicting option\n", \
374 } while (/*CONSTCOND*/0)
379 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
404 mparse_options |= MPARSE_QUICK;
407 if (strcmp(optarg, "utf8")) {
408 fprintf(stderr, "%s: -T%s: "
409 "Unsupported output format\n",
417 dup2(STDOUT_FILENO, STDERR_FILENO);
427 /* Compatibility with espie@'s makewhatis. */
436 if (OP_CONFFILE == op && argc > 0) {
437 fprintf(stderr, "%s: -C: Too many arguments\n",
442 exitcode = (int)MANDOCLEVEL_OK;
443 mchars = mchars_alloc();
444 mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL,
446 ohash_init(&mpages, 6, &mpages_info);
447 ohash_init(&mlinks, 6, &mlinks_info);
449 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
452 * Most of these deal with a specific directory.
453 * Jump into that directory first.
455 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
460 * The existing database is usable. Process
461 * all files specified on the command-line.
464 for (i = 0; i < argc; i++)
470 * Database missing or corrupt.
471 * Recreate from scratch.
473 exitcode = (int)MANDOCLEVEL_OK;
482 dbclose(OP_DEFAULT == op ? 0 : 1);
485 * If we have arguments, use them as our manpaths.
486 * If we don't, grok from manpath(1) or however else
487 * manpath_parse() wants to do it.
490 dirs.paths = mandoc_reallocarray(NULL,
491 argc, sizeof(char *));
492 dirs.sz = (size_t)argc;
493 for (i = 0; i < argc; i++)
494 dirs.paths[i] = mandoc_strdup(argv[i]);
496 manpath_parse(&dirs, path_arg, NULL, NULL);
499 exitcode = (int)MANDOCLEVEL_BADARG;
500 say("", "Empty manpath");
504 * First scan the tree rooted at a base directory, then
505 * build a new database and finally move it into place.
506 * Ignore zero-length directories and strip trailing
509 for (j = 0; j < dirs.sz; j++) {
510 sz = strlen(dirs.paths[j]);
511 if (sz && '/' == dirs.paths[j][sz - 1])
512 dirs.paths[j][--sz] = '\0';
517 ohash_init(&mpages, 6, &mpages_info);
518 ohash_init(&mlinks, 6, &mlinks_info);
521 if (0 == set_basedir(dirs.paths[j], argc > 0))
529 if (warnings && !nodb &&
530 ! (MPARSE_QUICK & mparse_options))
534 if (j + 1 < dirs.sz) {
536 ohash_delete(&mpages);
537 ohash_delete(&mlinks);
546 ohash_delete(&mpages);
547 ohash_delete(&mlinks);
550 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
551 " %s [-aDnpQ] [-Tutf8] dir ...\n"
552 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
553 " %s [-Dnp] -u dir [file ...]\n"
554 " %s [-Q] -t file ...\n",
555 progname, progname, progname,
558 return((int)MANDOCLEVEL_BADARG);
562 * Scan a directory tree rooted at "basedir" for manpages.
563 * We use fts(), scanning directory parts along the way for clues to our
564 * section and architecture.
566 * If use_all has been specified, grok all files.
567 * If not, sanitise paths to the following:
569 * [./]man*[/<arch>]/<name>.<section>
571 * [./]cat<section>[/<arch>]/<name>.0
573 * TODO: accomodate for multi-language directories.
583 char *dsec, *arch, *fsec, *cp;
588 argv[1] = (char *)NULL;
590 f = fts_open((char * const *)argv,
591 FTS_PHYSICAL | FTS_NOCHDIR, NULL);
593 exitcode = (int)MANDOCLEVEL_SYSERR;
594 say("", "&fts_open");
601 while (NULL != (ff = fts_read(f))) {
602 path = ff->fts_path + 2;
603 switch (ff->fts_info) {
606 * Symbolic links require various sanity checks,
607 * then get handled just like regular files.
610 if (NULL == realpath(path, buf)) {
612 say(path, "&realpath");
615 if (strstr(buf, basedir) != buf) {
616 if (warnings) say("",
617 "%s: outside base directory", buf);
620 /* Use logical inode to avoid mpages dupe. */
621 if (-1 == stat(path, ff->fts_statp)) {
629 * If we're a regular file, add an mlink by using the
630 * stored directory data and handling the filename.
633 if (0 == strcmp(path, MANDOC_DB))
635 if ( ! use_all && ff->fts_level < 2) {
637 say(path, "Extraneous file");
642 while (NULL == fsec) {
643 fsec = strrchr(ff->fts_name, '.');
644 if (NULL == fsec || strcmp(fsec+1, "gz"))
654 "No filename suffix");
657 } else if (0 == strcmp(++fsec, "html")) {
659 say(path, "Skip html");
661 } else if (0 == strcmp(fsec, "ps")) {
663 say(path, "Skip ps");
665 } else if (0 == strcmp(fsec, "pdf")) {
667 say(path, "Skip pdf");
669 } else if ( ! use_all &&
670 ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
671 (FORM_CAT == dform && strcmp(fsec, "0")))) {
673 say(path, "Wrong filename suffix");
678 mlink = mandoc_calloc(1, sizeof(struct mlink));
679 if (strlcpy(mlink->file, path,
680 sizeof(mlink->file)) >=
681 sizeof(mlink->file)) {
682 say(path, "Filename too long");
686 mlink->dform = dform;
689 mlink->name = ff->fts_name;
692 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 (FTS_DP == ff->fts_info)
720 if (0 == strncmp(cp, "man", 3)) {
723 } else if (0 == strncmp(cp, "cat", 3)) {
731 if (NULL != dsec || use_all)
735 say(path, "Unknown directory part");
736 fts_set(f, ff, FTS_SKIP);
740 * Possibly our architecture.
741 * If we're descending, keep tabs on it.
743 if (FTS_DP != ff->fts_info && NULL != dsec)
749 if (FTS_DP == ff->fts_info || use_all)
752 say(path, "Extraneous directory part");
753 fts_set(f, ff, FTS_SKIP);
763 * Add a file to the mlinks table.
764 * Do not verify that it's a "valid" looking manpage (we'll do that
767 * Try to infer the manual section, architecture, and page name from the
768 * path, assuming it looks like
770 * [./]man*[/<arch>]/<name>.<section>
772 * [./]cat<section>[/<arch>]/<name>.0
774 * See treescan() for the fts(3) version of this.
777 filescan(const char *file)
786 if (0 == strncmp(file, "./", 2))
790 * We have to do lstat(2) before realpath(3) loses
791 * the information whether this is a symbolic link.
792 * We need to know that because for symbolic links,
793 * we want to use the orginal file name, while for
794 * regular files, we want to use the real path.
796 if (-1 == lstat(file, &st)) {
797 exitcode = (int)MANDOCLEVEL_BADARG;
800 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
801 exitcode = (int)MANDOCLEVEL_BADARG;
802 say(file, "Not a regular file");
807 * We have to resolve the file name to the real path
808 * in any case for the base directory check.
810 if (NULL == realpath(file, buf)) {
811 exitcode = (int)MANDOCLEVEL_BADARG;
812 say(file, "&realpath");
818 else if (strstr(buf, basedir) == buf)
819 start = buf + strlen(basedir);
821 exitcode = (int)MANDOCLEVEL_BADARG;
822 say("", "%s: outside base directory", buf);
827 * Now we are sure the file is inside our tree.
828 * If it is a symbolic link, ignore the real path
829 * and use the original name.
830 * This implies passing stuff like "cat1/../man1/foo.1"
831 * on the command line won't work. So don't do that.
832 * Note the stat(2) can still fail if the link target
835 if (S_IFLNK & st.st_mode) {
836 if (-1 == stat(buf, &st)) {
837 exitcode = (int)MANDOCLEVEL_BADARG;
841 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
842 say(file, "Filename too long");
846 if (OP_TEST != op && strstr(buf, basedir) == buf)
847 start += strlen(basedir);
850 mlink = mandoc_calloc(1, sizeof(struct mlink));
851 mlink->dform = FORM_NONE;
852 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
853 sizeof(mlink->file)) {
854 say(start, "Filename too long");
859 * First try to guess our directory structure.
860 * If we find a separator, try to look for man* or cat*.
861 * If we find one of these and what's underneath is a directory,
862 * assume it's an architecture.
864 if (NULL != (p = strchr(start, '/'))) {
866 if (0 == strncmp(start, "man", 3)) {
867 mlink->dform = FORM_SRC;
868 mlink->dsec = start + 3;
869 } else if (0 == strncmp(start, "cat", 3)) {
870 mlink->dform = FORM_CAT;
871 mlink->dsec = start + 3;
875 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
883 * Now check the file suffix.
884 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
886 p = strrchr(start, '\0');
887 while (p-- > start && '/' != *p && '.' != *p)
896 * Now try to parse the name.
897 * Use the filename portion of the path.
900 if (NULL != (p = strrchr(start, '/'))) {
904 mlink_add(mlink, &st);
908 mlink_add(struct mlink *mlink, const struct stat *st)
910 struct inodev inodev;
914 assert(NULL != mlink->file);
916 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
917 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
918 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
919 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
921 if ('0' == *mlink->fsec) {
923 mlink->fsec = mandoc_strdup(mlink->dsec);
924 mlink->fform = FORM_CAT;
925 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
926 mlink->fform = FORM_SRC;
928 mlink->fform = FORM_NONE;
930 slot = ohash_qlookup(&mlinks, mlink->file);
931 assert(NULL == ohash_find(&mlinks, slot));
932 ohash_insert(&mlinks, slot, mlink);
934 inodev.st_ino = st->st_ino;
935 inodev.st_dev = st->st_dev;
936 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
937 sizeof(struct inodev), inodev.st_ino);
938 mpage = ohash_find(&mpages, slot);
940 mpage = mandoc_calloc(1, sizeof(struct mpage));
941 mpage->inodev.st_ino = inodev.st_ino;
942 mpage->inodev.st_dev = inodev.st_dev;
943 ohash_insert(&mpages, slot, mpage);
945 mlink->next = mpage->mlinks;
946 mpage->mlinks = mlink;
947 mlink->mpage = mpage;
951 mlink_free(struct mlink *mlink)
968 mpage = ohash_first(&mpages, &slot);
969 while (NULL != mpage) {
970 while (NULL != (mlink = mpage->mlinks)) {
971 mpage->mlinks = mlink->next;
979 mpage = ohash_next(&mpages, &slot);
984 * For each mlink to the mpage, check whether the path looks like
985 * it is formatted, and if it does, check whether a source manual
986 * exists by the same name, ignoring the suffix.
987 * If both conditions hold, drop the mlink.
990 mlinks_undupe(struct mpage *mpage)
997 mpage->form = FORM_CAT;
998 prev = &mpage->mlinks;
999 while (NULL != (mlink = *prev)) {
1000 if (FORM_CAT != mlink->dform) {
1001 mpage->form = FORM_NONE;
1004 (void)strlcpy(buf, mlink->file, sizeof(buf));
1005 bufp = strstr(buf, "cat");
1006 assert(NULL != bufp);
1007 memcpy(bufp, "man", 3);
1008 if (NULL != (bufp = strrchr(buf, '.')))
1010 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1011 if (NULL == ohash_find(&mlinks,
1012 ohash_qlookup(&mlinks, buf)))
1015 say(mlink->file, "Man source exists: %s", buf);
1018 *prev = mlink->next;
1022 prev = &(*prev)->next;
1027 mlink_check(struct mpage *mpage, struct mlink *mlink)
1033 * Check whether the manual section given in a file
1034 * agrees with the directory where the file is located.
1035 * Some manuals have suffixes like (3p) on their
1036 * section number either inside the file or in the
1037 * directory name, some are linked into more than one
1038 * section, like encrypt(1) = makekey(8).
1041 if (FORM_SRC == mpage->form &&
1042 strcasecmp(mpage->sec, mlink->dsec))
1043 say(mlink->file, "Section \"%s\" manual in %s directory",
1044 mpage->sec, mlink->dsec);
1047 * Manual page directories exist for each kernel
1048 * architecture as returned by machine(1).
1049 * However, many manuals only depend on the
1050 * application architecture as returned by arch(1).
1051 * For example, some (2/ARM) manuals are shared
1052 * across the "armish" and "zaurus" kernel
1054 * A few manuals are even shared across completely
1055 * different architectures, for example fdformat(1)
1056 * on amd64, i386, sparc, and sparc64.
1059 if (strcasecmp(mpage->arch, mlink->arch))
1060 say(mlink->file, "Architecture \"%s\" manual in "
1061 "\"%s\" directory", mpage->arch, mlink->arch);
1065 * parse_cat() doesn't set NAME_TITLE yet.
1068 if (FORM_CAT == mpage->form)
1072 * Check whether this mlink
1073 * appears as a name in the NAME section.
1076 slot = ohash_qlookup(&names, mlink->name);
1077 str = ohash_find(&names, slot);
1078 assert(NULL != str);
1079 if ( ! (NAME_TITLE & str->mask))
1080 say(mlink->file, "Name missing in NAME section");
1084 * Run through the files in the global vector "mpages"
1085 * and add them to the database specified in "basedir".
1087 * This handles the parsing scheme itself, using the cues of directory
1088 * and filename to determine whether the file is parsable or not.
1091 mpages_merge(struct mparse *mp)
1094 struct ohash_info str_info;
1095 struct mpage *mpage, *mpage_dest;
1096 struct mlink *mlink, *mlink_dest;
1103 enum mandoclevel lvl;
1105 str_info.alloc = hash_alloc;
1106 str_info.calloc = hash_calloc;
1107 str_info.free = hash_free;
1108 str_info.key_offset = offsetof(struct str, key);
1111 SQL_EXEC("BEGIN TRANSACTION");
1113 mpage = ohash_first(&mpages, &pslot);
1114 while (mpage != NULL) {
1115 mlinks_undupe(mpage);
1116 if (mpage->mlinks == NULL) {
1117 mpage = ohash_next(&mpages, &pslot);
1121 name_mask = NAME_MASK;
1122 ohash_init(&names, 4, &str_info);
1123 ohash_init(&strings, 6, &str_info);
1129 mparse_open(mp, &fd, mpage->mlinks->file);
1131 say(mpage->mlinks->file, "&open");
1136 * Try interpreting the file as mdoc(7) or man(7)
1137 * source code, unless it is already known to be
1138 * formatted. Fall back to formatted mode.
1140 if (mpage->mlinks->dform != FORM_CAT ||
1141 mpage->mlinks->fform != FORM_CAT) {
1142 lvl = mparse_readfd(mp, fd, mpage->mlinks->file);
1143 if (lvl < MANDOCLEVEL_FATAL)
1144 mparse_result(mp, &mdoc, &man, &sodest);
1147 if (sodest != NULL) {
1148 mlink_dest = ohash_find(&mlinks,
1149 ohash_qlookup(&mlinks, sodest));
1150 if (mlink_dest == NULL) {
1151 mandoc_asprintf(&cp, "%s.gz", sodest);
1152 mlink_dest = ohash_find(&mlinks,
1153 ohash_qlookup(&mlinks, cp));
1156 if (mlink_dest != NULL) {
1158 /* The .so target exists. */
1160 mpage_dest = mlink_dest->mpage;
1161 mlink = mpage->mlinks;
1163 mlink->mpage = mpage_dest;
1166 * If the target was already
1167 * processed, add the links
1168 * to the database now.
1169 * Otherwise, this will
1170 * happen when we come
1174 if (mpage_dest->pageid)
1175 dbadd_mlink_name(mlink);
1177 if (mlink->next == NULL)
1179 mlink = mlink->next;
1182 /* Move all links to the target. */
1184 mlink->next = mlink_dest->next;
1185 mlink_dest->next = mpage->mlinks;
1186 mpage->mlinks = NULL;
1189 } else if (mdoc != NULL) {
1190 mpage->form = FORM_SRC;
1191 mpage->sec = mdoc_meta(mdoc)->msec;
1192 mpage->sec = mandoc_strdup(
1193 mpage->sec == NULL ? "" : mpage->sec);
1194 mpage->arch = mdoc_meta(mdoc)->arch;
1195 mpage->arch = mandoc_strdup(
1196 mpage->arch == NULL ? "" : mpage->arch);
1198 mandoc_strdup(mdoc_meta(mdoc)->title);
1199 } else if (man != NULL) {
1200 mpage->form = FORM_SRC;
1202 mandoc_strdup(man_meta(man)->msec);
1204 mandoc_strdup(mpage->mlinks->arch);
1206 mandoc_strdup(man_meta(man)->title);
1208 mpage->form = FORM_CAT;
1210 mandoc_strdup(mpage->mlinks->dsec);
1212 mandoc_strdup(mpage->mlinks->arch);
1214 mandoc_strdup(mpage->mlinks->name);
1216 putkey(mpage, mpage->sec, TYPE_sec);
1217 if (*mpage->arch != '\0')
1218 putkey(mpage, mpage->arch, TYPE_arch);
1220 for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
1221 if ('\0' != *mlink->dsec)
1222 putkey(mpage, mlink->dsec, TYPE_sec);
1223 if ('\0' != *mlink->fsec)
1224 putkey(mpage, mlink->fsec, TYPE_sec);
1225 putkey(mpage, '\0' == *mlink->arch ?
1226 any : mlink->arch, TYPE_arch);
1227 putkey(mpage, mlink->name, NAME_FILE);
1230 assert(mpage->desc == NULL);
1232 parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1233 else if (man != NULL)
1234 parse_man(mpage, man_meta(man), man_node(man));
1236 parse_cat(mpage, fd);
1237 if (mpage->desc == NULL)
1238 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1240 if (warnings && !use_all)
1241 for (mlink = mpage->mlinks; mlink;
1242 mlink = mlink->next)
1243 mlink_check(mpage, mlink);
1248 if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1249 exitcode = (int)MANDOCLEVEL_SYSERR;
1250 say(mpage->mlinks->file, "&wait gunzip");
1252 ohash_delete(&strings);
1253 ohash_delete(&names);
1254 mpage = ohash_next(&mpages, &pslot);
1258 SQL_EXEC("END TRANSACTION");
1265 const char *name, *sec, *arch, *key;
1268 sqlite3_prepare_v2(db,
1269 "SELECT name, sec, arch, key FROM ("
1270 "SELECT name AS key, pageid FROM names "
1271 "WHERE bits & ? AND NOT EXISTS ("
1272 "SELECT pageid FROM mlinks "
1273 "WHERE mlinks.pageid == names.pageid "
1274 "AND mlinks.name == names.name"
1277 "SELECT sec, arch, name, pageid FROM mlinks "
1279 ") USING (pageid);",
1282 if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1283 say("", "%s", sqlite3_errmsg(db));
1285 while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1286 name = (const char *)sqlite3_column_text(stmt, 0);
1287 sec = (const char *)sqlite3_column_text(stmt, 1);
1288 arch = (const char *)sqlite3_column_text(stmt, 2);
1289 key = (const char *)sqlite3_column_text(stmt, 3);
1290 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1291 '\0' == *arch ? "" : "/",
1292 '\0' == *arch ? "" : arch, key);
1294 sqlite3_finalize(stmt);
1298 parse_cat(struct mpage *mpage, int fd)
1301 char *line, *p, *title;
1302 size_t len, plen, titlesz;
1304 stream = (-1 == fd) ?
1305 fopen(mpage->mlinks->file, "r") :
1307 if (NULL == stream) {
1311 say(mpage->mlinks->file, "&fopen");
1315 /* Skip to first blank line. */
1317 while (NULL != (line = fgetln(stream, &len)))
1322 * Assume the first line that is not indented
1323 * is the first section header. Skip to it.
1326 while (NULL != (line = fgetln(stream, &len)))
1327 if ('\n' != *line && ' ' != *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 (NULL != (line = fgetln(stream, &len))) {
1341 if (' ' != *line || '\n' != line[len - 1])
1343 while (len > 0 && isspace((unsigned char)*line)) {
1349 title = mandoc_realloc(title, titlesz + len);
1350 memcpy(title + titlesz, line, len);
1352 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 = mandoc_realloc(title, titlesz + 1);
1372 title[titlesz] = '\0';
1375 * Skip to the first dash.
1376 * Use the remaining line as the description (no more than 70
1380 if (NULL != (p = strstr(title, "- "))) {
1381 for (p += 2; ' ' == *p || '\b' == *p; p++)
1382 /* Skip to next word. */ ;
1385 say(mpage->mlinks->file,
1386 "No dash in title line");
1392 /* Strip backspace-encoding from line. */
1394 while (NULL != (line = memchr(p, '\b', plen))) {
1397 memmove(line, line + 1, plen--);
1400 memmove(line - 1, line + 1, plen - len);
1404 mpage->desc = mandoc_strdup(p);
1410 * Put a type/word pair into the word database for this particular file.
1413 putkey(const struct mpage *mpage, char *value, uint64_t type)
1417 assert(NULL != value);
1418 if (TYPE_arch == type)
1419 for (cp = value; *cp; cp++)
1420 if (isupper((unsigned char)*cp))
1421 *cp = _tolower((unsigned char)*cp);
1422 putkeys(mpage, value, strlen(value), type);
1426 * Grok all nodes at or below a certain mdoc node into putkey().
1429 putmdockey(const struct mpage *mpage,
1430 const struct mdoc_node *n, uint64_t m)
1433 for ( ; NULL != n; n = n->next) {
1434 if (NULL != n->child)
1435 putmdockey(mpage, n->child, m);
1436 if (MDOC_TEXT == n->type)
1437 putkey(mpage, n->string, m);
1442 parse_man(struct mpage *mpage, const struct man_meta *meta,
1443 const struct man_node *n)
1445 const struct man_node *head, *body;
1446 char *start, *title;
1454 * We're only searching for one thing: the first text child in
1455 * the BODY of a NAME section. Since we don't keep track of
1456 * sections in -man, run some hoops to find out whether we're in
1457 * the correct section or not.
1460 if (MAN_BODY == n->type && MAN_SH == n->tok) {
1462 assert(body->parent);
1463 if (NULL != (head = body->parent->head) &&
1464 1 == head->nchild &&
1465 NULL != (head = (head->child)) &&
1466 MAN_TEXT == head->type &&
1467 0 == strcmp(head->string, "NAME") &&
1468 NULL != body->child) {
1471 * Suck the entire NAME section into memory.
1472 * Yes, we might run away.
1473 * But too many manuals have big, spread-out
1474 * NAME sections over many lines.
1478 man_deroff(&title, body);
1483 * Go through a special heuristic dance here.
1484 * Conventionally, one or more manual names are
1485 * comma-specified prior to a whitespace, then a
1486 * dash, then a description. Try to puzzle out
1487 * the name parts here.
1492 sz = strcspn(start, " ,");
1493 if ('\0' == start[sz])
1500 * Assume a stray trailing comma in the
1501 * name list if a name begins with a dash.
1504 if ('-' == start[0] ||
1505 ('\\' == start[0] && '-' == start[1]))
1508 putkey(mpage, start, NAME_TITLE);
1509 if ( ! (mpage->name_head_done ||
1510 strcasecmp(start, meta->title))) {
1511 putkey(mpage, start, NAME_HEAD);
1512 mpage->name_head_done = 1;
1520 assert(',' == byte);
1522 while (' ' == *start)
1526 if (start == title) {
1527 putkey(mpage, start, NAME_TITLE);
1528 if ( ! (mpage->name_head_done ||
1529 strcasecmp(start, meta->title))) {
1530 putkey(mpage, start, NAME_HEAD);
1531 mpage->name_head_done = 1;
1537 while (isspace((unsigned char)*start))
1540 if (0 == strncmp(start, "-", 1))
1542 else if (0 == strncmp(start, "\\-\\-", 4))
1544 else if (0 == strncmp(start, "\\-", 2))
1546 else if (0 == strncmp(start, "\\(en", 4))
1548 else if (0 == strncmp(start, "\\(em", 4))
1551 while (' ' == *start)
1554 mpage->desc = mandoc_strdup(start);
1560 for (n = n->child; n; n = n->next) {
1561 if (NULL != mpage->desc)
1563 parse_man(mpage, meta, n);
1568 parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1569 const struct mdoc_node *n)
1573 for (n = n->child; NULL != n; n = n->next) {
1584 if (NULL != mdocs[n->tok].fp)
1585 if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1587 if (mdocs[n->tok].mask)
1588 putmdockey(mpage, n->child,
1589 mdocs[n->tok].mask);
1592 assert(MDOC_ROOT != n->type);
1595 if (NULL != n->child)
1596 parse_mdoc(mpage, meta, n);
1601 parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1602 const struct mdoc_node *n)
1607 if (SEC_SYNOPSIS != n->sec ||
1608 NULL == (n = n->child) ||
1609 MDOC_TEXT != n->type)
1613 * Only consider those `Fd' macro fields that begin with an
1614 * "inclusion" token (versus, e.g., #define).
1617 if (strcmp("#include", n->string))
1620 if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1624 * Strip away the enclosing angle brackets and make sure we're
1629 if ('<' == *start || '"' == *start)
1632 if (0 == (sz = strlen(start)))
1635 end = &start[(int)sz - 1];
1636 if ('>' == *end || '"' == *end)
1640 putkeys(mpage, start, end - start + 1, TYPE_In);
1645 parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1650 if (n->type != MDOC_TEXT)
1653 /* Skip function pointer punctuation. */
1656 while (*cp == '(' || *cp == '*')
1658 sz = strcspn(cp, "()");
1660 putkeys(mpage, cp, sz, TYPE_Fn);
1661 if (n->sec == SEC_SYNOPSIS)
1662 putkeys(mpage, cp, sz, NAME_SYN);
1666 parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1667 const struct mdoc_node *n)
1670 if (n->child == NULL)
1673 parse_mdoc_fname(mpage, n->child);
1675 for (n = n->child->next; n != NULL; n = n->next)
1676 if (n->type == MDOC_TEXT)
1677 putkey(mpage, n->string, TYPE_Fa);
1683 parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1684 const struct mdoc_node *n)
1687 if (n->type != MDOC_HEAD)
1690 if (n->child != NULL)
1691 parse_mdoc_fname(mpage, n->child);
1697 parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1698 const struct mdoc_node *n)
1702 if (NULL == (n = n->child))
1705 if (NULL == n->next) {
1706 putkey(mpage, n->string, TYPE_Xr);
1710 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1711 putkey(mpage, cp, TYPE_Xr);
1717 parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1718 const struct mdoc_node *n)
1721 if (MDOC_BODY == n->type)
1722 mdoc_deroff(&mpage->desc, n);
1727 parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1728 const struct mdoc_node *n)
1731 if (SEC_NAME == n->sec)
1732 putmdockey(mpage, n->child, NAME_TITLE);
1733 else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1734 if (n->child == NULL)
1735 putkey(mpage, meta->name, NAME_SYN);
1737 putmdockey(mpage, n->child, NAME_SYN);
1739 if ( ! (mpage->name_head_done ||
1740 n->child == NULL || n->child->string == NULL ||
1741 strcasecmp(n->child->string, meta->title))) {
1742 putkey(mpage, n->child->string, NAME_HEAD);
1743 mpage->name_head_done = 1;
1749 parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1750 const struct mdoc_node *n)
1753 return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1757 parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1758 const struct mdoc_node *n)
1761 return(MDOC_HEAD == n->type);
1765 parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1766 const struct mdoc_node *n)
1769 return(MDOC_BODY == n->type);
1773 * Add a string to the hash table for the current manual.
1774 * Each string has a bitmask telling which macros it belongs to.
1775 * When we finish the manual, we'll dump the table.
1778 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1789 mustfree = render_string(&cp, &sz);
1795 name_mask &= ~NAME_FIRST;
1797 say(mpage->mlinks->file,
1798 "Adding name %*s, bits=%d", sz, cp, v);
1802 for (i = 0; i < mansearch_keymax; i++)
1803 if ((uint64_t)1 << i & v)
1804 say(mpage->mlinks->file,
1805 "Adding key %s=%*s",
1806 mansearch_keynames[i], sz, cp);
1810 slot = ohash_qlookupi(htab, cp, &end);
1811 s = ohash_find(htab, slot);
1813 if (NULL != s && mpage == s->mpage) {
1816 } else if (NULL == s) {
1817 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1818 memcpy(s->key, cp, sz);
1819 ohash_insert(htab, slot, s);
1829 * Take a Unicode codepoint and produce its UTF-8 encoding.
1830 * This isn't the best way to do this, but it works.
1831 * The magic numbers are from the UTF-8 packaging.
1832 * They're not as scary as they seem: read the UTF-8 spec for details.
1835 utf8(unsigned int cp, char out[7])
1840 if (cp <= 0x0000007F) {
1843 } else if (cp <= 0x000007FF) {
1845 out[0] = (cp >> 6 & 31) | 192;
1846 out[1] = (cp & 63) | 128;
1847 } else if (cp <= 0x0000FFFF) {
1849 out[0] = (cp >> 12 & 15) | 224;
1850 out[1] = (cp >> 6 & 63) | 128;
1851 out[2] = (cp & 63) | 128;
1852 } else if (cp <= 0x001FFFFF) {
1854 out[0] = (cp >> 18 & 7) | 240;
1855 out[1] = (cp >> 12 & 63) | 128;
1856 out[2] = (cp >> 6 & 63) | 128;
1857 out[3] = (cp & 63) | 128;
1858 } else if (cp <= 0x03FFFFFF) {
1860 out[0] = (cp >> 24 & 3) | 248;
1861 out[1] = (cp >> 18 & 63) | 128;
1862 out[2] = (cp >> 12 & 63) | 128;
1863 out[3] = (cp >> 6 & 63) | 128;
1864 out[4] = (cp & 63) | 128;
1865 } else if (cp <= 0x7FFFFFFF) {
1867 out[0] = (cp >> 30 & 1) | 252;
1868 out[1] = (cp >> 24 & 63) | 128;
1869 out[2] = (cp >> 18 & 63) | 128;
1870 out[3] = (cp >> 12 & 63) | 128;
1871 out[4] = (cp >> 6 & 63) | 128;
1872 out[5] = (cp & 63) | 128;
1881 * If the string contains escape sequences,
1882 * replace it with an allocated rendering and return 1,
1883 * such that the caller can free it after use.
1884 * Otherwise, do nothing and return 0.
1887 render_string(char **public, size_t *psz)
1889 const char *src, *scp, *addcp, *seq;
1891 size_t ssz, dsz, addsz;
1892 char utfbuf[7], res[6];
1893 int seqlen, unicode;
1897 res[2] = ASCII_NBRSP;
1898 res[3] = ASCII_HYPH;
1899 res[4] = ASCII_BREAK;
1902 src = scp = *public;
1907 while (scp < src + *psz) {
1909 /* Leave normal characters unchanged. */
1911 if (strchr(res, *scp) == NULL) {
1919 * Found something that requires replacing,
1920 * make sure we have a destination buffer.
1924 dst = mandoc_malloc(ssz + 1);
1926 memcpy(dst, src, dsz);
1929 /* Handle single-char special characters. */
1951 * Found an escape sequence.
1952 * Read past the slash, then parse it.
1953 * Ignore everything except characters.
1957 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1961 * Render the special character
1962 * as either UTF-8 or ASCII.
1966 unicode = mchars_spec2cp(mchars, seq, seqlen);
1969 addsz = utf8(unicode, utfbuf);
1974 addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1977 if (*addcp == ASCII_NBRSP) {
1983 /* Copy the rendered glyph into the stream. */
1986 dst = mandoc_realloc(dst, ssz + 1);
1987 memcpy(dst + dsz, addcp, addsz);
1995 /* Trim trailing whitespace and NUL-terminate. */
1997 while (*psz > 0 && (*public)[*psz - 1] == ' ')
2000 (*public)[*psz] = '\0';
2007 dbadd_mlink(const struct mlink *mlink)
2012 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2013 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2014 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
2015 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
2016 SQL_STEP(stmts[STMT_INSERT_LINK]);
2017 sqlite3_reset(stmts[STMT_INSERT_LINK]);
2021 dbadd_mlink_name(const struct mlink *mlink)
2029 SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2030 bits = NAME_FILE & NAME_MASK;
2031 if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2032 bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2033 sqlite3_reset(stmts[STMT_SELECT_NAME]);
2037 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
2038 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2039 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2040 SQL_STEP(stmts[STMT_INSERT_NAME]);
2041 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2045 * Flush the current page's terms (and their bits) into the database.
2046 * Wrap the entire set of additions in a transaction to make sqlite be a
2048 * Also, handle escape sequences at the last possible moment.
2051 dbadd(struct mpage *mpage)
2053 struct mlink *mlink;
2060 mlink = mpage->mlinks;
2063 for (key = ohash_first(&names, &slot); NULL != key;
2064 key = ohash_next(&names, &slot))
2066 for (key = ohash_first(&strings, &slot); NULL != key;
2067 key = ohash_next(&strings, &slot))
2071 while (NULL != mlink) {
2072 fputs(mlink->name, stdout);
2073 if (NULL == mlink->next ||
2074 strcmp(mlink->dsec, mlink->next->dsec) ||
2075 strcmp(mlink->fsec, mlink->next->fsec) ||
2076 strcmp(mlink->arch, mlink->next->arch)) {
2078 if ('\0' == *mlink->dsec)
2079 fputs(mlink->fsec, stdout);
2081 fputs(mlink->dsec, stdout);
2082 if ('\0' != *mlink->arch)
2083 printf("/%s", mlink->arch);
2086 mlink = mlink->next;
2088 fputs(", ", stdout);
2090 printf(" - %s\n", mpage->desc);
2095 say(mlink->file, "Adding to database");
2099 mustfree = render_string(&cp, &i);
2101 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
2102 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
2103 SQL_STEP(stmts[STMT_INSERT_PAGE]);
2104 mpage->pageid = sqlite3_last_insert_rowid(db);
2105 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2109 while (NULL != mlink) {
2111 mlink = mlink->next;
2113 mlink = mpage->mlinks;
2115 for (key = ohash_first(&names, &slot); NULL != key;
2116 key = ohash_next(&names, &slot)) {
2117 assert(key->mpage == mpage);
2119 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2120 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
2121 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
2122 SQL_STEP(stmts[STMT_INSERT_NAME]);
2123 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2126 for (key = ohash_first(&strings, &slot); NULL != key;
2127 key = ohash_next(&strings, &slot)) {
2128 assert(key->mpage == mpage);
2130 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
2131 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
2132 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
2133 SQL_STEP(stmts[STMT_INSERT_KEY]);
2134 sqlite3_reset(stmts[STMT_INSERT_KEY]);
2142 struct mpage *mpage;
2143 struct mlink *mlink;
2148 SQL_EXEC("BEGIN TRANSACTION");
2150 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2151 mpage = ohash_next(&mpages, &slot)) {
2152 mlink = mpage->mlinks;
2154 say(mlink->file, "Deleting from database");
2157 for ( ; NULL != mlink; mlink = mlink->next) {
2159 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2161 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2163 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2165 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2166 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2171 SQL_EXEC("END TRANSACTION");
2175 * Close an existing database and its prepared statements.
2176 * If "real" is not set, rename the temporary file into the real one.
2188 for (i = 0; i < STMT__MAX; i++) {
2189 sqlite3_finalize(stmts[i]);
2199 if ('\0' == *tempfilename) {
2200 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2201 exitcode = (int)MANDOCLEVEL_SYSERR;
2202 say(MANDOC_DB, "&rename");
2207 switch (child = fork()) {
2209 exitcode = (int)MANDOCLEVEL_SYSERR;
2210 say("", "&fork cmp");
2213 execlp("cmp", "cmp", "-s",
2214 tempfilename, MANDOC_DB, NULL);
2215 say("", "&exec cmp");
2220 if (-1 == waitpid(child, &status, 0)) {
2221 exitcode = (int)MANDOCLEVEL_SYSERR;
2222 say("", "&wait cmp");
2223 } else if (WIFSIGNALED(status)) {
2224 exitcode = (int)MANDOCLEVEL_SYSERR;
2225 say("", "cmp died from signal %d", WTERMSIG(status));
2226 } else if (WEXITSTATUS(status)) {
2227 exitcode = (int)MANDOCLEVEL_SYSERR;
2229 "Data changed, but cannot replace database");
2232 *strrchr(tempfilename, '/') = '\0';
2233 switch (child = fork()) {
2235 exitcode = (int)MANDOCLEVEL_SYSERR;
2236 say("", "&fork rm");
2239 execlp("rm", "rm", "-rf", tempfilename, NULL);
2240 say("", "&exec rm");
2241 exit((int)MANDOCLEVEL_SYSERR);
2245 if (-1 == waitpid(child, &status, 0)) {
2246 exitcode = (int)MANDOCLEVEL_SYSERR;
2247 say("", "&wait rm");
2248 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2249 exitcode = (int)MANDOCLEVEL_SYSERR;
2250 say("", "%s: Cannot remove temporary directory",
2256 * This is straightforward stuff.
2257 * Open a database connection to a "temporary" database, then open a set
2258 * of prepared statements we'll use over and over again.
2259 * If "real" is set, we use the existing database; if not, we truncate a
2261 * Must be matched by dbclose().
2272 *tempfilename = '\0';
2273 ofl = SQLITE_OPEN_READWRITE;
2276 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2277 if (SQLITE_OK != rc) {
2278 exitcode = (int)MANDOCLEVEL_SYSERR;
2279 if (SQLITE_CANTOPEN != rc)
2280 say(MANDOC_DB, "%s", sqlite3_errstr(rc));
2283 goto prepare_statements;
2286 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2288 remove(MANDOC_DB "~");
2289 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2290 if (SQLITE_OK == rc)
2292 if (MPARSE_QUICK & mparse_options) {
2293 exitcode = (int)MANDOCLEVEL_SYSERR;
2294 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
2298 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2299 sizeof(tempfilename));
2300 if (NULL == mkdtemp(tempfilename)) {
2301 exitcode = (int)MANDOCLEVEL_SYSERR;
2302 say("", "&%s", tempfilename);
2305 (void)strlcat(tempfilename, "/" MANDOC_DB,
2306 sizeof(tempfilename));
2307 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2308 if (SQLITE_OK != rc) {
2309 exitcode = (int)MANDOCLEVEL_SYSERR;
2310 say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
2315 sql = "CREATE TABLE \"mpages\" (\n"
2316 " \"desc\" TEXT NOT NULL,\n"
2317 " \"form\" INTEGER NOT NULL,\n"
2318 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2321 "CREATE TABLE \"mlinks\" (\n"
2322 " \"sec\" TEXT NOT NULL,\n"
2323 " \"arch\" TEXT NOT NULL,\n"
2324 " \"name\" TEXT NOT NULL,\n"
2325 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2326 "ON DELETE CASCADE\n"
2328 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2330 "CREATE TABLE \"names\" (\n"
2331 " \"bits\" INTEGER NOT NULL,\n"
2332 " \"name\" TEXT NOT NULL,\n"
2333 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2334 "ON DELETE CASCADE,\n"
2335 " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
2338 "CREATE TABLE \"keys\" (\n"
2339 " \"bits\" INTEGER NOT NULL,\n"
2340 " \"key\" TEXT NOT NULL,\n"
2341 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2342 "ON DELETE CASCADE\n"
2344 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2346 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2347 exitcode = (int)MANDOCLEVEL_SYSERR;
2348 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2354 if (SQLITE_OK != sqlite3_exec(db,
2355 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2356 exitcode = (int)MANDOCLEVEL_SYSERR;
2357 say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2358 sqlite3_errmsg(db));
2363 sql = "DELETE FROM mpages WHERE pageid IN "
2364 "(SELECT pageid FROM mlinks WHERE "
2365 "sec=? AND arch=? AND name=?)";
2366 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2367 sql = "INSERT INTO mpages "
2368 "(desc,form) VALUES (?,?)";
2369 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2370 sql = "INSERT INTO mlinks "
2371 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2372 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2373 sql = "SELECT bits FROM names where pageid = ?";
2374 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
2375 sql = "INSERT INTO names "
2376 "(bits,name,pageid) VALUES (?,?,?)";
2377 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2378 sql = "INSERT INTO keys "
2379 "(bits,key,pageid) VALUES (?,?,?)";
2380 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2384 * When opening a new database, we can turn off
2385 * synchronous mode for much better performance.
2388 if (real && SQLITE_OK != sqlite3_exec(db,
2389 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2390 exitcode = (int)MANDOCLEVEL_SYSERR;
2391 say(MANDOC_DB, "PRAGMA synchronous: %s",
2392 sqlite3_errmsg(db));
2402 hash_calloc(size_t n, size_t sz, void *arg)
2405 return(mandoc_calloc(n, sz));
2409 hash_alloc(size_t sz, void *arg)
2412 return(mandoc_malloc(sz));
2416 hash_free(void *p, void *arg)
2423 set_basedir(const char *targetdir, int report_baddir)
2425 static char startdir[PATH_MAX];
2426 static int getcwd_status; /* 1 = ok, 2 = failure */
2427 static int chdir_status; /* 1 = changed directory */
2431 * Remember the original working directory, if possible.
2432 * This will be needed if the second or a later directory
2433 * on the command line is given as a relative path.
2434 * Do not error out if the current directory is not
2435 * searchable: Maybe it won't be needed after all.
2437 if (0 == getcwd_status) {
2438 if (NULL == getcwd(startdir, sizeof(startdir))) {
2440 (void)strlcpy(startdir, strerror(errno),
2447 * We are leaving the old base directory.
2448 * Do not use it any longer, not even for messages.
2453 * If and only if the directory was changed earlier and
2454 * the next directory to process is given as a relative path,
2455 * first go back, or bail out if that is impossible.
2457 if (chdir_status && '/' != *targetdir) {
2458 if (2 == getcwd_status) {
2459 exitcode = (int)MANDOCLEVEL_SYSERR;
2460 say("", "getcwd: %s", startdir);
2463 if (-1 == chdir(startdir)) {
2464 exitcode = (int)MANDOCLEVEL_SYSERR;
2465 say("", "&chdir %s", startdir);
2471 * Always resolve basedir to the canonicalized absolute
2472 * pathname and append a trailing slash, such that
2473 * we can reliably check whether files are inside.
2475 if (NULL == realpath(targetdir, basedir)) {
2476 if (report_baddir || errno != ENOENT) {
2477 exitcode = (int)MANDOCLEVEL_BADARG;
2478 say("", "&%s: realpath", targetdir);
2481 } else if (-1 == chdir(basedir)) {
2482 if (report_baddir || errno != ENOENT) {
2483 exitcode = (int)MANDOCLEVEL_BADARG;
2489 cp = strchr(basedir, '\0');
2490 if ('/' != cp[-1]) {
2491 if (cp - basedir >= PATH_MAX - 1) {
2492 exitcode = (int)MANDOCLEVEL_SYSERR;
2493 say("", "Filename too long");
2503 say(const char *file, const char *format, ...)
2508 if ('\0' != *basedir)
2509 fprintf(stderr, "%s", basedir);
2510 if ('\0' != *basedir && '\0' != *file)
2513 fprintf(stderr, "%s", file);
2516 if (NULL != format) {
2529 if (NULL != format) {
2530 if ('\0' != *basedir || '\0' != *file)
2531 fputs(": ", stderr);
2532 va_start(ap, format);
2533 vfprintf(stderr, format, ap);
2537 if ('\0' != *basedir || '\0' != *file || NULL != format)
2538 fputs(": ", stderr);
2541 fputc('\n', stderr);