1 /* $Id: mandocdb.c,v 1.186 2015/03/13 00:19:41 schwarze Exp $ */
3 * Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
4 * Copyright (c) 2011-2015 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;
354 mpages_info.data = mlinks_info.data = NULL;
356 mpages_info.key_offset = offsetof(struct mpage, inodev);
357 mlinks_info.key_offset = offsetof(struct mlink, file);
359 progname = strrchr(argv[0], '/');
360 if (progname == NULL)
366 * We accept a few different invocations.
367 * The CHECKOP macro makes sure that invocation styles don't
368 * clobber each other.
370 #define CHECKOP(_op, _ch) do \
371 if (OP_DEFAULT != (_op)) { \
372 fprintf(stderr, "%s: -%c: Conflicting option\n", \
375 } while (/*CONSTCOND*/0)
380 while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
405 mparse_options |= MPARSE_QUICK;
408 if (strcmp(optarg, "utf8")) {
409 fprintf(stderr, "%s: -T%s: "
410 "Unsupported output format\n",
418 dup2(STDOUT_FILENO, STDERR_FILENO);
428 /* Compatibility with espie@'s makewhatis. */
437 if (OP_CONFFILE == op && argc > 0) {
438 fprintf(stderr, "%s: -C: Too many arguments\n",
443 exitcode = (int)MANDOCLEVEL_OK;
444 mchars = mchars_alloc();
445 mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL,
447 ohash_init(&mpages, 6, &mpages_info);
448 ohash_init(&mlinks, 6, &mlinks_info);
450 if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
453 * Most of these deal with a specific directory.
454 * Jump into that directory first.
456 if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
461 * The existing database is usable. Process
462 * all files specified on the command-line.
465 for (i = 0; i < argc; i++)
471 * Database missing or corrupt.
472 * Recreate from scratch.
474 exitcode = (int)MANDOCLEVEL_OK;
483 dbclose(OP_DEFAULT == op ? 0 : 1);
486 * If we have arguments, use them as our manpaths.
487 * If we don't, grok from manpath(1) or however else
488 * manpath_parse() wants to do it.
491 dirs.paths = mandoc_reallocarray(NULL,
492 argc, sizeof(char *));
493 dirs.sz = (size_t)argc;
494 for (i = 0; i < argc; i++)
495 dirs.paths[i] = mandoc_strdup(argv[i]);
497 manpath_parse(&dirs, path_arg, NULL, NULL);
500 exitcode = (int)MANDOCLEVEL_BADARG;
501 say("", "Empty manpath");
505 * First scan the tree rooted at a base directory, then
506 * build a new database and finally move it into place.
507 * Ignore zero-length directories and strip trailing
510 for (j = 0; j < dirs.sz; j++) {
511 sz = strlen(dirs.paths[j]);
512 if (sz && '/' == dirs.paths[j][sz - 1])
513 dirs.paths[j][--sz] = '\0';
518 ohash_init(&mpages, 6, &mpages_info);
519 ohash_init(&mlinks, 6, &mlinks_info);
522 if (0 == set_basedir(dirs.paths[j], argc > 0))
530 if (warnings && !nodb &&
531 ! (MPARSE_QUICK & mparse_options))
535 if (j + 1 < dirs.sz) {
537 ohash_delete(&mpages);
538 ohash_delete(&mlinks);
547 ohash_delete(&mpages);
548 ohash_delete(&mlinks);
551 fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
552 " %s [-aDnpQ] [-Tutf8] dir ...\n"
553 " %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
554 " %s [-Dnp] -u dir [file ...]\n"
555 " %s [-Q] -t file ...\n",
556 progname, progname, progname,
559 return((int)MANDOCLEVEL_BADARG);
563 * Scan a directory tree rooted at "basedir" for manpages.
564 * We use fts(), scanning directory parts along the way for clues to our
565 * section and architecture.
567 * If use_all has been specified, grok all files.
568 * If not, sanitise paths to the following:
570 * [./]man*[/<arch>]/<name>.<section>
572 * [./]cat<section>[/<arch>]/<name>.0
574 * TODO: accomodate for multi-language directories.
584 char *dsec, *arch, *fsec, *cp;
589 argv[1] = (char *)NULL;
591 f = fts_open((char * const *)argv,
592 FTS_PHYSICAL | FTS_NOCHDIR, NULL);
594 exitcode = (int)MANDOCLEVEL_SYSERR;
595 say("", "&fts_open");
602 while (NULL != (ff = fts_read(f))) {
603 path = ff->fts_path + 2;
604 switch (ff->fts_info) {
607 * Symbolic links require various sanity checks,
608 * then get handled just like regular files.
611 if (NULL == realpath(path, buf)) {
613 say(path, "&realpath");
616 if (strstr(buf, basedir) != buf
618 && strstr(buf, HOMEBREWDIR) != buf
621 if (warnings) say("",
622 "%s: outside base directory", buf);
625 /* Use logical inode to avoid mpages dupe. */
626 if (-1 == stat(path, ff->fts_statp)) {
634 * If we're a regular file, add an mlink by using the
635 * stored directory data and handling the filename.
638 if (0 == strcmp(path, MANDOC_DB))
640 if ( ! use_all && ff->fts_level < 2) {
642 say(path, "Extraneous file");
647 while (NULL == fsec) {
648 fsec = strrchr(ff->fts_name, '.');
649 if (NULL == fsec || strcmp(fsec+1, "gz"))
659 "No filename suffix");
662 } else if (0 == strcmp(++fsec, "html")) {
664 say(path, "Skip html");
666 } else if (0 == strcmp(fsec, "ps")) {
668 say(path, "Skip ps");
670 } else if (0 == strcmp(fsec, "pdf")) {
672 say(path, "Skip pdf");
674 } else if ( ! use_all &&
675 ((FORM_SRC == dform &&
676 strncmp(fsec, dsec, strlen(dsec))) ||
677 (FORM_CAT == dform && strcmp(fsec, "0")))) {
679 say(path, "Wrong filename suffix");
684 mlink = mandoc_calloc(1, sizeof(struct mlink));
685 if (strlcpy(mlink->file, path,
686 sizeof(mlink->file)) >=
687 sizeof(mlink->file)) {
688 say(path, "Filename too long");
692 mlink->dform = dform;
695 mlink->name = ff->fts_name;
698 mlink_add(mlink, ff->fts_statp);
708 say(path, "Not a regular file");
712 switch (ff->fts_level) {
714 /* Ignore the root directory. */
718 * This might contain manX/ or catX/.
719 * Try to infer this from the name.
720 * If we're not in use_all, enforce it.
723 if (FTS_DP == ff->fts_info)
726 if (0 == strncmp(cp, "man", 3)) {
729 } else if (0 == strncmp(cp, "cat", 3)) {
737 if (NULL != dsec || use_all)
741 say(path, "Unknown directory part");
742 fts_set(f, ff, FTS_SKIP);
746 * Possibly our architecture.
747 * If we're descending, keep tabs on it.
749 if (FTS_DP != ff->fts_info && NULL != dsec)
755 if (FTS_DP == ff->fts_info || use_all)
758 say(path, "Extraneous directory part");
759 fts_set(f, ff, FTS_SKIP);
769 * Add a file to the mlinks table.
770 * Do not verify that it's a "valid" looking manpage (we'll do that
773 * Try to infer the manual section, architecture, and page name from the
774 * path, assuming it looks like
776 * [./]man*[/<arch>]/<name>.<section>
778 * [./]cat<section>[/<arch>]/<name>.0
780 * See treescan() for the fts(3) version of this.
783 filescan(const char *file)
792 if (0 == strncmp(file, "./", 2))
796 * We have to do lstat(2) before realpath(3) loses
797 * the information whether this is a symbolic link.
798 * We need to know that because for symbolic links,
799 * we want to use the orginal file name, while for
800 * regular files, we want to use the real path.
802 if (-1 == lstat(file, &st)) {
803 exitcode = (int)MANDOCLEVEL_BADARG;
806 } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
807 exitcode = (int)MANDOCLEVEL_BADARG;
808 say(file, "Not a regular file");
813 * We have to resolve the file name to the real path
814 * in any case for the base directory check.
816 if (NULL == realpath(file, buf)) {
817 exitcode = (int)MANDOCLEVEL_BADARG;
818 say(file, "&realpath");
824 else if (strstr(buf, basedir) == buf)
825 start = buf + strlen(basedir);
827 else if (strstr(buf, HOMEBREWDIR) == buf)
831 exitcode = (int)MANDOCLEVEL_BADARG;
832 say("", "%s: outside base directory", buf);
837 * Now we are sure the file is inside our tree.
838 * If it is a symbolic link, ignore the real path
839 * and use the original name.
840 * This implies passing stuff like "cat1/../man1/foo.1"
841 * on the command line won't work. So don't do that.
842 * Note the stat(2) can still fail if the link target
845 if (S_IFLNK & st.st_mode) {
846 if (-1 == stat(buf, &st)) {
847 exitcode = (int)MANDOCLEVEL_BADARG;
851 if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
852 say(file, "Filename too long");
856 if (OP_TEST != op && strstr(buf, basedir) == buf)
857 start += strlen(basedir);
860 mlink = mandoc_calloc(1, sizeof(struct mlink));
861 mlink->dform = FORM_NONE;
862 if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
863 sizeof(mlink->file)) {
864 say(start, "Filename too long");
870 * First try to guess our directory structure.
871 * If we find a separator, try to look for man* or cat*.
872 * If we find one of these and what's underneath is a directory,
873 * assume it's an architecture.
875 if (NULL != (p = strchr(start, '/'))) {
877 if (0 == strncmp(start, "man", 3)) {
878 mlink->dform = FORM_SRC;
879 mlink->dsec = start + 3;
880 } else if (0 == strncmp(start, "cat", 3)) {
881 mlink->dform = FORM_CAT;
882 mlink->dsec = start + 3;
886 if (NULL != mlink->dsec && NULL != (p = strchr(start, '/'))) {
894 * Now check the file suffix.
895 * Suffix of `.0' indicates a catpage, `.1-9' is a manpage.
897 p = strrchr(start, '\0');
898 while (p-- > start && '/' != *p && '.' != *p)
907 * Now try to parse the name.
908 * Use the filename portion of the path.
911 if (NULL != (p = strrchr(start, '/'))) {
915 mlink_add(mlink, &st);
919 mlink_add(struct mlink *mlink, const struct stat *st)
921 struct inodev inodev;
925 assert(NULL != mlink->file);
927 mlink->dsec = mandoc_strdup(mlink->dsec ? mlink->dsec : "");
928 mlink->arch = mandoc_strdup(mlink->arch ? mlink->arch : "");
929 mlink->name = mandoc_strdup(mlink->name ? mlink->name : "");
930 mlink->fsec = mandoc_strdup(mlink->fsec ? mlink->fsec : "");
932 if ('0' == *mlink->fsec) {
934 mlink->fsec = mandoc_strdup(mlink->dsec);
935 mlink->fform = FORM_CAT;
936 } else if ('1' <= *mlink->fsec && '9' >= *mlink->fsec)
937 mlink->fform = FORM_SRC;
939 mlink->fform = FORM_NONE;
941 slot = ohash_qlookup(&mlinks, mlink->file);
942 assert(NULL == ohash_find(&mlinks, slot));
943 ohash_insert(&mlinks, slot, mlink);
945 memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
946 inodev.st_ino = st->st_ino;
947 inodev.st_dev = st->st_dev;
948 slot = ohash_lookup_memory(&mpages, (char *)&inodev,
949 sizeof(struct inodev), inodev.st_ino);
950 mpage = ohash_find(&mpages, slot);
952 mpage = mandoc_calloc(1, sizeof(struct mpage));
953 mpage->inodev.st_ino = inodev.st_ino;
954 mpage->inodev.st_dev = inodev.st_dev;
955 ohash_insert(&mpages, slot, mpage);
957 mlink->next = mpage->mlinks;
958 mpage->mlinks = mlink;
959 mlink->mpage = mpage;
963 mlink_free(struct mlink *mlink)
980 mpage = ohash_first(&mpages, &slot);
981 while (NULL != mpage) {
982 while (NULL != (mlink = mpage->mlinks)) {
983 mpage->mlinks = mlink->next;
991 mpage = ohash_next(&mpages, &slot);
996 * For each mlink to the mpage, check whether the path looks like
997 * it is formatted, and if it does, check whether a source manual
998 * exists by the same name, ignoring the suffix.
999 * If both conditions hold, drop the mlink.
1002 mlinks_undupe(struct mpage *mpage)
1005 struct mlink **prev;
1006 struct mlink *mlink;
1009 mpage->form = FORM_CAT;
1010 prev = &mpage->mlinks;
1011 while (NULL != (mlink = *prev)) {
1012 if (FORM_CAT != mlink->dform) {
1013 mpage->form = FORM_NONE;
1016 (void)strlcpy(buf, mlink->file, sizeof(buf));
1017 bufp = strstr(buf, "cat");
1018 assert(NULL != bufp);
1019 memcpy(bufp, "man", 3);
1020 if (NULL != (bufp = strrchr(buf, '.')))
1022 (void)strlcat(buf, mlink->dsec, sizeof(buf));
1023 if (NULL == ohash_find(&mlinks,
1024 ohash_qlookup(&mlinks, buf)))
1027 say(mlink->file, "Man source exists: %s", buf);
1030 *prev = mlink->next;
1034 prev = &(*prev)->next;
1039 mlink_check(struct mpage *mpage, struct mlink *mlink)
1045 * Check whether the manual section given in a file
1046 * agrees with the directory where the file is located.
1047 * Some manuals have suffixes like (3p) on their
1048 * section number either inside the file or in the
1049 * directory name, some are linked into more than one
1050 * section, like encrypt(1) = makekey(8).
1053 if (FORM_SRC == mpage->form &&
1054 strcasecmp(mpage->sec, mlink->dsec))
1055 say(mlink->file, "Section \"%s\" manual in %s directory",
1056 mpage->sec, mlink->dsec);
1059 * Manual page directories exist for each kernel
1060 * architecture as returned by machine(1).
1061 * However, many manuals only depend on the
1062 * application architecture as returned by arch(1).
1063 * For example, some (2/ARM) manuals are shared
1064 * across the "armish" and "zaurus" kernel
1066 * A few manuals are even shared across completely
1067 * different architectures, for example fdformat(1)
1068 * on amd64, i386, sparc, and sparc64.
1071 if (strcasecmp(mpage->arch, mlink->arch))
1072 say(mlink->file, "Architecture \"%s\" manual in "
1073 "\"%s\" directory", mpage->arch, mlink->arch);
1077 * parse_cat() doesn't set NAME_TITLE yet.
1080 if (FORM_CAT == mpage->form)
1084 * Check whether this mlink
1085 * appears as a name in the NAME section.
1088 slot = ohash_qlookup(&names, mlink->name);
1089 str = ohash_find(&names, slot);
1090 assert(NULL != str);
1091 if ( ! (NAME_TITLE & str->mask))
1092 say(mlink->file, "Name missing in NAME section");
1096 * Run through the files in the global vector "mpages"
1097 * and add them to the database specified in "basedir".
1099 * This handles the parsing scheme itself, using the cues of directory
1100 * and filename to determine whether the file is parsable or not.
1103 mpages_merge(struct mparse *mp)
1106 struct ohash_info str_info;
1107 struct mpage *mpage, *mpage_dest;
1108 struct mlink *mlink, *mlink_dest;
1116 str_info.alloc = hash_alloc;
1117 str_info.calloc = hash_calloc;
1118 str_info.free = hash_free;
1119 str_info.data = NULL;
1120 str_info.key_offset = offsetof(struct str, key);
1123 SQL_EXEC("BEGIN TRANSACTION");
1125 mpage = ohash_first(&mpages, &pslot);
1126 while (mpage != NULL) {
1127 mlinks_undupe(mpage);
1128 if ((mlink = mpage->mlinks) == NULL) {
1129 mpage = ohash_next(&mpages, &pslot);
1133 name_mask = NAME_MASK;
1134 ohash_init(&names, 4, &str_info);
1135 ohash_init(&strings, 6, &str_info);
1141 mparse_open(mp, &fd, mlink->file);
1143 say(mlink->file, "&open");
1148 * Interpret the file as mdoc(7) or man(7) source
1149 * code, unless it is known to be formatted.
1151 if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
1152 mparse_readfd(mp, fd, mlink->file);
1153 mparse_result(mp, &mdoc, &man, &sodest);
1156 if (sodest != NULL) {
1157 mlink_dest = ohash_find(&mlinks,
1158 ohash_qlookup(&mlinks, sodest));
1159 if (mlink_dest == NULL) {
1160 mandoc_asprintf(&cp, "%s.gz", sodest);
1161 mlink_dest = ohash_find(&mlinks,
1162 ohash_qlookup(&mlinks, cp));
1165 if (mlink_dest != NULL) {
1167 /* The .so target exists. */
1169 mpage_dest = mlink_dest->mpage;
1171 mlink->mpage = mpage_dest;
1174 * If the target was already
1175 * processed, add the links
1176 * to the database now.
1177 * Otherwise, this will
1178 * happen when we come
1182 if (mpage_dest->pageid)
1183 dbadd_mlink_name(mlink);
1185 if (mlink->next == NULL)
1187 mlink = mlink->next;
1190 /* Move all links to the target. */
1192 mlink->next = mlink_dest->next;
1193 mlink_dest->next = mpage->mlinks;
1194 mpage->mlinks = NULL;
1197 } else if (mdoc != NULL) {
1198 mpage->form = FORM_SRC;
1199 mpage->sec = mdoc_meta(mdoc)->msec;
1200 mpage->sec = mandoc_strdup(
1201 mpage->sec == NULL ? "" : mpage->sec);
1202 mpage->arch = mdoc_meta(mdoc)->arch;
1203 mpage->arch = mandoc_strdup(
1204 mpage->arch == NULL ? "" : mpage->arch);
1206 mandoc_strdup(mdoc_meta(mdoc)->title);
1207 } else if (man != NULL) {
1208 mpage->form = FORM_SRC;
1209 mpage->sec = mandoc_strdup(man_meta(man)->msec);
1210 mpage->arch = mandoc_strdup(mlink->arch);
1211 mpage->title = mandoc_strdup(man_meta(man)->title);
1213 mpage->form = FORM_CAT;
1214 mpage->sec = mandoc_strdup(mlink->dsec);
1215 mpage->arch = mandoc_strdup(mlink->arch);
1216 mpage->title = mandoc_strdup(mlink->name);
1218 putkey(mpage, mpage->sec, TYPE_sec);
1219 if (*mpage->arch != '\0')
1220 putkey(mpage, mpage->arch, TYPE_arch);
1222 for ( ; mlink != NULL; mlink = mlink->next) {
1223 if ('\0' != *mlink->dsec)
1224 putkey(mpage, mlink->dsec, TYPE_sec);
1225 if ('\0' != *mlink->fsec)
1226 putkey(mpage, mlink->fsec, TYPE_sec);
1227 putkey(mpage, '\0' == *mlink->arch ?
1228 any : mlink->arch, TYPE_arch);
1229 putkey(mpage, mlink->name, NAME_FILE);
1232 assert(mpage->desc == NULL);
1234 parse_mdoc(mpage, mdoc_meta(mdoc), mdoc_node(mdoc));
1235 else if (man != NULL)
1236 parse_man(mpage, man_meta(man), man_node(man));
1238 parse_cat(mpage, fd);
1239 if (mpage->desc == NULL)
1240 mpage->desc = mandoc_strdup(mpage->mlinks->name);
1242 if (warnings && !use_all)
1243 for (mlink = mpage->mlinks; mlink;
1244 mlink = mlink->next)
1245 mlink_check(mpage, mlink);
1248 mlink = mpage->mlinks;
1251 if (mparse_wait(mp) != MANDOCLEVEL_OK) {
1252 exitcode = (int)MANDOCLEVEL_SYSERR;
1253 say(mlink->file, "&wait gunzip");
1255 ohash_delete(&strings);
1256 ohash_delete(&names);
1257 mpage = ohash_next(&mpages, &pslot);
1261 SQL_EXEC("END TRANSACTION");
1268 const char *name, *sec, *arch, *key;
1271 sqlite3_prepare_v2(db,
1272 "SELECT name, sec, arch, key FROM ("
1273 "SELECT name AS key, pageid FROM names "
1274 "WHERE bits & ? AND NOT EXISTS ("
1275 "SELECT pageid FROM mlinks "
1276 "WHERE mlinks.pageid == names.pageid "
1277 "AND mlinks.name == names.name"
1280 "SELECT sec, arch, name, pageid FROM mlinks "
1282 ") USING (pageid);",
1285 if (SQLITE_OK != sqlite3_bind_int64(stmt, 1, NAME_TITLE))
1286 say("", "%s", sqlite3_errmsg(db));
1288 while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
1289 name = (const char *)sqlite3_column_text(stmt, 0);
1290 sec = (const char *)sqlite3_column_text(stmt, 1);
1291 arch = (const char *)sqlite3_column_text(stmt, 2);
1292 key = (const char *)sqlite3_column_text(stmt, 3);
1293 say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
1294 '\0' == *arch ? "" : "/",
1295 '\0' == *arch ? "" : arch, key);
1297 sqlite3_finalize(stmt);
1301 parse_cat(struct mpage *mpage, int fd)
1304 char *line, *p, *title;
1305 size_t len, plen, titlesz;
1307 stream = (-1 == fd) ?
1308 fopen(mpage->mlinks->file, "r") :
1310 if (NULL == stream) {
1314 say(mpage->mlinks->file, "&fopen");
1318 /* Skip to first blank line. */
1320 while (NULL != (line = fgetln(stream, &len)))
1325 * Assume the first line that is not indented
1326 * is the first section header. Skip to it.
1329 while (NULL != (line = fgetln(stream, &len)))
1330 if ('\n' != *line && ' ' != *line)
1334 * Read up until the next section into a buffer.
1335 * Strip the leading and trailing newline from each read line,
1336 * appending a trailing space.
1337 * Ignore empty (whitespace-only) lines.
1343 while (NULL != (line = fgetln(stream, &len))) {
1344 if (' ' != *line || '\n' != line[len - 1])
1346 while (len > 0 && isspace((unsigned char)*line)) {
1352 title = mandoc_realloc(title, titlesz + len);
1353 memcpy(title + titlesz, line, len);
1355 title[titlesz - 1] = ' ';
1359 * If no page content can be found, or the input line
1360 * is already the next section header, or there is no
1361 * trailing newline, reuse the page title as the page
1365 if (NULL == title || '\0' == *title) {
1367 say(mpage->mlinks->file,
1368 "Cannot find NAME section");
1374 title = mandoc_realloc(title, titlesz + 1);
1375 title[titlesz] = '\0';
1378 * Skip to the first dash.
1379 * Use the remaining line as the description (no more than 70
1383 if (NULL != (p = strstr(title, "- "))) {
1384 for (p += 2; ' ' == *p || '\b' == *p; p++)
1385 /* Skip to next word. */ ;
1388 say(mpage->mlinks->file,
1389 "No dash in title line");
1395 /* Strip backspace-encoding from line. */
1397 while (NULL != (line = memchr(p, '\b', plen))) {
1400 memmove(line, line + 1, plen--);
1403 memmove(line - 1, line + 1, plen - len);
1407 mpage->desc = mandoc_strdup(p);
1413 * Put a type/word pair into the word database for this particular file.
1416 putkey(const struct mpage *mpage, char *value, uint64_t type)
1420 assert(NULL != value);
1421 if (TYPE_arch == type)
1422 for (cp = value; *cp; cp++)
1423 if (isupper((unsigned char)*cp))
1424 *cp = _tolower((unsigned char)*cp);
1425 putkeys(mpage, value, strlen(value), type);
1429 * Grok all nodes at or below a certain mdoc node into putkey().
1432 putmdockey(const struct mpage *mpage,
1433 const struct mdoc_node *n, uint64_t m)
1436 for ( ; NULL != n; n = n->next) {
1437 if (NULL != n->child)
1438 putmdockey(mpage, n->child, m);
1439 if (MDOC_TEXT == n->type)
1440 putkey(mpage, n->string, m);
1445 parse_man(struct mpage *mpage, const struct man_meta *meta,
1446 const struct man_node *n)
1448 const struct man_node *head, *body;
1449 char *start, *title;
1457 * We're only searching for one thing: the first text child in
1458 * the BODY of a NAME section. Since we don't keep track of
1459 * sections in -man, run some hoops to find out whether we're in
1460 * the correct section or not.
1463 if (MAN_BODY == n->type && MAN_SH == n->tok) {
1465 assert(body->parent);
1466 if (NULL != (head = body->parent->head) &&
1467 1 == head->nchild &&
1468 NULL != (head = (head->child)) &&
1469 MAN_TEXT == head->type &&
1470 0 == strcmp(head->string, "NAME") &&
1471 NULL != body->child) {
1474 * Suck the entire NAME section into memory.
1475 * Yes, we might run away.
1476 * But too many manuals have big, spread-out
1477 * NAME sections over many lines.
1481 man_deroff(&title, body);
1486 * Go through a special heuristic dance here.
1487 * Conventionally, one or more manual names are
1488 * comma-specified prior to a whitespace, then a
1489 * dash, then a description. Try to puzzle out
1490 * the name parts here.
1495 sz = strcspn(start, " ,");
1496 if ('\0' == start[sz])
1503 * Assume a stray trailing comma in the
1504 * name list if a name begins with a dash.
1507 if ('-' == start[0] ||
1508 ('\\' == start[0] && '-' == start[1]))
1511 putkey(mpage, start, NAME_TITLE);
1512 if ( ! (mpage->name_head_done ||
1513 strcasecmp(start, meta->title))) {
1514 putkey(mpage, start, NAME_HEAD);
1515 mpage->name_head_done = 1;
1523 assert(',' == byte);
1525 while (' ' == *start)
1529 if (start == title) {
1530 putkey(mpage, start, NAME_TITLE);
1531 if ( ! (mpage->name_head_done ||
1532 strcasecmp(start, meta->title))) {
1533 putkey(mpage, start, NAME_HEAD);
1534 mpage->name_head_done = 1;
1540 while (isspace((unsigned char)*start))
1543 if (0 == strncmp(start, "-", 1))
1545 else if (0 == strncmp(start, "\\-\\-", 4))
1547 else if (0 == strncmp(start, "\\-", 2))
1549 else if (0 == strncmp(start, "\\(en", 4))
1551 else if (0 == strncmp(start, "\\(em", 4))
1554 while (' ' == *start)
1557 mpage->desc = mandoc_strdup(start);
1563 for (n = n->child; n; n = n->next) {
1564 if (NULL != mpage->desc)
1566 parse_man(mpage, meta, n);
1571 parse_mdoc(struct mpage *mpage, const struct mdoc_meta *meta,
1572 const struct mdoc_node *n)
1576 for (n = n->child; NULL != n; n = n->next) {
1587 if (NULL != mdocs[n->tok].fp)
1588 if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
1590 if (mdocs[n->tok].mask)
1591 putmdockey(mpage, n->child,
1592 mdocs[n->tok].mask);
1595 assert(MDOC_ROOT != n->type);
1598 if (NULL != n->child)
1599 parse_mdoc(mpage, meta, n);
1604 parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_meta *meta,
1605 const struct mdoc_node *n)
1610 if (SEC_SYNOPSIS != n->sec ||
1611 NULL == (n = n->child) ||
1612 MDOC_TEXT != n->type)
1616 * Only consider those `Fd' macro fields that begin with an
1617 * "inclusion" token (versus, e.g., #define).
1620 if (strcmp("#include", n->string))
1623 if (NULL == (n = n->next) || MDOC_TEXT != n->type)
1627 * Strip away the enclosing angle brackets and make sure we're
1632 if ('<' == *start || '"' == *start)
1635 if (0 == (sz = strlen(start)))
1638 end = &start[(int)sz - 1];
1639 if ('>' == *end || '"' == *end)
1643 putkeys(mpage, start, end - start + 1, TYPE_In);
1648 parse_mdoc_fname(struct mpage *mpage, const struct mdoc_node *n)
1653 if (n->type != MDOC_TEXT)
1656 /* Skip function pointer punctuation. */
1659 while (*cp == '(' || *cp == '*')
1661 sz = strcspn(cp, "()");
1663 putkeys(mpage, cp, sz, TYPE_Fn);
1664 if (n->sec == SEC_SYNOPSIS)
1665 putkeys(mpage, cp, sz, NAME_SYN);
1669 parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_meta *meta,
1670 const struct mdoc_node *n)
1673 if (n->child == NULL)
1676 parse_mdoc_fname(mpage, n->child);
1678 for (n = n->child->next; n != NULL; n = n->next)
1679 if (n->type == MDOC_TEXT)
1680 putkey(mpage, n->string, TYPE_Fa);
1686 parse_mdoc_Fo(struct mpage *mpage, const struct mdoc_meta *meta,
1687 const struct mdoc_node *n)
1690 if (n->type != MDOC_HEAD)
1693 if (n->child != NULL)
1694 parse_mdoc_fname(mpage, n->child);
1700 parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_meta *meta,
1701 const struct mdoc_node *n)
1705 if (NULL == (n = n->child))
1708 if (NULL == n->next) {
1709 putkey(mpage, n->string, TYPE_Xr);
1713 mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
1714 putkey(mpage, cp, TYPE_Xr);
1720 parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_meta *meta,
1721 const struct mdoc_node *n)
1724 if (MDOC_BODY == n->type)
1725 mdoc_deroff(&mpage->desc, n);
1730 parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_meta *meta,
1731 const struct mdoc_node *n)
1734 if (SEC_NAME == n->sec)
1735 putmdockey(mpage, n->child, NAME_TITLE);
1736 else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type) {
1737 if (n->child == NULL)
1738 putkey(mpage, meta->name, NAME_SYN);
1740 putmdockey(mpage, n->child, NAME_SYN);
1742 if ( ! (mpage->name_head_done ||
1743 n->child == NULL || n->child->string == NULL ||
1744 strcasecmp(n->child->string, meta->title))) {
1745 putkey(mpage, n->child->string, NAME_HEAD);
1746 mpage->name_head_done = 1;
1752 parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_meta *meta,
1753 const struct mdoc_node *n)
1756 return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
1760 parse_mdoc_head(struct mpage *mpage, const struct mdoc_meta *meta,
1761 const struct mdoc_node *n)
1764 return(MDOC_HEAD == n->type);
1768 parse_mdoc_body(struct mpage *mpage, const struct mdoc_meta *meta,
1769 const struct mdoc_node *n)
1772 return(MDOC_BODY == n->type);
1776 * Add a string to the hash table for the current manual.
1777 * Each string has a bitmask telling which macros it belongs to.
1778 * When we finish the manual, we'll dump the table.
1781 putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
1792 mustfree = render_string(&cp, &sz);
1798 name_mask &= ~NAME_FIRST;
1800 say(mpage->mlinks->file,
1801 "Adding name %*s, bits=%d", sz, cp, v);
1805 for (i = 0; i < mansearch_keymax; i++)
1806 if ((uint64_t)1 << i & v)
1807 say(mpage->mlinks->file,
1808 "Adding key %s=%*s",
1809 mansearch_keynames[i], sz, cp);
1813 slot = ohash_qlookupi(htab, cp, &end);
1814 s = ohash_find(htab, slot);
1816 if (NULL != s && mpage == s->mpage) {
1819 } else if (NULL == s) {
1820 s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
1821 memcpy(s->key, cp, sz);
1822 ohash_insert(htab, slot, s);
1832 * Take a Unicode codepoint and produce its UTF-8 encoding.
1833 * This isn't the best way to do this, but it works.
1834 * The magic numbers are from the UTF-8 packaging.
1835 * They're not as scary as they seem: read the UTF-8 spec for details.
1838 utf8(unsigned int cp, char out[7])
1843 if (cp <= 0x0000007F) {
1846 } else if (cp <= 0x000007FF) {
1848 out[0] = (cp >> 6 & 31) | 192;
1849 out[1] = (cp & 63) | 128;
1850 } else if (cp <= 0x0000FFFF) {
1852 out[0] = (cp >> 12 & 15) | 224;
1853 out[1] = (cp >> 6 & 63) | 128;
1854 out[2] = (cp & 63) | 128;
1855 } else if (cp <= 0x001FFFFF) {
1857 out[0] = (cp >> 18 & 7) | 240;
1858 out[1] = (cp >> 12 & 63) | 128;
1859 out[2] = (cp >> 6 & 63) | 128;
1860 out[3] = (cp & 63) | 128;
1861 } else if (cp <= 0x03FFFFFF) {
1863 out[0] = (cp >> 24 & 3) | 248;
1864 out[1] = (cp >> 18 & 63) | 128;
1865 out[2] = (cp >> 12 & 63) | 128;
1866 out[3] = (cp >> 6 & 63) | 128;
1867 out[4] = (cp & 63) | 128;
1868 } else if (cp <= 0x7FFFFFFF) {
1870 out[0] = (cp >> 30 & 1) | 252;
1871 out[1] = (cp >> 24 & 63) | 128;
1872 out[2] = (cp >> 18 & 63) | 128;
1873 out[3] = (cp >> 12 & 63) | 128;
1874 out[4] = (cp >> 6 & 63) | 128;
1875 out[5] = (cp & 63) | 128;
1884 * If the string contains escape sequences,
1885 * replace it with an allocated rendering and return 1,
1886 * such that the caller can free it after use.
1887 * Otherwise, do nothing and return 0.
1890 render_string(char **public, size_t *psz)
1892 const char *src, *scp, *addcp, *seq;
1894 size_t ssz, dsz, addsz;
1895 char utfbuf[7], res[6];
1896 int seqlen, unicode;
1900 res[2] = ASCII_NBRSP;
1901 res[3] = ASCII_HYPH;
1902 res[4] = ASCII_BREAK;
1905 src = scp = *public;
1910 while (scp < src + *psz) {
1912 /* Leave normal characters unchanged. */
1914 if (strchr(res, *scp) == NULL) {
1922 * Found something that requires replacing,
1923 * make sure we have a destination buffer.
1927 dst = mandoc_malloc(ssz + 1);
1929 memcpy(dst, src, dsz);
1932 /* Handle single-char special characters. */
1954 * Found an escape sequence.
1955 * Read past the slash, then parse it.
1956 * Ignore everything except characters.
1960 if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
1964 * Render the special character
1965 * as either UTF-8 or ASCII.
1969 unicode = mchars_spec2cp(mchars, seq, seqlen);
1972 addsz = utf8(unicode, utfbuf);
1977 addcp = mchars_spec2str(mchars, seq, seqlen, &addsz);
1980 if (*addcp == ASCII_NBRSP) {
1986 /* Copy the rendered glyph into the stream. */
1989 dst = mandoc_realloc(dst, ssz + 1);
1990 memcpy(dst + dsz, addcp, addsz);
1998 /* Trim trailing whitespace and NUL-terminate. */
2000 while (*psz > 0 && (*public)[*psz - 1] == ' ')
2003 (*public)[*psz] = '\0';
2010 dbadd_mlink(const struct mlink *mlink)
2015 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
2016 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
2017 SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
2018 SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
2019 SQL_STEP(stmts[STMT_INSERT_LINK]);
2020 sqlite3_reset(stmts[STMT_INSERT_LINK]);
2024 dbadd_mlink_name(const struct mlink *mlink)
2032 SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
2033 bits = NAME_FILE & NAME_MASK;
2034 if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
2035 bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
2036 sqlite3_reset(stmts[STMT_SELECT_NAME]);
2040 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
2041 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
2042 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
2043 SQL_STEP(stmts[STMT_INSERT_NAME]);
2044 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2048 * Flush the current page's terms (and their bits) into the database.
2049 * Wrap the entire set of additions in a transaction to make sqlite be a
2051 * Also, handle escape sequences at the last possible moment.
2054 dbadd(struct mpage *mpage)
2056 struct mlink *mlink;
2063 mlink = mpage->mlinks;
2066 for (key = ohash_first(&names, &slot); NULL != key;
2067 key = ohash_next(&names, &slot))
2069 for (key = ohash_first(&strings, &slot); NULL != key;
2070 key = ohash_next(&strings, &slot))
2074 while (NULL != mlink) {
2075 fputs(mlink->name, stdout);
2076 if (NULL == mlink->next ||
2077 strcmp(mlink->dsec, mlink->next->dsec) ||
2078 strcmp(mlink->fsec, mlink->next->fsec) ||
2079 strcmp(mlink->arch, mlink->next->arch)) {
2081 if ('\0' == *mlink->dsec)
2082 fputs(mlink->fsec, stdout);
2084 fputs(mlink->dsec, stdout);
2085 if ('\0' != *mlink->arch)
2086 printf("/%s", mlink->arch);
2089 mlink = mlink->next;
2091 fputs(", ", stdout);
2093 printf(" - %s\n", mpage->desc);
2098 say(mlink->file, "Adding to database");
2102 mustfree = render_string(&cp, &i);
2104 SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
2105 SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
2106 SQL_STEP(stmts[STMT_INSERT_PAGE]);
2107 mpage->pageid = sqlite3_last_insert_rowid(db);
2108 sqlite3_reset(stmts[STMT_INSERT_PAGE]);
2112 while (NULL != mlink) {
2114 mlink = mlink->next;
2116 mlink = mpage->mlinks;
2118 for (key = ohash_first(&names, &slot); NULL != key;
2119 key = ohash_next(&names, &slot)) {
2120 assert(key->mpage == mpage);
2122 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
2123 SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
2124 SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
2125 SQL_STEP(stmts[STMT_INSERT_NAME]);
2126 sqlite3_reset(stmts[STMT_INSERT_NAME]);
2129 for (key = ohash_first(&strings, &slot); NULL != key;
2130 key = ohash_next(&strings, &slot)) {
2131 assert(key->mpage == mpage);
2133 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
2134 SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
2135 SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
2136 SQL_STEP(stmts[STMT_INSERT_KEY]);
2137 sqlite3_reset(stmts[STMT_INSERT_KEY]);
2145 struct mpage *mpage;
2146 struct mlink *mlink;
2151 SQL_EXEC("BEGIN TRANSACTION");
2153 for (mpage = ohash_first(&mpages, &slot); NULL != mpage;
2154 mpage = ohash_next(&mpages, &slot)) {
2155 mlink = mpage->mlinks;
2157 say(mlink->file, "Deleting from database");
2160 for ( ; NULL != mlink; mlink = mlink->next) {
2162 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2164 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2166 SQL_BIND_TEXT(stmts[STMT_DELETE_PAGE],
2168 SQL_STEP(stmts[STMT_DELETE_PAGE]);
2169 sqlite3_reset(stmts[STMT_DELETE_PAGE]);
2174 SQL_EXEC("END TRANSACTION");
2178 * Close an existing database and its prepared statements.
2179 * If "real" is not set, rename the temporary file into the real one.
2191 for (i = 0; i < STMT__MAX; i++) {
2192 sqlite3_finalize(stmts[i]);
2202 if ('\0' == *tempfilename) {
2203 if (-1 == rename(MANDOC_DB "~", MANDOC_DB)) {
2204 exitcode = (int)MANDOCLEVEL_SYSERR;
2205 say(MANDOC_DB, "&rename");
2210 switch (child = fork()) {
2212 exitcode = (int)MANDOCLEVEL_SYSERR;
2213 say("", "&fork cmp");
2216 execlp("cmp", "cmp", "-s",
2217 tempfilename, MANDOC_DB, NULL);
2218 say("", "&exec cmp");
2223 if (-1 == waitpid(child, &status, 0)) {
2224 exitcode = (int)MANDOCLEVEL_SYSERR;
2225 say("", "&wait cmp");
2226 } else if (WIFSIGNALED(status)) {
2227 exitcode = (int)MANDOCLEVEL_SYSERR;
2228 say("", "cmp died from signal %d", WTERMSIG(status));
2229 } else if (WEXITSTATUS(status)) {
2230 exitcode = (int)MANDOCLEVEL_SYSERR;
2232 "Data changed, but cannot replace database");
2235 *strrchr(tempfilename, '/') = '\0';
2236 switch (child = fork()) {
2238 exitcode = (int)MANDOCLEVEL_SYSERR;
2239 say("", "&fork rm");
2242 execlp("rm", "rm", "-rf", tempfilename, NULL);
2243 say("", "&exec rm");
2244 exit((int)MANDOCLEVEL_SYSERR);
2248 if (-1 == waitpid(child, &status, 0)) {
2249 exitcode = (int)MANDOCLEVEL_SYSERR;
2250 say("", "&wait rm");
2251 } else if (WIFSIGNALED(status) || WEXITSTATUS(status)) {
2252 exitcode = (int)MANDOCLEVEL_SYSERR;
2253 say("", "%s: Cannot remove temporary directory",
2259 * This is straightforward stuff.
2260 * Open a database connection to a "temporary" database, then open a set
2261 * of prepared statements we'll use over and over again.
2262 * If "real" is set, we use the existing database; if not, we truncate a
2264 * Must be matched by dbclose().
2275 *tempfilename = '\0';
2276 ofl = SQLITE_OPEN_READWRITE;
2279 rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
2280 if (SQLITE_OK != rc) {
2281 exitcode = (int)MANDOCLEVEL_SYSERR;
2282 if (SQLITE_CANTOPEN != rc)
2283 say(MANDOC_DB, "%s", sqlite3_errstr(rc));
2286 goto prepare_statements;
2289 ofl |= SQLITE_OPEN_CREATE | SQLITE_OPEN_EXCLUSIVE;
2291 remove(MANDOC_DB "~");
2292 rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
2293 if (SQLITE_OK == rc)
2295 if (MPARSE_QUICK & mparse_options) {
2296 exitcode = (int)MANDOCLEVEL_SYSERR;
2297 say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
2301 (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
2302 sizeof(tempfilename));
2303 if (NULL == mkdtemp(tempfilename)) {
2304 exitcode = (int)MANDOCLEVEL_SYSERR;
2305 say("", "&%s", tempfilename);
2308 (void)strlcat(tempfilename, "/" MANDOC_DB,
2309 sizeof(tempfilename));
2310 rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
2311 if (SQLITE_OK != rc) {
2312 exitcode = (int)MANDOCLEVEL_SYSERR;
2313 say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
2318 sql = "CREATE TABLE \"mpages\" (\n"
2319 " \"desc\" TEXT NOT NULL,\n"
2320 " \"form\" INTEGER NOT NULL,\n"
2321 " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
2324 "CREATE TABLE \"mlinks\" (\n"
2325 " \"sec\" TEXT NOT NULL,\n"
2326 " \"arch\" TEXT NOT NULL,\n"
2327 " \"name\" TEXT NOT NULL,\n"
2328 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2329 "ON DELETE CASCADE\n"
2331 "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
2333 "CREATE TABLE \"names\" (\n"
2334 " \"bits\" INTEGER NOT NULL,\n"
2335 " \"name\" TEXT NOT NULL,\n"
2336 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2337 "ON DELETE CASCADE,\n"
2338 " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
2341 "CREATE TABLE \"keys\" (\n"
2342 " \"bits\" INTEGER NOT NULL,\n"
2343 " \"key\" TEXT NOT NULL,\n"
2344 " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
2345 "ON DELETE CASCADE\n"
2347 "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
2349 if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
2350 exitcode = (int)MANDOCLEVEL_SYSERR;
2351 say(MANDOC_DB, "%s", sqlite3_errmsg(db));
2357 if (SQLITE_OK != sqlite3_exec(db,
2358 "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
2359 exitcode = (int)MANDOCLEVEL_SYSERR;
2360 say(MANDOC_DB, "PRAGMA foreign_keys: %s",
2361 sqlite3_errmsg(db));
2366 sql = "DELETE FROM mpages WHERE pageid IN "
2367 "(SELECT pageid FROM mlinks WHERE "
2368 "sec=? AND arch=? AND name=?)";
2369 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
2370 sql = "INSERT INTO mpages "
2371 "(desc,form) VALUES (?,?)";
2372 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
2373 sql = "INSERT INTO mlinks "
2374 "(sec,arch,name,pageid) VALUES (?,?,?,?)";
2375 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
2376 sql = "SELECT bits FROM names where pageid = ?";
2377 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
2378 sql = "INSERT INTO names "
2379 "(bits,name,pageid) VALUES (?,?,?)";
2380 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
2381 sql = "INSERT INTO keys "
2382 "(bits,key,pageid) VALUES (?,?,?)";
2383 sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
2387 * When opening a new database, we can turn off
2388 * synchronous mode for much better performance.
2391 if (real && SQLITE_OK != sqlite3_exec(db,
2392 "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
2393 exitcode = (int)MANDOCLEVEL_SYSERR;
2394 say(MANDOC_DB, "PRAGMA synchronous: %s",
2395 sqlite3_errmsg(db));
2405 hash_calloc(size_t n, size_t sz, void *arg)
2408 return(mandoc_calloc(n, sz));
2412 hash_alloc(size_t sz, void *arg)
2415 return(mandoc_malloc(sz));
2419 hash_free(void *p, void *arg)
2426 set_basedir(const char *targetdir, int report_baddir)
2428 static char startdir[PATH_MAX];
2429 static int getcwd_status; /* 1 = ok, 2 = failure */
2430 static int chdir_status; /* 1 = changed directory */
2434 * Remember the original working directory, if possible.
2435 * This will be needed if the second or a later directory
2436 * on the command line is given as a relative path.
2437 * Do not error out if the current directory is not
2438 * searchable: Maybe it won't be needed after all.
2440 if (0 == getcwd_status) {
2441 if (NULL == getcwd(startdir, sizeof(startdir))) {
2443 (void)strlcpy(startdir, strerror(errno),
2450 * We are leaving the old base directory.
2451 * Do not use it any longer, not even for messages.
2456 * If and only if the directory was changed earlier and
2457 * the next directory to process is given as a relative path,
2458 * first go back, or bail out if that is impossible.
2460 if (chdir_status && '/' != *targetdir) {
2461 if (2 == getcwd_status) {
2462 exitcode = (int)MANDOCLEVEL_SYSERR;
2463 say("", "getcwd: %s", startdir);
2466 if (-1 == chdir(startdir)) {
2467 exitcode = (int)MANDOCLEVEL_SYSERR;
2468 say("", "&chdir %s", startdir);
2474 * Always resolve basedir to the canonicalized absolute
2475 * pathname and append a trailing slash, such that
2476 * we can reliably check whether files are inside.
2478 if (NULL == realpath(targetdir, basedir)) {
2479 if (report_baddir || errno != ENOENT) {
2480 exitcode = (int)MANDOCLEVEL_BADARG;
2481 say("", "&%s: realpath", targetdir);
2484 } else if (-1 == chdir(basedir)) {
2485 if (report_baddir || errno != ENOENT) {
2486 exitcode = (int)MANDOCLEVEL_BADARG;
2492 cp = strchr(basedir, '\0');
2493 if ('/' != cp[-1]) {
2494 if (cp - basedir >= PATH_MAX - 1) {
2495 exitcode = (int)MANDOCLEVEL_SYSERR;
2496 say("", "Filename too long");
2506 say(const char *file, const char *format, ...)
2511 if ('\0' != *basedir)
2512 fprintf(stderr, "%s", basedir);
2513 if ('\0' != *basedir && '\0' != *file)
2516 fprintf(stderr, "%s", file);
2519 if (NULL != format) {
2532 if (NULL != format) {
2533 if ('\0' != *basedir || '\0' != *file)
2534 fputs(": ", stderr);
2535 va_start(ap, format);
2536 vfprintf(stderr, format, ap);
2540 if ('\0' != *basedir || '\0' != *file || NULL != format)
2541 fputs(": ", stderr);
2544 fputc('\n', stderr);