-/* $Id: mandocdb.c,v 1.131 2014/04/04 18:23:30 schwarze Exp $ */
+/* $Id: mandocdb.c,v 1.219 2016/07/15 18:03:45 schwarze Exp $ */
/*
* Copyright (c) 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2011, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2011-2016 Ingo Schwarze <schwarze@openbsd.org>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
- * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
-#ifdef HAVE_CONFIG_H
#include "config.h"
-#endif
+#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
#include <assert.h>
#include <ctype.h>
+#if HAVE_ERR
+#include <err.h>
+#endif
#include <errno.h>
#include <fcntl.h>
+#if HAVE_FTS
#include <fts.h>
-#include <getopt.h>
+#else
+#include "compat_fts.h"
+#endif
#include <limits.h>
+#if HAVE_SANDBOX_INIT
+#include <sandbox.h>
+#endif
#include <stddef.h>
#include <stdio.h>
#include <stdint.h>
#include <string.h>
#include <unistd.h>
-#ifdef HAVE_OHASH
-#include <ohash.h>
-#else
-#include "compat_ohash.h"
-#endif
#include <sqlite3.h>
+#include "mandoc_aux.h"
+#include "mandoc_ohash.h"
+#include "mandoc.h"
+#include "roff.h"
#include "mdoc.h"
#include "man.h"
-#include "mandoc.h"
-#include "mandoc_aux.h"
-#include "manpath.h"
+#include "manconf.h"
#include "mansearch.h"
extern int mansearch_keymax;
#define SQL_EXEC(_v) \
if (SQLITE_OK != sqlite3_exec(db, (_v), NULL, NULL, NULL)) \
- fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+ say("", "%s: %s", (_v), sqlite3_errmsg(db))
#define SQL_BIND_TEXT(_s, _i, _v) \
if (SQLITE_OK != sqlite3_bind_text \
((_s), (_i)++, (_v), -1, SQLITE_STATIC)) \
- fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+ say(mlink->file, "%s", sqlite3_errmsg(db))
#define SQL_BIND_INT(_s, _i, _v) \
if (SQLITE_OK != sqlite3_bind_int \
((_s), (_i)++, (_v))) \
- fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+ say(mlink->file, "%s", sqlite3_errmsg(db))
#define SQL_BIND_INT64(_s, _i, _v) \
if (SQLITE_OK != sqlite3_bind_int64 \
((_s), (_i)++, (_v))) \
- fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+ say(mlink->file, "%s", sqlite3_errmsg(db))
#define SQL_STEP(_s) \
if (SQLITE_DONE != sqlite3_step((_s))) \
- fprintf(stderr, "%s\n", sqlite3_errmsg(db))
+ say(mlink->file, "%s", sqlite3_errmsg(db))
enum op {
OP_DEFAULT = 0, /* new dbs from dir list or default config */
OP_TEST /* change no databases, report potential problems */
};
-enum form {
- FORM_NONE, /* format is unknown */
- FORM_SRC, /* format is -man or -mdoc */
- FORM_CAT /* format is cat */
-};
-
struct str {
- char *rendered; /* key in UTF-8 or ASCII form */
const struct mpage *mpage; /* if set, the owning parse */
uint64_t mask; /* bitmask in sequence */
- char key[]; /* may contain escape sequences */
+ char key[]; /* rendered text */
};
struct inodev {
struct mpage {
struct inodev inodev; /* used for hashing routine */
- int64_t recno; /* id in mpages SQL table */
- enum form form; /* format from file content */
+ int64_t pageid; /* pageid in mpages SQL table */
char *sec; /* section from file content */
char *arch; /* architecture from file content */
char *title; /* title from file content */
char *desc; /* description from file content */
struct mlink *mlinks; /* singly linked list */
+ int form; /* format from file content */
+ int name_head_done;
};
struct mlink {
char file[PATH_MAX]; /* filename rel. to manpath */
- enum form dform; /* format from directory */
- enum form fform; /* format from file name suffix */
char *dsec; /* section from directory */
char *arch; /* architecture from directory */
char *name; /* name from file name (not empty) */
char *fsec; /* section from file name suffix */
struct mlink *next; /* singly linked list */
struct mpage *mpage; /* parent */
+ int dform; /* format from directory */
+ int fform; /* format from file name suffix */
int gzip; /* filename has a .gz suffix */
};
STMT_DELETE_PAGE = 0, /* delete mpage */
STMT_INSERT_PAGE, /* insert mpage */
STMT_INSERT_LINK, /* insert mlink */
+ STMT_INSERT_NAME, /* insert name */
+ STMT_SELECT_NAME, /* retrieve existing name flags */
STMT_INSERT_KEY, /* insert parsed key */
STMT__MAX
};
-typedef int (*mdoc_fp)(struct mpage *, const struct mdoc_node *);
+typedef int (*mdoc_fp)(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
struct mdoc_handler {
mdoc_fp fp; /* optional handler */
uint64_t mask; /* set unless handler returns 0 */
};
+
+int mandocdb(int, char *[]);
+
static void dbclose(int);
-static void dbadd(struct mpage *, struct mchars *);
+static void dbadd(struct mpage *);
static void dbadd_mlink(const struct mlink *mlink);
+static void dbadd_mlink_name(const struct mlink *mlink);
static int dbopen(int);
static void dbprune(void);
static void filescan(const char *);
-static void *hash_alloc(size_t, void *);
-static void hash_free(void *, size_t, void *);
-static void *hash_halloc(size_t, void *);
static void mlink_add(struct mlink *, const struct stat *);
static void mlink_check(struct mpage *, struct mlink *);
static void mlink_free(struct mlink *);
static void mlinks_undupe(struct mpage *);
static void mpages_free(void);
-static void mpages_merge(struct mchars *, struct mparse *);
+static void mpages_merge(struct mparse *);
static void names_check(void);
static void parse_cat(struct mpage *, int);
-static void parse_man(struct mpage *, const struct man_node *);
-static void parse_mdoc(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_body(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_head(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Fd(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Fn(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Nd(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Nm(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Sh(struct mpage *, const struct mdoc_node *);
-static int parse_mdoc_Xr(struct mpage *, const struct mdoc_node *);
+static void parse_man(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static void parse_mdoc(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_head(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Fd(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static void parse_mdoc_fname(struct mpage *, const struct roff_node *);
+static int parse_mdoc_Fn(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Fo(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Nd(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Nm(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Sh(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Va(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
+static int parse_mdoc_Xr(struct mpage *, const struct roff_meta *,
+ const struct roff_node *);
static void putkey(const struct mpage *, char *, uint64_t);
-static void putkeys(const struct mpage *,
- const char *, size_t, uint64_t);
+static void putkeys(const struct mpage *, char *, size_t, uint64_t);
static void putmdockey(const struct mpage *,
- const struct mdoc_node *, uint64_t);
-static void render_key(struct mchars *, struct str *);
+ const struct roff_node *, uint64_t);
+static int render_string(char **, size_t *);
static void say(const char *, const char *, ...);
-static int set_basedir(const char *);
+static int set_basedir(const char *, int);
static int treescan(void);
static size_t utf8(unsigned int, char [7]);
static char tempfilename[32];
-static char *progname;
static int nodb; /* no database changes */
static int mparse_options; /* abort the parse early */
-static int use_all; /* use all found files */
-static int debug; /* print what we're doing */
-static int warnings; /* warn about crap */
+static int use_all; /* use all found files */
+static int debug; /* print what we're doing */
+static int warnings; /* warn about crap */
static int write_utf8; /* write UTF-8 output; else ASCII */
static int exitcode; /* to be returned by main */
-static enum op op; /* operational mode */
+static enum op op; /* operational mode */
static char basedir[PATH_MAX]; /* current base directory */
static struct ohash mpages; /* table of distinct manual pages */
static struct ohash mlinks; /* table of directory entries */
+static struct ohash names; /* table of all names */
static struct ohash strings; /* table of all strings */
static sqlite3 *db = NULL; /* current database */
static sqlite3_stmt *stmts[STMT__MAX]; /* current statements */
+static uint64_t name_mask;
static const struct mdoc_handler mdocs[MDOC_MAX] = {
{ NULL, 0 }, /* Ap */
{ NULL, TYPE_Ic }, /* Ic */
{ NULL, TYPE_In }, /* In */
{ NULL, TYPE_Li }, /* Li */
- { parse_mdoc_Nd, TYPE_Nd }, /* Nd */
+ { parse_mdoc_Nd, 0 }, /* Nd */
{ parse_mdoc_Nm, 0 }, /* Nm */
{ NULL, 0 }, /* Op */
{ NULL, 0 }, /* Ot */
{ NULL, TYPE_Pa }, /* Pa */
{ NULL, 0 }, /* Rv */
{ NULL, TYPE_St }, /* St */
- { NULL, TYPE_Va }, /* Va */
- { parse_mdoc_body, TYPE_Va }, /* Vt */
+ { parse_mdoc_Va, TYPE_Va }, /* Va */
+ { parse_mdoc_Va, TYPE_Vt }, /* Vt */
{ parse_mdoc_Xr, 0 }, /* Xr */
{ NULL, 0 }, /* %A */
{ NULL, 0 }, /* %B */
{ NULL, 0 }, /* Ux */
{ NULL, 0 }, /* Xc */
{ NULL, 0 }, /* Xo */
- { parse_mdoc_head, 0 }, /* Fo */
+ { parse_mdoc_Fo, 0 }, /* Fo */
{ NULL, 0 }, /* Fc */
{ NULL, 0 }, /* Oo */
{ NULL, 0 }, /* Oc */
{ NULL, 0 }, /* sp */
{ NULL, 0 }, /* %U */
{ NULL, 0 }, /* Ta */
+ { NULL, 0 }, /* ll */
};
+
int
-main(int argc, char *argv[])
+mandocdb(int argc, char *argv[])
{
- int ch, i;
- size_t j, sz;
- const char *path_arg;
- struct mchars *mc;
- struct manpaths dirs;
+ struct manconf conf;
struct mparse *mp;
- struct ohash_info mpages_info, mlinks_info;
-
- memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
- memset(&dirs, 0, sizeof(struct manpaths));
+ const char *path_arg, *progname;
+ size_t j, sz;
+ int ch, i;
- mpages_info.alloc = mlinks_info.alloc = hash_alloc;
- mpages_info.halloc = mlinks_info.halloc = hash_halloc;
- mpages_info.hfree = mlinks_info.hfree = hash_free;
+#if HAVE_PLEDGE
+ if (pledge("stdio rpath wpath cpath fattr flock proc exec", NULL) == -1) {
+ warn("pledge");
+ return (int)MANDOCLEVEL_SYSERR;
+ }
+#endif
- mpages_info.key_offset = offsetof(struct mpage, inodev);
- mlinks_info.key_offset = offsetof(struct mlink, file);
+#if HAVE_SANDBOX_INIT
+ if (sandbox_init(kSBXProfileNoInternet, SANDBOX_NAMED, NULL) == -1) {
+ warnx("sandbox_init");
+ return (int)MANDOCLEVEL_SYSERR;
+ }
+#endif
- progname = strrchr(argv[0], '/');
- if (progname == NULL)
- progname = argv[0];
- else
- ++progname;
+ memset(&conf, 0, sizeof(conf));
+ memset(stmts, 0, STMT__MAX * sizeof(sqlite3_stmt *));
/*
- * We accept a few different invocations.
+ * We accept a few different invocations.
* The CHECKOP macro makes sure that invocation styles don't
* clobber each other.
*/
#define CHECKOP(_op, _ch) do \
if (OP_DEFAULT != (_op)) { \
- fprintf(stderr, "-%c: Conflicting option\n", (_ch)); \
+ warnx("-%c: Conflicting option", (_ch)); \
goto usage; \
} while (/*CONSTCOND*/0)
while (-1 != (ch = getopt(argc, argv, "aC:Dd:npQT:tu:v")))
switch (ch) {
- case ('a'):
+ case 'a':
use_all = 1;
break;
- case ('C'):
+ case 'C':
CHECKOP(op, ch);
path_arg = optarg;
op = OP_CONFFILE;
break;
- case ('D'):
+ case 'D':
debug++;
break;
- case ('d'):
+ case 'd':
CHECKOP(op, ch);
path_arg = optarg;
op = OP_UPDATE;
break;
- case ('n'):
+ case 'n':
nodb = 1;
break;
- case ('p'):
+ case 'p':
warnings = 1;
break;
- case ('Q'):
+ case 'Q':
mparse_options |= MPARSE_QUICK;
break;
- case ('T'):
+ case 'T':
if (strcmp(optarg, "utf8")) {
- fprintf(stderr, "-T%s: Unsupported "
- "output format\n", optarg);
+ warnx("-T%s: Unsupported output format",
+ optarg);
goto usage;
}
write_utf8 = 1;
break;
- case ('t'):
+ case 't':
CHECKOP(op, ch);
dup2(STDOUT_FILENO, STDERR_FILENO);
op = OP_TEST;
nodb = warnings = 1;
break;
- case ('u'):
+ case 'u':
CHECKOP(op, ch);
path_arg = optarg;
op = OP_DELETE;
break;
- case ('v'):
+ case 'v':
/* Compatibility with espie@'s makewhatis. */
break;
default:
argc -= optind;
argv += optind;
+#if HAVE_PLEDGE
+ if (nodb) {
+ if (pledge("stdio rpath", NULL) == -1) {
+ warn("pledge");
+ return (int)MANDOCLEVEL_SYSERR;
+ }
+ }
+#endif
+
if (OP_CONFFILE == op && argc > 0) {
- fprintf(stderr, "-C: Too many arguments\n");
+ warnx("-C: Too many arguments");
goto usage;
}
exitcode = (int)MANDOCLEVEL_OK;
- mp = mparse_alloc(mparse_options, MANDOCLEVEL_FATAL, NULL, NULL);
- mc = mchars_alloc();
-
- ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&mlinks, 6, &mlinks_info);
+ mchars_alloc();
+ mp = mparse_alloc(mparse_options, MANDOCLEVEL_BADARG, NULL, NULL);
+ mandoc_ohash_init(&mpages, 6, offsetof(struct mpage, inodev));
+ mandoc_ohash_init(&mlinks, 6, offsetof(struct mlink, file));
if (OP_UPDATE == op || OP_DELETE == op || OP_TEST == op) {
- /*
- * Force processing all files.
- */
- use_all = 1;
/*
- * All of these deal with a specific directory.
- * Jump into that directory then collect files specified
- * on the command-line.
+ * Most of these deal with a specific directory.
+ * Jump into that directory first.
*/
- if (0 == set_basedir(path_arg))
+ if (OP_TEST != op && 0 == set_basedir(path_arg, 1))
goto out;
- for (i = 0; i < argc; i++)
- filescan(argv[i]);
- if (0 == dbopen(1))
- goto out;
- if (OP_TEST != op)
- dbprune();
+
+ if (dbopen(1)) {
+ /*
+ * The existing database is usable. Process
+ * all files specified on the command-line.
+ */
+#if HAVE_PLEDGE
+ if (!nodb) {
+ if (pledge("stdio rpath wpath cpath fattr flock", NULL) == -1) {
+ warn("pledge");
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ goto out;
+ }
+ }
+#endif
+ use_all = 1;
+ for (i = 0; i < argc; i++)
+ filescan(argv[i]);
+ if (OP_TEST != op)
+ dbprune();
+ } else {
+ /*
+ * Database missing or corrupt.
+ * Recreate from scratch.
+ */
+ exitcode = (int)MANDOCLEVEL_OK;
+ op = OP_DEFAULT;
+ if (0 == treescan())
+ goto out;
+ if (0 == dbopen(0))
+ goto out;
+ }
if (OP_DELETE != op)
- mpages_merge(mc, mp);
- dbclose(1);
+ mpages_merge(mp);
+ dbclose(OP_DEFAULT == op ? 0 : 1);
} else {
/*
* If we have arguments, use them as our manpaths.
* If we don't, grok from manpath(1) or however else
- * manpath_parse() wants to do it.
+ * manconf_parse() wants to do it.
*/
if (argc > 0) {
- dirs.paths = mandoc_calloc
- (argc, sizeof(char *));
- dirs.sz = (size_t)argc;
+ conf.manpath.paths = mandoc_reallocarray(NULL,
+ argc, sizeof(char *));
+ conf.manpath.sz = (size_t)argc;
for (i = 0; i < argc; i++)
- dirs.paths[i] = mandoc_strdup(argv[i]);
+ conf.manpath.paths[i] = mandoc_strdup(argv[i]);
} else
- manpath_parse(&dirs, path_arg, NULL, NULL);
+ manconf_parse(&conf, path_arg, NULL, NULL);
- if (0 == dirs.sz) {
+ if (conf.manpath.sz == 0) {
exitcode = (int)MANDOCLEVEL_BADARG;
say("", "Empty manpath");
}
* Ignore zero-length directories and strip trailing
* slashes.
*/
- for (j = 0; j < dirs.sz; j++) {
- sz = strlen(dirs.paths[j]);
- if (sz && '/' == dirs.paths[j][sz - 1])
- dirs.paths[j][--sz] = '\0';
+ for (j = 0; j < conf.manpath.sz; j++) {
+ sz = strlen(conf.manpath.paths[j]);
+ if (sz && conf.manpath.paths[j][sz - 1] == '/')
+ conf.manpath.paths[j][--sz] = '\0';
if (0 == sz)
continue;
if (j) {
- ohash_init(&mpages, 6, &mpages_info);
- ohash_init(&mlinks, 6, &mlinks_info);
+ mandoc_ohash_init(&mpages, 6,
+ offsetof(struct mpage, inodev));
+ mandoc_ohash_init(&mlinks, 6,
+ offsetof(struct mlink, file));
}
- if (0 == set_basedir(dirs.paths[j]))
- goto out;
+ if ( ! set_basedir(conf.manpath.paths[j], argc > 0))
+ continue;
if (0 == treescan())
- goto out;
- if (0 == set_basedir(dirs.paths[j]))
- goto out;
+ continue;
if (0 == dbopen(0))
- goto out;
+ continue;
- mpages_merge(mc, mp);
- if (warnings &&
+ mpages_merge(mp);
+ if (warnings && !nodb &&
! (MPARSE_QUICK & mparse_options))
names_check();
dbclose(0);
- if (j + 1 < dirs.sz) {
+ if (j + 1 < conf.manpath.sz) {
mpages_free();
ohash_delete(&mpages);
ohash_delete(&mlinks);
}
}
out:
- set_basedir(NULL);
- manpath_free(&dirs);
- mchars_free(mc);
+ manconf_free(&conf);
mparse_free(mp);
+ mchars_free();
mpages_free();
ohash_delete(&mpages);
ohash_delete(&mlinks);
- return(exitcode);
+ return exitcode;
usage:
+ progname = getprogname();
fprintf(stderr, "usage: %s [-aDnpQ] [-C file] [-Tutf8]\n"
" %s [-aDnpQ] [-Tutf8] dir ...\n"
" %s [-DnpQ] [-Tutf8] -d dir [file ...]\n"
" %s [-Dnp] -u dir [file ...]\n"
" %s [-Q] -t file ...\n",
- progname, progname, progname,
- progname, progname);
+ progname, progname, progname, progname, progname);
- return((int)MANDOCLEVEL_BADARG);
+ return (int)MANDOCLEVEL_BADARG;
}
/*
* If use_all has been specified, grok all files.
* If not, sanitise paths to the following:
*
- * [./]man*[/<arch>]/<name>.<section>
+ * [./]man*[/<arch>]/<name>.<section>
* or
* [./]cat<section>[/<arch>]/<name>.0
*
- * TODO: accomodate for multi-language directories.
+ * TODO: accommodate for multi-language directories.
*/
static int
treescan(void)
{
+ char buf[PATH_MAX];
FTS *f;
FTSENT *ff;
struct mlink *mlink;
argv[0] = ".";
argv[1] = (char *)NULL;
- /*
- * Walk through all components under the directory, using the
- * logical descent of files.
- */
- f = fts_open((char * const *)argv, FTS_LOGICAL, NULL);
- if (NULL == f) {
+ f = fts_open((char * const *)argv,
+ FTS_PHYSICAL | FTS_NOCHDIR, NULL);
+ if (f == NULL) {
exitcode = (int)MANDOCLEVEL_SYSERR;
say("", "&fts_open");
- return(0);
+ return 0;
}
dsec = arch = NULL;
dform = FORM_NONE;
- while (NULL != (ff = fts_read(f))) {
+ while ((ff = fts_read(f)) != NULL) {
path = ff->fts_path + 2;
+ switch (ff->fts_info) {
+
+ /*
+ * Symbolic links require various sanity checks,
+ * then get handled just like regular files.
+ */
+ case FTS_SL:
+ if (realpath(path, buf) == NULL) {
+ if (warnings)
+ say(path, "&realpath");
+ continue;
+ }
+ if (strstr(buf, basedir) != buf
+#ifdef HOMEBREWDIR
+ && strstr(buf, HOMEBREWDIR) != buf
+#endif
+ ) {
+ if (warnings) say("",
+ "%s: outside base directory", buf);
+ continue;
+ }
+ /* Use logical inode to avoid mpages dupe. */
+ if (stat(path, ff->fts_statp) == -1) {
+ if (warnings)
+ say(path, "&stat");
+ continue;
+ }
+ /* FALLTHROUGH */
+
/*
* If we're a regular file, add an mlink by using the
* stored directory data and handling the filename.
*/
- if (FTS_F == ff->fts_info) {
- if (0 == strcmp(path, MANDOC_DB))
+ case FTS_F:
+ if ( ! strcmp(path, MANDOC_DB))
continue;
if ( ! use_all && ff->fts_level < 2) {
if (warnings)
}
gzip = 0;
fsec = NULL;
- while (NULL == fsec) {
+ while (fsec == NULL) {
fsec = strrchr(ff->fts_name, '.');
- if (NULL == fsec || strcmp(fsec+1, "gz"))
+ if (fsec == NULL || strcmp(fsec+1, "gz"))
break;
gzip = 1;
*fsec = '\0';
fsec = NULL;
}
- if (NULL == fsec) {
+ if (fsec == NULL) {
if ( ! use_all) {
if (warnings)
say(path,
"No filename suffix");
continue;
}
- } else if (0 == strcmp(++fsec, "html")) {
+ } else if ( ! strcmp(++fsec, "html")) {
if (warnings)
say(path, "Skip html");
continue;
- } else if (0 == strcmp(fsec, "ps")) {
+ } else if ( ! strcmp(fsec, "ps")) {
if (warnings)
say(path, "Skip ps");
continue;
- } else if (0 == strcmp(fsec, "pdf")) {
+ } else if ( ! strcmp(fsec, "pdf")) {
if (warnings)
say(path, "Skip pdf");
continue;
} else if ( ! use_all &&
- ((FORM_SRC == dform && strcmp(fsec, dsec)) ||
- (FORM_CAT == dform && strcmp(fsec, "0")))) {
+ ((dform == FORM_SRC &&
+ strncmp(fsec, dsec, strlen(dsec))) ||
+ (dform == FORM_CAT && strcmp(fsec, "0")))) {
if (warnings)
say(path, "Wrong filename suffix");
continue;
fsec[-1] = '\0';
mlink = mandoc_calloc(1, sizeof(struct mlink));
- strlcpy(mlink->file, path, sizeof(mlink->file));
+ if (strlcpy(mlink->file, path,
+ sizeof(mlink->file)) >=
+ sizeof(mlink->file)) {
+ say(path, "Filename too long");
+ free(mlink);
+ continue;
+ }
mlink->dform = dform;
mlink->dsec = dsec;
mlink->arch = arch;
mlink->gzip = gzip;
mlink_add(mlink, ff->fts_statp);
continue;
- } else if (FTS_D != ff->fts_info &&
- FTS_DP != ff->fts_info) {
+
+ case FTS_D:
+ case FTS_DP:
+ break;
+
+ default:
if (warnings)
say(path, "Not a regular file");
continue;
}
switch (ff->fts_level) {
- case (0):
+ case 0:
/* Ignore the root directory. */
break;
- case (1):
+ case 1:
/*
* This might contain manX/ or catX/.
* Try to infer this from the name.
* If we're not in use_all, enforce it.
*/
cp = ff->fts_name;
- if (FTS_DP == ff->fts_info)
+ if (ff->fts_info == FTS_DP) {
+ dform = FORM_NONE;
+ dsec = NULL;
break;
+ }
- if (0 == strncmp(cp, "man", 3)) {
+ if ( ! strncmp(cp, "man", 3)) {
dform = FORM_SRC;
dsec = cp + 3;
- } else if (0 == strncmp(cp, "cat", 3)) {
+ } else if ( ! strncmp(cp, "cat", 3)) {
dform = FORM_CAT;
dsec = cp + 3;
} else {
dsec = NULL;
}
- if (NULL != dsec || use_all)
+ if (dsec != NULL || use_all)
break;
if (warnings)
say(path, "Unknown directory part");
fts_set(f, ff, FTS_SKIP);
break;
- case (2):
+ case 2:
/*
* Possibly our architecture.
* If we're descending, keep tabs on it.
*/
- if (FTS_DP != ff->fts_info && NULL != dsec)
+ if (ff->fts_info != FTS_DP && dsec != NULL)
arch = ff->fts_name;
else
arch = NULL;
break;
default:
- if (FTS_DP == ff->fts_info || use_all)
+ if (ff->fts_info == FTS_DP || use_all)
break;
if (warnings)
say(path, "Extraneous directory part");
}
fts_close(f);
- return(1);
+ return 1;
}
/*
* Try to infer the manual section, architecture, and page name from the
* path, assuming it looks like
*
- * [./]man*[/<arch>]/<name>.<section>
+ * [./]man*[/<arch>]/<name>.<section>
* or
* [./]cat<section>[/<arch>]/<name>.0
*
if (0 == strncmp(file, "./", 2))
file += 2;
+ /*
+ * We have to do lstat(2) before realpath(3) loses
+ * the information whether this is a symbolic link.
+ * We need to know that because for symbolic links,
+ * we want to use the orginal file name, while for
+ * regular files, we want to use the real path.
+ */
+ if (-1 == lstat(file, &st)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, "&lstat");
+ return;
+ } else if (0 == ((S_IFREG | S_IFLNK) & st.st_mode)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, "Not a regular file");
+ return;
+ }
+
+ /*
+ * We have to resolve the file name to the real path
+ * in any case for the base directory check.
+ */
if (NULL == realpath(file, buf)) {
exitcode = (int)MANDOCLEVEL_BADARG;
say(file, "&realpath");
return;
}
- if (strstr(buf, basedir) == buf)
- start = buf + strlen(basedir) + 1;
- else if (OP_TEST == op)
+ if (OP_TEST == op)
+ start = buf;
+ else if (strstr(buf, basedir) == buf)
+ start = buf + strlen(basedir);
+#ifdef HOMEBREWDIR
+ else if (strstr(buf, HOMEBREWDIR) == buf)
start = buf;
+#endif
else {
exitcode = (int)MANDOCLEVEL_BADARG;
say("", "%s: outside base directory", buf);
return;
}
- if (-1 == stat(buf, &st)) {
- exitcode = (int)MANDOCLEVEL_BADARG;
- say(file, "&stat");
- return;
- } else if ( ! (S_IFREG & st.st_mode)) {
- exitcode = (int)MANDOCLEVEL_BADARG;
- say(file, "Not a regular file");
- return;
+ /*
+ * Now we are sure the file is inside our tree.
+ * If it is a symbolic link, ignore the real path
+ * and use the original name.
+ * This implies passing stuff like "cat1/../man1/foo.1"
+ * on the command line won't work. So don't do that.
+ * Note the stat(2) can still fail if the link target
+ * doesn't exist.
+ */
+ if (S_IFLNK & st.st_mode) {
+ if (-1 == stat(buf, &st)) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say(file, "&stat");
+ return;
+ }
+ if (strlcpy(buf, file, sizeof(buf)) >= sizeof(buf)) {
+ say(file, "Filename too long");
+ return;
+ }
+ start = buf;
+ if (OP_TEST != op && strstr(buf, basedir) == buf)
+ start += strlen(basedir);
}
mlink = mandoc_calloc(1, sizeof(struct mlink));
- strlcpy(mlink->file, start, sizeof(mlink->file));
+ mlink->dform = FORM_NONE;
+ if (strlcpy(mlink->file, start, sizeof(mlink->file)) >=
+ sizeof(mlink->file)) {
+ say(start, "Filename too long");
+ free(mlink);
+ return;
+ }
/*
* First try to guess our directory structure.
assert(NULL == ohash_find(&mlinks, slot));
ohash_insert(&mlinks, slot, mlink);
+ memset(&inodev, 0, sizeof(inodev)); /* Clear padding. */
inodev.st_ino = st->st_ino;
inodev.st_dev = st->st_dev;
slot = ohash_lookup_memory(&mpages, (char *)&inodev,
mpage->form = FORM_NONE;
goto nextlink;
}
- if (strlcpy(buf, mlink->file, PATH_MAX) >= PATH_MAX) {
- if (warnings)
- say(mlink->file, "Filename too long");
- goto nextlink;
- }
+ (void)strlcpy(buf, mlink->file, sizeof(buf));
bufp = strstr(buf, "cat");
assert(NULL != bufp);
memcpy(bufp, "man", 3);
if (NULL != (bufp = strrchr(buf, '.')))
*++bufp = '\0';
- strlcat(buf, mlink->dsec, PATH_MAX);
+ (void)strlcat(buf, mlink->dsec, sizeof(buf));
if (NULL == ohash_find(&mlinks,
- ohash_qlookup(&mlinks, buf)))
+ ohash_qlookup(&mlinks, buf)))
goto nextlink;
if (warnings)
say(mlink->file, "Man source exists: %s", buf);
/*
* XXX
- * parse_cat() doesn't set TYPE_Nm and TYPE_NAME yet.
+ * parse_cat() doesn't set NAME_TITLE yet.
*/
if (FORM_CAT == mpage->form)
* appears as a name in the NAME section.
*/
- slot = ohash_qlookup(&strings, mlink->name);
- str = ohash_find(&strings, slot);
+ slot = ohash_qlookup(&names, mlink->name);
+ str = ohash_find(&names, slot);
assert(NULL != str);
- if ( ! (TYPE_NAME & str->mask))
+ if ( ! (NAME_TITLE & str->mask))
say(mlink->file, "Name missing in NAME section");
}
* and filename to determine whether the file is parsable or not.
*/
static void
-mpages_merge(struct mchars *mc, struct mparse *mp)
+mpages_merge(struct mparse *mp)
{
char any[] = "any";
- struct ohash_info str_info;
- int fd[2];
struct mpage *mpage, *mpage_dest;
struct mlink *mlink, *mlink_dest;
- struct mdoc *mdoc;
- struct man *man;
+ struct roff_man *man;
char *sodest;
char *cp;
- pid_t child_pid;
- int status;
+ int fd;
unsigned int pslot;
- enum mandoclevel lvl;
- str_info.alloc = hash_alloc;
- str_info.halloc = hash_halloc;
- str_info.hfree = hash_free;
- str_info.key_offset = offsetof(struct str, key);
-
- if (0 == nodb)
+ if ( ! nodb)
SQL_EXEC("BEGIN TRANSACTION");
mpage = ohash_first(&mpages, &pslot);
- while (NULL != mpage) {
+ while (mpage != NULL) {
mlinks_undupe(mpage);
- if (NULL == mpage->mlinks) {
+ if ((mlink = mpage->mlinks) == NULL) {
mpage = ohash_next(&mpages, &pslot);
continue;
}
- ohash_init(&strings, 6, &str_info);
+ name_mask = NAME_MASK;
+ mandoc_ohash_init(&names, 4, offsetof(struct str, key));
+ mandoc_ohash_init(&strings, 6, offsetof(struct str, key));
mparse_reset(mp);
- mdoc = NULL;
man = NULL;
sodest = NULL;
- child_pid = 0;
- fd[0] = -1;
- fd[1] = -1;
-
- if (mpage->mlinks->gzip) {
- if (-1 == pipe(fd)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say(mpage->mlinks->file, "&pipe gunzip");
- goto nextpage;
- }
- switch (child_pid = fork()) {
- case (-1):
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say(mpage->mlinks->file, "&fork gunzip");
- child_pid = 0;
- close(fd[1]);
- close(fd[0]);
- goto nextpage;
- case (0):
- close(fd[0]);
- if (-1 == dup2(fd[1], STDOUT_FILENO)) {
- say(mpage->mlinks->file,
- "&dup gunzip");
- exit(1);
- }
- execlp("gunzip", "gunzip", "-c",
- mpage->mlinks->file, NULL);
- say(mpage->mlinks->file, "&exec gunzip");
- exit(1);
- default:
- close(fd[1]);
- break;
- }
+
+ if ((fd = mparse_open(mp, mlink->file)) == -1) {
+ say(mlink->file, "&open");
+ goto nextpage;
}
/*
- * Try interpreting the file as mdoc(7) or man(7)
- * source code, unless it is already known to be
- * formatted. Fall back to formatted mode.
+ * Interpret the file as mdoc(7) or man(7) source
+ * code, unless it is known to be formatted.
*/
- if (FORM_CAT != mpage->mlinks->dform ||
- FORM_CAT != mpage->mlinks->fform) {
- lvl = mparse_readfd(mp, fd[0], mpage->mlinks->file);
- if (lvl < MANDOCLEVEL_FATAL)
- mparse_result(mp, &mdoc, &man, &sodest);
+ if (mlink->dform != FORM_CAT || mlink->fform != FORM_CAT) {
+ mparse_readfd(mp, fd, mlink->file);
+ close(fd);
+ mparse_result(mp, &man, &sodest);
}
- if (NULL != sodest) {
+ if (sodest != NULL) {
mlink_dest = ohash_find(&mlinks,
ohash_qlookup(&mlinks, sodest));
- if (NULL != mlink_dest) {
+ if (mlink_dest == NULL) {
+ mandoc_asprintf(&cp, "%s.gz", sodest);
+ mlink_dest = ohash_find(&mlinks,
+ ohash_qlookup(&mlinks, cp));
+ free(cp);
+ }
+ if (mlink_dest != NULL) {
/* The .so target exists. */
mpage_dest = mlink_dest->mpage;
- mlink = mpage->mlinks;
while (1) {
mlink->mpage = mpage_dest;
* to the target.
*/
- if (mpage_dest->recno)
- dbadd_mlink(mlink);
+ if (mpage_dest->pageid)
+ dbadd_mlink_name(mlink);
- if (NULL == mlink->next)
+ if (mlink->next == NULL)
break;
mlink = mlink->next;
}
mpage->mlinks = NULL;
}
goto nextpage;
- } else if (NULL != mdoc) {
+ } else if (man != NULL && man->macroset == MACROSET_MDOC) {
+ mdoc_validate(man);
mpage->form = FORM_SRC;
- mpage->sec =
- mandoc_strdup(mdoc_meta(mdoc)->msec);
- mpage->arch = mdoc_meta(mdoc)->arch;
+ mpage->sec = man->meta.msec;
+ mpage->sec = mandoc_strdup(
+ mpage->sec == NULL ? "" : mpage->sec);
+ mpage->arch = man->meta.arch;
mpage->arch = mandoc_strdup(
- NULL == mpage->arch ? "" : mpage->arch);
- mpage->title =
- mandoc_strdup(mdoc_meta(mdoc)->title);
- } else if (NULL != man) {
+ mpage->arch == NULL ? "" : mpage->arch);
+ mpage->title = mandoc_strdup(man->meta.title);
+ } else if (man != NULL && man->macroset == MACROSET_MAN) {
+ man_validate(man);
mpage->form = FORM_SRC;
- mpage->sec =
- mandoc_strdup(man_meta(man)->msec);
- mpage->arch =
- mandoc_strdup(mpage->mlinks->arch);
- mpage->title =
- mandoc_strdup(man_meta(man)->title);
+ mpage->sec = mandoc_strdup(man->meta.msec);
+ mpage->arch = mandoc_strdup(mlink->arch);
+ mpage->title = mandoc_strdup(man->meta.title);
} else {
mpage->form = FORM_CAT;
- mpage->sec =
- mandoc_strdup(mpage->mlinks->dsec);
- mpage->arch =
- mandoc_strdup(mpage->mlinks->arch);
- mpage->title =
- mandoc_strdup(mpage->mlinks->name);
+ mpage->sec = mandoc_strdup(mlink->dsec);
+ mpage->arch = mandoc_strdup(mlink->arch);
+ mpage->title = mandoc_strdup(mlink->name);
}
putkey(mpage, mpage->sec, TYPE_sec);
- putkey(mpage, '\0' == *mpage->arch ?
- any : mpage->arch, TYPE_arch);
+ if (*mpage->arch != '\0')
+ putkey(mpage, mpage->arch, TYPE_arch);
- for (mlink = mpage->mlinks; mlink; mlink = mlink->next) {
+ for ( ; mlink != NULL; mlink = mlink->next) {
if ('\0' != *mlink->dsec)
putkey(mpage, mlink->dsec, TYPE_sec);
if ('\0' != *mlink->fsec)
putkey(mpage, mlink->fsec, TYPE_sec);
putkey(mpage, '\0' == *mlink->arch ?
any : mlink->arch, TYPE_arch);
- putkey(mpage, mlink->name, TYPE_Nm);
+ putkey(mpage, mlink->name, NAME_FILE);
}
- if (NULL != mdoc) {
- if (NULL != (cp = mdoc_meta(mdoc)->name))
- putkey(mpage, cp, TYPE_Nm);
- assert(NULL == mpage->desc);
- parse_mdoc(mpage, mdoc_node(mdoc));
- putkey(mpage, NULL != mpage->desc ?
- mpage->desc : mpage->mlinks->name, TYPE_Nd);
- } else if (NULL != man)
- parse_man(mpage, man_node(man));
+ assert(mpage->desc == NULL);
+ if (man != NULL && man->macroset == MACROSET_MDOC)
+ parse_mdoc(mpage, &man->meta, man->first);
+ else if (man != NULL)
+ parse_man(mpage, &man->meta, man->first);
else
- parse_cat(mpage, fd[0]);
+ parse_cat(mpage, fd);
+ if (mpage->desc == NULL)
+ mpage->desc = mandoc_strdup(mpage->mlinks->name);
if (warnings && !use_all)
for (mlink = mpage->mlinks; mlink;
mlink = mlink->next)
mlink_check(mpage, mlink);
- dbadd(mpage, mc);
+ dbadd(mpage);
+ mlink = mpage->mlinks;
nextpage:
- if (child_pid) {
- if (-1 == waitpid(child_pid, &status, 0)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say(mpage->mlinks->file, "&wait gunzip");
- } else if (WIFSIGNALED(status)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say(mpage->mlinks->file,
- "gunzip died from signal %d",
- WTERMSIG(status));
- } else if (WEXITSTATUS(status)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say(mpage->mlinks->file,
- "gunzip failed with code %d",
- WEXITSTATUS(status));
- }
- }
ohash_delete(&strings);
+ ohash_delete(&names);
mpage = ohash_next(&mpages, &pslot);
}
{
sqlite3_stmt *stmt;
const char *name, *sec, *arch, *key;
- size_t i;
- int irc;
sqlite3_prepare_v2(db,
"SELECT name, sec, arch, key FROM ("
- "SELECT key, pageid FROM keys "
+ "SELECT name AS key, pageid FROM names "
"WHERE bits & ? AND NOT EXISTS ("
"SELECT pageid FROM mlinks "
- "WHERE mlinks.pageid == keys.pageid "
- "AND mlinks.name == keys.key"
+ "WHERE mlinks.pageid == names.pageid "
+ "AND mlinks.name == names.name"
")"
") JOIN ("
- "SELECT * FROM mlinks GROUP BY pageid"
+ "SELECT sec, arch, name, pageid FROM mlinks "
+ "GROUP BY pageid"
") USING (pageid);",
-1, &stmt, NULL);
- i = 1;
- SQL_BIND_INT64(stmt, i, TYPE_NAME);
+ if (sqlite3_bind_int64(stmt, 1, NAME_TITLE) != SQLITE_OK)
+ say("", "%s", sqlite3_errmsg(db));
- while (SQLITE_ROW == (irc = sqlite3_step(stmt))) {
- name = sqlite3_column_text(stmt, 0);
- sec = sqlite3_column_text(stmt, 1);
- arch = sqlite3_column_text(stmt, 2);
- key = sqlite3_column_text(stmt, 3);
+ while (sqlite3_step(stmt) == SQLITE_ROW) {
+ name = (const char *)sqlite3_column_text(stmt, 0);
+ sec = (const char *)sqlite3_column_text(stmt, 1);
+ arch = (const char *)sqlite3_column_text(stmt, 2);
+ key = (const char *)sqlite3_column_text(stmt, 3);
say("", "%s(%s%s%s) lacks mlink \"%s\"", name, sec,
'\0' == *arch ? "" : "/",
'\0' == *arch ? "" : arch, key);
{
FILE *stream;
char *line, *p, *title;
- size_t len, plen, titlesz;
+ size_t linesz, plen, titlesz;
+ ssize_t len;
+ int offs;
stream = (-1 == fd) ?
fopen(mpage->mlinks->file, "r") :
fdopen(fd, "r");
if (NULL == stream) {
+ if (-1 != fd)
+ close(fd);
if (warnings)
say(mpage->mlinks->file, "&fopen");
return;
}
+ line = NULL;
+ linesz = 0;
+
/* Skip to first blank line. */
- while (NULL != (line = fgetln(stream, &len)))
- if ('\n' == *line)
+ while (getline(&line, &linesz, stream) != -1)
+ if (*line == '\n')
break;
/*
* is the first section header. Skip to it.
*/
- while (NULL != (line = fgetln(stream, &len)))
- if ('\n' != *line && ' ' != *line)
+ while (getline(&line, &linesz, stream) != -1)
+ if (*line != '\n' && *line != ' ')
break;
-
+
/*
* Read up until the next section into a buffer.
* Strip the leading and trailing newline from each read line,
titlesz = 0;
title = NULL;
- while (NULL != (line = fgetln(stream, &len))) {
- if (' ' != *line || '\n' != line[len - 1])
+ while ((len = getline(&line, &linesz, stream)) != -1) {
+ if (*line != ' ')
break;
- while (len > 0 && isspace((unsigned char)*line)) {
- line++;
- len--;
- }
- if (1 == len)
+ offs = 0;
+ while (isspace((unsigned char)line[offs]))
+ offs++;
+ if (line[offs] == '\0')
continue;
- title = mandoc_realloc(title, titlesz + len);
- memcpy(title + titlesz, line, len);
- titlesz += len;
+ title = mandoc_realloc(title, titlesz + len - offs);
+ memcpy(title + titlesz, line + offs, len - offs);
+ titlesz += len - offs;
title[titlesz - 1] = ' ';
}
+ free(line);
/*
* If no page content can be found, or the input line
if (warnings)
say(mpage->mlinks->file,
"Cannot find NAME section");
- assert(NULL == mpage->desc);
- mpage->desc = mandoc_strdup(mpage->mlinks->name);
- putkey(mpage, mpage->mlinks->name, TYPE_Nd);
fclose(stream);
free(title);
return;
}
- title = mandoc_realloc(title, titlesz + 1);
- title[titlesz] = '\0';
+ title[titlesz - 1] = '\0';
/*
* Skip to the first dash.
if (0 == len) {
memmove(line, line + 1, plen--);
continue;
- }
+ }
memmove(line - 1, line + 1, plen - len);
plen -= 2;
}
- assert(NULL == mpage->desc);
mpage->desc = mandoc_strdup(p);
- putkey(mpage, mpage->desc, TYPE_Nd);
fclose(stream);
free(title);
}
*/
static void
putmdockey(const struct mpage *mpage,
- const struct mdoc_node *n, uint64_t m)
+ const struct roff_node *n, uint64_t m)
{
for ( ; NULL != n; n = n->next) {
if (NULL != n->child)
putmdockey(mpage, n->child, m);
- if (MDOC_TEXT == n->type)
+ if (n->type == ROFFT_TEXT)
putkey(mpage, n->string, m);
}
}
static void
-parse_man(struct mpage *mpage, const struct man_node *n)
+parse_man(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
- const struct man_node *head, *body;
+ const struct roff_node *head, *body;
char *start, *title;
char byte;
size_t sz;
- if (NULL == n)
+ if (n == NULL)
return;
/*
* the correct section or not.
*/
- if (MAN_BODY == n->type && MAN_SH == n->tok) {
+ if (n->type == ROFFT_BODY && n->tok == MAN_SH) {
body = n;
- assert(body->parent);
- if (NULL != (head = body->parent->head) &&
- 1 == head->nchild &&
- NULL != (head = (head->child)) &&
- MAN_TEXT == head->type &&
- 0 == strcmp(head->string, "NAME") &&
- NULL != body->child) {
+ if ((head = body->parent->head) != NULL &&
+ (head = head->child) != NULL &&
+ head->next == NULL &&
+ head->type == ROFFT_TEXT &&
+ strcmp(head->string, "NAME") == 0 &&
+ body->child != NULL) {
/*
* Suck the entire NAME section into memory.
*/
title = NULL;
- man_deroff(&title, body);
+ deroff(&title, body);
if (NULL == title)
return;
- /*
+ /*
* Go through a special heuristic dance here.
* Conventionally, one or more manual names are
* comma-specified prior to a whitespace, then a
('\\' == start[0] && '-' == start[1]))
break;
- putkey(mpage, start, TYPE_NAME | TYPE_Nm);
+ putkey(mpage, start, NAME_TITLE);
+ if ( ! (mpage->name_head_done ||
+ strcasecmp(start, meta->title))) {
+ putkey(mpage, start, NAME_HEAD);
+ mpage->name_head_done = 1;
+ }
if (' ' == byte) {
start += sz + 1;
}
if (start == title) {
- putkey(mpage, start, TYPE_NAME | TYPE_Nm);
+ putkey(mpage, start, NAME_TITLE);
+ if ( ! (mpage->name_head_done ||
+ strcasecmp(start, meta->title))) {
+ putkey(mpage, start, NAME_HEAD);
+ mpage->name_head_done = 1;
+ }
free(title);
return;
}
while (' ' == *start)
start++;
- assert(NULL == mpage->desc);
mpage->desc = mandoc_strdup(start);
- putkey(mpage, mpage->desc, TYPE_Nd);
free(title);
return;
}
for (n = n->child; n; n = n->next) {
if (NULL != mpage->desc)
break;
- parse_man(mpage, n);
+ parse_man(mpage, meta, n);
}
}
static void
-parse_mdoc(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
assert(NULL != n);
for (n = n->child; NULL != n; n = n->next) {
switch (n->type) {
- case (MDOC_ELEM):
- /* FALLTHROUGH */
- case (MDOC_BLOCK):
- /* FALLTHROUGH */
- case (MDOC_HEAD):
- /* FALLTHROUGH */
- case (MDOC_BODY):
- /* FALLTHROUGH */
- case (MDOC_TAIL):
+ case ROFFT_ELEM:
+ case ROFFT_BLOCK:
+ case ROFFT_HEAD:
+ case ROFFT_BODY:
+ case ROFFT_TAIL:
if (NULL != mdocs[n->tok].fp)
- if (0 == (*mdocs[n->tok].fp)(mpage, n))
+ if (0 == (*mdocs[n->tok].fp)(mpage, meta, n))
break;
if (mdocs[n->tok].mask)
putmdockey(mpage, n->child,
mdocs[n->tok].mask);
break;
default:
- assert(MDOC_ROOT != n->type);
+ assert(n->type != ROFFT_ROOT);
continue;
}
if (NULL != n->child)
- parse_mdoc(mpage, n);
+ parse_mdoc(mpage, meta, n);
}
}
static int
-parse_mdoc_Fd(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_Fd(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
- const char *start, *end;
+ char *start, *end;
size_t sz;
if (SEC_SYNOPSIS != n->sec ||
- NULL == (n = n->child) ||
- MDOC_TEXT != n->type)
- return(0);
+ NULL == (n = n->child) ||
+ n->type != ROFFT_TEXT)
+ return 0;
/*
* Only consider those `Fd' macro fields that begin with an
*/
if (strcmp("#include", n->string))
- return(0);
+ return 0;
- if (NULL == (n = n->next) || MDOC_TEXT != n->type)
- return(0);
+ if ((n = n->next) == NULL || n->type != ROFFT_TEXT)
+ return 0;
/*
* Strip away the enclosing angle brackets and make sure we're
start++;
if (0 == (sz = strlen(start)))
- return(0);
+ return 0;
end = &start[(int)sz - 1];
if ('>' == *end || '"' == *end)
if (end > start)
putkeys(mpage, start, end - start + 1, TYPE_In);
- return(0);
+ return 0;
}
-static int
-parse_mdoc_Fn(struct mpage *mpage, const struct mdoc_node *n)
+static void
+parse_mdoc_fname(struct mpage *mpage, const struct roff_node *n)
{
char *cp;
+ size_t sz;
- if (NULL == (n = n->child) || MDOC_TEXT != n->type)
- return(0);
-
- /*
- * Parse: .Fn "struct type *name" "char *arg".
- * First strip away pointer symbol.
- * Then store the function name, then type.
- * Finally, store the arguments.
- */
+ if (n->type != ROFFT_TEXT)
+ return;
- if (NULL == (cp = strrchr(n->string, ' ')))
- cp = n->string;
+ /* Skip function pointer punctuation. */
- while ('*' == *cp)
+ cp = n->string;
+ while (*cp == '(' || *cp == '*')
cp++;
+ sz = strcspn(cp, "()");
+
+ putkeys(mpage, cp, sz, TYPE_Fn);
+ if (n->sec == SEC_SYNOPSIS)
+ putkeys(mpage, cp, sz, NAME_SYN);
+}
- putkey(mpage, cp, TYPE_Fn);
+static int
+parse_mdoc_Fn(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
+{
- if (n->string < cp)
- putkeys(mpage, n->string, cp - n->string, TYPE_Ft);
+ if (n->child == NULL)
+ return 0;
- for (n = n->next; NULL != n; n = n->next)
- if (MDOC_TEXT == n->type)
+ parse_mdoc_fname(mpage, n->child);
+
+ for (n = n->child->next; n != NULL; n = n->next)
+ if (n->type == ROFFT_TEXT)
putkey(mpage, n->string, TYPE_Fa);
- return(0);
+ return 0;
}
static int
-parse_mdoc_Xr(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_Fo(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
+{
+
+ if (n->type != ROFFT_HEAD)
+ return 1;
+
+ if (n->child != NULL)
+ parse_mdoc_fname(mpage, n->child);
+
+ return 0;
+}
+
+static int
+parse_mdoc_Va(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
+{
+ char *cp;
+
+ if (n->type != ROFFT_ELEM && n->type != ROFFT_BODY)
+ return 0;
+
+ if (n->child != NULL &&
+ n->child->next == NULL &&
+ n->child->type == ROFFT_TEXT)
+ return 1;
+
+ cp = NULL;
+ deroff(&cp, n);
+ if (cp != NULL) {
+ putkey(mpage, cp, TYPE_Vt | (n->tok == MDOC_Va ||
+ n->type == ROFFT_BODY ? TYPE_Va : 0));
+ free(cp);
+ }
+
+ return 0;
+}
+
+static int
+parse_mdoc_Xr(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
char *cp;
if (NULL == (n = n->child))
- return(0);
+ return 0;
if (NULL == n->next) {
putkey(mpage, n->string, TYPE_Xr);
- return(0);
+ return 0;
}
mandoc_asprintf(&cp, "%s(%s)", n->string, n->next->string);
putkey(mpage, cp, TYPE_Xr);
free(cp);
- return(0);
+ return 0;
}
static int
-parse_mdoc_Nd(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_Nd(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
- if (MDOC_BODY == n->type)
- mdoc_deroff(&mpage->desc, n);
- return(0);
+ if (n->type == ROFFT_BODY)
+ deroff(&mpage->desc, n);
+ return 0;
}
static int
-parse_mdoc_Nm(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_Nm(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
if (SEC_NAME == n->sec)
- putmdockey(mpage, n->child, TYPE_NAME | TYPE_Nm);
- else if (SEC_SYNOPSIS == n->sec && MDOC_HEAD == n->type)
- putmdockey(mpage, n->child, TYPE_Nm);
- return(0);
-}
-
-static int
-parse_mdoc_Sh(struct mpage *mpage, const struct mdoc_node *n)
-{
-
- return(SEC_CUSTOM == n->sec && MDOC_HEAD == n->type);
+ putmdockey(mpage, n->child, NAME_TITLE);
+ else if (n->sec == SEC_SYNOPSIS && n->type == ROFFT_HEAD) {
+ if (n->child == NULL)
+ putkey(mpage, meta->name, NAME_SYN);
+ else
+ putmdockey(mpage, n->child, NAME_SYN);
+ }
+ if ( ! (mpage->name_head_done ||
+ n->child == NULL || n->child->string == NULL ||
+ strcasecmp(n->child->string, meta->title))) {
+ putkey(mpage, n->child->string, ROFFT_HEAD);
+ mpage->name_head_done = 1;
+ }
+ return 0;
}
static int
-parse_mdoc_head(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_Sh(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
- return(MDOC_HEAD == n->type);
+ return n->sec == SEC_CUSTOM && n->type == ROFFT_HEAD;
}
static int
-parse_mdoc_body(struct mpage *mpage, const struct mdoc_node *n)
+parse_mdoc_head(struct mpage *mpage, const struct roff_meta *meta,
+ const struct roff_node *n)
{
- return(MDOC_BODY == n->type);
+ return n->type == ROFFT_HEAD;
}
/*
* When we finish the manual, we'll dump the table.
*/
static void
-putkeys(const struct mpage *mpage,
- const char *cp, size_t sz, uint64_t v)
+putkeys(const struct mpage *mpage, char *cp, size_t sz, uint64_t v)
{
+ struct ohash *htab;
struct str *s;
const char *end;
- uint64_t mask;
unsigned int slot;
- int i;
+ int i, mustfree;
if (0 == sz)
return;
- if (debug > 1) {
- for (i = 0, mask = 1;
- i < mansearch_keymax;
- i++, mask <<= 1)
- if (mask & v)
- break;
- say(mpage->mlinks->file, "Adding key %s=%*s",
- mansearch_keynames[i], sz, cp);
+ mustfree = render_string(&cp, &sz);
+
+ if (TYPE_Nm & v) {
+ htab = &names;
+ v &= name_mask;
+ if (v & NAME_FIRST)
+ name_mask &= ~NAME_FIRST;
+ if (debug > 1)
+ say(mpage->mlinks->file,
+ "Adding name %*s, bits=%d", sz, cp, v);
+ } else {
+ htab = &strings;
+ if (debug > 1)
+ for (i = 0; i < mansearch_keymax; i++)
+ if ((uint64_t)1 << i & v)
+ say(mpage->mlinks->file,
+ "Adding key %s=%*s",
+ mansearch_keynames[i], sz, cp);
}
end = cp + sz;
- slot = ohash_qlookupi(&strings, cp, &end);
- s = ohash_find(&strings, slot);
+ slot = ohash_qlookupi(htab, cp, &end);
+ s = ohash_find(htab, slot);
if (NULL != s && mpage == s->mpage) {
s->mask |= v;
return;
} else if (NULL == s) {
- s = mandoc_calloc(sizeof(struct str) + sz + 1, 1);
+ s = mandoc_calloc(1, sizeof(struct str) + sz + 1);
memcpy(s->key, cp, sz);
- ohash_insert(&strings, slot, s);
+ ohash_insert(htab, slot, s);
}
s->mpage = mpage;
s->mask = v;
+
+ if (mustfree)
+ free(cp);
}
/*
out[4] = (cp >> 6 & 63) | 128;
out[5] = (cp & 63) | 128;
} else
- return(0);
+ return 0;
out[rc] = '\0';
- return(rc);
+ return rc;
}
/*
- * Store the rendered version of a key, or alias the pointer
- * if the key contains no escape sequences.
+ * If the string contains escape sequences,
+ * replace it with an allocated rendering and return 1,
+ * such that the caller can free it after use.
+ * Otherwise, do nothing and return 0.
*/
-static void
-render_key(struct mchars *mc, struct str *key)
+static int
+render_string(char **public, size_t *psz)
{
- size_t sz, bsz, pos;
+ const char *src, *scp, *addcp, *seq;
+ char *dst;
+ size_t ssz, dsz, addsz;
char utfbuf[7], res[6];
- char *buf;
- const char *seq, *cpp, *val;
- int len, u;
- enum mandoc_esc esc;
-
- assert(NULL == key->rendered);
+ int seqlen, unicode;
res[0] = '\\';
res[1] = '\t';
res[4] = ASCII_BREAK;
res[5] = '\0';
- val = key->key;
- bsz = strlen(val);
+ src = scp = *public;
+ ssz = *psz;
+ dst = NULL;
+ dsz = 0;
- /*
- * Pre-check: if we have no stop-characters, then set the
- * pointer as ourselvse and get out of here.
- */
- if (strcspn(val, res) == bsz) {
- key->rendered = key->key;
- return;
- }
+ while (scp < src + *psz) {
- /* Pre-allocate by the length of the input */
+ /* Leave normal characters unchanged. */
- buf = mandoc_malloc(++bsz);
- pos = 0;
+ if (strchr(res, *scp) == NULL) {
+ if (dst != NULL)
+ dst[dsz++] = *scp;
+ scp++;
+ continue;
+ }
- while ('\0' != *val) {
/*
- * Halt on the first escape sequence.
- * This also halts on the end of string, in which case
- * we just copy, fallthrough, and exit the loop.
+ * Found something that requires replacing,
+ * make sure we have a destination buffer.
*/
- if ((sz = strcspn(val, res)) > 0) {
- memcpy(&buf[pos], val, sz);
- pos += sz;
- val += sz;
+
+ if (dst == NULL) {
+ dst = mandoc_malloc(ssz + 1);
+ dsz = scp - src;
+ memcpy(dst, src, dsz);
}
- switch (*val) {
- case (ASCII_HYPH):
- buf[pos++] = '-';
- val++;
+ /* Handle single-char special characters. */
+
+ switch (*scp) {
+ case '\\':
+ break;
+ case '\t':
+ case ASCII_NBRSP:
+ dst[dsz++] = ' ';
+ scp++;
continue;
- case ('\t'):
- /* FALLTHROUGH */
- case (ASCII_NBRSP):
- buf[pos++] = ' ';
- val++;
+ case ASCII_HYPH:
+ dst[dsz++] = '-';
/* FALLTHROUGH */
- case (ASCII_BREAK):
+ case ASCII_BREAK:
+ scp++;
continue;
default:
- break;
+ abort();
}
- if ('\\' != *val)
- break;
-
- /* Read past the slash. */
-
- val++;
/*
- * Parse the escape sequence and see if it's a
- * predefined character or special character.
+ * Found an escape sequence.
+ * Read past the slash, then parse it.
+ * Ignore everything except characters.
*/
- esc = mandoc_escape
- ((const char **)&val, &seq, &len);
- if (ESCAPE_ERROR == esc)
- break;
- if (ESCAPE_SPECIAL != esc)
+ scp++;
+ if (mandoc_escape(&scp, &seq, &seqlen) != ESCAPE_SPECIAL)
continue;
/*
*/
if (write_utf8) {
- if (0 == (u = mchars_spec2cp(mc, seq, len)))
+ unicode = mchars_spec2cp(seq, seqlen);
+ if (unicode <= 0)
continue;
- cpp = utfbuf;
- if (0 == (sz = utf8(u, utfbuf)))
+ addsz = utf8(unicode, utfbuf);
+ if (addsz == 0)
continue;
- sz = strlen(cpp);
+ addcp = utfbuf;
} else {
- cpp = mchars_spec2str(mc, seq, len, &sz);
- if (NULL == cpp)
+ addcp = mchars_spec2str(seq, seqlen, &addsz);
+ if (addcp == NULL)
continue;
- if (ASCII_NBRSP == *cpp) {
- cpp = " ";
- sz = 1;
+ if (*addcp == ASCII_NBRSP) {
+ addcp = " ";
+ addsz = 1;
}
}
/* Copy the rendered glyph into the stream. */
- bsz += sz;
- buf = mandoc_realloc(buf, bsz);
- memcpy(&buf[pos], cpp, sz);
- pos += sz;
+ ssz += addsz;
+ dst = mandoc_realloc(dst, ssz + 1);
+ memcpy(dst + dsz, addcp, addsz);
+ dsz += addsz;
}
+ if (dst != NULL) {
+ *public = dst;
+ *psz = dsz;
+ }
+
+ /* Trim trailing whitespace and NUL-terminate. */
- buf[pos] = '\0';
- key->rendered = buf;
+ while (*psz > 0 && (*public)[*psz - 1] == ' ')
+ --*psz;
+ if (dst != NULL) {
+ (*public)[*psz] = '\0';
+ return 1;
+ } else
+ return 0;
}
static void
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->dsec);
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->arch);
SQL_BIND_TEXT(stmts[STMT_INSERT_LINK], i, mlink->name);
- SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->recno);
+ SQL_BIND_INT64(stmts[STMT_INSERT_LINK], i, mlink->mpage->pageid);
SQL_STEP(stmts[STMT_INSERT_LINK]);
sqlite3_reset(stmts[STMT_INSERT_LINK]);
}
+static void
+dbadd_mlink_name(const struct mlink *mlink)
+{
+ uint64_t bits;
+ size_t i;
+
+ dbadd_mlink(mlink);
+
+ i = 1;
+ SQL_BIND_INT64(stmts[STMT_SELECT_NAME], i, mlink->mpage->pageid);
+ bits = NAME_FILE & NAME_MASK;
+ if (sqlite3_step(stmts[STMT_SELECT_NAME]) == SQLITE_ROW) {
+ bits |= sqlite3_column_int64(stmts[STMT_SELECT_NAME], 0);
+ sqlite3_reset(stmts[STMT_SELECT_NAME]);
+ }
+
+ i = 1;
+ SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, bits);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, mlink->name);
+ SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mlink->mpage->pageid);
+ SQL_STEP(stmts[STMT_INSERT_NAME]);
+ sqlite3_reset(stmts[STMT_INSERT_NAME]);
+}
+
/*
* Flush the current page's terms (and their bits) into the database.
* Wrap the entire set of additions in a transaction to make sqlite be a
* Also, handle escape sequences at the last possible moment.
*/
static void
-dbadd(struct mpage *mpage, struct mchars *mc)
+dbadd(struct mpage *mpage)
{
struct mlink *mlink;
struct str *key;
+ char *cp;
size_t i;
unsigned int slot;
+ int mustfree;
mlink = mpage->mlinks;
if (nodb) {
+ for (key = ohash_first(&names, &slot); NULL != key;
+ key = ohash_next(&names, &slot))
+ free(key);
+ for (key = ohash_first(&strings, &slot); NULL != key;
+ key = ohash_next(&strings, &slot))
+ free(key);
+ if (0 == debug)
+ return;
while (NULL != mlink) {
fputs(mlink->name, stdout);
if (NULL == mlink->next ||
if (NULL != mlink)
fputs(", ", stdout);
}
- for (key = ohash_first(&strings, &slot); NULL != key;
- key = ohash_next(&strings, &slot)) {
- if (TYPE_Nd & key->mask) {
- if (NULL == key->rendered)
- render_key(mc, key);
- printf(" - %s", key->rendered);
- break;
- }
- }
- putchar('\n');
+ printf(" - %s\n", mpage->desc);
return;
}
if (debug)
say(mlink->file, "Adding to database");
+ cp = mpage->desc;
+ i = strlen(cp);
+ mustfree = render_string(&cp, &i);
i = 1;
- SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, FORM_SRC == mpage->form);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_PAGE], i, cp);
+ SQL_BIND_INT(stmts[STMT_INSERT_PAGE], i, mpage->form);
SQL_STEP(stmts[STMT_INSERT_PAGE]);
- mpage->recno = sqlite3_last_insert_rowid(db);
+ mpage->pageid = sqlite3_last_insert_rowid(db);
sqlite3_reset(stmts[STMT_INSERT_PAGE]);
+ if (mustfree)
+ free(cp);
while (NULL != mlink) {
dbadd_mlink(mlink);
mlink = mlink->next;
}
+ mlink = mpage->mlinks;
+ for (key = ohash_first(&names, &slot); NULL != key;
+ key = ohash_next(&names, &slot)) {
+ assert(key->mpage == mpage);
+ i = 1;
+ SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, key->mask);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_NAME], i, key->key);
+ SQL_BIND_INT64(stmts[STMT_INSERT_NAME], i, mpage->pageid);
+ SQL_STEP(stmts[STMT_INSERT_NAME]);
+ sqlite3_reset(stmts[STMT_INSERT_NAME]);
+ free(key);
+ }
for (key = ohash_first(&strings, &slot); NULL != key;
key = ohash_next(&strings, &slot)) {
assert(key->mpage == mpage);
- if (NULL == key->rendered)
- render_key(mc, key);
i = 1;
SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, key->mask);
- SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->rendered);
- SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->recno);
+ SQL_BIND_TEXT(stmts[STMT_INSERT_KEY], i, key->key);
+ SQL_BIND_INT64(stmts[STMT_INSERT_KEY], i, mpage->pageid);
SQL_STEP(stmts[STMT_INSERT_KEY]);
sqlite3_reset(stmts[STMT_INSERT_KEY]);
- if (key->rendered != key->key)
- free(key->rendered);
free(key);
}
}
}
switch (child = fork()) {
- case (-1):
+ case -1:
exitcode = (int)MANDOCLEVEL_SYSERR;
say("", "&fork cmp");
return;
- case (0):
+ case 0:
execlp("cmp", "cmp", "-s",
- tempfilename, MANDOC_DB, NULL);
+ tempfilename, MANDOC_DB, (char *)NULL);
say("", "&exec cmp");
exit(0);
default:
*strrchr(tempfilename, '/') = '\0';
switch (child = fork()) {
- case (-1):
+ case -1:
exitcode = (int)MANDOCLEVEL_SYSERR;
say("", "&fork rm");
return;
- case (0):
- execlp("rm", "rm", "-rf", tempfilename, NULL);
+ case 0:
+ execlp("rm", "rm", "-rf", tempfilename, (char *)NULL);
say("", "&exec rm");
exit((int)MANDOCLEVEL_SYSERR);
default:
const char *sql;
int rc, ofl;
- if (nodb)
- return(1);
+ if (nodb)
+ return 1;
*tempfilename = '\0';
ofl = SQLITE_OPEN_READWRITE;
rc = sqlite3_open_v2(MANDOC_DB, &db, ofl, NULL);
if (SQLITE_OK != rc) {
exitcode = (int)MANDOCLEVEL_SYSERR;
- say(MANDOC_DB, "%s", sqlite3_errmsg(db));
- return(0);
+ if (SQLITE_CANTOPEN != rc)
+ say(MANDOC_DB, "%s", sqlite3_errstr(rc));
+ return 0;
}
goto prepare_statements;
}
remove(MANDOC_DB "~");
rc = sqlite3_open_v2(MANDOC_DB "~", &db, ofl, NULL);
- if (SQLITE_OK == rc)
+ if (SQLITE_OK == rc)
goto create_tables;
if (MPARSE_QUICK & mparse_options) {
exitcode = (int)MANDOCLEVEL_SYSERR;
- say(MANDOC_DB "~", "%s", sqlite3_errmsg(db));
- return(0);
+ say(MANDOC_DB "~", "%s", sqlite3_errstr(rc));
+ return 0;
}
- if (strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
- sizeof(tempfilename)) >= sizeof(tempfilename)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say("", "/tmp/mandocdb.XXXXXX: Filename too long");
- return(0);
- }
+ (void)strlcpy(tempfilename, "/tmp/mandocdb.XXXXXX",
+ sizeof(tempfilename));
if (NULL == mkdtemp(tempfilename)) {
exitcode = (int)MANDOCLEVEL_SYSERR;
say("", "&%s", tempfilename);
- return(0);
- }
- if (strlcat(tempfilename, "/" MANDOC_DB,
- sizeof(tempfilename)) >= sizeof(tempfilename)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- say("", "%s/" MANDOC_DB ": Filename too long",
- tempfilename);
- return(0);
+ return 0;
}
+ (void)strlcat(tempfilename, "/" MANDOC_DB,
+ sizeof(tempfilename));
rc = sqlite3_open_v2(tempfilename, &db, ofl, NULL);
if (SQLITE_OK != rc) {
exitcode = (int)MANDOCLEVEL_SYSERR;
- say("", "%s: %s", tempfilename, sqlite3_errmsg(db));
- return(0);
+ say("", "%s: %s", tempfilename, sqlite3_errstr(rc));
+ return 0;
}
create_tables:
sql = "CREATE TABLE \"mpages\" (\n"
+ " \"desc\" TEXT NOT NULL,\n"
" \"form\" INTEGER NOT NULL,\n"
- " \"id\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
+ " \"pageid\" INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL\n"
");\n"
"\n"
"CREATE TABLE \"mlinks\" (\n"
" \"sec\" TEXT NOT NULL,\n"
" \"arch\" TEXT NOT NULL,\n"
" \"name\" TEXT NOT NULL,\n"
- " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
+ " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
"ON DELETE CASCADE\n"
");\n"
+ "CREATE INDEX mlinks_pageid_idx ON mlinks (pageid);\n"
+ "\n"
+ "CREATE TABLE \"names\" (\n"
+ " \"bits\" INTEGER NOT NULL,\n"
+ " \"name\" TEXT NOT NULL,\n"
+ " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
+ "ON DELETE CASCADE,\n"
+ " UNIQUE (\"name\", \"pageid\") ON CONFLICT REPLACE\n"
+ ");\n"
"\n"
"CREATE TABLE \"keys\" (\n"
" \"bits\" INTEGER NOT NULL,\n"
" \"key\" TEXT NOT NULL,\n"
- " \"pageid\" INTEGER NOT NULL REFERENCES mpages(id) "
+ " \"pageid\" INTEGER NOT NULL REFERENCES mpages(pageid) "
"ON DELETE CASCADE\n"
- ");\n";
+ ");\n"
+ "CREATE INDEX keys_pageid_idx ON keys (pageid);\n";
if (SQLITE_OK != sqlite3_exec(db, sql, NULL, NULL, NULL)) {
exitcode = (int)MANDOCLEVEL_SYSERR;
say(MANDOC_DB, "%s", sqlite3_errmsg(db));
- return(0);
+ sqlite3_close(db);
+ return 0;
}
prepare_statements:
- SQL_EXEC("PRAGMA foreign_keys = ON");
- sql = "DELETE FROM mpages WHERE id IN "
+ if (SQLITE_OK != sqlite3_exec(db,
+ "PRAGMA foreign_keys = ON", NULL, NULL, NULL)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB, "PRAGMA foreign_keys: %s",
+ sqlite3_errmsg(db));
+ sqlite3_close(db);
+ return 0;
+ }
+
+ sql = "DELETE FROM mpages WHERE pageid IN "
"(SELECT pageid FROM mlinks WHERE "
"sec=? AND arch=? AND name=?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_DELETE_PAGE], NULL);
sql = "INSERT INTO mpages "
- "(form) VALUES (?)";
+ "(desc,form) VALUES (?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_PAGE], NULL);
sql = "INSERT INTO mlinks "
"(sec,arch,name,pageid) VALUES (?,?,?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_LINK], NULL);
+ sql = "SELECT bits FROM names where pageid = ?";
+ sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_SELECT_NAME], NULL);
+ sql = "INSERT INTO names "
+ "(bits,name,pageid) VALUES (?,?,?)";
+ sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_NAME], NULL);
sql = "INSERT INTO keys "
"(bits,key,pageid) VALUES (?,?,?)";
sqlite3_prepare_v2(db, sql, -1, &stmts[STMT_INSERT_KEY], NULL);
* synchronous mode for much better performance.
*/
- if (real)
- SQL_EXEC("PRAGMA synchronous = OFF");
+ if (real && SQLITE_OK != sqlite3_exec(db,
+ "PRAGMA synchronous = OFF", NULL, NULL, NULL)) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say(MANDOC_DB, "PRAGMA synchronous: %s",
+ sqlite3_errmsg(db));
+ sqlite3_close(db);
+ return 0;
+ }
#endif
- return(1);
-}
-
-static void *
-hash_halloc(size_t sz, void *arg)
-{
-
- return(mandoc_calloc(sz, 1));
-}
-
-static void *
-hash_alloc(size_t sz, void *arg)
-{
-
- return(mandoc_malloc(sz));
-}
-
-static void
-hash_free(void *p, size_t sz, void *arg)
-{
-
- free(p);
+ return 1;
}
static int
-set_basedir(const char *targetdir)
+set_basedir(const char *targetdir, int report_baddir)
{
static char startdir[PATH_MAX];
- static int fd;
+ static int getcwd_status; /* 1 = ok, 2 = failure */
+ static int chdir_status; /* 1 = changed directory */
+ char *cp;
/*
- * Remember where we started by keeping a fd open to the origin
- * path component: throughout this utility, we chdir() a lot to
- * handle relative paths, and by doing this, we can return to
- * the starting point.
+ * Remember the original working directory, if possible.
+ * This will be needed if the second or a later directory
+ * on the command line is given as a relative path.
+ * Do not error out if the current directory is not
+ * searchable: Maybe it won't be needed after all.
*/
- if ('\0' == *startdir) {
- if (NULL == getcwd(startdir, PATH_MAX)) {
- exitcode = (int)MANDOCLEVEL_SYSERR;
- if (NULL != targetdir)
- say("", "&getcwd");
- return(0);
- }
- if (-1 == (fd = open(startdir, O_RDONLY, 0))) {
+ if (0 == getcwd_status) {
+ if (NULL == getcwd(startdir, sizeof(startdir))) {
+ getcwd_status = 2;
+ (void)strlcpy(startdir, strerror(errno),
+ sizeof(startdir));
+ } else
+ getcwd_status = 1;
+ }
+
+ /*
+ * We are leaving the old base directory.
+ * Do not use it any longer, not even for messages.
+ */
+ *basedir = '\0';
+
+ /*
+ * If and only if the directory was changed earlier and
+ * the next directory to process is given as a relative path,
+ * first go back, or bail out if that is impossible.
+ */
+ if (chdir_status && '/' != *targetdir) {
+ if (2 == getcwd_status) {
exitcode = (int)MANDOCLEVEL_SYSERR;
- say("", "&open %s", startdir);
- return(0);
+ say("", "getcwd: %s", startdir);
+ return 0;
}
- if (NULL == targetdir)
- targetdir = startdir;
- } else {
- if (-1 == fd)
- return(0);
- if (-1 == fchdir(fd)) {
- close(fd);
- basedir[0] = '\0';
+ if (-1 == chdir(startdir)) {
exitcode = (int)MANDOCLEVEL_SYSERR;
say("", "&chdir %s", startdir);
- return(0);
- }
- if (NULL == targetdir) {
- close(fd);
- return(1);
+ return 0;
}
}
+
+ /*
+ * Always resolve basedir to the canonicalized absolute
+ * pathname and append a trailing slash, such that
+ * we can reliably check whether files are inside.
+ */
if (NULL == realpath(targetdir, basedir)) {
- basedir[0] = '\0';
- exitcode = (int)MANDOCLEVEL_BADARG;
- say("", "&%s: realpath", targetdir);
- return(0);
+ if (report_baddir || errno != ENOENT) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say("", "&%s: realpath", targetdir);
+ }
+ return 0;
} else if (-1 == chdir(basedir)) {
- exitcode = (int)MANDOCLEVEL_BADARG;
- say("", "&chdir");
- return(0);
+ if (report_baddir || errno != ENOENT) {
+ exitcode = (int)MANDOCLEVEL_BADARG;
+ say("", "&chdir");
+ }
+ return 0;
+ }
+ chdir_status = 1;
+ cp = strchr(basedir, '\0');
+ if ('/' != cp[-1]) {
+ if (cp - basedir >= PATH_MAX - 1) {
+ exitcode = (int)MANDOCLEVEL_SYSERR;
+ say("", "Filename too long");
+ return 0;
+ }
+ *cp++ = '/';
+ *cp = '\0';
}
- return(1);
+ return 1;
}
static void
if ('\0' != *basedir)
fprintf(stderr, "%s", basedir);
if ('\0' != *basedir && '\0' != *file)
- fputs("//", stderr);
+ fputc('/', stderr);
if ('\0' != *file)
fprintf(stderr, "%s", file);
use_errno = 1;
if (NULL != format) {
switch (*format) {
- case ('&'):
+ case '&':
format++;
break;
- case ('\0'):
+ case '\0':
format = NULL;
break;
default:
git.ckatri.com / mandoc.git / blobdiff