#include "configuration.h"
#include "syscalls.h"
#include "wstat.h"
-#include "words.h"
#include "printf.h"
#include "filepart.h"
#include "trackname.h"
#include "cache.h"
#include "eventlog.h"
#include "hash.h"
+#include "unicode.h"
+#include "unidata.h"
#define RESCAN "disorder-rescan"
#define DEADLOCK "disorder-deadlock"
const struct kvp *p,
int *used_db);
static int trackdb_alltags_tid(DB_TXN *tid, char ***taglistp);
-static int trackdb_get_global_tid(const char *name,
- DB_TXN *tid,
- const char **rp);
static char **trackdb_new_tid(int *ntracksp,
int maxtracks,
DB_TXN *tid);
static const char *home; /* home had better not change */
DB_ENV *trackdb_env; /* db environment */
-DB *trackdb_tracksdb; /* the db itself */
-DB *trackdb_prefsdb; /* preferences */
-DB *trackdb_searchdb; /* the search database */
+
+/** @brief The tracks database
+ * - Keys are UTF-8(NFC(unicode(path name)))
+ * - Values are encoded key-value pairs
+ * - Data is reconstructable data about tracks that currently exist
+ */
+DB *trackdb_tracksdb;
+
+/** @brief The preferences database
+ *
+ * - Keys are UTF-8(NFC(unicode(path name)))
+ * - Values are encoded key-value pairs
+ * - Data is user data about tracks (that might not exist any more)
+ * and cannot be reconstructed
+ */
+DB *trackdb_prefsdb;
+
+/** @brief The search database
+ *
+ * - Keys are UTF-8(NFKC(casefold(search term)))
+ * - Values are UTF-8(NFC(unicode(path name)))
+ * - There can be more than one value per key
+ * - Presence of key,value means that path matches the search terms
+ * - Only tracks fond in @ref tracks_tracksdb are represented here
+ * - This database can be reconstructed, it contains no user data
+ */
+DB *trackdb_searchdb;
+
+/** @brief The tags database
+ *
+ * - Keys are UTF-8(NFKC(casefold(tag)))
+ * - Values are UTF-8(NFC(unicode(path name)))
+ * - There can be more than one value per key
+ * - Presence of key,value means that path matches the tag
+ * - This is always in sync with the tags preference
+ * - This database can be reconstructed, it contains no user data
+ */
DB *trackdb_tagsdb; /* the tags database */
+
+/** @brief The global preferences database
+ * - Keys are UTF-8(NFC(preference))
+ * - Values are global preference values
+ * - Data is user data and cannot be reconstructed
+ */
DB *trackdb_globaldb; /* global preferences */
+
+/** @brief The noticed database
+ * - Keys are 64-bit big-endian timestamps
+ * - Values are UTF-8(NFC(unicode(path name)))
+ * - There can be more than one value per key
+ * - Presence of key,value means that path was added at the given time
+ * - Data cannot be reconstructed (but isn't THAT important)
+ */
DB *trackdb_noticeddb; /* when track noticed */
static pid_t db_deadlock_pid = -1; /* deadlock manager PID */
static pid_t rescan_pid = -1; /* rescanner PID */
static pid_t subprogram(ev_source *ev, const char *prog,
int outputfd) {
pid_t pid;
- int lfd;
/* If we're in the background then trap subprocess stdout/stderr */
- if(!isatty(2))
- lfd = logfd(ev, prog);
- else
- lfd = -1;
if(!(pid = xfork())) {
exitfn = _exit;
ev_signal_atfork(ev);
signal(SIGPIPE, SIG_DFL);
- if(lfd != -1) {
- xdup2(lfd, 1);
- xdup2(lfd, 2);
- xclose(lfd);
- }
if(outputfd != -1) {
xdup2(outputfd, 1);
xclose(outputfd);
setpriority(PRIO_PROCESS, 0, 0);
execlp(prog, prog, "--config", configfile,
debugging ? "--debug" : "--no-debug",
+ log_default == &log_syslog ? "--syslog" : "--no-syslog",
(char *)0);
fatal(errno, "error invoking %s", prog);
}
- if(lfd != -1) xclose(lfd);
return pid;
}
if((err = db->set_bt_compare(db, compare)))
fatal(0, "db->set_bt_compare %s: %s", path, db_strerror(err));
if((err = db->open(db, 0, path, 0, dbtype,
- openflags | DB_AUTO_COMMIT, mode)))
- fatal(0, "db->open %s: %s", path, db_strerror(err));
+ openflags | DB_AUTO_COMMIT, mode))) {
+ if((openflags & DB_CREATE) || errno != ENOENT)
+ fatal(0, "db->open %s: %s", path, db_strerror(err));
+ db->close(db, 0);
+ db = 0;
+ }
return db;
}
-/* open track databases */
-void trackdb_open(void) {
+/** @brief Open track databases
+ * @param dbupgrade Non-0 to allow non-current database versions
+ */
+void trackdb_open(int dbupgrade) {
+ int newdb, err;
+
/* sanity checks */
assert(opened == 0);
++opened;
+ /* check the database version first */
+ trackdb_globaldb = open_db("global.db", 0, DB_HASH, 0, 0666);
+ if(trackdb_globaldb) {
+ /* This is an existing database */
+ const char *s;
+ long oldversion;
+
+ s = trackdb_get_global("_dbversion");
+ oldversion = s ? atol(s) : 1;
+ if(oldversion > config->dbversion) {
+ /* Database is from the future */
+ fatal(0, "this version of DisOrder is too old for database version %ld",
+ oldversion);
+ }
+ if(oldversion < config->dbversion && !dbupgrade) {
+ /* This database needs upgrading. This isn't implemented yet so we just
+ * fail. */
+ fatal(0, "database needs upgrading from %ld to %ld",
+ oldversion, config->dbversion);
+ }
+ if(oldversion == config->dbversion && dbupgrade) {
+ /* This doesn't make any sense */
+ fatal(0, "database is already at current version");
+ }
+ newdb = 0;
+ /* Close the database again, we'll open it property below */
+ if((err = trackdb_globaldb->close(trackdb_globaldb, 0)))
+ fatal(0, "error closing global.db: %s", db_strerror(err));
+ trackdb_globaldb = 0;
+ } else {
+ if(dbupgrade) {
+ /* Cannot upgrade a new database */
+ fatal(0, "cannot upgrade a database that does not exist");
+ }
+ /* This is a brand new database */
+ newdb = 1;
+ }
/* open the databases */
trackdb_tracksdb = open_db("tracks.db",
DB_RECNUM, DB_BTREE, DB_CREATE, 0666);
trackdb_globaldb = open_db("global.db", 0, DB_HASH, DB_CREATE, 0666);
trackdb_noticeddb = open_db("noticed.db",
DB_DUPSORT, DB_BTREE, DB_CREATE, 0666);
+ if(newdb && !dbupgrade) {
+ /* Stash the database version */
+ char buf[32];
+
+ snprintf(buf, sizeof buf, "%ld", config->dbversion);
+ trackdb_set_global("_dbversion", buf, 0);
+ }
D(("opened databases"));
}
return !strncmp(name, prefix, (sizeof prefix) - 1);
}
+/** @brief Word_Break property tailor that treats underscores as spaces */
+static int tailor_underscore_Word_Break_Other(uint32_t c) {
+ switch(c) {
+ default:
+ return -1;
+ case 0x005F: /* LOW LINE (SPACING UNDERSCORE) */
+ return unicode_Word_Break_Other;
+ }
+}
+
+/** @brief Normalize and split a string using a given tailoring */
+static void word_split(struct vector *v,
+ const char *s,
+ unicode_property_tailor *pt) {
+ size_t nw, nt32, i;
+ uint32_t *t32, **w32;
+
+ /* Convert to UTF-32 */
+ if(!(t32 = utf8_to_utf32(s, strlen(s), &nt32)))
+ return;
+ /* Erase case distinctions */
+ if(!(t32 = utf32_casefold_compat(t32, nt32, &nt32)))
+ return;
+ /* Split into words, treating _ as a space */
+ w32 = utf32_word_split(t32, nt32, &nw, pt);
+ /* Convert words back to UTF-8 and append to result */
+ for(i = 0; i < nw; ++i)
+ vector_append(v, utf32_to_utf8(w32[i], utf32_len(w32[i]), 0));
+}
+
/* compute the words of a track name */
static char **track_to_words(const char *track,
const struct kvp *p) {
struct vector v;
- char **w;
- int nw;
const char *rootless = track_rootless(track);
if(!rootless)
rootless = track; /* bodge */
vector_init(&v);
- if((w = words(casefold(strip_extension(rootless)), &nw)))
- vector_append_many(&v, w, nw);
-
+ rootless = strip_extension(rootless);
+ word_split(&v, strip_extension(rootless), tailor_underscore_Word_Break_Other);
for(; p; p = p->next)
if(is_display_pref(p->name))
- if((w = words(casefold(p->value), &nw)))
- vector_append_many(&v, w, nw);
+ word_split(&v, p->value, 0);
vector_terminate(&v);
return dedupe(v.vec, v.nvec);
}
}
/** @brief notice a possibly new track
+ * @param track NFC UTF-8 track name
+ * @param path Raw path name
+ * @param tid Transaction ID
* @return @c DB_NOTFOUND if new, 0 if already known, @c DB_LOCK_DEADLOCK also
*/
int trackdb_notice_tid(const char *track,
static int stats_read(ev_source attribute((unused)) *ev,
ev_reader *reader,
- int attribute((unused)) fd,
void *ptr,
size_t bytes,
int eof,
}
static int stats_error(ev_source attribute((unused)) *ev,
- int attribute((unused)) fd,
int errno_value,
void *u) {
struct stats_details *const d = u;
/* casefold all the words */
w = xmalloc(nwordlist * sizeof (char *));
for(n = 0; n < nwordlist; ++n) {
- w[n] = casefold(wordlist[n]);
+ w[n] = utf8_casefold_compat(wordlist[n], strlen(wordlist[n]), 0);
if(checktag(w[n])) ++ntags; /* count up tags */
}
/* find the longest non-stopword */
const char *value,
const char *who) {
DB_TXN *tid;
- DBT k, d;
int err;
int state;
- memset(&k, 0, sizeof k);
- memset(&d, 0, sizeof d);
- k.data = (void *)name;
- k.size = strlen(name);
- if(value) {
- d.data = (void *)value;
- d.size = strlen(value);
- }
for(;;) {
tid = trackdb_begin_transaction();
- if(value)
- err = trackdb_globaldb->put(trackdb_globaldb, tid, &k, &d, 0);
- else
- err = trackdb_globaldb->del(trackdb_globaldb, tid, &k, 0);
- if(!err || err == DB_NOTFOUND) break;
- if(err != DB_LOCK_DEADLOCK)
- fatal(0, "error updating database: %s", db_strerror(err));
+ if(!(err = trackdb_set_global_tid(name, value, tid)))
+ break;
trackdb_abort_transaction(tid);
}
trackdb_commit_transaction(tid);
reqtracks = 0;
}
+int trackdb_set_global_tid(const char *name,
+ const char *value,
+ DB_TXN *tid) {
+ DBT k, d;
+ int err;
+
+ memset(&k, 0, sizeof k);
+ memset(&d, 0, sizeof d);
+ k.data = (void *)name;
+ k.size = strlen(name);
+ if(value) {
+ d.data = (void *)value;
+ d.size = strlen(value);
+ }
+ if(value)
+ err = trackdb_globaldb->put(trackdb_globaldb, tid, &k, &d, 0);
+ else
+ err = trackdb_globaldb->del(trackdb_globaldb, tid, &k, 0);
+ if(err == DB_LOCK_DEADLOCK) return err;
+ if(err)
+ fatal(0, "error updating database: %s", db_strerror(err));
+ return 0;
+}
+
const char *trackdb_get_global(const char *name) {
DB_TXN *tid;
int err;
return r;
}
-static int trackdb_get_global_tid(const char *name,
- DB_TXN *tid,
- const char **rp) {
+int trackdb_get_global_tid(const char *name,
+ DB_TXN *tid,
+ const char **rp) {
DBT k, d;
int err;
case DB_LOCK_DEADLOCK:
return err;
default:
- fatal(0, "error updating database: %s", db_strerror(err));
+ fatal(0, "error reading database: %s", db_strerror(err));
}
}
~mdw / disorder / blobdiff