|
|
@@ -0,0 +1,314 @@
|
|
|
+#include <endian.h>
|
|
|
+#include <sys/stat.h>
|
|
|
+#include <errno.h>
|
|
|
+
|
|
|
+#include "store_mdb.h"
|
|
|
+
|
|
|
+
|
|
|
+#if __BYTE_ORDER == __BIG_ENDIAN // FIXME not portable.
|
|
|
+ #define REVERSEKEY 0
|
|
|
+ #define REVERSEDUP 0
|
|
|
+#else
|
|
|
+ #define REVERSEKEY MDB_REVERSEKEY
|
|
|
+ #define REVERSEDUP MDB_REVERSEDUP
|
|
|
+#endif
|
|
|
+
|
|
|
+#define INT_KEY_MASK MDB_INTEGERKEY | REVERSEKEY
|
|
|
+#define INT_DUP_KEY_MASK MDB_DUPSORT | MDB_DUPFIXED \
|
|
|
+ | MDB_INTEGERKEY | REVERSEKEY | REVERSEDUP
|
|
|
+
|
|
|
+#define INT_DUP_MASK MDB_DUPSORT | MDB_DUPFIXED \
|
|
|
+ | MDB_INTEGERDUP | REVERSEKEY | REVERSEDUP
|
|
|
+
|
|
|
+/**
|
|
|
+ * Number of DBs defined.
|
|
|
+ */
|
|
|
+#define N_DB 13
|
|
|
+
|
|
|
+#define ENV_DIR_MODE 0750
|
|
|
+#define ENV_FILE_MODE 0640
|
|
|
+
|
|
|
+
|
|
|
+typedef char DbLabel[8];
|
|
|
+
|
|
|
+
|
|
|
+/**
|
|
|
+ * Static handles.
|
|
|
+ */
|
|
|
+static char *env_path = NULL;
|
|
|
+static bool env_init = false;
|
|
|
+static bool db_init = false;
|
|
|
+
|
|
|
+/**
|
|
|
+ * Main DB labels
|
|
|
+ */
|
|
|
+// Term key to serialized term
|
|
|
+static const DbLabel DB_T_ST = "t:st";
|
|
|
+// Joined triple keys to context key
|
|
|
+static const DbLabel DB_SPO_C = "spo:c";
|
|
|
+// This has empty values and is used to keep track of empty contexts.
|
|
|
+static const DbLabel DB_C_ = "c:";
|
|
|
+// Prefix to namespace
|
|
|
+static const DbLabel DB_PFX_NS = "pfx:ns";
|
|
|
+
|
|
|
+/**
|
|
|
+ * Indices
|
|
|
+ */
|
|
|
+// Namespace to prefix
|
|
|
+static const DbLabel DB_NS_PFX = "ns:pfx";
|
|
|
+// Term hash to term key
|
|
|
+static const DbLabel DB_TH_T = "th:t";
|
|
|
+// 1-bound lookups
|
|
|
+static const DbLabel DB_S_PO = "s:po";
|
|
|
+static const DbLabel DB_P_SO = "p:so";
|
|
|
+static const DbLabel DB_O_SP = "o:sp";
|
|
|
+// 2-bound lookups
|
|
|
+static const DbLabel DB_PO_S = "po:s";
|
|
|
+static const DbLabel DB_SO_P = "so:p";
|
|
|
+static const DbLabel DB_SP_O = "sp:o";
|
|
|
+// Context lookup
|
|
|
+static const DbLabel DB_C_SPO = "c:spo";
|
|
|
+
|
|
|
+/**
|
|
|
+ * Order in which keys are looked up if two terms are bound.
|
|
|
+ * The indices with the smallest average number of values per key should be
|
|
|
+ * looked up first.
|
|
|
+ *
|
|
|
+ * 0 = s:po
|
|
|
+ * 1 = p:so
|
|
|
+ * 2 = o:sp
|
|
|
+ */
|
|
|
+static const uint8_t lookup_rank[3] = {0, 2, 1};
|
|
|
+
|
|
|
+static const uint8_t lookup_ordering_1bound[3][3] = {
|
|
|
+ {0, 1, 2}, // spo
|
|
|
+ {1, 0, 2}, // pso
|
|
|
+ {2, 0, 1}, // osp
|
|
|
+};
|
|
|
+
|
|
|
+static const uint8_t lookup_ordering_2bound[3][3] = {
|
|
|
+ {1, 2, 0}, // po:s
|
|
|
+ {0, 2, 1}, // so:p
|
|
|
+ {0, 1, 2}, // sp:o
|
|
|
+};
|
|
|
+
|
|
|
+/**
|
|
|
+ * Index of each DB in the following constants.
|
|
|
+ */
|
|
|
+typedef enum {
|
|
|
+ IDX_T_ST,
|
|
|
+ IDX_SPO_C,
|
|
|
+ IDX_C_,
|
|
|
+ IDX_PFX_NS,
|
|
|
+ IDX_NS_PFX,
|
|
|
+ IDX_TH_T,
|
|
|
+ IDX_S_PO,
|
|
|
+ IDX_P_SO,
|
|
|
+ IDX_O_SP,
|
|
|
+ IDX_PO_S,
|
|
|
+ IDX_SO_P,
|
|
|
+ IDX_SP_O,
|
|
|
+ IDX_C_SPO,
|
|
|
+} DBIdx;
|
|
|
+
|
|
|
+/**
|
|
|
+ * DB indices. These are populated on init.
|
|
|
+ */
|
|
|
+static MDB_dbi dbis[N_DB];
|
|
|
+
|
|
|
+/**
|
|
|
+ * DB labels.
|
|
|
+ */
|
|
|
+static const char *db_labels[N_DB] = {
|
|
|
+ DB_T_ST,
|
|
|
+ DB_SPO_C,
|
|
|
+ DB_C_,
|
|
|
+ DB_PFX_NS,
|
|
|
+ DB_NS_PFX,
|
|
|
+ DB_TH_T,
|
|
|
+ DB_S_PO,
|
|
|
+ DB_P_SO,
|
|
|
+ DB_O_SP,
|
|
|
+ DB_PO_S,
|
|
|
+ DB_SO_P,
|
|
|
+ DB_SP_O,
|
|
|
+ DB_C_SPO,
|
|
|
+};
|
|
|
+
|
|
|
+/**
|
|
|
+ * DB flags. These are aligned with the dbi_labels index.
|
|
|
+ */
|
|
|
+static const unsigned int db_flags[N_DB] = {
|
|
|
+ INT_KEY_MASK,
|
|
|
+ INT_DUP_KEY_MASK,
|
|
|
+ 0,
|
|
|
+ INT_DUP_KEY_MASK,
|
|
|
+ 0,
|
|
|
+ INT_DUP_MASK,
|
|
|
+ INT_DUP_KEY_MASK,
|
|
|
+ INT_DUP_MASK,
|
|
|
+ INT_DUP_MASK,
|
|
|
+ INT_DUP_MASK,
|
|
|
+ INT_DUP_KEY_MASK,
|
|
|
+ 0,
|
|
|
+ INT_KEY_MASK,
|
|
|
+};
|
|
|
+
|
|
|
+/**
|
|
|
+ * 1-bound and 2-bound lookup indices.
|
|
|
+ */
|
|
|
+static DBIdx lookup_indices[6] = {
|
|
|
+ IDX_S_PO,
|
|
|
+ IDX_P_SO,
|
|
|
+ IDX_O_SP,
|
|
|
+ IDX_PO_S,
|
|
|
+ IDX_SO_P,
|
|
|
+ IDX_SP_O,
|
|
|
+};
|
|
|
+
|
|
|
+
|
|
|
+/**
|
|
|
+ * Static prototypes.
|
|
|
+ */
|
|
|
+static int _dbi_init(bool create);
|
|
|
+static int _mdbenv_init(bool create);
|
|
|
+static void _env_cleanup();
|
|
|
+
|
|
|
+
|
|
|
+/**
|
|
|
+ * API.
|
|
|
+ */
|
|
|
+int
|
|
|
+LSUP_store_open(bool create)
|
|
|
+{
|
|
|
+ if(UNLIKELY(!env_init)) _mdbenv_init(create);
|
|
|
+
|
|
|
+ atexit(_env_cleanup);
|
|
|
+
|
|
|
+ int rc = mdb_env_open(LSUP_mdbenv, env_path, 0, ENV_FILE_MODE);
|
|
|
+
|
|
|
+ if (UNLIKELY(!db_init)) rc |= _dbi_init(create);
|
|
|
+
|
|
|
+ return rc;
|
|
|
+}
|
|
|
+
|
|
|
+
|
|
|
+int
|
|
|
+LSUP_store_stats()
|
|
|
+{
|
|
|
+ // TODO
|
|
|
+ MDB_stat env_stat, db_stats[N_DB];
|
|
|
+ return 0;
|
|
|
+}
|
|
|
+
|
|
|
+
|
|
|
+size_t
|
|
|
+LSUP_store_size(MDB_txn *txn)
|
|
|
+{
|
|
|
+ if (!env_init) return 0;
|
|
|
+
|
|
|
+ MDB_stat stat;
|
|
|
+ mdb_stat(txn, dbis[IDX_SPO_C], &stat);
|
|
|
+
|
|
|
+ return stat.ms_entries;
|
|
|
+}
|
|
|
+
|
|
|
+
|
|
|
+/**
|
|
|
+ * Static functions.
|
|
|
+ */
|
|
|
+
|
|
|
+/**
|
|
|
+ * @brief Create and initialize the MDB env.
|
|
|
+ *
|
|
|
+ * This function takes care of creaating the environment path if not existing,
|
|
|
+ * and checking that it's a writable directory. If the path is not specified
|
|
|
+ * in the LSUP_STORE_PATH environment variable, a temporary file is used.
|
|
|
+ *
|
|
|
+ * This should be only called once per program run, therefore it's best to let
|
|
|
+ * LSUP_mdb_env_open() call it when needed.
|
|
|
+ *
|
|
|
+ * TODO create is not used; either use it or get rid of it.
|
|
|
+ */
|
|
|
+static int
|
|
|
+_mdbenv_init(bool create)
|
|
|
+{
|
|
|
+ env_path = getenv("LSUP_STORE_PATH");
|
|
|
+ struct stat path_stat;
|
|
|
+ int rc;
|
|
|
+
|
|
|
+ // Set environment path.
|
|
|
+ if (env_path == NULL) {
|
|
|
+ printf(
|
|
|
+ "WARNING: `LSUP_STORE_PATH' environment variable is not set. "
|
|
|
+ "A temporary store path will be open and will be DESTROYED "
|
|
|
+ "at the closing of the program.");
|
|
|
+ env_path = mkdtemp("lsup_mdb-XXXXXX");
|
|
|
+ } else {
|
|
|
+
|
|
|
+ rc = stat(env_path, &path_stat);
|
|
|
+ if (rc == ENOENT) {
|
|
|
+ if (mkdir(env_path, ENV_DIR_MODE) != 0) abort();
|
|
|
+ } else {
|
|
|
+ if (!S_ISDIR(path_stat.st_mode)) abort();
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ mdb_env_create(&LSUP_mdbenv);
|
|
|
+
|
|
|
+ // Set map size.
|
|
|
+ size_t mapsize;
|
|
|
+ char *env_mapsize = getenv("LSUP_MDB_MAPSIZE");
|
|
|
+
|
|
|
+ if (env_mapsize == NULL) {
|
|
|
+ mapsize = 1024LU << 30; // 1Tb
|
|
|
+ } else {
|
|
|
+ sscanf(env_mapsize, "%lu", &mapsize);
|
|
|
+ }
|
|
|
+
|
|
|
+ mdb_env_set_mapsize(LSUP_mdbenv, mapsize);
|
|
|
+
|
|
|
+ mdb_env_set_maxdbs(LSUP_mdbenv, N_DB);
|
|
|
+
|
|
|
+ env_init = true;
|
|
|
+
|
|
|
+ return rc;
|
|
|
+
|
|
|
+}
|
|
|
+
|
|
|
+
|
|
|
+/**
|
|
|
+ * @brief Open and allocate DB handles in an array.
|
|
|
+ *
|
|
|
+ * @param bool create [in]: If true, the DBs are created. This is only needed
|
|
|
+ * on bootstrap.
|
|
|
+ */
|
|
|
+static int
|
|
|
+_dbi_init(bool create)
|
|
|
+{
|
|
|
+ MDB_txn *txn;
|
|
|
+ unsigned int txn_flags = create ? 0 : MDB_RDONLY;
|
|
|
+ unsigned int create_flag = create ? MDB_CREATE : 0;
|
|
|
+
|
|
|
+ mdb_txn_begin(LSUP_mdbenv, NULL, txn_flags, &txn);
|
|
|
+
|
|
|
+ for (int i = 0; i < N_DB; i++) {
|
|
|
+ mdb_dbi_open(txn, db_labels[i], db_flags[i] | create_flag, dbis + i);
|
|
|
+ }
|
|
|
+
|
|
|
+ mdb_txn_commit(txn);
|
|
|
+
|
|
|
+ return 0;
|
|
|
+}
|
|
|
+
|
|
|
+
|
|
|
+/* TODO */
|
|
|
+static void
|
|
|
+_env_cleanup()
|
|
|
+{
|
|
|
+ if (env_init) {
|
|
|
+ printf("Cleaning up MDB env.\n");
|
|
|
+ mdb_env_close(LSUP_mdbenv);
|
|
|
+ }
|
|
|
+}
|
|
|
+
|
Stefano Cossu