#include "zmalloc.h" /* total memory usage aware version of malloc/free */
#include "lzf.h" /* LZF compression library */
#include "pqsort.h" /* Partial qsort for SORT+LIMIT */
+#include "zipmap.h"
/* Error codes */
#define REDIS_OK 0
#define REDIS_ZSET 3
#define REDIS_HASH 4
-/* Objects encoding */
+/* Objects encoding. Some kind of objects like Strings and Hashes can be
+ * internally represented in multiple ways. The 'encoding' field of the object
+ * is set to one of this fields for this object. */
#define REDIS_ENCODING_RAW 0 /* Raw representation */
#define REDIS_ENCODING_INT 1 /* Encoded as integer */
+#define REDIS_ENCODING_ZIPMAP 2 /* Encoded as zipmap */
+#define REDIS_ENCODING_HT 3 /* Encoded as an hash table */
/* Object types only used for dumping to disk */
#define REDIS_EXPIRETIME 253
static void appendCommand(redisClient *c);
static void substrCommand(redisClient *c);
static void zrankCommand(redisClient *c);
+static void hsetCommand(redisClient *c);
+static void hgetCommand(redisClient *c);
/*================================= Globals ================================= */
{"zcard",zcardCommand,2,REDIS_CMD_INLINE,1,1,1},
{"zscore",zscoreCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
{"zrank",zrankCommand,3,REDIS_CMD_INLINE,1,1,1},
+ {"hset",hsetCommand,4,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
+ {"hget",hgetCommand,3,REDIS_CMD_BULK,1,1,1},
{"incrby",incrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
{"decrby",decrbyCommand,3,REDIS_CMD_INLINE|REDIS_CMD_DENYOOM,1,1,1},
{"getset",getsetCommand,3,REDIS_CMD_BULK|REDIS_CMD_DENYOOM,1,1,1},
};
/* Db->dict */
-static dictType hashDictType = {
+static dictType dbDictType = {
dictObjHash, /* hash function */
NULL, /* key dup */
NULL, /* val dup */
NULL /* val destructor */
};
+/* Hash type hash table (note that small hashes are represented with zimpaps) */
+static dictType hashDictType = {
+ dictEncObjHash, /* hash function */
+ NULL, /* key dup */
+ NULL, /* val dup */
+ dictEncObjKeyCompare, /* key compare */
+ dictRedisObjectDestructor, /* key destructor */
+ dictRedisObjectDestructor /* val destructor */
+};
+
/* Keylist hash table type has unencoded redis objects as keys and
* lists as values. It's used for blocking operations (BLPOP) and to
* map swapped keys to a list of clients waiting for this keys to be loaded. */
exit(1);
}
for (j = 0; j < server.dbnum; j++) {
- server.db[j].dict = dictCreate(&hashDictType,NULL);
+ server.db[j].dict = dictCreate(&dbDictType,NULL);
server.db[j].expires = dictCreate(&keyptrDictType,NULL);
server.db[j].blockingkeys = dictCreate(&keylistDictType,NULL);
if (server.vm_enabled)
return createObject(REDIS_SET,d);
}
+static robj *createHashObject(void) {
+ /* All the Hashes start as zipmaps. Will be automatically converted
+ * into hash tables if there are enough elements or big elements
+ * inside. */
+ unsigned char *zm = zipmapNew();
+ robj *o = createObject(REDIS_HASH,zm);
+ o->encoding = REDIS_ENCODING_ZIPMAP;
+ return o;
+}
+
static robj *createZsetObject(void) {
zset *zs = zmalloc(sizeof(*zs));
* to overwrite the old. So we delete the old key in the database.
* This will also make sure that swap pages about the old object
* will be marked as free. */
- if (deleteIfSwapped(c->db,c->argv[1]))
+ if (server.vm_enabled && deleteIfSwapped(c->db,c->argv[1]))
incrRefCount(c->argv[1]);
dictReplace(c->db->dict,c->argv[1],c->argv[2]);
incrRefCount(c->argv[2]);
lobj = lookupKeyWrite(c->db,c->argv[1]);
if (lobj == NULL) {
if (handleClientsWaitingListPush(c,c->argv[1],c->argv[2])) {
- addReply(c,shared.ok);
+ addReply(c,shared.cone);
return;
}
lobj = createListObject();
return;
}
if (handleClientsWaitingListPush(c,c->argv[1],c->argv[2])) {
- addReply(c,shared.ok);
+ addReply(c,shared.cone);
return;
}
list = lobj->ptr;
incrRefCount(c->argv[2]);
}
server.dirty++;
- addReply(c,shared.ok);
+ addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",listLength(list)));
}
static void lpushCommand(redisClient *c) {
zskiplistNode *zn = zmalloc(sizeof(*zn));
zn->forward = zmalloc(sizeof(zskiplistNode*) * level);
- zn->span = zmalloc(sizeof(unsigned int) * level);
+ if (level > 0)
+ zn->span = zmalloc(sizeof(unsigned int) * (level - 1));
zn->score = score;
zn->obj = obj;
return zn;
zsl->header = zslCreateNode(ZSKIPLIST_MAXLEVEL,0,NULL);
for (j = 0; j < ZSKIPLIST_MAXLEVEL; j++) {
zsl->header->forward[j] = NULL;
- zsl->header->span[j] = 0;
+
+ /* span has space for ZSKIPLIST_MAXLEVEL-1 elements */
+ if (j < ZSKIPLIST_MAXLEVEL-1)
+ zsl->header->span[j] = 0;
}
zsl->header->backward = NULL;
zsl->tail = NULL;
static void zslInsert(zskiplist *zsl, double score, robj *obj) {
zskiplistNode *update[ZSKIPLIST_MAXLEVEL], *x;
- unsigned int span[ZSKIPLIST_MAXLEVEL];
+ unsigned int rank[ZSKIPLIST_MAXLEVEL];
int i, level;
x = zsl->header;
for (i = zsl->level-1; i >= 0; i--) {
- /* store span that is crossed to reach the insert position */
- span[i] = i == (zsl->level-1) ? 0 : span[i+1];
+ /* store rank that is crossed to reach the insert position */
+ rank[i] = i == (zsl->level-1) ? 0 : rank[i+1];
while (x->forward[i] &&
(x->forward[i]->score < score ||
(x->forward[i]->score == score &&
compareStringObjects(x->forward[i]->obj,obj) < 0))) {
- span[i] += x->span[i];
+ rank[i] += i > 0 ? x->span[i-1] : 1;
x = x->forward[i];
}
update[i] = x;
level = zslRandomLevel();
if (level > zsl->level) {
for (i = zsl->level; i < level; i++) {
- span[i] = 0;
+ rank[i] = 0;
update[i] = zsl->header;
- update[i]->span[i] = zsl->length;
+ update[i]->span[i-1] = zsl->length;
}
zsl->level = level;
}
update[i]->forward[i] = x;
/* update span covered by update[i] as x is inserted here */
- x->span[i] = update[i]->span[i] - (span[0] - span[i]);
- update[i]->span[i] = (span[0] - span[i]) + 1;
+ if (i > 0) {
+ x->span[i-1] = update[i]->span[i-1] - (rank[0] - rank[i]);
+ update[i]->span[i-1] = (rank[0] - rank[i]) + 1;
+ }
}
/* increment span for untouched levels */
for (i = level; i < zsl->level; i++) {
- update[i]->span[i]++;
+ update[i]->span[i-1]++;
}
x->backward = (update[0] == zsl->header) ? NULL : update[0];
if (x && score == x->score && compareStringObjects(x->obj,obj) == 0) {
for (i = 0; i < zsl->level; i++) {
if (update[i]->forward[i] == x) {
- update[i]->span[i] += x->span[i] - 1;
+ if (i > 0) {
+ update[i]->span[i-1] += x->span[i-1] - 1;
+ }
update[i]->forward[i] = x->forward[i];
} else {
- update[i]->span[i] -= 1;
+ /* invariant: i > 0, because update[0]->forward[0]
+ * is always equal to x */
+ update[i]->span[i-1] -= 1;
}
}
if (x->forward[0]) {
for (i = 0; i < zsl->level; i++) {
if (update[i]->forward[i] == x) {
- update[i]->span[i] += x->span[i] - 1;
+ if (i > 0) {
+ update[i]->span[i-1] += x->span[i-1] - 1;
+ }
update[i]->forward[i] = x->forward[i];
} else {
- update[i]->span[i] -= 1;
+ /* invariant: i > 0, because update[0]->forward[0]
+ * is always equal to x */
+ update[i]->span[i-1] -= 1;
}
}
if (x->forward[0]) {
return x->forward[0];
}
+/* Find the rank for an element by both score and key.
+ * Returns 0 when the element cannot be found, rank otherwise.
+ * Note that the rank is 1-based due to the span of zsl->header to the
+ * first element. */
+static unsigned long zslGetRank(zskiplist *zsl, double score, robj *o) {
+ zskiplistNode *x;
+ unsigned long rank = 0;
+ int i;
+
+ x = zsl->header;
+ for (i = zsl->level-1; i >= 0; i--) {
+ while (x->forward[i] &&
+ (x->forward[i]->score < score ||
+ (x->forward[i]->score == score &&
+ compareStringObjects(x->forward[i]->obj,o) <= 0))) {
+ rank += i > 0 ? x->span[i-1] : 1;
+ x = x->forward[i];
+ }
+
+ /* x might be equal to zsl->header, so test if obj is non-NULL */
+ if (x->obj && compareStringObjects(x->obj,o) == 0) {
+ return rank;
+ }
+ }
+ return 0;
+}
+
+/* Finds an element by its rank. The rank argument needs to be 1-based. */
+zskiplistNode* zslGetElementByRank(zskiplist *zsl, unsigned long rank) {
+ zskiplistNode *x;
+ unsigned long traversed = 0;
+ int i;
+
+ x = zsl->header;
+ for (i = zsl->level-1; i >= 0; i--) {
+ while (x->forward[i] && (traversed + (i > 0 ? x->span[i-1] : 1)) <= rank) {
+ traversed += i > 0 ? x->span[i-1] : 1;
+ x = x->forward[i];
+ }
+
+ if (traversed == rank) {
+ return x;
+ }
+ }
+ return NULL;
+}
+
/* The actual Z-commands implementations */
/* This generic command implements both ZADD and ZINCRBY.
if (end >= llen) end = llen-1;
rangelen = (end-start)+1;
- /* Return the result in form of a multi-bulk reply */
+ /* check if starting point is trivial, before searching
+ * the element in log(N) time */
if (reverse) {
- ln = zsl->tail;
- while (start--)
- ln = ln->backward;
+ ln = start == 0 ? zsl->tail : zslGetElementByRank(zsl, llen - start);
} else {
- ln = zsl->header->forward[0];
- while (start--)
- ln = ln->forward[0];
+ ln = start == 0 ? zsl->header->forward[0] : zslGetElementByRank(zsl, start + 1);
}
+ /* Return the result in form of a multi-bulk reply */
addReplySds(c,sdscatprintf(sdsempty(),"*%d\r\n",
withscores ? (rangelen*2) : rangelen));
for (j = 0; j < rangelen; j++) {
}
if (o->type != REDIS_ZSET) {
addReply(c,shared.wrongtypeerr);
- return;
- }
+ } else {
+ zset *zs = o->ptr;
+ zskiplist *zsl = zs->zsl;
+ dictEntry *de;
+ unsigned long rank;
- zset *zs = o->ptr;
- zskiplist *zsl = zs->zsl;
- dictEntry *de = dictFind(zs->dict,c->argv[2]);
- if (!de) {
- addReply(c,shared.nullbulk);
- return;
+ de = dictFind(zs->dict,c->argv[2]);
+ if (!de) {
+ addReply(c,shared.nullbulk);
+ return;
+ }
+
+ double *score = dictGetEntryVal(de);
+ rank = zslGetRank(zsl, *score, c->argv[2]);
+ if (rank) {
+ addReplyLong(c, rank-1);
+ } else {
+ addReply(c,shared.nullbulk);
+ }
}
+}
- double *score = dictGetEntryVal(de);
- zskiplistNode *x;
- unsigned int rank = 0;
- int i;
+/* ==================================== Hash ================================ */
+static void hsetCommand(redisClient *c) {
+ int update = 0;
+ robj *o = lookupKeyWrite(c->db,c->argv[1]);
- x = zsl->header;
- for (i = zsl->level-1; i >= 0; i--) {
- while (x->forward[i] &&
- (x->forward[i]->score < *score ||
- (x->forward[i]->score == *score &&
- compareStringObjects(x->forward[i]->obj,c->argv[2]) <= 0))) {
- rank += x->span[i];
- x = x->forward[i];
+ if (o == NULL) {
+ o = createHashObject();
+ dictAdd(c->db->dict,c->argv[1],o);
+ incrRefCount(c->argv[1]);
+ } else {
+ if (o->type != REDIS_HASH) {
+ addReply(c,shared.wrongtypeerr);
+ return;
}
+ }
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *zm = o->ptr;
- /* x might be equal to zsl->header, so test if obj is non-NULL */
- if (x->obj && compareStringObjects(x->obj,c->argv[2]) == 0) {
- /* the pointer from zsl->header to the first element also spans one,
- * which makes the rank 1-based */
- addReplyLong(c, rank-1);
- return;
+ zm = zipmapSet(zm,c->argv[2]->ptr,sdslen(c->argv[2]->ptr),
+ c->argv[3]->ptr,sdslen(c->argv[3]->ptr),&update);
+ } else {
+ if (dictAdd(o->ptr,c->argv[2],c->argv[3]) == DICT_OK) {
+ incrRefCount(c->argv[2]);
+ } else {
+ update = 1;
}
+ incrRefCount(c->argv[3]);
}
+ server.dirty++;
+ addReplySds(c,sdscatprintf(sdsempty(),":%d\r\n",update == 0));
+}
+
+static void hgetCommand(redisClient *c) {
+ robj *o = lookupKeyRead(c->db,c->argv[1]);
+
+ if (o == NULL) {
+ addReply(c,shared.nullbulk);
+ return;
+ } else {
+ if (o->encoding == REDIS_ENCODING_ZIPMAP) {
+ unsigned char *zm = o->ptr;
+ unsigned char *val;
+ unsigned int vlen;
+
+ if (zipmapGet(zm,c->argv[2]->ptr,sdslen(c->argv[2]->ptr), &val,&vlen)) {
+ addReplySds(c,sdscatprintf(sdsempty(),"$%u\r\n", vlen));
+ addReplySds(c,sdsnewlen(val,vlen));
+ addReply(c,shared.crlf);
+ return;
+ } else {
+ addReply(c,shared.nullbulk);
+ return;
+ }
+ } else {
+ struct dictEntry *de;
- addReply(c,shared.nullbulk);
+ de = dictFind(o->ptr,c->argv[2]);
+ if (de == NULL) {
+ addReply(c,shared.nullbulk);
+ } else {
+ robj *e = dictGetEntryVal(de);
+
+ addReplyBulkLen(c,e);
+ addReply(c,e);
+ addReply(c,shared.crlf);
+ }
+ }
+ }
}
/* ========================= Non type-specific commands ==================== */