void replicationFeedSlaves(list *slaves, int dictid, robj **argv, int argc) {
//propagateExpire删除超时key，或者propagate函数分发同步客户端写操作时会调用这里。
//将这条指令发送给所有的slaves，放到其缓冲里面以待发送。
listNode *ln;
listIter li;
int j;
listRewind(slaves,&li);//获取slaves的第一个节点
while((ln = listNext(&li))) {
redisClient *slave = ln->value;
//下面不发送给这个slave，那么数据怎么办? 后面记着了的。这个状态的客户端是个从库，并且sync后rdb还没有开始，没必要为其追加写操作日志。
/* Don't feed slaves that are still waiting for BGSAVE to start */
if (slave->replstate == REDIS_REPL_WAIT_BGSAVE_START) continue;
/* Feed slaves that are waiting for the initial SYNC (so these commands
* are queued in the output buffer until the initial SYNC completes),
* or are already in sync with the master. */
if (slave->slaveseldb != dictid) {
robj *selectcmd;
if (dictid >= 0 && dictid < REDIS_SHARED_SELECT_CMDS) {
selectcmd = shared.select[dictid];
incrRefCount(selectcmd);
} else {
selectcmd = createObject(REDIS_STRING,
sdscatprintf(sdsempty(),"select %d\r\n",dictid));
}
addReply(slave,selectcmd);//追加一条选择db的指令给当前这个slave
decrRefCount(selectcmd);
slave->slaveseldb = dictid;
}
addReplyMultiBulkLen(slave,argc);//增加参数数目指令
for (j = 0; j < argc; j++) //一个个将参数追加到slave上面去。
addReplyBulk(slave,argv[j]);
}
}
void lsCommand(redisClient* c)
{
DIR* pstDir = opendir((const char*)c->argv[1]->ptr);
if (pstDir == NULL) {
addReplyErrorFormat(c, "opendir fail, %s", strerror(errno));
return;
}
list* dir = listCreate();
struct dirent* pstDirent = NULL;
while ((pstDirent = readdir(pstDir)) != NULL) {
if(strcmp(pstDirent->d_name, ".") == 0 ||
strcmp(pstDirent->d_name, "..") == 0)
continue;
sds path = sdsdup(c->argv[1]->ptr);
size_t len = sdslen(path);
if (path[len - 1] != '/')
path = sdscat(path, "/");
path = sdscat(path, pstDirent->d_name);
listAddNodeTail(dir, path);
}
closedir(pstDir);
addReplyMultiBulkLen(c, listLength(dir));
listIter* iter = listGetIterator(dir, AL_START_HEAD);
listNode* node = NULL;
while ((node = listNext(iter))) {
addReplyBulkCString(c, node->value);
sdsfree(node->value);
listDelNode(dir, node);
}
listRelease(dir);
}
// 执行事务
void execCommand(redisClient *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
// 如果没执行过 MULTI ，报错
if (!(c->flags & REDIS_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC if some WATCHed key was touched.
* A failed EXEC will return a multi bulk nil object. */
// 如果在执行事务之前，有监视中（WATCHED）的 key 被改变
// 那么取消这个事务
if (c->flags & REDIS_DIRTY_CAS) {
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS);
unwatchAllKeys(c);
addReply(c,shared.nullmultibulk);
return;
}
/* Replicate a MULTI request now that we are sure the block is executed.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
// 为保证一致性和原子性
// 如果处在 AOF 模式中，向 AOF 文件发送 MULTI
// 如果处在复制模式中，向附属节点发送 MULTI
execCommandReplicateMulti(c);
/* Exec all the queued commands */
// 开始执行所有事务中的命令（FIFO 方式）
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
// 备份所有参数和命令
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count);
for (j = 0; j < c->mstate.count; j++) {
c->argc = c->mstate.commands[j].argc; // 取出参数数量
c->argv = c->mstate.commands[j].argv; // 取出参数
c->cmd = c->mstate.commands[j].cmd; // 取出要执行的命令
call(c,REDIS_CALL_FULL); // 执行命令
/* Commands may alter argc/argv, restore mstate. */
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
// 恢复所有参数和命令
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
// 重置事务状态
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS);
/* Make sure the EXEC command is always replicated / AOF, since we
* always send the MULTI command (we can't know beforehand if the
* next operations will contain at least a modification to the DB). */
// 更新状态值，确保事务执行之后的状态为脏
server.dirty++;
}
/*
* EXEC 命令的实现
*/
void execCommand(redisClient *c)
{
int j;
// 用于保存执行命令、命令的参数和参数数量的副本
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
// 只能在 MULTI 已启用的情况下执行
if (!(c->flags & REDIS_MULTI))
{
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC because:
* 以下情况发生时，取消事务
*
* 1) Some WATCHed key was touched.
* 某些被监视的键已被修改（状态为 REDIS_DIRTY_CAS）
*
* 2) There was a previous error while queueing commands.
* 有命令在入队时发生错误（状态为 REDIS_DIRTY_EXEC）
*
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned.
*
* 第一种情况返回多个空白 NULL 对象，
* 第二种情况返回一个 EXECABORT 错误。
*/
if (c->flags & (REDIS_DIRTY_CAS|REDIS_DIRTY_EXEC))
{
// 根据状态，决定返回的错误的类型
addReply(c, c->flags & REDIS_DIRTY_EXEC ? shared.execaborterr :
shared.nullmultibulk);
// 以下四句可以用 discardTransaction() 来替换
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS|REDIS_DIRTY_EXEC);
unwatchAllKeys(c);
goto handle_monitor;
}
/* Replicate a MULTI request now that we are sure the block is executed.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
// 向所有附属节点和 AOF 文件发送 MULTI 命令
execCommandReplicateMulti(c);
/* Exec all the queued commands */
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
// 将三个原始参数备份起来
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count);
// 执行所有入队的命令
for (j = 0; j < c->mstate.count; j++)
{
// 因为 call 可能修改命令，而命令需要传送给其他同步节点
// 所以这里将要执行的命令（及其参数）先备份起来
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
// 执行命令
call(c,REDIS_CALL_FULL);
/* Commands may alter argc/argv, restore mstate. */
// 还原原始的参数到队列里
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
// 还原三个原始命令
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
// 以下三句也可以用 discardTransaction() 来替换
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS|REDIS_DIRTY_EXEC);
/* Make sure the EXEC command is always replicated / AOF, since we
* always send the MULTI command (we can't know beforehand if the
* next operations will contain at least a modification to the DB). */
server.dirty++;
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as REDIS_CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering. */
// 向同步节点发送命令
if (listLength(server.monitors) && !server.loading)
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
void zrangeGenericCommand(redisClient *c, int reverse) {
robj *o;
long start;
long end;
int withscores = 0;
int llen;
int rangelen, j;
zset *zsetobj;
zskiplist *zsl;
zskiplistNode *ln;
robj *ele;
if ((getLongFromObjectOrReply(c, c->argv[2], &start, NULL) != REDIS_OK) ||
(getLongFromObjectOrReply(c, c->argv[3], &end, NULL) != REDIS_OK)) return;
if (c->argc == 5 && !strcasecmp(c->argv[4]->ptr,"withscores")) {
withscores = 1;
} else if (c->argc >= 5) {
addReply(c,shared.syntaxerr);
return;
}
if ((o = lookupKeyReadOrReply(c,c->argv[1],shared.emptymultibulk)) == NULL
|| checkType(c,o,REDIS_ZSET)) return;
zsetobj = o->ptr;
zsl = zsetobj->zsl;
llen = zsl->length;
/* convert negative indexes */
if (start < 0) start = llen+start;
if (end < 0) end = llen+end;
if (start < 0) start = 0;
/* Invariant: start >= 0, so this test will be true when end < 0.
* The range is empty when start > end or start >= length. */
if (start > end || start >= llen) {
addReply(c,shared.emptymultibulk);
return;
}
if (end >= llen) end = llen-1;
rangelen = (end-start)+1;
/* check if starting point is trivial, before searching
* the element in log(N) time */
if (reverse) {
ln = start == 0 ? zsl->tail : zslGetElementByRank(zsl, llen-start);
} else {
ln = start == 0 ?
zsl->header->level[0].forward : zslGetElementByRank(zsl, start+1);
}
/* Return the result in form of a multi-bulk reply */
addReplyMultiBulkLen(c,withscores ? (rangelen*2) : rangelen);
for (j = 0; j < rangelen; j++) {
ele = ln->obj;
addReplyBulk(c,ele);
if (withscores)
addReplyDouble(c,ln->score);
ln = reverse ? ln->backward : ln->level[0].forward;
}
}
/* This command implements SCAN, HSCAN and SSCAN commands.
* If object 'o' is passed, then it must be a Hash or Set object, otherwise
* if 'o' is NULL the command will operate on the dictionary associated with
* the current database.
*
* When 'o' is not NULL the function assumes that the first argument in
* the client arguments vector is a key so it skips it before iterating
* in order to parse options.
*
* In the case of a Hash object the function returns both the field and value
* of every element on the Hash. */
void scanGenericCommand(client *c, robj *o, unsigned long cursor) {
int i, j;
list *keys = listCreate();
listNode *node, *nextnode;
long count = 10;
sds pat = NULL;
int patlen = 0, use_pattern = 0;
dict *ht;
/* Object must be NULL (to iterate keys names), or the type of the object
* must be Set, Sorted Set, or Hash. */
serverAssert(o == NULL || o->type == OBJ_SET || o->type == OBJ_HASH ||
o->type == OBJ_ZSET);
/* Set i to the first option argument. The previous one is the cursor. */
i = (o == NULL) ? 2 : 3; /* Skip the key argument if needed. */
/* Step 1: Parse options. */
while (i < c->argc) {
j = c->argc - i;
if (!strcasecmp(c->argv[i]->ptr, "count") && j >= 2) {
if (getLongFromObjectOrReply(c, c->argv[i+1], &count, NULL)
!= C_OK)
{
goto cleanup;
}
if (count < 1) {
addReply(c,shared.syntaxerr);
goto cleanup;
}
i += 2;
} else if (!strcasecmp(c->argv[i]->ptr, "match") && j >= 2) {
pat = c->argv[i+1]->ptr;
patlen = sdslen(pat);
/* The pattern always matches if it is exactly "*", so it is
* equivalent to disabling it. */
use_pattern = !(pat[0] == '*' && patlen == 1);
i += 2;
} else {
addReply(c,shared.syntaxerr);
goto cleanup;
}
}
/* Step 2: Iterate the collection.
*
* Note that if the object is encoded with a ziplist, intset, or any other
* representation that is not a hash table, we are sure that it is also
* composed of a small number of elements. So to avoid taking state we
* just return everything inside the object in a single call, setting the
* cursor to zero to signal the end of the iteration. */
/* Handle the case of a hash table. */
ht = NULL;
if (o == NULL) {
ht = c->db->dict;
} else if (o->type == OBJ_SET && o->encoding == OBJ_ENCODING_HT) {
ht = o->ptr;
} else if (o->type == OBJ_HASH && o->encoding == OBJ_ENCODING_HT) {
ht = o->ptr;
count *= 2; /* We return key / value for this type. */
} else if (o->type == OBJ_ZSET && o->encoding == OBJ_ENCODING_SKIPLIST) {
zset *zs = o->ptr;
ht = zs->dict;
count *= 2; /* We return key / value for this type. */
}
if (ht) {
void *privdata[2];
/* We set the max number of iterations to ten times the specified
* COUNT, so if the hash table is in a pathological state (very
* sparsely populated) we avoid to block too much time at the cost
* of returning no or very few elements. */
long maxiterations = count*10;
/* We pass two pointers to the callback: the list to which it will
* add new elements, and the object containing the dictionary so that
* it is possible to fetch more data in a type-dependent way. */
privdata[0] = keys;
privdata[1] = o;
do {
cursor = dictScan(ht, cursor, scanCallback, privdata);
} while (cursor &&
maxiterations-- &&
listLength(keys) < (unsigned long)count);
} else if (o->type == OBJ_SET) {
int pos = 0;
int64_t ll;
while(intsetGet(o->ptr,pos++,&ll))
listAddNodeTail(keys,createStringObjectFromLongLong(ll));
cursor = 0;
} else if (o->type == OBJ_HASH || o->type == OBJ_ZSET) {
unsigned char *p = ziplistIndex(o->ptr,0);
unsigned char *vstr;
unsigned int vlen;
long long vll;
while(p) {
ziplistGet(p,&vstr,&vlen,&vll);
listAddNodeTail(keys,
(vstr != NULL) ? createStringObject((char*)vstr,vlen) :
createStringObjectFromLongLong(vll));
p = ziplistNext(o->ptr,p);
}
cursor = 0;
} else {
serverPanic("Not handled encoding in SCAN.");
}
/* Step 3: Filter elements. */
node = listFirst(keys);
while (node) {
robj *kobj = listNodeValue(node);
nextnode = listNextNode(node);
int filter = 0;
/* Filter element if it does not match the pattern. */
if (!filter && use_pattern) {
if (sdsEncodedObject(kobj)) {
if (!stringmatchlen(pat, patlen, kobj->ptr, sdslen(kobj->ptr), 0))
filter = 1;
} else {
char buf[LONG_STR_SIZE];
int len;
serverAssert(kobj->encoding == OBJ_ENCODING_INT);
len = ll2string(buf,sizeof(buf),(long)kobj->ptr);
if (!stringmatchlen(pat, patlen, buf, len, 0)) filter = 1;
}
}
/* Filter element if it is an expired key. */
if (!filter && o == NULL && expireIfNeeded(c->db, kobj)) filter = 1;
/* Remove the element and its associted value if needed. */
if (filter) {
decrRefCount(kobj);
listDelNode(keys, node);
}
/* If this is a hash or a sorted set, we have a flat list of
* key-value elements, so if this element was filtered, remove the
* value, or skip it if it was not filtered: we only match keys. */
if (o && (o->type == OBJ_ZSET || o->type == OBJ_HASH)) {
node = nextnode;
nextnode = listNextNode(node);
if (filter) {
kobj = listNodeValue(node);
decrRefCount(kobj);
listDelNode(keys, node);
}
}
node = nextnode;
}
/* Step 4: Reply to the client. */
addReplyMultiBulkLen(c, 2);
addReplyBulkLongLong(c,cursor);
addReplyMultiBulkLen(c, listLength(keys));
while ((node = listFirst(keys)) != NULL) {
robj *kobj = listNodeValue(node);
addReplyBulk(c, kobj);
decrRefCount(kobj);
listDelNode(keys, node);
}
cleanup:
listSetFreeMethod(keys,decrRefCountVoid);
listRelease(keys);
}
void srandmemberWithCountCommand(redisClient *c) {
long l;
unsigned long count, size;
int uniq = 1;
robj *set, *ele;
int64_t llele;
int encoding;
dict *d;
if (getLongFromObjectOrReply(c,c->argv[2],&l,NULL) != REDIS_OK) return;
if (l >= 0) {
count = (unsigned) l;
} else {
/* A negative count means: return the same elements multiple times
* (i.e. don't remove the extracted element after every extraction). */
count = -l;
uniq = 0;
}
if ((set = lookupKeyReadOrReply(c,c->argv[1],shared.emptymultibulk))
== NULL || checkType(c,set,REDIS_SET)) return;
size = setTypeSize(set);
/* If count is zero, serve it ASAP to avoid special cases later. */
if (count == 0) {
addReply(c,shared.emptymultibulk);
return;
}
/* CASE 1: The count was negative, so the extraction method is just:
* "return N random elements" sampling the whole set every time.
* This case is trivial and can be served without auxiliary data
* structures. */
if (!uniq) {
addReplyMultiBulkLen(c,count);
while(count--) {
encoding = setTypeRandomElement(set,&ele,&llele);
if (encoding == REDIS_ENCODING_INTSET) {
addReplyBulkLongLong(c,llele);
} else {
addReplyBulk(c,ele);
}
}
return;
}
/* CASE 2:
* The number of requested elements is greater than the number of
* elements inside the set: simply return the whole set. */
if (count >= size) {
sunionDiffGenericCommand(c,c->argv+1,1,NULL,REDIS_OP_UNION);
return;
}
/* For CASE 3 and CASE 4 we need an auxiliary dictionary. */
d = dictCreate(&setDictType,NULL);
/* CASE 3:
* The number of elements inside the set is not greater than
* SRANDMEMBER_SUB_STRATEGY_MUL times the number of requested elements.
* In this case we create a set from scratch with all the elements, and
* subtract random elements to reach the requested number of elements.
*
* This is done because if the number of requsted elements is just
* a bit less than the number of elements in the set, the natural approach
* used into CASE 3 is highly inefficient. */
if (count*SRANDMEMBER_SUB_STRATEGY_MUL > size) {
setTypeIterator *si;
/* Add all the elements into the temporary dictionary. */
si = setTypeInitIterator(set);
while((encoding = setTypeNext(si,&ele,&llele)) != -1) {
int retval = DICT_ERR;
if (encoding == REDIS_ENCODING_INTSET) {
retval = dictAdd(d,createStringObjectFromLongLong(llele),NULL);
} else if (ele->encoding == REDIS_ENCODING_RAW) {
retval = dictAdd(d,dupStringObject(ele),NULL);
} else if (ele->encoding == REDIS_ENCODING_INT) {
retval = dictAdd(d,
createStringObjectFromLongLong((long)ele->ptr),NULL);
}
redisAssert(retval == DICT_OK);
}
setTypeReleaseIterator(si);
redisAssert(dictSize(d) == size);
/* Remove random elements to reach the right count. */
while(size > count) {
dictEntry *de;
de = dictGetRandomKey(d);
dictDelete(d,dictGetKey(de));
size--;
}
}
/* CASE 4: We have a big set compared to the requested number of elements.
* In this case we can simply get random elements from the set and add
* to the temporary set, trying to eventually get enough unique elements
* to reach the specified count. */
else {
unsigned long added = 0;
while(added < count) {
encoding = setTypeRandomElement(set,&ele,&llele);
if (encoding == REDIS_ENCODING_INTSET) {
ele = createStringObjectFromLongLong(llele);
} else if (ele->encoding == REDIS_ENCODING_RAW) {
ele = dupStringObject(ele);
} else if (ele->encoding == REDIS_ENCODING_INT) {
ele = createStringObjectFromLongLong((long)ele->ptr);
}
/* Try to add the object to the dictionary. If it already exists
* free it, otherwise increment the number of objects we have
* in the result dictionary. */
if (dictAdd(d,ele,NULL) == DICT_OK)
added++;
else
decrRefCount(ele);
}
}
/* CASE 3 & 4: send the result to the user. */
{
dictIterator *di;
dictEntry *de;
addReplyMultiBulkLen(c,count);
di = dictGetIterator(d);
while((de = dictNext(di)) != NULL)
addReplyBulk(c,dictGetKey(de));
dictReleaseIterator(di);
dictRelease(d);
}
}
/* PUBSUB command for Pub/Sub introspection. */
void pubsubCommand(redisClient *c) {
// PUBSUB CHANNELS [pattern] 子命令
if (!strcasecmp(c->argv[1]->ptr,"channels") &&
(c->argc == 2 || c->argc ==3))
{
/* PUBSUB CHANNELS [<pattern>] */
// 检查命令请求是否给定了 pattern 参数
// 如果没有给定的话，就设为 NULL
sds pat = (c->argc == 2) ? NULL : c->argv[2]->ptr;
// 创建 pubsub_channels 的字典迭代器
// 该字典的键为频道，值为链表
// 链表中保存了所有订阅键所对应的频道的客户端
dictIterator *di = dictGetIterator(server.pubsub_channels);
dictEntry *de;
long mblen = 0;
void *replylen;
replylen = addDeferredMultiBulkLength(c);
// 从迭代器中获取一个客户端
while((de = dictNext(di)) != NULL) {
// 从字典中取出客户端所订阅的频道
robj *cobj = dictGetKey(de);
sds channel = cobj->ptr;
// 顺带一提
// 因为 Redis 的字典实现只能遍历字典的值（客户端）
// 所以这里才会有遍历字典值然后通过字典值取出字典键（频道）的蹩脚用法
// 如果没有给定 pattern 参数，那么打印所有找到的频道
// 如果给定了 pattern 参数，那么只打印和 pattern 相匹配的频道
if (!pat || stringmatchlen(pat, sdslen(pat),
channel, sdslen(channel),0))
{
// 向客户端输出频道
addReplyBulk(c,cobj);
mblen++;
}
}
// 释放字典迭代器
dictReleaseIterator(di);
setDeferredMultiBulkLength(c,replylen,mblen);
// PUBSUB NUMSUB [channel-1 channel-2 ... channel-N] 子命令
} else if (!strcasecmp(c->argv[1]->ptr,"numsub") && c->argc >= 2) {
/* PUBSUB NUMSUB [Channel_1 ... Channel_N] */
int j;
addReplyMultiBulkLen(c,(c->argc-2)*2);
for (j = 2; j < c->argc; j++) {
// c->argv[j] 也即是客户端输入的第 N 个频道名字
// pubsub_channels 的字典为频道名字
// 而值则是保存了 c->argv[j] 频道所有订阅者的链表
// 而调用 dictFetchValue 也就是取出所有订阅给定频道的客户端
list *l = dictFetchValue(server.pubsub_channels,c->argv[j]);
addReplyBulk(c,c->argv[j]);
// 向客户端返回链表的长度属性
// 这个属性就是某个频道的订阅者数量
// 例如：如果一个频道有三个订阅者，那么链表的长度就是 3
// 而返回给客户端的数字也是三
addReplyBulkLongLong(c,l ? listLength(l) : 0);
}
// PUBSUB NUMPAT 子命令
} else if (!strcasecmp(c->argv[1]->ptr,"numpat") && c->argc == 2) {
/* PUBSUB NUMPAT */
// pubsub_patterns 链表保存了服务器中所有被订阅的模式
// pubsub_patterns 的长度就是服务器中被订阅模式的数量
addReplyLongLong(c,listLength(server.pubsub_patterns));
// 错误处理
} else {
addReplyErrorFormat(c,
"Unknown PUBSUB subcommand or wrong number of arguments for '%s'",
(char*)c->argv[1]->ptr);
}
}
void spopWithCountCommand(redisClient *c) {
long l;
unsigned long count, size;
robj *set;
/* Get the count argument */
if (getLongFromObjectOrReply(c,c->argv[2],&l,NULL) != REDIS_OK) return;
if (l >= 0) {
count = (unsigned) l;
} else {
addReply(c,shared.outofrangeerr);
return;
}
/* Make sure a key with the name inputted exists, and that it's type is
* indeed a set. Otherwise, return nil */
if ((set = lookupKeyReadOrReply(c,c->argv[1],shared.emptymultibulk))
== NULL || checkType(c,set,REDIS_SET)) return;
/* If count is zero, serve an empty multibulk ASAP to avoid special
* cases later. */
if (count == 0) {
addReply(c,shared.emptymultibulk);
return;
}
size = setTypeSize(set);
/* Generate an SPOP keyspace notification */
notifyKeyspaceEvent(REDIS_NOTIFY_SET,"spop",c->argv[1],c->db->id);
server.dirty += count;
/* CASE 1:
* The number of requested elements is greater than or equal to
* the number of elements inside the set: simply return the whole set. */
if (count >= size) {
/* We just return the entire set */
sunionDiffGenericCommand(c,c->argv+1,1,NULL,REDIS_OP_UNION);
/* Delete the set as it is now empty */
dbDelete(c->db,c->argv[1]);
notifyKeyspaceEvent(REDIS_NOTIFY_GENERIC,"del",c->argv[1],c->db->id);
/* Propagate this command as an DEL operation */
rewriteClientCommandVector(c,2,shared.del,c->argv[1]);
signalModifiedKey(c->db,c->argv[1]);
server.dirty++;
return;
}
/* Case 2 and 3 require to replicate SPOP as a set of SERM commands.
* Prepare our replication argument vector. Also send the array length
* which is common to both the code paths. */
robj *propargv[3];
propargv[0] = createStringObject("SREM",4);
propargv[1] = c->argv[1];
addReplyMultiBulkLen(c,count);
/* Common iteration vars. */
robj *objele;
int encoding;
int64_t llele;
unsigned long remaining = size-count; /* Elements left after SPOP. */
/* If we are here, the number of requested elements is less than the
* number of elements inside the set. Also we are sure that count < size.
* Use two different strategies.
*
* CASE 2: The number of elements to return is small compared to the
* set size. We can just extract random elements and return them to
* the set. */
if (remaining*SPOP_MOVE_STRATEGY_MUL > count) {
while(count--) {
encoding = setTypeRandomElement(set,&objele,&llele);
if (encoding == REDIS_ENCODING_INTSET) {
objele = createStringObjectFromLongLong(llele);
} else {
incrRefCount(objele);
}
/* Return the element to the client and remove from the set. */
addReplyBulk(c,objele);
setTypeRemove(set,objele);
/* Replicate/AOF this command as an SREM operation */
propargv[2] = objele;
alsoPropagate(server.sremCommand,c->db->id,propargv,3,
REDIS_PROPAGATE_AOF|REDIS_PROPAGATE_REPL);
decrRefCount(objele);
}
} else {
/* CASE 3: The number of elements to return is very big, approaching
* the size of the set itself. After some time extracting random elements
* from such a set becomes computationally expensive, so we use
* a different strategy, we extract random elements that we don't
* want to return (the elements that will remain part of the set),
* creating a new set as we do this (that will be stored as the original
* set). Then we return the elements left in the original set and
* release it. */
robj *newset = NULL;
/* Create a new set with just the remaining elements. */
while(remaining--) {
encoding = setTypeRandomElement(set,&objele,&llele);
if (encoding == REDIS_ENCODING_INTSET) {
objele = createStringObjectFromLongLong(llele);
} else {
incrRefCount(objele);
}
if (!newset) newset = setTypeCreate(objele);
setTypeAdd(newset,objele);
setTypeRemove(set,objele);
decrRefCount(objele);
}
/* Assign the new set as the key value. */
incrRefCount(set); /* Protect the old set value. */
dbOverwrite(c->db,c->argv[1],newset);
/* Tranfer the old set to the client and release it. */
setTypeIterator *si;
si = setTypeInitIterator(set);
while((encoding = setTypeNext(si,&objele,&llele)) != -1) {
if (encoding == REDIS_ENCODING_INTSET) {
objele = createStringObjectFromLongLong(llele);
} else {
incrRefCount(objele);
}
addReplyBulk(c,objele);
/* Replicate/AOF this command as an SREM operation */
propargv[2] = objele;
alsoPropagate(server.sremCommand,c->db->id,propargv,3,
REDIS_PROPAGATE_AOF|REDIS_PROPAGATE_REPL);
decrRefCount(objele);
}
setTypeReleaseIterator(si);
decrRefCount(set);
}
/* Don't propagate the command itself even if we incremented the
* dirty counter. We don't want to propagate an SPOP command since
* we propagated the command as a set of SREMs operations using
* the alsoPropagate() API. */
decrRefCount(propargv[0]);
preventCommandPropagation(c);
}
//receiver是被阻塞的客户端，key是造成阻塞的键，db是key所在的数据库，value是被提供给客户端的值
//如果dstkey不为空，则将value推入到dstkey中
int serveClientBlockedOnList(client *receiver, robj *key, robj *dstkey, redisDb *db, robj *value, int where)
{
robj *argv[3];
//如果dstkey为空，则执行的是BLPOP或BRPOP
if (dstkey == NULL) {
/* Propagate the [LR]POP operation. */
//根据where判断出是LPOP还是RPOP命令
argv[0] = (where == LIST_HEAD) ? shared.lpop :
shared.rpop;
//弹出元素的key
argv[1] = key;
//将[LR]POP命令传播到AOF和REPL
propagate((where == LIST_HEAD) ?
server.lpopCommand : server.rpopCommand,
db->id,argv,2,PROPAGATE_AOF|PROPAGATE_REPL);
/* BRPOP/BLPOP */
//发送回复信息
addReplyMultiBulkLen(receiver,2);
addReplyBulk(receiver,key);
addReplyBulk(receiver,value);
//dstkey不为空，执行BRPOPLPUSH命令
} else {
/* BRPOPLPUSH */
//以读操作将dstkey对象的值读出来
robj *dstobj =
lookupKeyWrite(receiver->db,dstkey);
//如果dstobj对象是列表类型，将BRPOPLPUSH命令分为RPOP和LPUSH分别处理
if (!(dstobj &&
checkType(receiver,dstobj,OBJ_LIST)))
{
/* Propagate the RPOP operation. */
//保存RPOP命令和被弹出元素的键
argv[0] = shared.rpop;
argv[1] = key;
//将RPOP命令传播到AOF和REPL
propagate(server.rpopCommand,
db->id,argv,2,
PROPAGATE_AOF|
PROPAGATE_REPL);
//将一个value推入到目标列表dstobj头部
rpoplpushHandlePush(receiver,dstkey,dstobj,
value);
/* Propagate the LPUSH operation. */
//保存LPOP命令和目标键和弹出元素的键
argv[0] = shared.lpush;
argv[1] = dstkey;
argv[2] = value;
//将LPUSH命令传播到AOF和REPL
propagate(server.lpushCommand,
db->id,argv,3,
PROPAGATE_AOF|
PROPAGATE_REPL);
} else {
/* BRPOPLPUSH failed because of wrong
* destination type. */
return C_ERR;
}
}
return C_OK;
}
//LRANGE key start stop
//LRANGE命令的实现
void lrangeCommand(client *c) {
robj *o;
long start, end, llen, rangelen;
//将字符串类型起始地址start和结束地址end转换为long类型保存在start和end中
//如果任意失败，则直接返回
if ((getLongFromObjectOrReply(c, c->argv[2], &start, NULL) != C_OK) ||
(getLongFromObjectOrReply(c, c->argv[3], &end, NULL) != C_OK)) return;
//以读操作取出key大小的value值，如果value对象不是列表类型，直接返回
if ((o = lookupKeyReadOrReply(c,c->argv[1],shared.emptymultibulk)) == NULL
|| checkType(c,o,OBJ_LIST)) return;
//获取列表元素数量
llen = listTypeLength(o);
/* convert negative indexes */
//将负数范围转换成合法范围
if (start < 0) start = llen+start;
if (end < 0) end = llen+end;
if (start < 0) start = 0;
/* Invariant: start >= 0, so this test will be true when end < 0.
* The range is empty when start > end or start >= length. */
//不合理的范围，发送空信息
if (start > end || start >= llen) {
addReply(c,shared.emptymultibulk);
return;
}
//end不能超过元素个数
if (end >= llen) end = llen-1;
rangelen = (end-start)+1;
/* Return the result in form of a multi-bulk reply */
//发送最后的范围值给client
addReplyMultiBulkLen(c,rangelen);
//只对编码为quicklist类型的value对象操作
if (o->encoding == OBJ_ENCODING_QUICKLIST) {
//创建迭代器，指向start起始的位置
listTypeIterator *iter = listTypeInitIterator(o, start, LIST_TAIL);
//遍历要找范围的大小次
while(rangelen--) {
listTypeEntry entry;
//保存当前指向的entry节点值到entry中，并且指向下一个entry节点
listTypeNext(iter, &entry);
quicklistEntry *qe = &entry.entry;
//若是是字符串类型的vlaue
if (qe->value) {
//发送字符串类型的值给client
addReplyBulkCBuffer(c,qe->value,qe->sz);
} else {
//否则，发送整型的值给client
addReplyBulkLongLong(c,qe->longval);
}
}
//释放迭代器
listTypeReleaseIterator(iter);
} else {
serverPanic("List encoding is not QUICKLIST!");
}
}
// BRPOP BLPOP 命令的底层实现
// BLPOP key [key ...] timeout
void blockingPopGenericCommand(client *c, int where) {
robj *o;
mstime_t timeout;
int j;
// 以秒为单位取出timeout值
if (getTimeoutFromObjectOrReply(c,c->argv[c->argc-1],&timeout,UNIT_SECONDS)
!= C_OK) return;
//遍历所有的key
for (j = 1; j < c->argc-1; j++) {
//以写操作取出当前key的值
o = lookupKeyWrite(c->db,c->argv[j]);
// value对象不为空
if (o != NULL) {
// 如果value对象的类型不是列表类型，发送类型错误信息
if (o->type != OBJ_LIST) {
addReply(c,shared.wrongtypeerr);
return;
} else {
// 列表长度不为0
if (listTypeLength(o) != 0) {
/* Non empty list, this is like a non normal [LR]POP. */
// 保存事件名称
char *event = (where == LIST_HEAD) ? "lpop" : "rpop";
// 保存弹出的value对象
robj *value = listTypePop(o,where);
serverAssert(value != NULL);
// 发送回复给client
addReplyMultiBulkLen(c,2);
addReplyBulk(c,c->argv[j]);
addReplyBulk(c,value);
// 释放value
decrRefCount(value);
// 发送事件通知
notifyKeyspaceEvent(NOTIFY_LIST,event,
c->argv[j],c->db->id);
//如果弹出元素后列表为空
if (listTypeLength(o) == 0) {
//从数据库中删除当前的key
dbDelete(c->db,c->argv[j]);
// 发送"del"的事件通知
notifyKeyspaceEvent(NOTIFY_GENERIC,"del",
c->argv[j],c->db->id);
}
//数据库的键被修改，发送信号
signalModifiedKey(c->db,c->argv[j]);
//更新脏键
server.dirty++;
/* Replicate it as an [LR]POP instead of B[LR]POP. */
// 传播一个[LR]POP 而不是B[LR]POP
rewriteClientCommandVector(c,2,
(where == LIST_HEAD) ? shared.lpop : shared.rpop,
c->argv[j]);
return;
}
}
}
}
/* If we are inside a MULTI/EXEC and the list is empty the only thing
* we can do is treating it as a timeout (even with timeout 0). */
// 如果命令在一个事务中执行，则发送一个空回复以避免死等待
if (c->flags & CLIENT_MULTI) {
addReply(c,shared.nullmultibulk);
return;
}
/* If the list is empty or the key does not exists we must block */
// 参数中的所有键都不存在，则阻塞这些键
blockForKeys(c, c->argv + 1, c->argc - 2, timeout, NULL);
}
void scriptCommand(redisClient *c)
{
if (c->argc == 2 && !strcasecmp(c->argv[1]->ptr,"flush"))
{
scriptingReset();
addReply(c,shared.ok);
server.dirty++; /* Replicating this command is a good idea. */
}
else if (c->argc >= 2 && !strcasecmp(c->argv[1]->ptr,"exists"))
{
int j;
addReplyMultiBulkLen(c, c->argc-2);
for (j = 2; j < c->argc; j++)
{
if (dictFind(server.lua_scripts,c->argv[j]->ptr))
addReply(c,shared.cone);
else
addReply(c,shared.czero);
}
}
else if (c->argc == 3 && !strcasecmp(c->argv[1]->ptr,"load"))
{
char funcname[43];
sds sha;
// 生成函数名
funcname[0] = 'f';
funcname[1] = '_';
// 计算脚本的 SHA1 值
sha1hex(funcname+2,c->argv[2]->ptr,sdslen(c->argv[2]->ptr));
sha = sdsnewlen(funcname+2,40);
// 根据 SHA1 值，在字典中查找脚本
if (dictFind(server.lua_scripts,sha) == NULL)
{
// 如果脚本未定义，在 Lua 中创建新函数
if (luaCreateFunction(c,server.lua,funcname,c->argv[2])
== REDIS_ERR)
{
sdsfree(sha);
return;
}
}
// 返回脚本的 SHA1 值
addReplyBulkCBuffer(c,funcname+2,40);
sdsfree(sha);
}
else if (c->argc == 2 && !strcasecmp(c->argv[1]->ptr,"kill"))
{
if (server.lua_caller == NULL)
{
addReplySds(c,sdsnew("-NOTBUSY No scripts in execution right now.\r\n"));
}
else if (server.lua_write_dirty)
{
addReplySds(c,sdsnew("-UNKILLABLE Sorry the script already executed write commands against the dataset. You can either wait the script termination or kill the server in an hard way using the SHUTDOWN NOSAVE command.\r\n"));
}
else
{
server.lua_kill = 1;
addReply(c,shared.ok);
}
}
else
{
addReplyError(c, "Unknown SCRIPT subcommand or wrong # of args.");
}
}
/* Blocking RPOP/LPOP */
void blockingPopGenericCommand(redisClient *c, int where) {
robj *o;
time_t timeout;
int j;
if (getTimeoutFromObjectOrReply(c,c->argv[c->argc-1],&timeout) != REDIS_OK)
return;
for (j = 1; j < c->argc-1; j++) {
o = lookupKeyWrite(c->db,c->argv[j]);
if (o != NULL) {
if (o->type != REDIS_LIST) {
addReply(c,shared.wrongtypeerr);
return;
} else {
if (listTypeLength(o) != 0) {
/* If the list contains elements fall back to the usual
* non-blocking POP operation */
struct redisCommand *orig_cmd;
robj *argv[2], **orig_argv;
int orig_argc;
/* We need to alter the command arguments before to call
* popGenericCommand() as the command takes a single key. */
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
argv[1] = c->argv[j];
c->argv = argv;
c->argc = 2;
/* Also the return value is different, we need to output
* the multi bulk reply header and the key name. The
* "real" command will add the last element (the value)
* for us. If this souds like an hack to you it's just
* because it is... */
addReplyMultiBulkLen(c,2);
addReplyBulk(c,argv[1]);
popGenericCommand(c,where);
/* Fix the client structure with the original stuff */
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
return;
}
}
}
}
/* If we are inside a MULTI/EXEC and the list is empty the only thing
* we can do is treating it as a timeout (even with timeout 0). */
if (c->flags & REDIS_MULTI) {
addReply(c,shared.nullmultibulk);
return;
}
/* If the list is empty or the key does not exists we must block */
blockForKeys(c, c->argv + 1, c->argc - 2, timeout, NULL);
}
// 执行事务内的所有命令
void execCommand(redisClient *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
int must_propagate = 0; /* Need to propagate MULTI/EXEC to AOF / slaves? */
// 必须设置多命令标记
if (!(c->flags & REDIS_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
// 停止执行事务命令的情况：
// 1. 被监视的数据被修改
// 2. 命令队列中的命令执行失败
/* Check if we need to abort the EXEC because:
* 1) Some WATCHed key was touched.
* 2) There was a previous error while queueing commands.
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned. */
if (c->flags & (REDIS_DIRTY_CAS|REDIS_DIRTY_EXEC)) {
addReply(c, c->flags & REDIS_DIRTY_EXEC ? shared.execaborterr :
shared.nullmultibulk);
discardTransaction(c);
goto handle_monitor;
}
// 执行队列中的所有命令
/* Exec all the queued commands */
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
// 保存当前的命令，一般为 MULTI，在执行完所有的命令后会恢复。
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count);
for (j = 0; j < c->mstate.count; j++) {
// 命令队列中的命令被赋值给当前的命令
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
// 遇到包含写操作的命令需要将 MULTI 命令写入 AOF 文件
/* Propagate a MULTI request once we encounter the first write op.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
if (!must_propagate && !(c->cmd->flags & REDIS_CMD_READONLY)) {
execCommandPropagateMulti(c);
must_propagate = 1;
}
// 调用 call() 执行
call(c,REDIS_CALL_FULL);
// 这几句是多余的
/* Commands may alter argc/argv, restore mstate. */
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
// 恢复当前的命令，一般为 MULTI
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
// 事务已经执行完毕，清理与此事务相关的信息，如命令队列和客户端标记
discardTransaction(c);
/* Make sure the EXEC command will be propagated as well if MULTI
* was already propagated. */
if (must_propagate) server.dirty++;
// 和监视器相关，后续提到
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as REDIS_CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering. */
if (listLength(server.monitors) && !server.loading)
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
// EXEC 命令实现
void execCommand(client *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
// 传播的标识
int must_propagate = 0; /* Need to propagate MULTI/EXEC to AOF / slaves? */
// 如果客户端当前不处于事务状态，回复错误后返回
if (!(c->flags & CLIENT_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC because:
* 1) Some WATCHed key was touched.
* 2) There was a previous error while queueing commands.
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned. */
// 检查是否需要中断EXEC的执行因为：
/*
1. 被监控的key被修改
2. 入队命令时发生了错误
*/
// 第一种情况返回空回复对象，第二种情况返回一个EXECABORT错误
// 如果客户的处于 1.命令入队时错误或者2.被监控的key被修改
if (c->flags & (CLIENT_DIRTY_CAS|CLIENT_DIRTY_EXEC)) {
// 回复错误信息
addReply(c, c->flags & CLIENT_DIRTY_EXEC ? shared.execaborterr :
shared.nullmultibulk);
// 取消事务
discardTransaction(c);
// 跳转到处理监控器代码
goto handle_monitor;
}
/* Exec all the queued commands */
// 执行队列数组中的命令
// 因为所有的命令都是安全的，因此取消对客户端的所有的键的监视
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
// 备份EXEC命令
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
// 回复一个事务命令的个数
addReplyMultiBulkLen(c,c->mstate.count);
// 遍历执行所有事务命令
for (j = 0; j < c->mstate.count; j++) {
// 设置一个当前事务命令给客户端
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
/* Propagate a MULTI request once we encounter the first write op.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
// 当执行到第一个写命令时，传播事务状态
if (!must_propagate && !(c->cmd->flags & CMD_READONLY)) {
// 发送一个MULTI命令给所有的从节点和AOF文件
execCommandPropagateMulti(c);
// 设置已经传播过的标识
must_propagate = 1;
}
// 执行该命令
call(c,CMD_CALL_FULL);
/* Commands may alter argc/argv, restore mstate. */
// 命令可能会被修改，重新存储在事务命令队列中
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
// 还原命令和参数
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
// 取消事务状态
discardTransaction(c);
/* Make sure the EXEC command will be propagated as well if MULTI
* was already propagated. */
// 如果传播了EXEC命令，表示执行了写命令，更新数据库脏键数
if (must_propagate) server.dirty++;
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering. */
// 如果服务器设置了监控器，并且服务器不处于载入文件的状态
if (listLength(server.monitors) && !server.loading)
// 将参数列表中的参数发送给监控器
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
//监视机制触发见touchWatchedKey决定是否触发REDIS_DIRTY_CAS 取消事物函数更具watch的键是否有触发REDIS_DIRTY_CAS来决定是否继续
//执行事物中的命令，见execCommand。//取消watch见unwatchAllKeys
void execCommand(redisClient *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
int must_propagate = 0; /* Need to propagate MULTI/EXEC to AOF / slaves? */
// 客户端没有执行事务
if (!(c->flags & REDIS_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC because:
*
* 检查是否需要阻止事务执行，因为：
*
* 1) Some WATCHed key was touched.
* 有被监视的键已经被修改了
*
* 2) There was a previous error while queueing commands.
* 命令在入队时发生错误
* （注意这个行为是 2.6.4 以后才修改的，之前是静默处理入队出错命令）
*
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned.
*
* 第一种情况返回多个批量回复的空对象
* 而第二种情况则返回一个 EXECABORT 错误
原子性:事务具有原子性指的是，数据库将事务中的多个操作当作一个整体来执行，服务器要么
就执行事务中的所有操作，要么就一个操作也不执行。
*/
if (c->flags & (REDIS_DIRTY_CAS|REDIS_DIRTY_EXEC)) {//入队命令中的任何一个出错，或者命令key被watch监视，并且该key在执行watch和exec命令之间发生了变化
addReply(c, c->flags & REDIS_DIRTY_EXEC ? shared.execaborterr :
shared.nullmultibulk);
// 取消事务
discardTransaction(c);
goto handle_monitor;
}
/* Exec all the queued commands */
// 已经可以保证安全性了，取消客户端对所有键的监视
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
// 因为事务中的命令在执行时可能会修改命令和命令的参数
// 所以为了正确地传播命令，需要现备份这些命令和参数
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count);
// 执行事务中的命令
for (j = 0; j < c->mstate.count; j++) {
// 因为 Redis 的命令必须在客户端的上下文中执行
// 所以要将事务队列中的命令、命令参数等设置给客户端
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
/* Propagate a MULTI request once we encounter the first write op.
*
* 当遇上第一个写命令时，传播 MULTI 命令。
*
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees.
*
* 这可以确保服务器和 AOF 文件以及附属节点的数据一致性。
*/
if (!must_propagate && !(c->cmd->flags & REDIS_CMD_READONLY)) {
// 传播 MULTI 命令
execCommandPropagateMulti(c);
// 计数器，只发送一次
must_propagate = 1;
}
// 执行命令
call(c,REDIS_CALL_FULL);
/* Commands may alter argc/argv, restore mstate. */
// 因为执行后命令、命令参数可能会被改变
// 比如 SPOP 会被改写为 SREM
// 所以这里需要更新事务队列中的命令和参数
// 确保附属节点和 AOF 的数据一致性
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
// 还原命令、命令参数
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
// 清理事务状态
discardTransaction(c);
/* Make sure the EXEC command will be propagated as well if MULTI
* was already propagated. */
// 将服务器设为脏，确保 EXEC 命令也会被传播
if (must_propagate) server.dirty++;
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as REDIS_CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering. */
if (listLength(server.monitors) && !server.loading)
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
/* The SORT command is the most complex command in Redis. Warning: this code
* is optimized for speed and a bit less for readability */
void sortCommand(redisClient *c) {
list *operations;
unsigned int outputlen = 0;
int desc = 0, alpha = 0;
long limit_start = 0, limit_count = -1, start, end;
int j, dontsort = 0, vectorlen;
int getop = 0; /* GET operation counter */
int int_convertion_error = 0;
robj *sortval, *sortby = NULL, *storekey = NULL;
redisSortObject *vector; /* Resulting vector to sort */
/* Lookup the key to sort. It must be of the right types */
sortval = lookupKeyRead(c->db,c->argv[1]);
if (sortval && sortval->type != REDIS_SET &&
sortval->type != REDIS_LIST &&
sortval->type != REDIS_ZSET)
{
addReply(c,shared.wrongtypeerr);
return;
}
/* Create a list of operations to perform for every sorted element.
* Operations can be GET/DEL/INCR/DECR */
operations = listCreate();
listSetFreeMethod(operations,zfree);
j = 2; /* options start at argv[2] */
/* Now we need to protect sortval incrementing its count, in the future
* SORT may have options able to overwrite/delete keys during the sorting
* and the sorted key itself may get destroyed */
if (sortval)
incrRefCount(sortval);
else
sortval = createListObject();
/* The SORT command has an SQL-alike syntax, parse it */
while(j < c->argc) {
int leftargs = c->argc-j-1;
if (!strcasecmp(c->argv[j]->ptr,"asc")) {
desc = 0;
} else if (!strcasecmp(c->argv[j]->ptr,"desc")) {
desc = 1;
} else if (!strcasecmp(c->argv[j]->ptr,"alpha")) {
alpha = 1;
} else if (!strcasecmp(c->argv[j]->ptr,"limit") && leftargs >= 2) {
if ((getLongFromObjectOrReply(c, c->argv[j+1], &limit_start, NULL) != REDIS_OK) ||
(getLongFromObjectOrReply(c, c->argv[j+2], &limit_count, NULL) != REDIS_OK)) return;
j+=2;
} else if (!strcasecmp(c->argv[j]->ptr,"store") && leftargs >= 1) {
storekey = c->argv[j+1];
j++;
} else if (!strcasecmp(c->argv[j]->ptr,"by") && leftargs >= 1) {
sortby = c->argv[j+1];
/* If the BY pattern does not contain '*', i.e. it is constant,
* we don't need to sort nor to lookup the weight keys. */
if (strchr(c->argv[j+1]->ptr,'*') == NULL) dontsort = 1;
j++;
} else if (!strcasecmp(c->argv[j]->ptr,"get") && leftargs >= 1) {
listAddNodeTail(operations,createSortOperation(
REDIS_SORT_GET,c->argv[j+1]));
getop++;
j++;
} else {
decrRefCount(sortval);
listRelease(operations);
addReply(c,shared.syntaxerr);
return;
}
j++;
}
/* For the STORE option, or when SORT is called from a Lua script,
* we want to force a specific ordering even when no explicit ordering
* was asked (SORT BY nosort). This guarantees that replication / AOF
* is deterministic.
*
* However in the case 'dontsort' is true, but the type to sort is a
* sorted set, we don't need to do anything as ordering is guaranteed
* in this special case. */
if ((storekey || c->flags & REDIS_LUA_CLIENT) &&
(dontsort && sortval->type != REDIS_ZSET))
{
/* Force ALPHA sorting */
dontsort = 0;
alpha = 1;
sortby = NULL;
}
/* Destructively convert encoded sorted sets for SORT. */
if (sortval->type == REDIS_ZSET)
zsetConvert(sortval, REDIS_ENCODING_SKIPLIST);
/* Objtain the length of the object to sort. */
switch(sortval->type) {
case REDIS_LIST: vectorlen = listTypeLength(sortval); break;
case REDIS_SET: vectorlen = setTypeSize(sortval); break;
case REDIS_ZSET: vectorlen = dictSize(((zset*)sortval->ptr)->dict); break;
default: vectorlen = 0; redisPanic("Bad SORT type"); /* Avoid GCC warning */
}
/* Perform LIMIT start,count sanity checking. */
start = (limit_start < 0) ? 0 : limit_start;
end = (limit_count < 0) ? vectorlen-1 : start+limit_count-1;
if (start >= vectorlen) {
start = vectorlen-1;
end = vectorlen-2;
}
if (end >= vectorlen) end = vectorlen-1;
/* Optimization:
*
* 1) if the object to sort is a sorted set.
* 2) There is nothing to sort as dontsort is true (BY <constant string>).
* 3) We have a LIMIT option that actually reduces the number of elements
* to fetch.
*
* In this case to load all the objects in the vector is a huge waste of
* resources. We just allocate a vector that is big enough for the selected
* range length, and make sure to load just this part in the vector. */
if (sortval->type == REDIS_ZSET &&
dontsort &&
(start != 0 || end != vectorlen-1))
{
vectorlen = end-start+1;
}
/* Load the sorting vector with all the objects to sort */
vector = zmalloc(sizeof(redisSortObject)*vectorlen);
j = 0;
if (sortval->type == REDIS_LIST) {
listTypeIterator *li = listTypeInitIterator(sortval,0,REDIS_TAIL);
listTypeEntry entry;
while(listTypeNext(li,&entry)) {
vector[j].obj = listTypeGet(&entry);
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
listTypeReleaseIterator(li);
} else if (sortval->type == REDIS_SET) {
setTypeIterator *si = setTypeInitIterator(sortval);
robj *ele;
while((ele = setTypeNextObject(si)) != NULL) {
vector[j].obj = ele;
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
setTypeReleaseIterator(si);
} else if (sortval->type == REDIS_ZSET && dontsort) {
/* Special handling for a sorted set, if 'dontsort' is true.
* This makes sure we return elements in the sorted set original
* ordering, accordingly to DESC / ASC options.
*
* Note that in this case we also handle LIMIT here in a direct
* way, just getting the required range, as an optimization. */
zset *zs = sortval->ptr;
zskiplist *zsl = zs->zsl;
zskiplistNode *ln;
robj *ele;
int rangelen = vectorlen;
/* Check if starting point is trivial, before doing log(N) lookup. */
if (desc) {
long zsetlen = dictSize(((zset*)sortval->ptr)->dict);
ln = zsl->tail;
if (start > 0)
ln = zslGetElementByRank(zsl,zsetlen-start);
} else {
ln = zsl->header->level[0].forward;
if (start > 0)
ln = zslGetElementByRank(zsl,start+1);
}
while(rangelen--) {
redisAssertWithInfo(c,sortval,ln != NULL);
ele = ln->obj;
vector[j].obj = ele;
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
ln = desc ? ln->backward : ln->level[0].forward;
}
/* The code producing the output does not know that in the case of
* sorted set, 'dontsort', and LIMIT, we are able to get just the
* range, already sorted, so we need to adjust "start" and "end"
* to make sure start is set to 0. */
end -= start;
start = 0;
} else if (sortval->type == REDIS_ZSET) {
dict *set = ((zset*)sortval->ptr)->dict;
dictIterator *di;
dictEntry *setele;
di = dictGetIterator(set);
while((setele = dictNext(di)) != NULL) {
vector[j].obj = dictGetKey(setele);
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
dictReleaseIterator(di);
} else {
redisPanic("Unknown type");
}
redisAssertWithInfo(c,sortval,j == vectorlen);
/* Now it's time to load the right scores in the sorting vector */
if (dontsort == 0) {
for (j = 0; j < vectorlen; j++) {
robj *byval;
if (sortby) {
/* lookup value to sort by */
byval = lookupKeyByPattern(c->db,sortby,vector[j].obj);
if (!byval) continue;
} else {
/* use object itself to sort by */
byval = vector[j].obj;
}
if (alpha) {
if (sortby) vector[j].u.cmpobj = getDecodedObject(byval);
} else {
if (byval->encoding == REDIS_ENCODING_RAW) {
char *eptr;
vector[j].u.score = strtod(byval->ptr,&eptr);
if (eptr[0] != '\0' || errno == ERANGE ||
isnan(vector[j].u.score))
{
int_convertion_error = 1;
}
} else if (byval->encoding == REDIS_ENCODING_INT) {
/* Don't need to decode the object if it's
* integer-encoded (the only encoding supported) so
* far. We can just cast it */
vector[j].u.score = (long)byval->ptr;
} else {
redisAssertWithInfo(c,sortval,1 != 1);
}
}
/* when the object was retrieved using lookupKeyByPattern,
* its refcount needs to be decreased. */
if (sortby) {
decrRefCount(byval);
}
}
}
if (dontsort == 0) {
server.sort_desc = desc;
server.sort_alpha = alpha;
server.sort_bypattern = sortby ? 1 : 0;
server.sort_store = storekey ? 1 : 0;
if (sortby && (start != 0 || end != vectorlen-1))
pqsort(vector,vectorlen,sizeof(redisSortObject),sortCompare, start,end);
else
qsort(vector,vectorlen,sizeof(redisSortObject),sortCompare);
}
/* Send command output to the output buffer, performing the specified
* GET/DEL/INCR/DECR operations if any. */
outputlen = getop ? getop*(end-start+1) : end-start+1;
if (int_convertion_error) {
addReplyError(c,"One or more scores can't be converted into double");
} else if (storekey == NULL) {
/* STORE option not specified, sent the sorting result to client */
addReplyMultiBulkLen(c,outputlen);
for (j = start; j <= end; j++) {
listNode *ln;
listIter li;
if (!getop) addReplyBulk(c,vector[j].obj);
listRewind(operations,&li);
while((ln = listNext(&li))) {
redisSortOperation *sop = ln->value;
robj *val = lookupKeyByPattern(c->db,sop->pattern,
vector[j].obj);
if (sop->type == REDIS_SORT_GET) {
if (!val) {
addReply(c,shared.nullbulk);
} else {
addReplyBulk(c,val);
decrRefCount(val);
}
} else {
/* Always fails */
redisAssertWithInfo(c,sortval,sop->type == REDIS_SORT_GET);
}
}
}
} else {
robj *sobj = createZiplistObject();
/* STORE option specified, set the sorting result as a List object */
for (j = start; j <= end; j++) {
listNode *ln;
listIter li;
if (!getop) {
listTypePush(sobj,vector[j].obj,REDIS_TAIL);
} else {
listRewind(operations,&li);
while((ln = listNext(&li))) {
redisSortOperation *sop = ln->value;
robj *val = lookupKeyByPattern(c->db,sop->pattern,
vector[j].obj);
if (sop->type == REDIS_SORT_GET) {
if (!val) val = createStringObject("",0);
/* listTypePush does an incrRefCount, so we should take care
* care of the incremented refcount caused by either
* lookupKeyByPattern or createStringObject("",0) */
listTypePush(sobj,val,REDIS_TAIL);
decrRefCount(val);
} else {
/* Always fails */
redisAssertWithInfo(c,sortval,sop->type == REDIS_SORT_GET);
}
}
}
}
if (outputlen) {
setKey(c->db,storekey,sobj);
notifyKeyspaceEvent(REDIS_NOTIFY_LIST,"sortstore",storekey,
c->db->id);
server.dirty += outputlen;
} else if (dbDelete(c->db,storekey)) {
signalModifiedKey(c->db,storekey);
notifyKeyspaceEvent(REDIS_NOTIFY_GENERIC,"del",storekey,c->db->id);
server.dirty++;
}
decrRefCount(sobj);
addReplyLongLong(c,outputlen);
}
/* Cleanup */
if (sortval->type == REDIS_LIST || sortval->type == REDIS_SET)
for (j = 0; j < vectorlen; j++)
decrRefCount(vector[j].obj);
decrRefCount(sortval);
listRelease(operations);
for (j = 0; j < vectorlen; j++) {
if (alpha && vector[j].u.cmpobj)
decrRefCount(vector[j].u.cmpobj);
}
zfree(vector);
}
void execCommand(redisClient *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
if (!(c->flags & REDIS_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC if some WATCHed key was touched.
* A failed EXEC will return a multi bulk nil object. */
if (c->flags & REDIS_DIRTY_CAS) {
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS);
unwatchAllKeys(c);
addReply(c,shared.nullmultibulk);
goto handle_monitor;
}
/* Replicate a MULTI request now that we are sure the block is executed.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
execCommandReplicateMulti(c);
/* Exec all the queued commands */
unwatchAllKeys(c); /* Unwatch ASAP otherwise we'll waste CPU cycles */
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count);
for (j = 0; j < c->mstate.count; j++) {
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
call(c,REDIS_CALL_FULL);
/* Commands may alter argc/argv, restore mstate. */
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
freeClientMultiState(c);
initClientMultiState(c);
c->flags &= ~(REDIS_MULTI|REDIS_DIRTY_CAS);
/* Make sure the EXEC command is always replicated / AOF, since we
* always send the MULTI command (we can't know beforehand if the
* next operations will contain at least a modification to the DB). */
server.dirty++;
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as REDIS_CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering. */
if (listLength(server.monitors) && !server.loading)
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum, robj *dstkey, int op) {
robj **sets = zmalloc(sizeof(robj*)*setnum);
setTypeIterator *si;
robj *ele, *dstset = NULL;
int j, cardinality = 0;
int diff_algo = 1;
for (j = 0; j < setnum; j++) {
robj *setobj = dstkey ?
lookupKeyWrite(c->db,setkeys[j]) :
lookupKeyRead(c->db,setkeys[j]);
if (!setobj) {
sets[j] = NULL;
continue;
}
if (checkType(c,setobj,REDIS_SET)) {
zfree(sets);
return;
}
sets[j] = setobj;
}
/* Select what DIFF algorithm to use.
*
* Algorithm 1 is O(N*M) where N is the size of the element first set
* and M the total number of sets.
*
* Algorithm 2 is O(N) where N is the total number of elements in all
* the sets.
*
* We compute what is the best bet with the current input here. */
if (op == REDIS_OP_DIFF && sets[0]) {
long long algo_one_work = 0, algo_two_work = 0;
for (j = 0; j < setnum; j++) {
if (sets[j] == NULL) continue;
algo_one_work += setTypeSize(sets[0]);
algo_two_work += setTypeSize(sets[j]);
}
/* Algorithm 1 has better constant times and performs less operations
* if there are elements in common. Give it some advantage. */
algo_one_work /= 2;
diff_algo = (algo_one_work <= algo_two_work) ? 1 : 2;
if (diff_algo == 1 && setnum > 1) {
/* With algorithm 1 it is better to order the sets to subtract
* by decreasing size, so that we are more likely to find
* duplicated elements ASAP. */
qsort(sets+1,setnum-1,sizeof(robj*),
qsortCompareSetsByRevCardinality);
}
}
/* We need a temp set object to store our union. If the dstkey
* is not NULL (that is, we are inside an SUNIONSTORE operation) then
* this set object will be the resulting object to set into the target key*/
dstset = createIntsetObject();
if (op == REDIS_OP_UNION) {
/* Union is trivial, just add every element of every set to the
* temporary set. */
for (j = 0; j < setnum; j++) {
if (!sets[j]) continue; /* non existing keys are like empty sets */
si = setTypeInitIterator(sets[j]);
while((ele = setTypeNextObject(si)) != NULL) {
if (setTypeAdd(dstset,ele)) cardinality++;
decrRefCount(ele);
}
setTypeReleaseIterator(si);
}
} else if (op == REDIS_OP_DIFF && sets[0] && diff_algo == 1) {
/* DIFF Algorithm 1:
*
* We perform the diff by iterating all the elements of the first set,
* and only adding it to the target set if the element does not exist
* into all the other sets.
*
* This way we perform at max N*M operations, where N is the size of
* the first set, and M the number of sets. */
si = setTypeInitIterator(sets[0]);
while((ele = setTypeNextObject(si)) != NULL) {
for (j = 1; j < setnum; j++) {
if (!sets[j]) continue; /* no key is an empty set. */
if (sets[j] == sets[0]) break; /* same set! */
if (setTypeIsMember(sets[j],ele)) break;
}
if (j == setnum) {
/* There is no other set with this element. Add it. */
setTypeAdd(dstset,ele);
cardinality++;
}
decrRefCount(ele);
}
setTypeReleaseIterator(si);
} else if (op == REDIS_OP_DIFF && sets[0] && diff_algo == 2) {
/* DIFF Algorithm 2:
*
* Add all the elements of the first set to the auxiliary set.
* Then remove all the elements of all the next sets from it.
*
* This is O(N) where N is the sum of all the elements in every
* set. */
for (j = 0; j < setnum; j++) {
if (!sets[j]) continue; /* non existing keys are like empty sets */
si = setTypeInitIterator(sets[j]);
while((ele = setTypeNextObject(si)) != NULL) {
if (j == 0) {
if (setTypeAdd(dstset,ele)) cardinality++;
} else {
if (setTypeRemove(dstset,ele)) cardinality--;
}
decrRefCount(ele);
}
setTypeReleaseIterator(si);
/* Exit if result set is empty as any additional removal
* of elements will have no effect. */
if (cardinality == 0) break;
}
}
/* Output the content of the resulting set, if not in STORE mode */
if (!dstkey) {
addReplyMultiBulkLen(c,cardinality);
si = setTypeInitIterator(dstset);
while((ele = setTypeNextObject(si)) != NULL) {
addReplyBulk(c,ele);
decrRefCount(ele);
}
setTypeReleaseIterator(si);
decrRefCount(dstset);
} else {
/* If we have a target key where to store the resulting set
* create this key with the result set inside */
int deleted = dbDelete(c->db,dstkey);
if (setTypeSize(dstset) > 0) {
dbAdd(c->db,dstkey,dstset);
addReplyLongLong(c,setTypeSize(dstset));
notifyKeyspaceEvent(REDIS_NOTIFY_SET,
op == REDIS_OP_UNION ? "sunionstore" : "sdiffstore",
dstkey,c->db->id);
} else {
decrRefCount(dstset);
addReply(c,shared.czero);
if (deleted)
notifyKeyspaceEvent(REDIS_NOTIFY_GENERIC,"del",
dstkey,c->db->id);
}
signalModifiedKey(c->db,dstkey);
server.dirty++;
}
zfree(sets);
}
/* Returns ID(s) of documents added */
void jsondocsetbyjsonCommand(redisClient *c) {
/* args 0-N: ["jsondocsetbyjson", id-field-name, json] */
jsonSyncClients(c);
sds field_name = c->argv[1]->ptr;
struct jsonObj *root = yajl_decode(c->argv[2]->ptr, NULL);
if (!root) {
addReply(c, g.err_parse);
return;
}
robj *key;
if (root->type == JSON_TYPE_LIST) {
robj *keys[root->content.obj.elements];
D("Procesing list!\n");
/* First, process all documents for names to make sure they are valid
* documents. We don't want to add half the documents then reach a
* failure scenario. */
for (int i = 0; i < root->content.obj.elements; i++) {
struct jsonObj *o = root->content.obj.fields[i];
key = jsonObjFindId(field_name, o);
keys[i] = key;
if (!key) {
/* Free any allocated keys so far */
for (int j = 0; i < j; j++)
decrRefCount(keys[j]);
jsonObjFree(root);
addReplyErrorFormat(
c,
"field '%s' not found or unusable as key for document %d",
field_name, i);
return;
}
}
/* Now actually add all the documents */
/* Note how a multi-set gets a multibulk reply while
* a regular one-document set gets just one bulk result. */
addReplyMultiBulkLen(c, root->content.obj.elements);
for (int i = 0; i < root->content.obj.elements; i++) {
struct jsonObj *o = root->content.obj.fields[i];
jsonObjAddToDB(keys[i], o);
addReplyBulkCBuffer(c, keys[i]->ptr, sdslen(keys[i]->ptr));
decrRefCount(keys[i]);
}
} else if (root->type == JSON_TYPE_MAP) {
key = jsonObjFindId(field_name, root);
if (key) {
jsonObjAddToDB(key, root);
addReplyBulkCBuffer(c, key->ptr, sdslen(key->ptr));
decrRefCount(key);
} else {
addReplyErrorFormat(
c, "field '%s' not found or unusable as key for document",
field_name);
}
} else {
addReplyError(c, "JSON isn't map or array of maps.");
}
jsonObjFree(root);
}
void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum, robj *dstkey, int op) {
robj **sets = zmalloc(sizeof(robj*)*setnum);
setTypeIterator *si;
robj *ele, *dstset = NULL;
int j, cardinality = 0;
for (j = 0; j < setnum; j++) {
robj *setobj = dstkey ?
lookupKeyWrite(c->db,setkeys[j]) :
lookupKeyRead(c->db,setkeys[j]);
if (!setobj) {
sets[j] = NULL;
continue;
}
if (checkType(c,setobj,REDIS_SET)) {
zfree(sets);
return;
}
sets[j] = setobj;
}
/* We need a temp set object to store our union. If the dstkey
* is not NULL (that is, we are inside an SUNIONSTORE operation) then
* this set object will be the resulting object to set into the target key*/
dstset = createIntsetObject();
/* Iterate all the elements of all the sets, add every element a single
* time to the result set */
for (j = 0; j < setnum; j++) {
if (op == REDIS_OP_DIFF && j == 0 && !sets[j]) break; /* result set is empty */
if (!sets[j]) continue; /* non existing keys are like empty sets */
si = setTypeInitIterator(sets[j]);
while((ele = setTypeNextObject(si)) != NULL) {
if (op == REDIS_OP_UNION || j == 0) {
if (setTypeAdd(dstset,ele)) {
cardinality++;
}
} else if (op == REDIS_OP_DIFF) {
if (setTypeRemove(dstset,ele)) {
cardinality--;
}
}
decrRefCount(ele);
}
setTypeReleaseIterator(si);
/* Exit when result set is empty. */
if (op == REDIS_OP_DIFF && cardinality == 0) break;
}
/* Output the content of the resulting set, if not in STORE mode */
if (!dstkey) {
addReplyMultiBulkLen(c,cardinality);
si = setTypeInitIterator(dstset);
while((ele = setTypeNextObject(si)) != NULL) {
addReplyBulk(c,ele);
decrRefCount(ele);
}
setTypeReleaseIterator(si);
decrRefCount(dstset);
} else {
/* If we have a target key where to store the resulting set
* create this key with the result set inside */
dbDelete(c->db,dstkey);
if (setTypeSize(dstset) > 0) {
dbAdd(c->db,dstkey,dstset);
addReplyLongLong(c,setTypeSize(dstset));
} else {
decrRefCount(dstset);
addReply(c,shared.czero);
}
touchWatchedKey(c->db,dstkey);
server.dirty++;
}
zfree(sets);
}
/* 对应事务指令exec, 执行已经设置的事务指令 */
void execCommand(client *c) {
int j;
robj **orig_argv;
int orig_argc;
struct redisCommand *orig_cmd;
int must_propagate = 0; /* Need to propagate MULTI/EXEC to AOF / slaves? */
if (!(c->flags & CLIENT_MULTI)) {
addReplyError(c,"EXEC without MULTI");
return;
}
/* Check if we need to abort the EXEC because:
* 1) Some WATCHed key was touched.
* 2) There was a previous error while queueing commands.
* A failed EXEC in the first case returns a multi bulk nil object
* (technically it is not an error but a special behavior), while
* in the second an EXECABORT error is returned.
*
* 检查是否需要放弃事务指令集
* 1) 触发了一些WATCH的键
* 2) 存储事务指令时发生了错误 */
if (c->flags & (CLIENT_DIRTY_CAS|CLIENT_DIRTY_EXEC)) {
addReply(c, c->flags & CLIENT_DIRTY_EXEC ? shared.execaborterr :
shared.nullmultibulk);
discardTransaction(c); /* 放弃事务 */
goto handle_monitor;
}
/* 逐个执行所有的事务指令集, Exec all the queued commands */
unwatchAllKeys(c); /* 执行事务前释放监控的键 */
orig_argv = c->argv;
orig_argc = c->argc;
orig_cmd = c->cmd;
addReplyMultiBulkLen(c,c->mstate.count); /* 多应答块儿个数 */
for (j = 0; j < c->mstate.count; j++) {
c->argc = c->mstate.commands[j].argc;
c->argv = c->mstate.commands[j].argv;
c->cmd = c->mstate.commands[j].cmd;
/* Propagate a MULTI request once we encounter the first write op.
* This way we'll deliver the MULTI/..../EXEC block as a whole and
* both the AOF and the replication link will have the same consistency
* and atomicity guarantees. */
if (!must_propagate && !(c->cmd->flags & CMD_READONLY)) {
execCommandPropagateMulti(c);
must_propagate = 1;
}
/* 执行事务指令 */
call(c,CMD_CALL_FULL);
/* Commands may alter argc/argv, restore mstate. */
c->mstate.commands[j].argc = c->argc;
c->mstate.commands[j].argv = c->argv;
c->mstate.commands[j].cmd = c->cmd;
}
c->argv = orig_argv;
c->argc = orig_argc;
c->cmd = orig_cmd;
discardTransaction(c);
/* Make sure the EXEC command will be propagated as well if MULTI
* was already propagated. */
if (must_propagate) server.dirty++;
handle_monitor:
/* Send EXEC to clients waiting data from MONITOR. We do it here
* since the natural order of commands execution is actually:
* MUTLI, EXEC, ... commands inside transaction ...
* Instead EXEC is flagged as CMD_SKIP_MONITOR in the command
* table, and we do it here with correct ordering.
*
* 向监控客户端发送执行的指令, 因为multi和exec之间的部分仅仅暂存指令,
* 而没有执行, 因此此处补发指令的执行 */
if (listLength(server.monitors) && !server.loading)
replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
}
/* The SORT command is the most complex command in Redis. Warning: this code
* is optimized for speed and a bit less for readability */
void sortCommand(redisClient *c) {
list *operations;
unsigned int outputlen = 0;
int desc = 0, alpha = 0;
long limit_start = 0, limit_count = -1, start, end;
int j, dontsort = 0, vectorlen;
int getop = 0; /* GET operation counter */
int int_convertion_error = 0;
robj *sortval, *sortby = NULL, *storekey = NULL;
redisSortObject *vector; /* Resulting vector to sort */
/* Lookup the key to sort. It must be of the right types */
sortval = lookupKeyRead(c->db,c->argv[1]);
if (sortval && sortval->type != REDIS_SET && sortval->type != REDIS_LIST &&
sortval->type != REDIS_ZSET)
{
addReply(c,shared.wrongtypeerr);
return;
}
/* Create a list of operations to perform for every sorted element.
* Operations can be GET/DEL/INCR/DECR */
operations = listCreate();
listSetFreeMethod(operations,zfree);
j = 2;
/* Now we need to protect sortval incrementing its count, in the future
* SORT may have options able to overwrite/delete keys during the sorting
* and the sorted key itself may get destroied */
if (sortval)
incrRefCount(sortval);
else
sortval = createListObject();
/* The SORT command has an SQL-alike syntax, parse it */
while(j < c->argc) {
int leftargs = c->argc-j-1;
if (!strcasecmp(c->argv[j]->ptr,"asc")) {
desc = 0;
} else if (!strcasecmp(c->argv[j]->ptr,"desc")) {
desc = 1;
} else if (!strcasecmp(c->argv[j]->ptr,"alpha")) {
alpha = 1;
} else if (!strcasecmp(c->argv[j]->ptr,"limit") && leftargs >= 2) {
if ((getLongFromObjectOrReply(c, c->argv[j+1], &limit_start, NULL) != REDIS_OK) ||
(getLongFromObjectOrReply(c, c->argv[j+2], &limit_count, NULL) != REDIS_OK)) return;
j+=2;
} else if (!strcasecmp(c->argv[j]->ptr,"store") && leftargs >= 1) {
storekey = c->argv[j+1];
j++;
} else if (!strcasecmp(c->argv[j]->ptr,"by") && leftargs >= 1) {
sortby = c->argv[j+1];
/* If the BY pattern does not contain '*', i.e. it is constant,
* we don't need to sort nor to lookup the weight keys. */
if (strchr(c->argv[j+1]->ptr,'*') == NULL) dontsort = 1;
j++;
} else if (!strcasecmp(c->argv[j]->ptr,"get") && leftargs >= 1) {
listAddNodeTail(operations,createSortOperation(
REDIS_SORT_GET,c->argv[j+1]));
getop++;
j++;
} else {
decrRefCount(sortval);
listRelease(operations);
addReply(c,shared.syntaxerr);
return;
}
j++;
}
/* If we have STORE we need to force sorting for deterministic output
* and replication. We use alpha sorting since this is guaranteed to
* work with any input. */
if (storekey && dontsort) {
dontsort = 0;
alpha = 1;
sortby = NULL;
}
/* Destructively convert encoded sorted sets for SORT. */
if (sortval->type == REDIS_ZSET)
zsetConvert(sortval, REDIS_ENCODING_SKIPLIST);
/* Load the sorting vector with all the objects to sort */
switch(sortval->type) {
case REDIS_LIST: vectorlen = listTypeLength(sortval); break;
case REDIS_SET: vectorlen = setTypeSize(sortval); break;
case REDIS_ZSET: vectorlen = dictSize(((zset*)sortval->ptr)->dict); break;
default: vectorlen = 0; redisPanic("Bad SORT type"); /* Avoid GCC warning */
}
vector = zmalloc(sizeof(redisSortObject)*vectorlen);
j = 0;
if (sortval->type == REDIS_LIST) {
listTypeIterator *li = listTypeInitIterator(sortval,0,REDIS_TAIL);
listTypeEntry entry;
while(listTypeNext(li,&entry)) {
vector[j].obj = listTypeGet(&entry);
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
listTypeReleaseIterator(li);
} else if (sortval->type == REDIS_SET) {
setTypeIterator *si = setTypeInitIterator(sortval);
robj *ele;
while((ele = setTypeNextObject(si)) != NULL) {
vector[j].obj = ele;
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
setTypeReleaseIterator(si);
} else if (sortval->type == REDIS_ZSET) {
dict *set = ((zset*)sortval->ptr)->dict;
dictIterator *di;
dictEntry *setele;
di = dictGetIterator(set);
while((setele = dictNext(di)) != NULL) {
vector[j].obj = dictGetKey(setele);
vector[j].u.score = 0;
vector[j].u.cmpobj = NULL;
j++;
}
dictReleaseIterator(di);
} else {
redisPanic("Unknown type");
}
redisAssertWithInfo(c,sortval,j == vectorlen);
/* Now it's time to load the right scores in the sorting vector */
if (dontsort == 0) {
for (j = 0; j < vectorlen; j++) {
robj *byval;
if (sortby) {
/* lookup value to sort by */
byval = lookupKeyByPattern(c->db,sortby,vector[j].obj);
if (!byval) continue;
} else {
/* use object itself to sort by */
byval = vector[j].obj;
}
if (alpha) {
if (sortby) vector[j].u.cmpobj = getDecodedObject(byval);
} else {
if (byval->encoding == REDIS_ENCODING_RAW) {
char *eptr;
vector[j].u.score = strtod(byval->ptr,&eptr);
if (eptr[0] != '\0' || errno == ERANGE ||
isnan(vector[j].u.score))
{
int_convertion_error = 1;
}
} else if (byval->encoding == REDIS_ENCODING_INT) {
/* Don't need to decode the object if it's
* integer-encoded (the only encoding supported) so
* far. We can just cast it */
vector[j].u.score = (long)byval->ptr;
} else {
redisAssertWithInfo(c,sortval,1 != 1);
}
}
/* when the object was retrieved using lookupKeyByPattern,
* its refcount needs to be decreased. */
if (sortby) {
decrRefCount(byval);
}
}
}
/* We are ready to sort the vector... perform a bit of sanity check
* on the LIMIT option too. We'll use a partial version of quicksort. */
start = (limit_start < 0) ? 0 : limit_start;
end = (limit_count < 0) ? vectorlen-1 : start+limit_count-1;
if (start >= vectorlen) {
start = vectorlen-1;
end = vectorlen-2;
}
if (end >= vectorlen) end = vectorlen-1;
server.sort_dontsort = dontsort;
if (dontsort == 0) {
server.sort_desc = desc;
server.sort_alpha = alpha;
server.sort_bypattern = sortby ? 1 : 0;
if (sortby && (start != 0 || end != vectorlen-1))
pqsort(vector,vectorlen,sizeof(redisSortObject),sortCompare, start,end);
else
qsort(vector,vectorlen,sizeof(redisSortObject),sortCompare);
}
/* Send command output to the output buffer, performing the specified
* GET/DEL/INCR/DECR operations if any. */
outputlen = getop ? getop*(end-start+1) : end-start+1;
if (int_convertion_error) {
addReplyError(c,"One or more scores can't be converted into double");
} else if (storekey == NULL) {
/* STORE option not specified, sent the sorting result to client */
addReplyMultiBulkLen(c,outputlen);
for (j = start; j <= end; j++) {
listNode *ln;
listIter li;
if (!getop) addReplyBulk(c,vector[j].obj);
listRewind(operations,&li);
while((ln = listNext(&li))) {
redisSortOperation *sop = ln->value;
robj *val = lookupKeyByPattern(c->db,sop->pattern,
vector[j].obj);
if (sop->type == REDIS_SORT_GET) {
if (!val) {
addReply(c,shared.nullbulk);
} else {
addReplyBulk(c,val);
decrRefCount(val);
}
} else {
/* Always fails */
redisAssertWithInfo(c,sortval,sop->type == REDIS_SORT_GET);
}
}
}
} else {
robj *sobj = createZiplistObject();
/* STORE option specified, set the sorting result as a List object */
for (j = start; j <= end; j++) {
listNode *ln;
listIter li;
if (!getop) {
listTypePush(sobj,vector[j].obj,REDIS_TAIL);
} else {
listRewind(operations,&li);
while((ln = listNext(&li))) {
redisSortOperation *sop = ln->value;
robj *val = lookupKeyByPattern(c->db,sop->pattern,
vector[j].obj);
if (sop->type == REDIS_SORT_GET) {
if (!val) val = createStringObject("",0);
/* listTypePush does an incrRefCount, so we should take care
* care of the incremented refcount caused by either
* lookupKeyByPattern or createStringObject("",0) */
listTypePush(sobj,val,REDIS_TAIL);
decrRefCount(val);
} else {
/* Always fails */
redisAssertWithInfo(c,sortval,sop->type == REDIS_SORT_GET);
}
}
}
}
if (outputlen) {
setKey(c->db,storekey,sobj);
server.dirty += outputlen;
} else if (dbDelete(c->db,storekey)) {
signalModifiedKey(c->db,storekey);
server.dirty++;
}
decrRefCount(sobj);
addReplyLongLong(c,outputlen);
}
/* Cleanup */
if (sortval->type == REDIS_LIST || sortval->type == REDIS_SET)
for (j = 0; j < vectorlen; j++)
decrRefCount(vector[j].obj);
decrRefCount(sortval);
listRelease(operations);
for (j = 0; j < vectorlen; j++) {
if (alpha && vector[j].u.cmpobj)
decrRefCount(vector[j].u.cmpobj);
}
zfree(vector);
}
/* This function should be called by Redis every time a single command,
* a MULTI/EXEC block, or a Lua script, terminated its execution after
* being called by a client.
*
* All the keys with at least one client blocked that received at least
* one new element via some PUSH/XADD operation are accumulated into
* the server.ready_keys list. This function will run the list and will
* serve clients accordingly. Note that the function will iterate again and
* again as a result of serving BRPOPLPUSH we can have new blocking clients
* to serve because of the PUSH side of BRPOPLPUSH. */
void handleClientsBlockedOnKeys(void) {
while(listLength(server.ready_keys) != 0) {
list *l;
/* Point server.ready_keys to a fresh list and save the current one
* locally. This way as we run the old list we are free to call
* signalKeyAsReady() that may push new elements in server.ready_keys
* when handling clients blocked into BRPOPLPUSH. */
l = server.ready_keys;
server.ready_keys = listCreate();
while(listLength(l) != 0) {
listNode *ln = listFirst(l);
readyList *rl = ln->value;
/* First of all remove this key from db->ready_keys so that
* we can safely call signalKeyAsReady() against this key. */
dictDelete(rl->db->ready_keys,rl->key);
/* Serve clients blocked on list key. */
robj *o = lookupKeyWrite(rl->db,rl->key);
if (o != NULL && o->type == OBJ_LIST) {
dictEntry *de;
/* We serve clients in the same order they blocked for
* this key, from the first blocked to the last. */
de = dictFind(rl->db->blocking_keys,rl->key);
if (de) {
list *clients = dictGetVal(de);
int numclients = listLength(clients);
while(numclients--) {
listNode *clientnode = listFirst(clients);
client *receiver = clientnode->value;
if (receiver->btype != BLOCKED_LIST) {
/* Put on the tail, so that at the next call
* we'll not run into it again. */
listDelNode(clients,clientnode);
listAddNodeTail(clients,receiver);
continue;
}
robj *dstkey = receiver->bpop.target;
int where = (receiver->lastcmd &&
receiver->lastcmd->proc == blpopCommand) ?
LIST_HEAD : LIST_TAIL;
robj *value = listTypePop(o,where);
if (value) {
/* Protect receiver->bpop.target, that will be
* freed by the next unblockClient()
* call. */
if (dstkey) incrRefCount(dstkey);
unblockClient(receiver);
if (serveClientBlockedOnList(receiver,
rl->key,dstkey,rl->db,value,
where) == C_ERR)
{
/* If we failed serving the client we need
* to also undo the POP operation. */
listTypePush(o,value,where);
}
if (dstkey) decrRefCount(dstkey);
decrRefCount(value);
} else {
break;
}
}
}
if (listTypeLength(o) == 0) {
dbDelete(rl->db,rl->key);
notifyKeyspaceEvent(NOTIFY_GENERIC,"del",rl->key,rl->db->id);
}
/* We don't call signalModifiedKey() as it was already called
* when an element was pushed on the list. */
}
/* Serve clients blocked on stream key. */
else if (o != NULL && o->type == OBJ_STREAM) {
dictEntry *de = dictFind(rl->db->blocking_keys,rl->key);
stream *s = o->ptr;
/* We need to provide the new data arrived on the stream
* to all the clients that are waiting for an offset smaller
* than the current top item. */
if (de) {
list *clients = dictGetVal(de);
listNode *ln;
listIter li;
listRewind(clients,&li);
while((ln = listNext(&li))) {
client *receiver = listNodeValue(ln);
if (receiver->btype != BLOCKED_STREAM) continue;
streamID *gt = dictFetchValue(receiver->bpop.keys,
rl->key);
if (s->last_id.ms > gt->ms ||
(s->last_id.ms == gt->ms &&
s->last_id.seq > gt->seq))
{
streamID start = *gt;
start.seq++; /* Can't overflow, it's an uint64_t */
/* If we blocked in the context of a consumer
* group, we need to resolve the group and
* consumer here. */
streamCG *group = NULL;
streamConsumer *consumer = NULL;
if (receiver->bpop.xread_group) {
group = streamLookupCG(s,
receiver->bpop.xread_group->ptr);
/* In theory if the group is not found we
* just perform the read without the group,
* but actually when the group, or the key
* itself is deleted (triggering the removal
* of the group), we check for blocked clients
* and send them an error. */
}
if (group) {
consumer = streamLookupConsumer(group,
receiver->bpop.xread_consumer->ptr,
1);
}
/* Note that after we unblock the client, 'gt'
* and other receiver->bpop stuff are no longer
* valid, so we must do the setup above before
* this call. */
unblockClient(receiver);
/* Emit the two elements sub-array consisting of
* the name of the stream and the data we
* extracted from it. Wrapped in a single-item
* array, since we have just one key. */
addReplyMultiBulkLen(receiver,1);
addReplyMultiBulkLen(receiver,2);
addReplyBulk(receiver,rl->key);
streamPropInfo pi = {
rl->key,
receiver->bpop.xread_group
};
streamReplyWithRange(receiver,s,&start,NULL,
receiver->bpop.xread_count,
0, group, consumer, 0, &pi);
}
}
}
}
/* Free this item. */
decrRefCount(rl->key);
zfree(rl);
listDelNode(l,ln);
}
listRelease(l); /* We have the new list on place at this point. */
}
}
