系列目录
redis源码分析序
（一）Redis结构解析
（二）结构体分析
（三）dict哈希结构1
（三）dict哈希结构2
（三）dict哈希结构3
（三）dict哈希结构4
（四）sds字符串
（五）sparkline微线图
（六）ziplist压缩列表
（七）zipmap压缩图
（八）t_hash哈希转换
（九）t_list，t_string的分析
（十）testhelp.h小型测试框架和redis-check-aof.c日志检测
（十一）memtest内存检测
（十二）redis-check-dump本地数据库检测
（十三）redis-benchmark性能测试
（十四）rdb.c本地数据库操
（十五）aof-append only file解析
（十六）config配置文件
（十七）multi事务操作
（十八）db.c内存数据库操作
（十九）replication主从数据复制的实现
（二十） ae事件驱动
（二十一）anet网络通信的封装
（二十二）networking网络协议传输
（二十三）CRC循环冗余算法和RAND随机数算法
（二十四）tool工具类
（二十五）zmalloc内存分配实现
（二十六）slowLog和hyperloglog
（二十七）rio系统I/O的封装
（二十八）object创建和释放redisObject对象
（二十九）bio后台I/O服务的实现
（三十） pubsub发布订阅模式
（三十一）latency延迟分析处理
（三十二）redis-cli.c客户端命令行接口的实现(1)
（三十三）redis-cli.c客户端命令行接口的实现(2)
（三十四）redis.h服务端的实现分析（1）
（三十五）redis.c服务端的实现分析（2）
（三十六）Redis中的优秀设计之处
如果有看过之前我分析的ziplist压缩列表的分析的话，理解这个我觉得不是什么特别的难题。ziplist压缩列表和zipmap都采用了动态分配字节的做法表示长度，比如通过固定的字节表示节省了不少的空间。同样带来的问题就是复杂的指针移动，和字符位置移动。但总的来说，一定是利大于弊了，要不然设计者也不会这么做。ziplist保存的使用一个列表，zipmap就保存的则是一个个键值对，通过key:value key:value的形式连着。下面我给出zipmap的结构构成，zipmap其实也就是一个超级长的字符串。
里面涉及了几个变量zmlen，len,free，下面给出完整的解释：
1/* String -> String Map data structure optimized for size.
2 * This file implements a data structure mapping strings to other strings
3 * implementing an O(n) lookup data structure designed to be very memory
4 * efficient.
5 *
6 * The Redis Hash type uses this data structure for hashes composed of a small
7 * number of elements, to switch to a hash table once a given number of
8 * elements is reached.
9 *
10 * Given that many times Redis Hashes are used to represent objects composed
11 * of few fields, this is a very big win in terms of used memory.
12 *
13 * zipmap压缩表和ziplist十分类似，都做到了内存操作效率比较高的
14 * --------------------------------------------------------------------------
15 *
16 * Copyright (c) 2009-2010, Salvatore Sanfilippo
17 * All rights reserved.
18 *
19 * Redistribution and use in source and binary forms, with or without
20 * modification, are permitted provided that the following conditions are met:
21 *
22 * * Redistributions of source code must retain the above copyright notice,
23 * this list of conditions and the following disclaimer.
24 * * Redistributions in binary form must reproduce the above copyright
25 * notice, this list of conditions and the following disclaimer in the
26 * documentation and/or other materials provided with the distribution.
27 * * Neither the name of Redis nor the names of its contributors may be used
28 * to endorse or promote products derived from this software without
29 * specific prior written permission.
30 *
31 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 'AS IS'
32 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
33 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
34 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
35 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
36 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
37 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
38 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
39 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
40 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
41 * POSSIBILITY OF SUCH DAMAGE.
42 */
1/* Memory layout of a zipmap, for the map 'foo' => 'bar', 'hello' => 'world':
2 *
3 * 'foo''bar''hello''world'
4 *
5 * is 1 byte length that holds the current size of the zipmap.
6 * When the zipmap length is greater than or equal to 254, this value
7 * is not used and the zipmap needs to be traversed to find out the length.
8 * 占有着1个字节，所以他的最多可代表的数量是254,当zipmap中的元素记录超过这个数时，
9 * 那只能从前往后后遍历算大小了，和ziplist是不一样的。
10 *
11 * is the length of the following string (key or value).
12 * lengths are encoded in a single value or in a 5 bytes value.
13 * If the first byte value (as an unsigned 8 bit value) is between 0 and
14 * 252, it's a single-byte length. If it is 253 then a four bytes unsigned
15 * integer follows (in the host byte ordering). A value of 255 is used to
16 * signal the end of the hash. The special value 254 is used to mark
17 * empty space that can be used to add new key/value pairs.
18 * 代表了后面字符串key 或 value的值的长度，长度一般被编码1个字节或5个字节表示，这个和ziplist类似
19 * 如果后面的字符串长度小于等于252个，可与用单字节表示，其他253,254等长度被用来表示其他作用了，当超过这个数时候
20 * 则直接按5字节的方式存储长度。
21 *
22 * is the number of free unused bytes after the string, resulting
23 * from modification of values associated to a key. For instance if 'foo'
24 * is set to 'bar', and later 'foo' will be set to 'hi', it will have a
25 * free byte to use if the value will enlarge again later, or even in
26 * order to add a key/value pair if it fits.
27 * 一般来表示后面的value长度的空闲值，当key：value=“foo”:'bar',后来被改为“foo”:'hi'，空闲长度就为1了
28 *
29 * is always an unsigned 8 bit number, because if after an
30 * update operation there are more than a few free bytes, the zipmap will be
31 * reallocated to make sure it is as small as possible.
32 * 的数字一般比较小，如果空闲太大，zipmap会进行调整大小使map整体变得尽可能小
33 *
34 * The most compact representation of the above two elements hash is actually:
35 * 这是一个例子：
36 * 'foo''bar''hello''world'
37 * <总键值对数><第一个key的长度>key字符<第一个value的长度><空闲长度开始都为0>后面同前
38 * '\x02\x03foo\x03\x00bar\x05hello\x05\x00world\xff'
39 *
40 * Note that because keys and values are prefixed length 'objects',
41 * the lookup will take O(N) where N is the number of elements
42 * in the zipmap and *not* the number of bytes needed to represent the zipmap.
43 * This lowers the constant times considerably.
44 */
说到键值对，里面最最重要的方法当然是根据key ，setValue的方法了，方法如下：
1/* Set key to value, creating the key if it does not already exist.
2 * If 'update' is not NULL, *update is set to 1 if the key was
3 * already preset, otherwise to 0. */
4unsigned char *zipmapSet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char *val, unsigned int vlen, int *update) {
5 unsigned int zmlen, offset;
6 unsigned int freelen, reqlen = zipmapRequiredLength(klen,vlen);
7 unsigned int empty, vempty;
8 unsigned char *p;
9
10 freelen = reqlen;
11 if (update) *update = 0;
12 //寻找key的位置
13 p = zipmapLookupRaw(zm,key,klen,&zmlen);
14 if (p == NULL) {
15 /* Key not found: enlarge */
16 //key的位置没有找到，调整zipmap的大小，准备添加操作
17 zm = zipmapResize(zm, zmlen+reqlen);
18 p = zm+zmlen-1;
19 zmlen = zmlen+reqlen;
20
21 /* Increase zipmap length (this is an insert) */
22 //如果头字节还没有达到最大值，则递增
23 if (zm[0] < ZIPMAP_BIGLEN) zm[0]++;
24 } else {
25 /* Key found. Is there enough space for the new value? */
26 /* Compute the total length: */
27 if (update) *update = 1;
28 //key的位置以及找到，判断是否有空间插入新的值
29 freelen = zipmapRawEntryLength(p);
30 if (freelen < reqlen) {
31 /* Store the offset of this key within the current zipmap, so
32 * it can be resized. Then, move the tail backwards so this
33 * pair fits at the current position. */
34 //如果没有空间插入新的值，则调整大小
35 offset = p-zm;
36 zm = zipmapResize(zm, zmlen-freelen+reqlen);
37 p = zm+offset;
38
39 /* The +1 in the number of bytes to be moved is caused by the
40 * end-of-zipmap byte. Note: the *original* zmlen is used. */
41 //移动空间以便增加新的值
42 memmove(p+reqlen, p+freelen, zmlen-(offset+freelen+1));
43 zmlen = zmlen-freelen+reqlen;
44 freelen = reqlen;
45 }
46 }
47
48 /* We now have a suitable block where the key/value entry can
49 * be written. If there is too much free space, move the tail
50 * of the zipmap a few bytes to the front and shrink the zipmap,
51 * as we want zipmaps to be very space efficient. */
52 empty = freelen-reqlen;
53 if (empty >= ZIPMAP_VALUE_MAX_FREE) {
54 /* First, move the tail bytes to the front, then resize
55 * the zipmap to be bytes smaller. */
56 offset = p-zm;
57 memmove(p+reqlen, p+freelen, zmlen-(offset+freelen+1));
58 zmlen -= empty;
59 zm = zipmapResize(zm, zmlen);
60 p = zm+offset;
61 vempty = 0;
62 } else {
63 vempty = empty;
64 }
65
66 /* Just write the key + value and we are done. */
67 /* Key: */
68 //定位到插入的位置，首先写入key值
69 p += zipmapEncodeLength(p,klen);
70 memcpy(p,key,klen);
71 p += klen;
72 /* Value: */
73 //key值后面是value值，再次写入
74 p += zipmapEncodeLength(p,vlen);
75 *p++ = vempty;
76 memcpy(p,val,vlen);
77 return zm;
78}
map里返回长度的方法有点特别，就直接定位了就用一个字节存储长度：
1/* Return the number of entries inside a zipmap */
2/* 返回map的长度 */
3unsigned int zipmapLen(unsigned char *zm) {
4 unsigned int len = 0;
5 //如果第一个长度小于最大值，则直接返回
6 if (zm[0] < ZIPMAP_BIGLEN) {
7 len = zm[0];
8 } else {
9 //否则变量计算长度
10 unsigned char *p = zipmapRewind(zm);
11 while((p = zipmapNext(p,NULL,NULL,NULL,NULL)) != NULL) len++;
12
13 /* Re-store length if small enough */
14 if (len < ZIPMAP_BIGLEN) zm[0] = len;
15 }
16 return len;
17}
平常我们在redis客户端执行set key 'value'命令的时候，调用的其实就是set方法，如下：
1zm = zipmapSet(zm,(unsigned char*) 'name',4, (unsigned char*) 'foo',3,NULL);
2zm = zipmapSet(zm,(unsigned char*) 'surname',7, (unsigned char*) 'foo',3,NULL);
3zm = zipmapSet(zm,(unsigned char*) 'age',3, (unsigned char*) 'foo',3,NULL);
比ziplist方法简单许多了，最后给出头文件
1/* String -> String Map data structure optimized for size.
2 *
3 * See zipmap.c for more info.
4 *
5 * --------------------------------------------------------------------------
6 *
7 * Copyright (c) 2009-2010, Salvatore Sanfilippo
8 * All rights reserved.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions are met:
12 *
13 * * Redistributions of source code must retain the above copyright notice,
14 * this list of conditions and the following disclaimer.
15 * * Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * * Neither the name of Redis nor the names of its contributors may be used
19 * to endorse or promote products derived from this software without
20 * specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 'AS IS'
23 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
26 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
27 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
28 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
29 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
30 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
31 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
32 * POSSIBILITY OF SUCH DAMAGE.
33 */
1#ifndef _ZIPMAP_H
2#define _ZIPMAP_H
3
4unsigned char *zipmapNew(void); //创建一个新的压缩图
5unsigned char *zipmapSet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char *val, unsigned int vlen, int *update); //设置压缩图中的某个键值对
6unsigned char *zipmapDel(unsigned char *zm, unsigned char *key, unsigned int klen, int *deleted); //删除压缩图上的某个键值对
7unsigned char *zipmapRewind(unsigned char *zm); //将在zipmapNext中被调用到
8unsigned char *zipmapNext(unsigned char *zm, unsigned char **key, unsigned int *klen, unsigned char **value, unsigned int *vlen); //取得此键值对的下一个键值对
9int zipmapGet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char **value, unsigned int *vlen); //获取某个键值对
10int zipmapExists(unsigned char *zm, unsigned char *key, unsigned int klen); //某个key值在zipmap中是否存在
11unsigned int zipmapLen(unsigned char *zm); //zipmap压缩图的总键值对数
12size_t zipmapBlobLen(unsigned char *zm); //压缩图的序列化到文件中所需大小
13void zipmapRepr(unsigned char *p); //输出的压缩图的具体信息，用于测试
14
15#endif
最后，基于本人对redis源代码分析有一段时间了，我把分析好的代码，同步到了我的个人github上了，放上地址大家可以一起学习：
github：https://github.com/linyiqun/Redis-Code
