X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/8ec0832157587a1d19ee0b6514c6b33d258fd51b..c22b2ec863386476c52e7442aa497151a8986af1:/zipmap.c diff --git a/zipmap.c b/zipmap.c index d5cfa162..6fd6ca86 100644 --- a/zipmap.c +++ b/zipmap.c @@ -42,10 +42,11 @@ /* Memory layout of a zipmap, for the map "foo" => "bar", "hello" => "world": * - * "foo""bar""hello""world" + * "foo""bar""hello""world" * - * is 1 byte status. Currently only 1 bit is used: if the least - * significant bit is set, it means the zipmap needs to be defragmented. + * is 1 byte length that holds the current size of the zipmap. + * When the zipmap length is greater than or equal to 254, this value + * is not used and the zipmap needs to be traversed to find out the length. * * is the length of the following string (key or value). * lengths are encoded in a single value or in a 5 bytes value. @@ -62,22 +63,15 @@ * or even in order to add a key/value pair if it fits. * * is always an unsigned 8 bit number, because if after an - * update operation there are more than a few free bytes, they'll be converted - * into empty space prefixed by the special value 254. + * update operation there are more than a few free bytes, the zipmap will be + * reallocated to make sure it is as small as possible. * * The most compact representation of the above two elements hash is actually: * - * "\x00\x03\x00foo\x03\x00bar\x05\x00hello\x05\x00world\xff" + * "\x02\x03foo\x03\x00bar\x05hello\x05\x00world\xff" * - * Empty space is marked using a 254 bytes + a (coded as already - * specified). The length includes the 254 bytes in the count and the - * space taken by the field. So for instance removing the "foo" key - * from the zipmap above will lead to the following representation: - * - * "\xfd\x10........\x05\x00hello\x05\x00world\xff" - * - * Note that because empty space, keys, values, are all prefixed length - * "objects", the lookup will take O(N) where N is the numeber of elements + * Note that because keys and values are prefixed length "objects", + * the lookup will take O(N) where N is the number of elements * in the zipmap and *not* the number of bytes needed to represent the zipmap. * This lowers the constant times considerably. */ @@ -87,21 +81,23 @@ #include #include "zmalloc.h" -#define ZIPMAP_BIGLEN 253 -#define ZIPMAP_EMPTY 254 +#define ZIPMAP_BIGLEN 254 #define ZIPMAP_END 255 -#define ZIPMAP_STATUS_FRAGMENTED 1 - /* The following defines the max value for the field described in the * comments above, that is, the max number of trailing bytes in a value. */ -#define ZIPMAP_VALUE_MAX_FREE 5 +#define ZIPMAP_VALUE_MAX_FREE 4 + +/* The following macro returns the number of bytes needed to encode the length + * for the integer value _l, that is, 1 byte for lengths < ZIPMAP_BIGLEN and + * 5 bytes for all the other lengths. */ +#define ZIPMAP_LEN_BYTES(_l) (((_l) < ZIPMAP_BIGLEN) ? 1 : sizeof(unsigned int)+1) /* Create a new empty zipmap. */ unsigned char *zipmapNew(void) { unsigned char *zm = zmalloc(2); - zm[0] = 0; /* Status */ + zm[0] = 0; /* Length */ zm[1] = ZIPMAP_END; return zm; } @@ -111,7 +107,7 @@ static unsigned int zipmapDecodeLength(unsigned char *p) { unsigned int len = *p; if (len < ZIPMAP_BIGLEN) return len; - memcpy(&len,p,sizeof(unsigned int)); + memcpy(&len,p+1,sizeof(unsigned int)); return len; } @@ -119,7 +115,7 @@ static unsigned int zipmapDecodeLength(unsigned char *p) { * the amount of bytes required to encode such a length. */ static unsigned int zipmapEncodeLength(unsigned char *p, unsigned int len) { if (p == NULL) { - return (len < ZIPMAP_BIGLEN) ? 1 : 1+sizeof(unsigned int); + return ZIPMAP_LEN_BYTES(len); } else { if (len < ZIPMAP_BIGLEN) { p[0] = len; @@ -137,50 +133,35 @@ static unsigned int zipmapEncodeLength(unsigned char *p, unsigned int len) { * * If NULL is returned, and totlen is not NULL, it is set to the entire * size of the zimap, so that the calling function will be able to - * reallocate the original zipmap to make room for more entries. - * - * If NULL is returned, and freeoff and freelen are not NULL, they are set - * to the offset of the first empty space that can hold '*freelen' bytes - * (freelen is an integer pointer used both to signal the required length - * and to get the reply from the function). If there is not a suitable - * free space block to hold the requested bytes, *freelen is set to 0. */ -static unsigned char *zipmapLookupRaw(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned int *totlen, unsigned int *freeoff, unsigned int *freelen) { - unsigned char *p = zm+1; - unsigned int l; - unsigned int reqfreelen; + * reallocate the original zipmap to make room for more entries. */ +static unsigned char *zipmapLookupRaw(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned int *totlen) { + unsigned char *p = zm+1, *k = NULL; + unsigned int l,llen; - if (freelen) { - reqfreelen = *freelen; - *freelen = 0; - assert(reqfreelen != 0); - } while(*p != ZIPMAP_END) { - if (*p == ZIPMAP_EMPTY) { - l = zipmapDecodeLength(p+1); - /* if the user want a free space report, and this space is - * enough, and we did't already found a suitable space... */ - if (freelen && l >= reqfreelen && *freelen == 0) { - *freelen = l; - *freeoff = p-zm; - } - p += l; - zm[0] |= ZIPMAP_STATUS_FRAGMENTED; - } else { - unsigned char free; + unsigned char free; - /* Match or skip the key */ - l = zipmapDecodeLength(p); - if (l == klen && !memcmp(p+1,key,l)) return p; - p += zipmapEncodeLength(NULL,l) + l; - /* Skip the value as well */ - l = zipmapDecodeLength(p); - p += zipmapEncodeLength(NULL,l); - free = p[0]; - p += l+1+free; /* +1 to skip the free byte */ + /* Match or skip the key */ + l = zipmapDecodeLength(p); + llen = zipmapEncodeLength(NULL,l); + if (k == NULL && l == klen && !memcmp(p+llen,key,l)) { + /* Only return when the user doesn't care + * for the total length of the zipmap. */ + if (totlen != NULL) { + k = p; + } else { + return p; + } } + p += llen+l; + /* Skip the value as well */ + l = zipmapDecodeLength(p); + p += zipmapEncodeLength(NULL,l); + free = p[0]; + p += l+1+free; /* +1 to skip the free byte */ } if (totlen != NULL) *totlen = (unsigned int)(p-zm)+1; - return NULL; + return k; } static unsigned long zipmapRequiredLength(unsigned int klen, unsigned int vlen) { @@ -210,57 +191,75 @@ static unsigned int zipmapRawValueLength(unsigned char *p) { return used; } -/* Set key to value, creating the key if it does not already exist. */ -unsigned char *zipmapSet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char *val, unsigned int vlen) { - unsigned int oldlen = 0, freeoff = 0, freelen; - unsigned int reqlen = zipmapRequiredLength(klen,vlen); +/* If 'p' points to a key, this function returns the total amount of + * bytes used to store this entry (entry = key + associated value + trailing + * free space if any). */ +static unsigned int zipmapRawEntryLength(unsigned char *p) { + unsigned int l = zipmapRawKeyLength(p); + return l + zipmapRawValueLength(p+l); +} + +static inline unsigned char *zipmapResize(unsigned char *zm, unsigned int len) { + zm = zrealloc(zm, len); + zm[len-1] = ZIPMAP_END; + return zm; +} + +/* Set key to value, creating the key if it does not already exist. + * If 'update' is not NULL, *update is set to 1 if the key was + * already preset, otherwise to 0. */ +unsigned char *zipmapSet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char *val, unsigned int vlen, int *update) { + unsigned int zmlen, offset; + unsigned int freelen, reqlen = zipmapRequiredLength(klen,vlen); unsigned int empty, vempty; unsigned char *p; freelen = reqlen; - p = zipmapLookupRaw(zm,key,klen,&oldlen,&freeoff,&freelen); - if (p == NULL && freelen == 0) { - printf("HERE oldlen:%u required:%u\n",oldlen,reqlen); - /* Key not found, and not space for the new key. Enlarge */ - zm = zrealloc(zm,oldlen+reqlen); - p = zm+oldlen-1; - zm[oldlen+reqlen-1] = ZIPMAP_END; - freelen = reqlen; - printf("New total length is: %u\n", oldlen+reqlen); - } else if (p == NULL) { - /* Key not found, but there is enough free space. */ - p = zm+freeoff; - /* note: freelen is already set in this case */ - } else { - unsigned char *b = p; + if (update) *update = 0; + p = zipmapLookupRaw(zm,key,klen,&zmlen); + if (p == NULL) { + /* Key not found: enlarge */ + zm = zipmapResize(zm, zmlen+reqlen); + p = zm+zmlen-1; + zmlen = zmlen+reqlen; + /* Increase zipmap length (this is an insert) */ + if (zm[0] < ZIPMAP_BIGLEN) zm[0]++; + } else { /* Key found. Is there enough space for the new value? */ /* Compute the total length: */ - freelen = zipmapRawKeyLength(b); - b += freelen; - freelen += zipmapRawValueLength(b); + if (update) *update = 1; + freelen = zipmapRawEntryLength(p); if (freelen < reqlen) { - /* Mark this blog as free and recurse */ - p[0] = ZIPMAP_EMPTY; - zipmapEncodeLength(p+1,freelen); - return zipmapSet(zm,key,klen,val,vlen); + /* Store the offset of this key within the current zipmap, so + * it can be resized. Then, move the tail backwards so this + * pair fits at the current position. */ + offset = p-zm; + zm = zipmapResize(zm, zmlen-freelen+reqlen); + p = zm+offset; + + /* The +1 in the number of bytes to be moved is caused by the + * end-of-zipmap byte. Note: the *original* zmlen is used. */ + memmove(p+reqlen, p+freelen, zmlen-(offset+freelen+1)); + zmlen = zmlen-freelen+reqlen; + freelen = reqlen; } } - /* Ok we have a suitable block where to write the new key/value - * entry. */ + /* We now have a suitable block where the key/value entry can + * be written. If there is too much free space, move the tail + * of the zipmap a few bytes to the front and shrink the zipmap, + * as we want zipmaps to be very space efficient. */ empty = freelen-reqlen; - /* If there is too much free space mark it as a free block instead - * of adding it as trailing empty space for the value, as we want - * zipmaps to be very space efficient. */ - if (empty > ZIPMAP_VALUE_MAX_FREE) { - unsigned char *e; - - e = p+reqlen; - e[0] = ZIPMAP_EMPTY; - zipmapEncodeLength(e+1,empty); + if (empty >= ZIPMAP_VALUE_MAX_FREE) { + /* First, move the tail bytes to the front, then resize + * the zipmap to be bytes smaller. */ + offset = p-zm; + memmove(p+reqlen, p+freelen, zmlen-(offset+freelen+1)); + zmlen -= empty; + zm = zipmapResize(zm, zmlen); + p = zm+offset; vempty = 0; - zm[0] |= ZIPMAP_STATUS_FRAGMENTED; } else { vempty = empty; } @@ -277,6 +276,91 @@ unsigned char *zipmapSet(unsigned char *zm, unsigned char *key, unsigned int kle return zm; } +/* Remove the specified key. If 'deleted' is not NULL the pointed integer is + * set to 0 if the key was not found, to 1 if it was found and deleted. */ +unsigned char *zipmapDel(unsigned char *zm, unsigned char *key, unsigned int klen, int *deleted) { + unsigned int zmlen, freelen; + unsigned char *p = zipmapLookupRaw(zm,key,klen,&zmlen); + if (p) { + freelen = zipmapRawEntryLength(p); + memmove(p, p+freelen, zmlen-((p-zm)+freelen+1)); + zm = zipmapResize(zm, zmlen-freelen); + + /* Decrease zipmap length */ + if (zm[0] < ZIPMAP_BIGLEN) zm[0]--; + + if (deleted) *deleted = 1; + } else { + if (deleted) *deleted = 0; + } + return zm; +} + +/* Call it before to iterate trought elements via zipmapNext() */ +unsigned char *zipmapRewind(unsigned char *zm) { + return zm+1; +} + +/* This function is used to iterate through all the zipmap elements. + * In the first call the first argument is the pointer to the zipmap + 1. + * In the next calls what zipmapNext returns is used as first argument. + * Example: + * + * unsigned char *i = zipmapRewind(my_zipmap); + * while((i = zipmapNext(i,&key,&klen,&value,&vlen)) != NULL) { + * printf("%d bytes key at $p\n", klen, key); + * printf("%d bytes value at $p\n", vlen, value); + * } + */ +unsigned char *zipmapNext(unsigned char *zm, unsigned char **key, unsigned int *klen, unsigned char **value, unsigned int *vlen) { + if (zm[0] == ZIPMAP_END) return NULL; + if (key) { + *key = zm; + *klen = zipmapDecodeLength(zm); + *key += ZIPMAP_LEN_BYTES(*klen); + } + zm += zipmapRawKeyLength(zm); + if (value) { + *value = zm+1; + *vlen = zipmapDecodeLength(zm); + *value += ZIPMAP_LEN_BYTES(*vlen); + } + zm += zipmapRawValueLength(zm); + return zm; +} + +/* Search a key and retrieve the pointer and len of the associated value. + * If the key is found the function returns 1, otherwise 0. */ +int zipmapGet(unsigned char *zm, unsigned char *key, unsigned int klen, unsigned char **value, unsigned int *vlen) { + unsigned char *p; + + if ((p = zipmapLookupRaw(zm,key,klen,NULL)) == NULL) return 0; + p += zipmapRawKeyLength(p); + *vlen = zipmapDecodeLength(p); + *value = p + ZIPMAP_LEN_BYTES(*vlen) + 1; + return 1; +} + +/* Return 1 if the key exists, otherwise 0 is returned. */ +int zipmapExists(unsigned char *zm, unsigned char *key, unsigned int klen) { + return zipmapLookupRaw(zm,key,klen,NULL) != NULL; +} + +/* Return the number of entries inside a zipmap */ +unsigned int zipmapLen(unsigned char *zm) { + unsigned int len = 0; + if (zm[0] < ZIPMAP_BIGLEN) { + len = zm[0]; + } else { + unsigned char *p = zipmapRewind(zm); + while((p = zipmapNext(p,NULL,NULL,NULL,NULL)) != NULL) len++; + + /* Re-store length if small enough */ + if (len < ZIPMAP_BIGLEN) zm[0] = len; + } + return len; +} + void zipmapRepr(unsigned char *p) { unsigned int l; @@ -285,10 +369,6 @@ void zipmapRepr(unsigned char *p) { if (p[0] == ZIPMAP_END) { printf("{end}"); break; - } else if (p[0] == ZIPMAP_EMPTY) { - l = zipmapDecodeLength(p+1); - printf("{%u empty block}", l); - p += l; } else { unsigned char e; @@ -314,13 +394,63 @@ void zipmapRepr(unsigned char *p) { printf("\n"); } +#ifdef ZIPMAP_TEST_MAIN int main(void) { unsigned char *zm; zm = zipmapNew(); - zm = zipmapSet(zm,(unsigned char*) "hello",5, (unsigned char*) "world!",6); - zm = zipmapSet(zm,(unsigned char*) "foo",3, (unsigned char*) "bar",3); - zm = zipmapSet(zm,(unsigned char*) "foo",3, (unsigned char*) "!",1); + + zm = zipmapSet(zm,(unsigned char*) "name",4, (unsigned char*) "foo",3,NULL); + zm = zipmapSet(zm,(unsigned char*) "surname",7, (unsigned char*) "foo",3,NULL); + zm = zipmapSet(zm,(unsigned char*) "age",3, (unsigned char*) "foo",3,NULL); + zipmapRepr(zm); + + zm = zipmapSet(zm,(unsigned char*) "hello",5, (unsigned char*) "world!",6,NULL); + zm = zipmapSet(zm,(unsigned char*) "foo",3, (unsigned char*) "bar",3,NULL); + zm = zipmapSet(zm,(unsigned char*) "foo",3, (unsigned char*) "!",1,NULL); + zipmapRepr(zm); + zm = zipmapSet(zm,(unsigned char*) "foo",3, (unsigned char*) "12345",5,NULL); + zipmapRepr(zm); + zm = zipmapSet(zm,(unsigned char*) "new",3, (unsigned char*) "xx",2,NULL); + zm = zipmapSet(zm,(unsigned char*) "noval",5, (unsigned char*) "",0,NULL); zipmapRepr(zm); + zm = zipmapDel(zm,(unsigned char*) "new",3,NULL); + zipmapRepr(zm); + + printf("\nLook up large key:\n"); + { + unsigned char buf[512]; + unsigned char *value; + unsigned int vlen, i; + for (i = 0; i < 512; i++) buf[i] = 'a'; + + zm = zipmapSet(zm,buf,512,(unsigned char*) "long",4,NULL); + if (zipmapGet(zm,buf,512,&value,&vlen)) { + printf(" is associated to the %d bytes value: %.*s\n", + vlen, vlen, value); + } + } + + printf("\nPerform a direct lookup:\n"); + { + unsigned char *value; + unsigned int vlen; + + if (zipmapGet(zm,(unsigned char*) "foo",3,&value,&vlen)) { + printf(" foo is associated to the %d bytes value: %.*s\n", + vlen, vlen, value); + } + } + printf("\nIterate trought elements:\n"); + { + unsigned char *i = zipmapRewind(zm); + unsigned char *key, *value; + unsigned int klen, vlen; + + while((i = zipmapNext(i,&key,&klen,&value,&vlen)) != NULL) { + printf(" %d:%.*s => %d:%.*s\n", klen, klen, key, vlen, vlen, value); + } + } return 0; } +#endif