obitools/obitools3
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 66 Wiki Activity

untested (and no possible retrieval) of CRC used to represent data in

Browse Source 
AVL trees
This commit is contained in:
Celine Mercier
2016-03-24 16:38:11 +01:00
parent 58ac860cc7
commit 69856f18dd
5 changed files with 252 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
src/obiavl.c
View File

@ -19,9 +19,8 @@
#include <fcntl.h>
#include <math.h>
//#include <libbloom.h>
#include "bloom.h"
#include "crc64.h"
#include "obiavl.h"
#include "obierrno.h"
#include "obitypes.h"
@ -692,6 +691,7 @@ AVL_node_p avl_create_node(OBIDMS_avl_p avl, index_t node_idx)
node->right_child = -1;
node->balance_factor = 0;
node->value = -1;
node->crc64 = 0; // TODO
return node;
}
@ -1018,7 +1018,7 @@ int remap_an_avl(OBIDMS_avl_p avl)
(avl->data)->data = mmap(NULL,
((avl->data)->header)->data_size_max,
PROT_READ,
MAP_SHARED,
MAP_SHARED, // TODO test MAP_PRIVATE?
avl->data_fd,
((avl->data)->header)->header_size);
if ((avl->data)->data == NULL)
@ -1027,7 +1027,7 @@ int remap_an_avl(OBIDMS_avl_p avl)
avl->tree = mmap(NULL,
((avl->header)->nb_items_max) * sizeof(AVL_node_t),
PROT_READ,
MAP_SHARED,
MAP_SHARED, // TODO test MAP_PRIVATE?
avl->avl_fd,
(avl->header)->header_size);
if (avl->tree == NULL)
@ -1054,8 +1054,6 @@ int obi_add_new_avl_in_group(OBIDMS_avl_group_p avl_group) // TODO check for err
return -1;
}
//fprintf(stderr, "\nindex length = %d, file name = %s\n", avl_idx_length, avl_name_with_idx);
(avl_group->sub_avls)[avl_group->current_avl_idx] = obi_create_avl(avl_group->dms, avl_name_with_idx);
if ((avl_group->sub_avls)[avl_group->current_avl_idx] == NULL)
{
@ -1578,7 +1576,8 @@ index_t insert_in_avl_group(OBIDMS_avl_group_p avl_group, byte_t* value) // TODO
if ((((avl_group->sub_avls)[avl_group->current_avl_idx])->header)->nb_items == NODE_COUNT_PER_AVL) // TODO add condition with data size
obi_add_new_avl_in_group(avl_group);
bloom_add(&((((avl_group->sub_avls)[avl_group->current_avl_idx])->header)->bloom_filter), value, (BYTE_ARRAY_HEADER_SIZE + *((int32_t*)(value+1))));
bloom_add(&((((avl_group->sub_avls)[avl_group->current_avl_idx])->header)->bloom_filter), value, BYTE_ARRAY_HEADER_SIZE + *((int32_t*)(value+1)));
return obi_avl_add((avl_group->sub_avls)[avl_group->current_avl_idx], value);
}
@ -1596,6 +1595,10 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
int n = 0;
int depth = 0;
uint64_t crc;
crc = crc64(value, BYTE_ARRAY_HEADER_SIZE + ((uint64_t) (*((int32_t*)(value+1))))); // TODO warning
// Check if first node
if (!((avl->header)->nb_items))
{
@ -1604,6 +1607,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
// Add the value in the data array and store its index
value_data_idx = avl_add_value_in_data_array(avl, value);
node_to_add->value = value_data_idx;
node_to_add->crc64 = crc;
// Update the number of items
((avl->header)->nb_items)++;
@ -1635,8 +1639,15 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
parent = next;
// Compare value with value of current node
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
//to_compare = obi_avl_get(avl, current_node->value);
//comp = byte_array_compare(to_compare, value);
comp = (current_node->crc64) - crc;
if (comp == 0)
{ // check if really same value
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
}
if (comp > 0)
// Go to left child
@ -1647,7 +1658,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
else if (comp == 0)
// Value already stored
{
//fprintf(stderr, "\n>>>ALREADY IN, %s, %lld\n", obi_obibytes_to_seq(value), (avl->header)->nb_items);
fprintf(stderr, "\n>>>ALREADY IN, %s, %lld\n", obi_obibytes_to_seq(value), (avl->header)->nb_items);
return current_node->value; // TODO should trigger error if using bloom filters
}
@ -1676,6 +1687,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
// Add the value in the data array and store its index
value_data_idx = avl_add_value_in_data_array(avl, value);
node_to_add->value = value_data_idx;
node_to_add->crc64 = crc;
// Update the number of items
((avl->header)->nb_items)++;
@ -1706,7 +1718,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, byte_t* value)
}
// Find if a value is already in an AVL tree TODO use bloom
// Find if a value is already in an AVL tree
index_t obi_avl_find(OBIDMS_avl_p avl, byte_t* value)
{
int comp;
@ -1714,14 +1726,25 @@ index_t obi_avl_find(OBIDMS_avl_p avl, byte_t* value)
byte_t* to_compare;
AVL_node_p current_node;
uint64_t crc;
crc = crc64(value, BYTE_ARRAY_HEADER_SIZE + ((uint64_t) (*((int32_t*)(value+1))))); // TODO warning
next = (avl->header)->root_idx;
while (next != -1)
{
current_node = (avl->tree)+next;
// Compare value with value of current node
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
//to_compare = obi_avl_get(avl, current_node->value);
//comp = byte_array_compare(to_compare, value);
comp = (current_node->crc64) - crc;
if (comp == 0)
{ // check if really same value
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
}
if (comp > 0)
// Go to left child
@ -1730,8 +1753,10 @@ index_t obi_avl_find(OBIDMS_avl_p avl, byte_t* value)
// Go to right child
next = current_node->right_child;
else if (comp == 0)
// Value found
{ // Value found
fprintf(stderr, "\n>>>ALREADY IN in find, %s, %lld\n", obi_obibytes_to_seq(value), (avl->header)->nb_items);
return current_node->value;
}
}
// Value not found
return -1;