Skip to content

GitLab

O
OBITools3
Commit 375bfcce authored by Celine Mercier's avatar Celine Mercier
Browse files
Options

Renamed "Obi_byte_arrays" to "Obiblobs" and moved Obiblob functions to 

separate obiblob.c and obiblob.h files
parent c225cfd8
Hide whitespace changes
Inline Side-by-side
Showing with 304 additions and 249 deletions
src/encode.c
View file @ 375bfcce
......@@ -349,111 +349,6 @@ char* decode_seq_on_4_bits(byte_t* seq_b, int32_t length_seq)
}
Obi_byte_array_p obi_byte_array(byte_t* encoded_value, uint8_t element_size, int32_t length_encoded_value, int32_t length_decoded_value)
{
Obi_byte_array_p byte_array;
// Allocate the memory for the byte array structure
byte_array = (Obi_byte_array_p) malloc(sizeof(Obi_byte_array_t) + length_encoded_value);
if (byte_array == NULL)
{
obi_set_errno(OBI_MALLOC_ERROR); // TODO
obidebug(1, "\nError allocating memory for a byte array");
return NULL;
}
// Store the number of bits on which each element is encoded
byte_array->element_size = element_size;
// Store the length (in bytes) of the encoded value
byte_array->length_encoded_value = length_encoded_value;
// Store the initial length (in bytes) of the decoded value
byte_array->length_decoded_value = length_decoded_value;
// Store the encoded value
memcpy(byte_array->value, encoded_value, length_encoded_value);
return byte_array;
}
Obi_byte_array_p obi_str_to_obibytes(char* value)
{
Obi_byte_array_p value_b;
int32_t length;
// Compute the number of bytes on which the value will be encoded
length = strlen(value) + 1; // +1 to store \0 at the end (makes retrieving faster)
value_b = obi_byte_array(value, ELEMENT_SIZE_STR, length, length);
if (value_b == NULL)
{
obidebug(1, "\nError encoding a character string in a byte array");
return NULL;
}
return value_b;
}
const char* obi_obibytes_to_str(Obi_byte_array_p value_b)
{
fprintf(stderr, "\n%s", value_b->value);
return value_b->value;
}
Obi_byte_array_p obi_seq_to_obibytes(char* seq)
{
Obi_byte_array_p value_b;
int32_t length_encoded_seq; // length of the encoded sequence in bytes
int32_t seq_length;
byte_t* encoded_seq;
seq_length = strlen(seq);
// Check if just ATGC and encode accordingly
if (only_ATGC(seq))
{
// Compute the length (in bytes) of the encoded sequence
length_encoded_seq = ceil((double) seq_length / (double) 4.0);
// Encode
encoded_seq = encode_seq_on_2_bits(seq, seq_length);
if (encoded_seq == NULL)
return NULL;
value_b = obi_byte_array(encoded_seq, ELEMENT_SIZE_SEQ_2, length_encoded_seq, seq_length);
}
else
{
// Compute the length (in bytes) of the encoded sequence
length_encoded_seq = ceil((double) seq_length / (double) 2.0);
// Encode
encoded_seq = encode_seq_on_4_bits(seq, seq_length);
if (encoded_seq == NULL)
return NULL;
value_b = obi_byte_array(encoded_seq, ELEMENT_SIZE_SEQ_4, length_encoded_seq, seq_length);
}
free(encoded_seq);
return value_b;
}
const char* obi_obibytes_to_seq(Obi_byte_array_p value_b)
{
// Decode
if (value_b->element_size == 2)
return decode_seq_on_2_bits(value_b->value, value_b->length_decoded_value);
else
return decode_seq_on_4_bits(value_b->value, value_b->length_decoded_value);
}
// TODO same for int
///////////////////// FOR DEBUGGING ///////////////////////////
//NOTE: The first byte is printed the first (at the left-most).
......
src/encode.h
View file @ 375bfcce
......@@ -26,27 +26,6 @@
*/
#define NUC_MASK_4B 0xF /**< Binary: 1111 to use when decoding 4 bits sequences
*/
#define ELEMENT_SIZE_STR (8) /**< The size of an element from a value of type character string.
*/
#define ELEMENT_SIZE_SEQ_2 (2) /**< The size of an element from a value of type DNA sequence encoded on 2 bits.
*/
#define ELEMENT_SIZE_SEQ_4 (4) /**< The size of an element from a value of type DNA sequence encoded on 4 bits.
*/
/**
* @brief Byte array structure.
*/
typedef struct Obi_byte_array {
uint8_t element_size; /**< Size in bits of one element from the value.
*/
int32_t length_encoded_value; /**< Length in bytes of the encoded value.
*/
int32_t length_decoded_value; /**< Length in bytes of the decoded value.
*/
byte_t value[]; /**< Encoded value.
*/
} Obi_byte_array_t, *Obi_byte_array_p;
/**
......@@ -201,65 +180,6 @@ byte_t* encode_seq_on_4_bits(char* seq, int32_t length);
char* decode_seq_on_4_bits(byte_t* seq_b, int32_t length_seq);
/**
* @brief Converts a character string to a byte array with a header.
*
* @warning The byte array must be freed by the caller.
*
* @param value The character string to convert.
*
* @returns A pointer to the byte array created.
* @retval NULL if an error occurred.
*
* @since October 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
Obi_byte_array_p obi_str_to_obibytes(char* value);
/**
* @brief Converts a byte array to a character string.
*
* @param value_b The byte array to convert.
*
* @returns A pointer to the character string contained in the byte array.
*
* @since October 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
const char* obi_obibytes_to_str(Obi_byte_array_p value_b);
/**
* @brief Converts a DNA sequence to a byte array with a header.
*
* @warning The byte array must be freed by the caller.
*
* @param value The DNA sequence to convert.
*
* @returns A pointer to the byte array created.
* @retval NULL if an error occurred.
*
* @since November 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
Obi_byte_array_p obi_seq_to_obibytes(char* seq);
/**
* @brief Converts a byte array to a DNA sequence.
*
* @param value_b The byte array to convert.
*
* @returns A pointer to the DNA sequence contained in the byte array.
* @retval NULL if an error occurred.
*
* @since November 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
const char* obi_obibytes_to_seq(Obi_byte_array_p value_b); // TODO move to encode source files
////////// FOR DEBUGGING ///////////
// little endian
......
src/obiavl.c
View file @ 375bfcce
......@@ -22,6 +22,7 @@
#include "bloom.h"
#include "crc64.h"
#include "obiavl.h"
#include "obiblob.h"
#include "obierrno.h"
#include "obitypes.h"
#include "obidebug.h"
......@@ -270,7 +271,7 @@ int add_new_avl_in_group(OBIDMS_avl_group_p avl_group);
* The function checks a bloom filter. No false negatives, possible false positives.
*
* @param avl A pointer to the AVL tree structure.
* @param value A pointer to the byte array structure.
* @param value A pointer to the blob structure.
*
* @retval 0 if the value is definitely not already stored in the AVL tree.
* @retval 1 if the value might already be stored in the AVL tree.
......@@ -278,17 +279,17 @@ int add_new_avl_in_group(OBIDMS_avl_group_p avl_group);
* @since April 2016
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
int maybe_in_avl(OBIDMS_avl_p avl, Obi_byte_array_p value);
int maybe_in_avl(OBIDMS_avl_p avl, Obi_blob_p value);
/**
* @brief Internal function comparing two byte arrays.
* @brief Internal function comparing two blobs.
*
* The encoding is compared first, then the length of the
* values, then the values themselves.
*
* @param value_1 A pointer to the first byte array structure.
* @param value_2 A pointer to the second byte array structure.
* @param value_1 A pointer to the first blob structure.
* @param value_2 A pointer to the second blob structure.
*
* @returns A value < 0 if value_1 < value_2,
* a value > 0 if value_1 > value_2,
......@@ -297,27 +298,27 @@ int maybe_in_avl(OBIDMS_avl_p avl, Obi_byte_array_p value);
* @since October 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
int byte_array_compare(Obi_byte_array_p value_1, Obi_byte_array_p value_2);
int blob_compare(Obi_blob_p value_1, Obi_blob_p value_2);
/**
* @brief Internal function calculating the size in bytes of a byte array.
* @brief Internal function calculating the size in bytes of a blob.
*
* @param value A pointer to the byte array structure.
* @param value A pointer to the blob structure.
*
* @returns The size of the byte array in bytes.
* @returns The size of the blob in bytes.
*
* @since October 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
int byte_array_sizeof(Obi_byte_array_p value);
int blob_sizeof(Obi_blob_p value);
/**
* @brief Internal function storing a value (byte array) in the data array referred to by an AVL tree.
* @brief Internal function storing a value (blob) in the data array referred to by an AVL tree.
*
* @param avl A pointer to the AVL tree structure.
* @param value A pointer to the value (byte array structure).
* @param value A pointer to the value (blob structure).
*
* @returns The index of the stored value.
* @retval -1 if an error occurred.
......@@ -325,7 +326,7 @@ int byte_array_sizeof(Obi_byte_array_p value);
* @since December 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_byte_array_p value);
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_blob_p value);
/**
......@@ -979,13 +980,13 @@ int add_new_avl_in_group(OBIDMS_avl_group_p avl_group)
}
int maybe_in_avl(OBIDMS_avl_p avl, Obi_byte_array_p value)
int maybe_in_avl(OBIDMS_avl_p avl, Obi_blob_p value)
{
return (bloom_check(&((avl->header)->bloom_filter), value, byte_array_sizeof(value)));
return (bloom_check(&((avl->header)->bloom_filter), value, blob_sizeof(value)));
}
int byte_array_compare(Obi_byte_array_p value_1, Obi_byte_array_p value_2)
int blob_compare(Obi_blob_p value_1, Obi_blob_p value_2)
{
int comp;
int32_t b;
......@@ -1013,13 +1014,13 @@ int byte_array_compare(Obi_byte_array_p value_1, Obi_byte_array_p value_2)
}
int byte_array_sizeof(Obi_byte_array_p value)
int blob_sizeof(Obi_blob_p value)
{
return (sizeof(Obi_byte_array_t) + (value->length_encoded_value));
return (sizeof(Obi_blob_t) + (value->length_encoded_value));
}
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_byte_array_p value)
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_blob_p value)
{
index_t value_idx;
int value_size;
......@@ -1027,7 +1028,7 @@ index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_byte_array_p value)
value_idx = ((avl->data)->header)->data_size_used;
// Grow the data if needed
value_size = byte_array_sizeof(value);
value_size = blob_sizeof(value);
while (((avl->data)->header)->data_size_max < (value_idx + value_size))
{
if (grow_avl_data(avl->data) < 0)
......@@ -2047,20 +2048,20 @@ int obi_close_avl_group(OBIDMS_avl_group_p avl_group)
}
Obi_byte_array_p obi_avl_get(OBIDMS_avl_p avl, index_t idx)
Obi_blob_p obi_avl_get(OBIDMS_avl_p avl, index_t idx)
{
return ((Obi_byte_array_p)(((avl->data)->data)+idx));
return ((Obi_blob_p)(((avl->data)->data)+idx));
}
index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value)
index_t obi_avl_add(OBIDMS_avl_p avl, Obi_blob_p value)
{
AVL_node_p node_to_add = NULL;
AVL_node_p current_node;
index_t next, parent;
index_t value_data_idx;
index_t node_idx;
Obi_byte_array_p to_compare;
Obi_blob_p to_compare;
int comp;
int n;
int depth;
......@@ -2068,7 +2069,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value)
n = 0;
depth = 0;
crc = crc64((byte_t*)value, byte_array_sizeof(value));
crc = crc64((byte_t*)value, blob_sizeof(value));
// Check if first node
if (!((avl->header)->nb_items))
......@@ -2115,7 +2116,7 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value)
if (comp == 0)
{ // check if really same value
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
comp = blob_compare(to_compare, value);
}
if (comp > 0)
......@@ -2189,15 +2190,15 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value)
// Find if a value is already in an AVL tree
index_t obi_avl_find(OBIDMS_avl_p avl, Obi_byte_array_p value)
index_t obi_avl_find(OBIDMS_avl_p avl, Obi_blob_p value)
{
int comp;
index_t next;
Obi_byte_array_p to_compare;
Obi_blob_p to_compare;
AVL_node_p current_node;
uint64_t crc;
crc = crc64((byte_t*)value, byte_array_sizeof(value));
crc = crc64((byte_t*)value, blob_sizeof(value));
next = (avl->header)->root_idx;
while (next != -1)
......@@ -2210,7 +2211,7 @@ index_t obi_avl_find(OBIDMS_avl_p avl, Obi_byte_array_p value)
if (comp == 0)
{ // Check if really same value
to_compare = obi_avl_get(avl, current_node->value);
comp = byte_array_compare(to_compare, value);
comp = blob_compare(to_compare, value);
}
if (comp > 0)
......@@ -2229,7 +2230,7 @@ index_t obi_avl_find(OBIDMS_avl_p avl, Obi_byte_array_p value)
}
Obi_byte_array_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx)
Obi_blob_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx)
{
int32_t avl_idx;
index_t idx_in_avl;
......@@ -2241,7 +2242,7 @@ Obi_byte_array_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx)
}
index_t obi_avl_group_add(OBIDMS_avl_group_p avl_group, Obi_byte_array_p value)
index_t obi_avl_group_add(OBIDMS_avl_group_p avl_group, Obi_blob_p value)
{
int32_t index_in_avl;
index_t index_with_avl;
......@@ -2286,7 +2287,7 @@ index_t obi_avl_group_add(OBIDMS_avl_group_p avl_group, Obi_byte_array_p value)
}
// Add in the current AVL
bloom_add(&((((avl_group->sub_avls)[avl_group->current_avl_idx])->header)->bloom_filter), value, byte_array_sizeof(value));
bloom_add(&((((avl_group->sub_avls)[avl_group->current_avl_idx])->header)->bloom_filter), value, blob_sizeof(value));
index_in_avl = (int32_t) obi_avl_add((avl_group->sub_avls)[avl_group->current_avl_idx], value);
// Build the index containing the AVL index
......
src/obiavl.h
View file @ 375bfcce
......@@ -6,7 +6,7 @@
* @file obiavl.h
* @author Celine Mercier
* @date December 3rd 2015
* @brief Header file for handling AVL trees for storing and retrieving byte arrays (i.e. coding for character strings).
* @brief Header file for handling AVL trees for storing and retrieving blobs (i.e. coding for character strings).
*/
......@@ -23,6 +23,7 @@
#include <stdbool.h>
#include "obidms.h"
#include "obiblob.h"
#include "obitypes.h"
#include "bloom.h"
#include "utils.h"
......@@ -314,28 +315,28 @@ int obi_close_avl_group(OBIDMS_avl_group_p avl_group);
/**
* @brief Recovers a value (byte array) in an AVL tree.
* @brief Recovers a value (blob) in an AVL tree.
*
* @warning The byte array recovered must be decoded to get the original value.
* @warning The blob recovered must be decoded to get the original value.
*
* @param avl A pointer to the AVL tree.
* @param index The index of the value in the data array.
*
* @returns A pointer to the byte array recovered.
* @returns A pointer to the blob recovered.
*
* @since December 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
Obi_byte_array_p obi_avl_get(OBIDMS_avl_p avl, index_t index);
Obi_blob_p obi_avl_get(OBIDMS_avl_p avl, index_t index);
/**
* @brief Adds a value (byte array) in an AVL tree NOT checking first if it is already in it. // TODO to discuss
* @brief Adds a value (blob) in an AVL tree NOT checking first if it is already in it. // TODO to discuss
*
* @warning The value given must be already be encoded into a byte array structure (Obi_byte_array_t).
* @warning The value given must be already be encoded into a blob structure (Obi_blob_t).
*
* @param avl A pointer to the AVL tree.
* @param value The byte array to add in the AVL tree.
* @param value The blob to add in the AVL tree.
*
* @returns The index of the value newly added in the AVL tree.
* @retval -1 if an error occurred.
......@@ -343,16 +344,16 @@ Obi_byte_array_p obi_avl_get(OBIDMS_avl_p avl, index_t index);
* @since December 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value);
index_t obi_avl_add(OBIDMS_avl_p avl, Obi_blob_p value);
/**
* @brief Finds a value (byte array) in an AVL tree.
* @brief Finds a value (blob) in an AVL tree.
*
* @warning The value given must be already be encoded into a byte array structure (Obi_byte_array_t).
* @warning The value given must be already be encoded into a blob structure (Obi_blob_t).
*
* @param avl A pointer to the AVL tree.
* @param value The byte array to add in the AVL tree.
* @param value The blob to add in the AVL tree.
*
* @returns The data index of the value.
* @retval -1 if the value is not in the tree.
......@@ -360,32 +361,32 @@ index_t obi_avl_add(OBIDMS_avl_p avl, Obi_byte_array_p value);
* @since December 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
index_t obi_avl_find(OBIDMS_avl_p avl, Obi_byte_array_p value);
index_t obi_avl_find(OBIDMS_avl_p avl, Obi_blob_p value);
/**
* @brief Recovers a value (byte array) in an AVL tree.
* @brief Recovers a value (blob) in an AVL tree.
*
* @warning The byte array recovered must be decoded to get the original value.
* @warning The blob recovered must be decoded to get the original value.
*
* @param avl_group A pointer to the AVL tree.
* @param index The index of the value in the data array.
*
* @returns A pointer to the byte array recovered.
* @returns A pointer to the blob recovered.
*
* @since April 2016
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
Obi_byte_array_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx);
Obi_blob_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx);
/**
* @brief Adds a value (byte array) in an AVL tree group, checking if it is already in it.
* @brief Adds a value (blob) in an AVL tree group, checking if it is already in it.
*
* @warning The value given must be already be encoded into a byte array structure (Obi_byte_array_t).
* @warning The value given must be already be encoded into a blob structure (Obi_blob_t).
*
* @param avl_group A pointer to the AVL tree group.
* @param value The byte array to add in the AVL tree group.
* @param value The blob to add in the AVL tree group.
*
* @returns The index of the value newly added in the AVL tree group.
* @retval -1 if an error occurred.
......@@ -393,7 +394,7 @@ Obi_byte_array_p obi_avl_group_get(OBIDMS_avl_group_p avl_group, index_t idx);
* @since April 2016
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
index_t obi_avl_group_add(OBIDMS_avl_group_p avl_group, Obi_byte_array_p value);
index_t obi_avl_group_add(OBIDMS_avl_group_p avl_group, Obi_blob_p value);
#endif /* OBIAVL_H_ */
......
src/obiblob.c 0 → 100644
View file @ 375bfcce
/****************************************************************************
* Obiblob functions *
****************************************************************************/
/**
* @file obiblob.c
* @author Celine Mercier
* @date April 11th 2016
* @brief Functions handling Obiblob structures.
*/
#include <stdlib.h>
#include <stdio.h>
#include <stdint.h>
#include "obierrno.h"
#include "obitypes.h" // For byte_t type
#include "obidebug.h"
#define DEBUG_LEVEL 0 // TODO has to be defined somewhere else (cython compil flag?)
// TODO: endianness problem?
Obi_blob_p obi_blob(byte_t* encoded_value, uint8_t element_size, int32_t length_encoded_value, int32_t length_decoded_value)
{
Obi_blob_p blob;
// Allocate the memory for the blob structure
blob = (Obi_blob_p) malloc(sizeof(Obi_blob_t) + length_encoded_value);
if (blob == NULL)
{
obi_set_errno(OBI_MALLOC_ERROR);
obidebug(1, "\nError allocating memory for a blob");
return NULL;
}
// Store the number of bits on which each element is encoded
blob->element_size = element_size;
// Store the length (in bytes) of the encoded value
blob->length_encoded_value = length_encoded_value;
// Store the initial length (in bytes) of the decoded value
blob->length_decoded_value = length_decoded_value;
// Store the encoded value
memcpy(blob->value, encoded_value, length_encoded_value);
return blob;
}
Obi_blob_p obi_str_to_blob(char* value)
{
Obi_blob_p value_b;
int32_t length;
// Compute the number of bytes on which the value will be encoded
length = strlen(value) + 1; // +1 to store \0 at the end (makes retrieving faster)
value_b = obi_blob(value, ELEMENT_SIZE_STR, length, length);
if (value_b == NULL)
{
obidebug(1, "\nError encoding a character string in a blob");
return NULL;
}
return value_b;
}
const char* obi_blob_to_str(Obi_blob_p value_b)
{
return value_b->value;
}
Obi_blob_p obi_seq_to_blob(char* seq)
{
Obi_blob_p value_b;
int32_t length_encoded_seq; // length of the encoded sequence in bytes
int32_t seq_length;
byte_t* encoded_seq;
seq_length = strlen(seq);
// Check if just ATGC and encode accordingly
if (only_ATGC(seq))
{
// Compute the length (in bytes) of the encoded sequence
length_encoded_seq = ceil((double) seq_length / (double) 4.0);
// Encode
encoded_seq = encode_seq_on_2_bits(seq, seq_length);
if (encoded_seq == NULL)
return NULL;
value_b = obi_blob(encoded_seq, ELEMENT_SIZE_SEQ_2, length_encoded_seq, seq_length);
}