Columns: elements names informations are now kept in a memory arena of

adapted size in the header, and added a boolean in the header indicating
whether the values should be evaluated (typically character strings to
be evaluated in Python)
This commit is contained in:
Celine Mercier
2017-12-13 22:46:50 +01:00
parent 2df5932b67
commit 1fd3323372
10 changed files with 177 additions and 111 deletions

View File

@ -139,12 +139,11 @@ static char* build_default_elements_names(index_t nb_elements_per_line);
* column with '\0' as separator (e.g. "0\01\02\0...\0n\0").
*
* @param elements_names The character string formatted with ';' as separator (e.g. "0;1;2;...;n\0").
* @param elts_names_length A pointer on an integer where the function will store the length of the character string.
*
* @since January 2017
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
static void format_elements_names(char* elements_names, int* elts_names_length);
static void format_elements_names(char* elements_names);
/**
@ -188,7 +187,7 @@ static int cmp_elements_names_with_name_and_idx(const void* name1, const void* n
*
* @param column A pointer as returned by obi_create_column().
* @param elements_names The names of the elements as formatted by format_elements_names().
* @param elts_names_length The length of elements_names.
* @param elts_names_length The length of elements_names including the two terminal '\0's.
*
* @retval 0 if the operation was successfully completed.
* @retval -1 if an error occurred.
@ -196,7 +195,19 @@ static int cmp_elements_names_with_name_and_idx(const void* name1, const void* n
* @since July 2015
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
static int set_elements_names(OBIDMS_column_p column, char* elements_names, int elts_names_length);
static int set_elements_names(OBIDMS_column_p column, char* elements_names, int64_t elts_names_length, index_t nb_elements_per_line);
/**
* @brief Internal function reading the informations related to the elements names
* of the lines of a column in the header of the OBIDMS column structure.
*
* @param header A pointer on the header of the column.
*
* @since December 2017
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
static void read_elt_names_informations(OBIDMS_column_header_p header);
/**
@ -224,7 +235,7 @@ static index_t check_elt_names_count(const char* elements_names, bool elt_names_
* @since January 2017
* @author Celine Mercier (celine.mercier@metabarcoding.org)
*/
static int get_formatted_elt_names_length(const char* elements_names);
static int get_formatted_elt_names_length(const char* elements_names, index_t nb_elements);
/**
@ -521,7 +532,15 @@ static char* build_default_elements_names(index_t nb_elements_per_line)
int i;
int len;
elements_names = (char*) malloc(ELEMENTS_NAMES_MAX * sizeof(char));
// if (nb_elements_per_line > NB_ELTS_MAX_IF_DEFAULT_NAME)
// {
// obi_set_errno(OBICOL_UNKNOWN_ERROR);
// obidebug(1, "\nError: too many elements per line to use the default names (max = %d elements)", NB_ELTS_MAX_IF_DEFAULT_NAME);
// return NULL;
// }
// TODO
elements_names = (char*) malloc(nb_elements_per_line * 10 * sizeof(char));
if (elements_names == NULL)
{
obi_set_errno(OBI_MALLOC_ERROR);
@ -529,13 +548,6 @@ static char* build_default_elements_names(index_t nb_elements_per_line)
return NULL;
}
if (nb_elements_per_line > NB_ELTS_MAX_IF_DEFAULT_NAME)
{
obi_set_errno(OBICOL_UNKNOWN_ERROR);
obidebug(1, "\nError: too many elements per line to use the default names (max = %d elements)", NB_ELTS_MAX_IF_DEFAULT_NAME);
return NULL;
}
len = 0;
for (i = 0; i < nb_elements_per_line; i++)
len += sprintf(elements_names+len, "%d;", i);
@ -549,14 +561,15 @@ static char* build_default_elements_names(index_t nb_elements_per_line)
static void format_elements_names(char* elements_names, int* elts_names_length)
static void format_elements_names(char* elements_names)
{
int i;
int i;
int64_t elts_names_length;
*elts_names_length = strlen(elements_names);
elts_names_length = strlen(elements_names);
// Replace the ';' with '\0'
for (i=0; i < *elts_names_length; i++)
for (i=0; i < elts_names_length; i++)
{
if (elements_names[i] == ';')
elements_names[i] = '\0';
@ -570,11 +583,11 @@ static int cmp_elements_names_with_idx(const void* n1_sort_idx, const void* n2_s
char* name1=NULL;
char* name2=NULL;
int name1_idx;
int name2_idx;
index_t name1_idx;
index_t name2_idx;
int name1_sort_idx = *((int*)n1_sort_idx);
int name2_sort_idx = *((int*)n2_sort_idx);
index_t name1_sort_idx = *((index_t*)n1_sort_idx);
index_t name2_sort_idx = *((index_t*)n2_sort_idx);
OBIDMS_column_header_p header = (OBIDMS_column_header_p) h;
name1_idx = (header->elements_names_idx)[name1_sort_idx];
@ -590,10 +603,10 @@ static int cmp_elements_names_with_idx(const void* n1_sort_idx, const void* n2_s
static int cmp_elements_names_with_name_and_idx(const void* name1, const void* n2_sort_idx, const void* h)
{
char* name2=NULL;
int name2_idx;
char* name2=NULL;
index_t name2_idx;
int name2_sort_idx = *((int*)n2_sort_idx);
index_t name2_sort_idx = *((index_t*)n2_sort_idx);
OBIDMS_column_header_p header = (OBIDMS_column_header_p) h;
name2_idx = (header->elements_names_idx)[name2_sort_idx];
@ -603,56 +616,67 @@ static int cmp_elements_names_with_name_and_idx(const void* name1, const void* n
}
static int set_elements_names(OBIDMS_column_p column, char* elements_names, int elts_names_length)
static int set_elements_names(OBIDMS_column_p column, char* elements_names, int64_t elts_names_length, index_t nb_elements)
{
int i, j;
OBIDMS_column_header_p header;
index_t i, j;
// Check that the elements names are not too long
if (elts_names_length+2 > ELEMENTS_NAMES_MAX)
{
obi_set_errno(OBICOL_UNKNOWN_ERROR);
obidebug(1, "\nError: element names too long (%d, max: %d)", elts_names_length+2, ELEMENTS_NAMES_MAX);
return -1;
}
// Copy the elements names in the header
memcpy((column->header)->elements_names, elements_names, elts_names_length*sizeof(char));
// Terminal characters
(column->header)->elements_names[elts_names_length] = '\0';
(column->header)->elements_names[elts_names_length + 1] = '\0';
header = column->header;
// Store the length of the character array containing the elements names
(column->header)->elements_names_length = elts_names_length;
header->elements_names_length = elts_names_length;
// Store the pointers pointing to the different elements stored in the memory arena
header->elements_names = (char*)&(header->mem_arena)[0];
header->elements_names_idx = (index_t*)&((char*)(header->mem_arena) + elts_names_length)[0];
header->sorted_elements_idx = (header->elements_names_idx) + nb_elements;
// Copy the elements names in the header
memcpy(header->elements_names, elements_names, (elts_names_length-2)*sizeof(char));
// Terminal characters
header->elements_names[elts_names_length - 2] = '\0';
header->elements_names[elts_names_length - 1] = '\0';
// Build the elements names index
i = 0;
j = 0;
// Index the first element name
((column->header)->elements_names_idx)[j] = i;
((column->header)->sorted_elements_idx)[j] = j;
(header->elements_names_idx)[j] = i;
(header->sorted_elements_idx)[j] = j;
i++;
j++;
while (i < elts_names_length)
while (i < elts_names_length-2)
{
if (elements_names[i] == '\0')
{ // Index new element name
((column->header)->elements_names_idx)[j] = i+1;
((column->header)->sorted_elements_idx)[j] = j;
(header->elements_names_idx)[j] = i+1;
(header->sorted_elements_idx)[j] = j;
j++;
}
i++;
}
// Build the sorted index
qsort_user_data((column->header)->sorted_elements_idx, j, sizeof(int), column->header, cmp_elements_names_with_idx);
qsort_user_data(header->sorted_elements_idx, j, sizeof(index_t), column->header, cmp_elements_names_with_idx);
return 0;
}
static void read_elt_names_informations(OBIDMS_column_header_p header)
{
int64_t elts_names_length;
elts_names_length = header->elements_names_length;
header->elements_names = (char*)&(header->mem_arena)[0];
header->elements_names_idx = (index_t*)&((char*)(header->mem_arena) + elts_names_length)[0];
header->sorted_elements_idx = (index_t*)&((header->elements_names_idx) + (header->nb_elements_per_line))[0];
}
static index_t check_elt_names_count(const char* elements_names, bool elt_names_formatted)
{
char sep;
@ -680,20 +704,19 @@ static index_t check_elt_names_count(const char* elements_names, bool elt_names_
static int get_formatted_elt_names_length(const char* elements_names)
static int get_formatted_elt_names_length(const char* elements_names, index_t nb_elements)
{
int i = 0;
bool stop = false;
index_t n = 0;
while (! stop)
while (n < nb_elements)
{
if ((elements_names[i] == '\0') && (elements_names[i+1] == '\0'))
stop = true;
else
i++;
if (elements_names[i] == '\0')
n++;
i++;
}
return i;
return i+1;
}
@ -866,14 +889,15 @@ obiversion_t obi_column_get_latest_version_from_name(OBIDMS_p dms, const char* c
}
size_t obi_get_platform_header_size()
// TODO make private
size_t obi_calculate_header_size(index_t nb_elements_per_line, int64_t elts_names_length)
{
size_t header_size;
size_t rounded_header_size;
double multiple;
header_size = sizeof(OBIDMS_column_header_t);
header_size = header_size + (nb_elements_per_line*2)*sizeof(int64_t) + elts_names_length*sizeof(char);
multiple = ceil((double) header_size / (double) getpagesize());
@ -891,6 +915,7 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
char* elements_names,
bool elt_names_formatted,
bool tuples,
bool to_eval,
const char* indexer_name,
const char* associated_column_name,
obiversion_t associated_column_version,
@ -911,7 +936,7 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
OBIType_t stored_data_type;
char* final_indexer_name;
char* built_elements_names = NULL;
int elts_names_length;
int64_t elts_names_length;
new_column = NULL;
@ -1012,12 +1037,11 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
// Format the elements names string
if (! elt_names_formatted)
format_elements_names(elements_names, &elts_names_length);
else
elts_names_length = get_formatted_elt_names_length(elements_names);
format_elements_names(elements_names);
elts_names_length = get_formatted_elt_names_length(elements_names, nb_elements_per_line);
// Calculate the size needed
header_size = obi_get_platform_header_size();
header_size = obi_calculate_header_size(nb_elements_per_line, elts_names_length);
data_size = obi_array_sizeof(stored_data_type, nb_lines, nb_elements_per_line);
file_size = header_size + data_size;
@ -1107,12 +1131,14 @@ OBIDMS_column_p obi_create_column(OBIDMS_p dms,
header->stored_data_type = stored_data_type;
header->returned_data_type = returned_data_type;
header->tuples = tuples;
header->to_eval = to_eval;
header->creation_date = time(NULL);
header->version = version_number;
header->cloned_from = -1;
header->finished = false;
set_elements_names(new_column, elements_names, elts_names_length);
set_elements_names(new_column, elements_names, elts_names_length, nb_elements_per_line);
read_elt_names_informations(header);
// Free the element names if they were built
if (built_elements_names != NULL)
@ -1237,7 +1263,7 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms,
}
// Open the column file, ALWAYS READ-ONLY
column_file_descriptor = openat(column_directory->dir_fd, column_file_name, O_RDONLY);
column_file_descriptor = openat(column_directory->dir_fd, column_file_name, O_RDWR);
if (column_file_descriptor < 0)
{
obi_set_errno(OBICOL_UNKNOWN_ERROR);
@ -1273,7 +1299,7 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms,
column->column_directory = column_directory;
column->header = mmap(NULL,
header_size,
PROT_READ,
PROT_READ | PROT_WRITE,
MAP_SHARED,
column_file_descriptor,
0
@ -1288,6 +1314,9 @@ OBIDMS_column_p obi_open_column(OBIDMS_p dms,
return NULL;
}
// Set the pointers relative to elements names informations in the header
read_elt_names_informations(column->header);
// Map the data
column->data = mmap(NULL,
(column->header)->data_size,
@ -1380,6 +1409,7 @@ OBIDMS_column_p obi_clone_column(OBIDMS_p dms,
(column_to_clone->header)->elements_names,
true,
(column_to_clone->header)->tuples,
(column_to_clone->header)->to_eval,
(column_to_clone->header)->indexer_name,
((column_to_clone->header)->associated_column).column_name,
((column_to_clone->header)->associated_column).version,
@ -1821,8 +1851,8 @@ OBIDMS_column_header_p obi_column_get_header_from_name(OBIDMS_p dms, const char*
return NULL;
}
// Open the column file (READ-ONLY)
column_file_descriptor = openat(column_directory->dir_fd, column_file_name, O_RDONLY);
// Open the column file
column_file_descriptor = openat(column_directory->dir_fd, column_file_name, O_RDWR);
if (column_file_descriptor < 0)
{
obidebug(1, "\nError opening a column file");
@ -1842,10 +1872,10 @@ OBIDMS_column_header_p obi_column_get_header_from_name(OBIDMS_p dms, const char*
return NULL;
}
// Fill the header structure
// Map the header structure
header = mmap(NULL,
header_size,
PROT_READ,
PROT_READ | PROT_WRITE,
MAP_SHARED,
column_file_descriptor,
0
@ -1859,6 +1889,9 @@ OBIDMS_column_header_p obi_column_get_header_from_name(OBIDMS_p dms, const char*
return NULL;
}
// Read the element names informations (storing pointers on informations)
read_elt_names_informations(header);
if (close(column_file_descriptor) < 0)
{
obi_set_errno(OBICOL_UNKNOWN_ERROR);
@ -1888,7 +1921,7 @@ index_t obi_column_get_element_index_from_name(OBIDMS_column_p column, const cha
{
int* elt_names_idx;
elt_names_idx = bsearch_user_data(element_name, (column->header)->sorted_elements_idx, (column->header)->nb_elements_per_line, sizeof(int), column->header, cmp_elements_names_with_name_and_idx);
elt_names_idx = bsearch_user_data(element_name, (column->header)->sorted_elements_idx, (column->header)->nb_elements_per_line, sizeof(index_t), column->header, cmp_elements_names_with_name_and_idx);
if (elt_names_idx != NULL)
return (index_t)(*elt_names_idx);
@ -1906,7 +1939,7 @@ char* obi_get_elements_names(OBIDMS_column_p column)
int elt_idx;
int len;
elements_names = (char*) malloc(ELEMENTS_NAMES_MAX * sizeof(char));
elements_names = (char*) malloc((column->header)->elements_names_length * sizeof(char));
if (elements_names == NULL)
{
obi_set_errno(OBI_MALLOC_ERROR);
@ -2148,6 +2181,8 @@ int obi_clean_unfinished_columns(OBIDMS_p dms)
// dir_to_delete[ddir] = column_dir_path;
// ddir++;
// }
// else
// free(column_dir_path);
free(col_name);
}