obiavl.c 63.2 KB
Newer Older
1 2 3 4 5 6 7 8
/****************************************************************************
 * OBIDMS AVL tree functions	                                            *
 ****************************************************************************/

/**
 * @file obiavl.c
 * @author Celine Mercier
 * @date December 3rd 2015
9
 * @brief Functions handling AVL trees for storing and retrieving blobs.
10 11 12 13 14 15 16 17 18 19 20 21
 */


#include <stdlib.h>
#include <stdio.h>
#include <stdint.h>
#include <string.h>
#include <time.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <math.h>

22
#include "bloom.h"
23
#include "crc64.h"
24
#include "obiavl.h"
25
#include "obiblob.h"
26 27 28 29
#include "obierrno.h"
#include "obitypes.h"
#include "obidebug.h"
#include "encode.h"
30
#include "utils.h"
31 32 33 34 35


#define DEBUG_LEVEL 0	// TODO has to be defined somewhere else (cython compil flag?)


36

37 38 39 40 41 42
/**************************************************************************
 *
 * D E C L A R A T I O N   O F   T H E   P R I V A T E   F U N C T I O N S
 *
 **************************************************************************/

43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

/**
 * @brief Internal function building the full path of an AVL directory containing an AVL or an AVL group.
 *
 * @warning The returned pointer has to be freed by the caller.
 *
 * @param dms A pointer to the OBIDMS to which the AVL tree belongs.
 * @param avl_name The name of the AVL tree or the base name of the AVL tree group.
 *
 * @returns A pointer to the full path of the AVL directory.
 * @retval NULL if an error occurred.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
char* get_full_path_of_avl_dir(OBIDMS_p dms, const char* avl_name);


61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
/**
 * @brief Internal function building the file name for an AVL tree file.
 *
 * @warning The returned pointer has to be freed by the caller.
 *
 * @param avl_name The name of the AVL tree.
 *
 * @returns A pointer to the name of the file where the AVL tree is stored.
 * @retval NULL if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
static char* build_avl_file_name(const char* avl_name);


/**
 * @brief Internal function building the file name for an AVL tree file.
 *
 * @warning The returned pointer has to be freed by the caller.
 *
 * @param avl_name The name of the AVL tree.
 *
 * @returns A pointer to the name of the file where the data referred to by the AVL tree is stored.
 * @retval NULL if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
static char* build_avl_data_file_name(const char* avl_name);


/**
94 95
 * @brief Internal function returning the size of an AVL tree header on this platform,
 *        including the size of the bloom filter associated with the AVL tree.
96 97 98 99 100 101
 *
 * @returns The size of an AVL tree header in bytes.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
102
size_t get_avl_header_size(void);
103 104 105 106 107 108 109 110 111 112


/**
 * @brief Internal function returning the initial size of an AVL tree on this platform.
 *
 * @returns The initial size of an AVL tree in bytes.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
113
size_t get_initial_avl_size(void);
114 115 116 117 118 119 120 121 122 123 124


/**
 * @brief Internal function returning the size, on this platform, of the header of the data
 *        referred to by an AVL tree.
 *
 * @returns The size of an AVL data header in bytes.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
125
size_t get_avl_data_header_size(void);
126 127 128 129 130 131 132 133 134 135 136


/**
 * @brief Internal function returning the initial size, on this platform, of the data
 *        referred to by an AVL tree.
 *
 * @returns The initial size of an AVL data array in bytes.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
137
size_t get_initial_avl_data_size(void);
138 139 140


/**
141
 * @brief Internal function truncating an AVL tree file to the minimum size that is a multiple of the page size.
142
 *
143
 * @param avl A pointer to the AVL tree structure.
144 145 146 147
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
148
 * @since April 2016
149 150
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
int truncate_avl_to_size_used(OBIDMS_avl_p avl);


/**
 * @brief Internal function truncating an AVL tree data file to the minimum size that is a multiple of the page size.
 *
 * @param avl A pointer to the the data structure referred to by an AVL tree.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int truncate_avl_data_to_size_used(OBIDMS_avl_data_p avl_data);
166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192


/**
 * @brief Internal function enlarging an AVL tree.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int grow_avl(OBIDMS_avl_p avl);


/**
 * @brief Internal function enlarging the data array referred to by an AVL tree.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
193
int grow_avl_data(OBIDMS_avl_data_p avl_data);
194 195 196


/**
197
 * @brief Internal function closing an AVL data structure where the data referred to by an AVL tree is stored.
198
 *
199
 * @param avl_data A pointer to the data structure referred to by an AVL tree.
200
 * @param writable Whether the AVL is writable or not (and therefore if the files can and should be truncated to the used size).
201
 *
202
 * @retval 0 if the operation was successfully completed.
203 204 205 206 207
 * @retval -1 if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
208
int close_avl_data(OBIDMS_avl_data_p avl_data, bool writable);
209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239


/**
 * @brief Internal function unmapping the tree and data parts of an AVL tree structure.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int unmap_an_avl(OBIDMS_avl_p avl);


/**
 * @brief Internal function (re)mapping the tree and data parts of an AVL tree structure.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int remap_an_avl(OBIDMS_avl_p avl);


/**
240
 * @brief Internal function creating and adding a new AVL in an AVL group.
241
 *
242 243 244
 * @warning The previous AVL in the list of the group is unmapped,
 * 			if it's not the 1st AVL being added.
 *
245
 * @param avl_group A pointer on the AVL tree group structure.
246 247 248 249 250 251 252 253 254 255
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int add_new_avl_in_group(OBIDMS_avl_group_p avl_group);


256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
/**
 * @brief Internal function adding an existing AVL in an AVL group.
 *
 * The AVL is hard-linked in the AVL group directory, and opened for that group.
 *
 * @param avl_group_dest A pointer on the destination AVL group to which the AVL should be added.
 * @param avl_group_source A pointer on the source AVL group where the AVL already exists.
 * @param avl_idx Index of the AVL in the source AVL group.
 *
 * @retval 0 if the operation was successfully completed.
 * @retval -1 if an error occurred.
 *
 * @since June 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
int add_existing_avl_in_group(OBIDMS_avl_group_p avl_group_dest, OBIDMS_avl_group_p avl_group_source, int avl_idx);


274 275 276 277 278 279
/**
 * @brief Internal function testing if a value might already be stored in an AVL tree.
 *
 * The function checks a bloom filter. No false negatives, possible false positives.
 *
 * @param avl A pointer to the AVL tree structure.
280
 * @param value A pointer to the blob structure.
281 282 283 284 285 286 287
 *
 * @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.
 *
 * @since April 2016
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
288
int maybe_in_avl(OBIDMS_avl_p avl, Obi_blob_p value);
289 290


291
/**
292
 * @brief Internal function storing a value (blob) in the data array referred to by an AVL tree.
293 294
 *
 * @param avl A pointer to the AVL tree structure.
295
 * @param value A pointer to the value (blob structure).
296 297 298 299 300 301 302
 *
 * @returns The index of the stored value.
 * @retval -1 if an error occurred.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
303
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_blob_p value);
304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323


/**
 * @brief Internal function initializing a node in an AVL tree.
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node_idx The index of the node to initialize in the mmapped AVL tree.
 *
 * @returns The node structure initialized.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
AVL_node_p avl_create_node(OBIDMS_avl_p avl, index_t node_idx);


/**
 * @brief Internal function updating the balance factors in an AVL tree
 *        after adding a node, only in the subtree that will have to be balanced.
 *        That subtree is found using the avl->path_idx array and the directions taken
324
 *        down the tree to add the new node are stored in the avl->path_dir array.
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
void avl_update_balance_factors(OBIDMS_avl_p avl);


/**
 * @brief Internal function rotating a node with a "left left rotation".
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node that has to be rotated.
 * @param node_idx The index of the node that has to be rotated.
 *
 * @returns The new root of the subtree.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
index_t avl_rotate_leftleft(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx);

/**
 * @brief Internal function rotating a node with a "left right rotation".
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node that has to be rotated.
 * @param node_idx The index of the node that has to be rotated.
 *
 * @returns The new root of the subtree.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
index_t avl_rotate_leftright(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx);


/**
 * @brief Internal function rotating a node with a "right left rotation".
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node that has to be rotated.
 * @param node_idx The index of the node that has to be rotated.
 *
 * @returns The new root of the subtree.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
index_t avl_rotate_rightleft(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx);


/**
 * @brief Internal function rotating a node with a "right right rotation".
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node that has to be rotated.
 * @param node_idx The index of the node that has to be rotated.
 *
 * @returns The new root of the subtree.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
index_t avl_rotate_rightright(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx);


/**
 * @brief Internal function balancing one node.
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node that has to be balanced.
 * @param node_idx The index of the node that has to be balanced.
 *
 * @returns The new root of the subtree.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
index_t avl_balance_node(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx);


/**
 * @brief Internal function balancing the nodes of an AVL tree after adding a node,
 *        only in the subtree that eventually has to be balanced.
 *        That subtree is found using the avl->path_idx array.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
void avl_balance(OBIDMS_avl_p avl);


/**
 * @brief Internal function printing a depth first traverse of a node.
 *
 * @param avl A pointer to the AVL tree structure.
 * @param node A pointer to the node.
 * @param node_idx The index of the node.
 * @param depth The depth of the node.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
void avl_print_node(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx, int depth);


/**
 * @brief Internal function printing a depth first traverse of an AVL tree.
 *
 * @param avl A pointer to the AVL tree structure.
 *
 * @since December 2015
 * @author Celine Mercier (celine.mercier@metabarcoding.org)
 */
void avl_print(OBIDMS_avl_p avl);


/************************************************************************
 *
 * D E F I N I T I O N   O F   T H E   P R I V A T E   F U N C T I O N S
 *
 ************************************************************************/

452 453 454 455 456

char* get_full_path_of_avl_dir(OBIDMS_p dms, const char* avl_name)
{
	char* avl_dir_name;

457
	avl_dir_name = obi_dms_get_full_path(dms, INDEXER_DIR_NAME);
458 459 460 461 462 463 464 465 466 467 468 469
	if (avl_dir_name == NULL)
	{
		obidebug(1, "\nError getting path for the DMS AVL directory");
		return NULL;
	}
	strcat(avl_dir_name, "/");
	strcat(avl_dir_name, avl_name);

	return avl_dir_name;
}


470 471 472 473
static char* build_avl_file_name(const char* avl_name)
{
	char* file_name;

474 475
	// Test if the AVL name is not too long
	if (strlen(avl_name) >= AVL_MAX_NAME)
476 477
	{
		obi_set_errno(OBI_AVL_ERROR);
478
		obidebug(1, "\nError due to AVL tree name too long");
479 480 481
		return NULL;
	}

482 483 484 485 486 487 488 489 490
	// Build the file name
	file_name = (char*) malloc((strlen(avl_name) + 5)*sizeof(char));
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for an AVL file name");
		return NULL;
	}
	if (sprintf(file_name,"%s.oda", avl_name) < 0)
491 492
	{
		obi_set_errno(OBI_AVL_ERROR);
493
		obidebug(1, "\nError building an AVL tree file name");
494 495 496 497 498 499 500 501 502 503 504 505 506
		free(file_name);
		return NULL;
	}

	return file_name;
}


static char* build_avl_data_file_name(const char* avl_name)
{
	char* file_name;

	// Build the file name
Celine Mercier's avatar
Celine Mercier committed
507
	file_name = (char*) malloc((strlen(avl_name) + 5)*sizeof(char));
508 509 510 511 512 513
	if (file_name == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for an AVL data file name");
		return NULL;
	}
Celine Mercier's avatar
Celine Mercier committed
514
	if (sprintf(file_name,"%s.odd", avl_name) < 0)
515 516
	{
		obi_set_errno(OBI_AVL_ERROR);
517
		obidebug(1, "\nError building an AVL tree data file name");
518 519 520 521 522 523 524 525 526 527 528 529 530
		return NULL;
	}

	return file_name;
}


size_t get_avl_header_size()
{
	size_t header_size;
	size_t rounded_header_size;
	double multiple;

531
	header_size = sizeof(OBIDMS_avl_header_t) + bloom_filter_size(MAX_NODE_COUNT_PER_AVL, BLOOM_FILTER_ERROR_RATE);
532

533
	multiple = ceil((double) header_size / (double) getpagesize());
534 535 536 537 538 539 540 541 542

	rounded_header_size = multiple * getpagesize();

	return rounded_header_size;
}


size_t get_initial_avl_size()
{
543 544 545 546 547 548 549
	size_t s;
	size_t m;

	m = 1;
	s = getpagesize() * m;

	return s;
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570
}


size_t get_avl_data_header_size()
{
	size_t header_size;
	size_t rounded_header_size;
	double multiple;

	header_size = sizeof(OBIDMS_avl_data_header_t);

	multiple = 	ceil((double) header_size / (double) getpagesize());

	rounded_header_size = multiple * getpagesize();

	return rounded_header_size;
}


size_t get_initial_avl_data_size()
{
571 572 573 574 575 576 577
	size_t s;
	size_t m;

	m = 1;
	s = getpagesize() * m;

	return s;
578 579 580
}


581
int truncate_avl_to_size_used(OBIDMS_avl_p avl)	// TODO is it necessary to unmap/remap?
582
{
583 584 585 586 587 588 589
	size_t  			file_size;
	size_t  			new_data_size;
	double  			multiple;
	int 				file_descriptor;

	// Compute the new size: used size rounded to the nearest greater multiple of page size greater than 0
	multiple = ceil((double) (ONE_IF_ZERO((avl->header)->nb_items * sizeof(AVL_node_t))) / (double) getpagesize());
590
	new_data_size = ((size_t) multiple) * getpagesize();
591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614

	// Check that it is actually greater than the current size of the file, otherwise no need to truncate
	if ((avl->header)->avl_size == new_data_size)
		return 0;

	// Get the file descriptor
	file_descriptor = avl->avl_fd;

	// Unmap the tree before truncating the file
	if (munmap(avl->tree, (avl->header)->avl_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError munmapping the tree of an AVL before truncating");
		return -1;
	}

	// Truncate the file
	file_size = (avl->header)->header_size + new_data_size;
	if (ftruncate(file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError truncating an AVL file");
		return -1;
	}
615

616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632
	// Remap the data
	avl->tree = mmap(NULL,
			         new_data_size,
					 PROT_READ | PROT_WRITE,
					 MAP_SHARED,
					 file_descriptor,
					 (avl->header)->header_size
					);
	if (avl->tree == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError re-mmapping the tree of an AVL after truncating");
		return -1;
	}

	// Set new data size and new max node count
	(avl->header)->avl_size     = new_data_size;
633
	(avl->header)->nb_items_max	= (index_t) floor(((double) (avl->header)->avl_size) / ((double) sizeof(AVL_node_t)));
634 635 636 637 638 639 640 641 642 643 644 645 646 647

	return 0;
}


int truncate_avl_data_to_size_used(OBIDMS_avl_data_p avl_data)	// TODO is it necessary to unmap/remap?
{
	size_t  			file_size;
	index_t  			new_data_size;
	double  			multiple;
	int 				file_descriptor;

	// Compute the new size: used size rounded to the nearest greater multiple of page size greater than 0
	multiple = ceil((double) (ONE_IF_ZERO((avl_data->header)->data_size_used)) / (double) getpagesize());
648
	new_data_size = ((index_t) multiple) * getpagesize();
649 650 651 652 653 654 655 656 657

	// Check that it is actually greater than the current size of the file, otherwise no need to truncate
	if ((avl_data->header)->data_size_max == new_data_size)
		return 0;

	// Get the file descriptor
	file_descriptor = avl_data->data_fd;

	// Unmap the data before truncating the file
658 659 660
	if (munmap(avl_data->data, (avl_data->header)->data_size_max) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
661 662
		obidebug(1, "\nError munmapping the data of an AVL before truncating");
		return -1;
663 664
	}

665 666 667
	// Truncate the file
	file_size = (avl_data->header)->header_size + new_data_size;
	if (ftruncate(file_descriptor, file_size) < 0)
668 669
	{
		obi_set_errno(OBI_AVL_ERROR);
670 671
		obidebug(1, "\nError truncating an AVL data file");
		return -1;
672 673
	}

674 675 676 677 678 679 680 681
	// Remap the data
	avl_data->data = mmap(NULL,
					      new_data_size,
						  PROT_READ | PROT_WRITE,
						  MAP_SHARED,
						  file_descriptor,
						  (avl_data->header)->header_size
					     );
682

683 684 685 686 687 688 689 690 691 692 693
	if (avl_data->data == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError re-mmapping the data of an AVL after truncating");
		return -1;
	}

	// Set new data size
	(avl_data->header)->data_size_max = new_data_size;

	return 0;
694 695 696 697 698 699 700 701 702 703 704
}


int grow_avl(OBIDMS_avl_p avl)		// TODO Lock when needed
{
	size_t	file_size;
	size_t 	old_data_size;
	size_t 	new_data_size;
	size_t 	header_size;
	int 	avl_file_descriptor;

705
	avl_file_descriptor = avl->avl_fd;
706 707 708 709 710 711 712 713 714 715 716

	// Calculate the new file size
	old_data_size = (avl->header)->avl_size;
	new_data_size = old_data_size * AVL_GROWTH_FACTOR;
	header_size = (avl->header)->header_size;
	file_size = header_size + new_data_size;

	// Enlarge the file
	if (ftruncate(avl_file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
717
		obidebug(1, "\nError enlarging an AVL tree file");
718 719 720 721 722 723 724 725
		return -1;
	}

	// Unmap and re-map the data

	if (munmap(avl->tree, old_data_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
726
		obidebug(1, "\nError munmapping the tree of an AVL tree file before enlarging");
727 728 729 730
		return -1;
	}

	avl->tree = mmap(NULL,
731 732 733 734 735 736
					 new_data_size,
					 PROT_READ | PROT_WRITE,
					 MAP_SHARED,
					 avl_file_descriptor,
					 header_size
					);
737 738 739 740

	if (avl->tree == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
741
		obidebug(1, "\nError re-mmapping the tree of an AVL tree file after enlarging the file");
742 743 744 745 746 747
		return -1;
	}

	// Set the new avl size
	(avl->header)->avl_size = new_data_size;

748 749 750
	// Set new maximum number of items
	(avl->header)->nb_items_max	 = (index_t) floor(((double) (avl->header)->avl_size) / ((double) sizeof(AVL_node_t)));

751 752 753 754
	return 0;
}


755
int grow_avl_data(OBIDMS_avl_data_p avl_data)		// TODO Lock when needed
756 757 758 759 760 761 762
{
	size_t 	file_size;
	index_t old_data_size;
	index_t new_data_size;
	size_t 	header_size;
	int 	avl_data_file_descriptor;

763
	avl_data_file_descriptor = avl_data->data_fd;
764 765

	// Calculate the new file size
766
	old_data_size = (avl_data->header)->data_size_max;
767
	new_data_size = old_data_size * AVL_GROWTH_FACTOR;
768
	header_size = (avl_data->header)->header_size;
769 770 771 772 773 774
	file_size = header_size + new_data_size;

	// Enlarge the file
	if (ftruncate(avl_data_file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
775
		obidebug(1, "\nError enlarging an AVL tree data file");
776 777 778 779 780
		return -1;
	}

	// Unmap and re-map the data

781
	if (munmap(avl_data->data, old_data_size) < 0)
782 783
	{
		obi_set_errno(OBI_AVL_ERROR);
784
		obidebug(1, "\nError munmapping the data of an AVL tree data file before enlarging");
785 786 787
		return -1;
	}

788 789 790 791 792 793 794
	avl_data->data = mmap(NULL,
						  new_data_size,
						  PROT_READ | PROT_WRITE,
						  MAP_SHARED,
						  avl_data_file_descriptor,
						  header_size
						 );
795

796
	if (avl_data->data == MAP_FAILED)
797 798
	{
		obi_set_errno(OBI_AVL_ERROR);
799
		obidebug(1, "\nError re-mmapping the data of an AVL tree data file after enlarging the file");
800 801 802 803
		return -1;
	}

	// Set new data size
804
	(avl_data->header)->data_size_max = new_data_size;
805

806
	// Initialize new data to 0
807
	memset((avl_data->data)+old_data_size, 0, new_data_size - old_data_size);
808 809 810 811 812

	return 0;
}


813
int close_avl_data(OBIDMS_avl_data_p avl_data, bool writable)
814
{
815
	int ret_val = 0;
816

817 818
	if (writable)
		ret_val = truncate_avl_data_to_size_used(avl_data);
819

820
	if (munmap(avl_data->data, (avl_data->header)->data_size_max) < 0)
821
	{
822 823 824
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError munmapping the data of an AVL tree data file");
		ret_val = -1;
825 826
	}

827 828 829 830 831 832
	if (munmap(avl_data->header, (avl_data->header)->header_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError munmapping the header of an AVL tree data file");
		ret_val = -1;
	}
833

834 835 836 837 838 839
	if (close(avl_data->data_fd) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError closing an AVL tree data file");
		ret_val = -1;
	}
840

841
	free(avl_data);
842

843
	return ret_val;
844 845 846
}


847
int unmap_an_avl(OBIDMS_avl_p avl)
848
{
849 850 851 852 853 854
	if (munmap((avl->data)->data, ((avl->data)->header)->data_size_max) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError unmapping the data of an AVL tree");
		return -1;
	}
855
	if (munmap(avl->tree, (avl->header)->avl_size) < 0)
856 857 858 859 860 861 862
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError unmapping the tree of an AVL tree");
		return -1;
	}
	return 0;
}
863 864


865 866 867 868 869 870 871 872 873 874 875 876 877 878
int remap_an_avl(OBIDMS_avl_p avl)
{
	(avl->data)->data = mmap(NULL,
							 ((avl->data)->header)->data_size_max,
							 PROT_READ,
							 MAP_SHARED,
							 (avl->data)->data_fd,
							 ((avl->data)->header)->header_size);
	if ((avl->data)->data == NULL)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError mapping the data of an AVL tree");
		return -1;
	}
879

880
	avl->tree = mmap(NULL,
881
					 (avl->header)->avl_size,
882 883 884 885 886 887 888 889 890 891 892 893 894
					 PROT_READ,
					 MAP_SHARED,
					 avl->avl_fd,
					 (avl->header)->header_size);
	if (avl->tree == NULL)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError mapping the tree of an AVL tree");
		return -1;
	}

	return 0;
}
895 896


897 898 899
int add_new_avl_in_group(OBIDMS_avl_group_p avl_group)
{
	// Check that maximum number of AVLs in a group was not reached
900
	if (avl_group->last_avl_idx == (MAX_NB_OF_AVLS_IN_GROUP - 1))
901
	{
902 903 904 905 906
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError: Trying to add new AVL in AVL group but maximum number of AVLs in a group reached");
		return -1;
	}

907
	// Unmap the previous AVL if it's not the 1st
908 909 910 911
	// Not done anymore, currently keeping all mapped for efficiency reasons
//	if (avl_group->last_avl_idx > 0)
//		if (unmap_an_avl((avl_group->sub_avls)[avl_group->last_avl_idx]) < 0)
//			return -1;
912 913

	// Increment current AVL index
914
	(avl_group->last_avl_idx)++;
915

916
	// Create the new AVL
917 918
	(avl_group->sub_avls)[avl_group->last_avl_idx] = obi_create_avl(avl_group->dms, avl_group->name, avl_group->last_avl_idx);
	if ((avl_group->sub_avls)[avl_group->last_avl_idx] == NULL)
919 920 921
	{
		obidebug(1, "\nError creating a new AVL tree in a group");
		return -1;
922 923
	}

924 925 926 927
	return 0;
}


928 929
int add_existing_avl_in_group(OBIDMS_avl_group_p avl_group_dest, OBIDMS_avl_group_p avl_group_source, int avl_idx)
{
930
	if (link(obi_get_full_path_of_avl_file_name(avl_group_source->dms, avl_group_source->name, avl_idx), obi_get_full_path_of_avl_file_name(avl_group_dest->dms, avl_group_dest->name, avl_idx)) < 0)
931 932 933 934 935
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError creating a hard link to an existing AVL tree file");
		return -1;
	}
936
	if (link(obi_get_full_path_of_avl_data_file_name(avl_group_source->dms, avl_group_source->name, avl_idx), obi_get_full_path_of_avl_data_file_name(avl_group_dest->dms, avl_group_dest->name, avl_idx)) < 0)
937 938 939 940 941 942 943 944 945
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError creating a hard link to an existing AVL data file");
		return -1;
	}

	// Increment current AVL index
	(avl_group_dest->last_avl_idx)++;

946
	// Open AVL for that group 		TODO ideally not needed because AVL open twice, but needed for now
947 948 949 950 951 952 953 954 955 956
	avl_group_dest->sub_avls[avl_group_dest->last_avl_idx] = obi_open_avl(avl_group_source->dms, avl_group_source->name, avl_idx);
	if ((avl_group_dest->sub_avls)[avl_group_dest->last_avl_idx] == NULL)
	{
		obidebug(1, "\nError opening an AVL to add in an AVL group");
		return -1;
	}

	return 0;
}

957

958
int maybe_in_avl(OBIDMS_avl_p avl, Obi_blob_p value)
959
{
960
	return (bloom_check(&((avl->header)->bloom_filter), value, obi_blob_sizeof(value)));
961 962 963
}


964
index_t avl_add_value_in_data_array(OBIDMS_avl_p avl, Obi_blob_p value)
965 966
{
	index_t value_idx;
967
	int value_size;
968 969 970 971

	value_idx = ((avl->data)->header)->data_size_used;

	// Grow the data if needed
972
	value_size = obi_blob_sizeof(value);
973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988
	while (((avl->data)->header)->data_size_max < (value_idx + value_size))
	{
		if (grow_avl_data(avl->data) < 0)
			return -1;
	}

	// Store the value itself at the end of the data
	memcpy((((avl->data)->data)+value_idx), value, value_size);

	// Update the data size
	((avl->data)->header)->data_size_used = value_idx + value_size;

	// Update the number of items
	(((avl->data)->header)->nb_items)++;

	return value_idx;
989 990 991 992 993 994 995 996 997
}


AVL_node_p avl_create_node(OBIDMS_avl_p avl, index_t node_idx)
{
	AVL_node_p node;

	node = (avl->tree)+node_idx;

998 999
	node->left_child     = -1;
	node->right_child    = -1;
1000
	node->balance_factor = 0;
1001 1002
	node->value          = -1;
	node->crc64          = 0; 	// TODO no NA value
1003 1004 1005 1006 1007 1008 1009 1010

	return node;
}


// Update the balance factors of the nodes from the node that will need balancing
void avl_update_balance_factors(OBIDMS_avl_p avl)
{
1011
	uint8_t    n;
1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
	AVL_node_p node;

	// Update balance factors from the node where balancing might be needed
	node=(avl->tree)+((avl->path_idx)[1]);

	for (n=1; (avl->path_dir)[n] != -1; n++)
	{
		if ((avl->path_dir)[n]) // Went right
		{
			(node->balance_factor)--;
			node=RIGHT_CHILD(node);
		}
		else 	// Went left
		{
			(node->balance_factor)++;
			node=LEFT_CHILD(node);
		}
	}
}


// Left Left Rotate
index_t avl_rotate_leftleft(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx)
{
	AVL_node_p left_child = LEFT_CHILD(node);
	index_t left_child_idx = node->left_child;

	node->left_child = left_child->right_child;
	left_child->right_child = node_idx;

	node->balance_factor = 0;
	left_child->balance_factor = 0;

	return left_child_idx;
}


// Left Right Rotate
index_t avl_rotate_leftright(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx)
{
	AVL_node_p left_child = LEFT_CHILD(node);
	index_t left_child_idx = node->left_child;
	AVL_node_p rc_of_lc = RIGHT_CHILD(left_child);
	index_t rc_of_lc_idx = left_child->right_child;

	node->left_child = rc_of_lc->right_child;
	left_child->right_child = rc_of_lc->left_child;
	rc_of_lc->left_child = left_child_idx;
	rc_of_lc->right_child = node_idx;

	if (rc_of_lc->balance_factor == -1)
	{
		left_child->balance_factor = 1;
		node->balance_factor = 0;
	}
	else if (rc_of_lc->balance_factor == 0)
	{
		left_child->balance_factor = 0;
		node->balance_factor = 0;
	}
	else	// if (rc_of_lc->balance_factor == 1)
	{
		left_child->balance_factor = 0;
		node->balance_factor = -1;
	}

	rc_of_lc->balance_factor = 0;

	return rc_of_lc_idx;
}


// Right Left Rotate
index_t avl_rotate_rightleft(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx)
{
	AVL_node_p right_child = RIGHT_CHILD(node);
	index_t right_child_idx = node->right_child;;
	AVL_node_p lc_of_rc = LEFT_CHILD(right_child);
	index_t lc_of_rc_idx = right_child->left_child;

	node->right_child = lc_of_rc->left_child;
	right_child->left_child = lc_of_rc->right_child;
	lc_of_rc->right_child = right_child_idx;
	lc_of_rc->left_child = node_idx;

	if (lc_of_rc->balance_factor == 1)
	{
		right_child->balance_factor = 1;
		node->balance_factor = 0;
	}
	else if (lc_of_rc->balance_factor == 0)
	{
		right_child->balance_factor = 0;
		node->balance_factor = 0;
	}
	else	// if (lc_of_rc->balance_factor == -1)
	{
		right_child->balance_factor = 0;
		node->balance_factor = 1;
	}

	lc_of_rc->balance_factor = 0;

	return lc_of_rc_idx;
}


// Right Right Rotate
index_t avl_rotate_rightright(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx)
{
	AVL_node_p right_child = RIGHT_CHILD(node);
	index_t right_child_idx = node->right_child;

	node->right_child = right_child->left_child;
	right_child->left_child = node_idx;

	node->balance_factor = 0;
	right_child->balance_factor = 0;

	return right_child_idx;
}


// Balance a given node
index_t avl_balance_node(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx)
{
	index_t new_root = 0;

	if (node->balance_factor == 2)
	{ // Left Heavy
		if ((LEFT_CHILD(node))->balance_factor == -1)
			new_root = avl_rotate_leftright(avl, node, node_idx);
		else
			new_root = avl_rotate_leftleft(avl, node, node_idx);
	}
	else if (node->balance_factor == -2)
	{ // Right Heavy
		if ((RIGHT_CHILD(node))->balance_factor == 1)
			new_root = avl_rotate_rightleft(avl, node, node_idx);
		else
			new_root = avl_rotate_rightright(avl, node, node_idx);
	}
	else
	// Node is balanced
		new_root = node_idx;

	return new_root;
}


// Balance a given tree
void avl_balance(OBIDMS_avl_p avl)
{
1165 1166
	index_t    new_root;
	index_t    node_index;
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
	AVL_node_p node_to_balance;
	AVL_node_p parent_of_node_to_balance;

	node_index = (avl->path_idx)[1];
	node_to_balance = (avl->tree)+node_index;
	parent_of_node_to_balance = (avl->tree)+((avl->path_idx)[0]);

	// Balance the 2nd node stored in the path (the first is only kept to connect the new root
	// of the subtree if needed).
	new_root = avl_balance_node(avl, node_to_balance, node_index);

	if (new_root != node_index)
	// If the root of the subtree has changed
	{
		// If the subtree's root is the tree's root, store the new root
		if (node_index == (avl->header)->root_idx)
			(avl->header)->root_idx = new_root;
		// Else, connect the new subtree's root to the parent of the subtree
		else if ((avl->path_dir)[0])	   // Subtree is the right child of its parent
			parent_of_node_to_balance->right_child = new_root;
		else	   // Subtree is the left child of its parent
			parent_of_node_to_balance->left_child = new_root;
	}
}


// Print a depth first traverse of a node
void avl_print_node(OBIDMS_avl_p avl, AVL_node_p node, index_t node_idx, int depth)
{
	int i = 0;

	if (node->left_child != -1)
		avl_print_node(avl, LEFT_CHILD(node), node->left_child, depth+2);

	for (i = 0; i < depth; i++)
		putchar(' ');

	fprintf(stderr, "Node idx: %lld, Value idx: %lld, Left child: %lld, Right child: %lld, "
1205 1206
			"Balance factor: %d, CRC: %llu\nValue:", node_idx, node->value, node->left_child, node->right_child, node->balance_factor, node->crc64);
	print_bits(((avl->data)->data)+(node->value), obi_blob_sizeof((Obi_blob_p)(((avl->data)->data)+(node->value))));
1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226

	if (node->right_child != -1)
		avl_print_node(avl, RIGHT_CHILD(node), node->right_child, depth+2);
}


// Print a depth first traverse of a tree
void avl_print(OBIDMS_avl_p avl)
{
	fprintf(stderr, "\nRoot index: %lld\n", (avl->header)->root_idx);
	avl_print_node(avl, (avl->tree)+((avl->header)->root_idx), (avl->header)->root_idx, 0);
}


/**********************************************************************
 *
 * D E F I N I T I O N   O F   T H E   P U B L I C   F U N C T I O N S
 *
 **********************************************************************/

1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354

char* obi_build_avl_name_with_idx(const char* avl_name, int avl_idx)
{
	char* avl_name_with_idx;
	int   avl_idx_length;

	if (avl_idx < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError building an AVL tree name with index: index < 0");
		return NULL;
	}

	avl_idx_length = avl_idx == 0 ? 1 : (int)(log10(avl_idx)+1);
	avl_name_with_idx = malloc((strlen(avl_name) + avl_idx_length + 2)*sizeof(char));
	if (avl_name_with_idx == NULL)
	{
		obi_set_errno(OBI_MALLOC_ERROR);
		obidebug(1, "\nError allocating memory for an AVL name");
		return NULL;
	}
	if (sprintf(avl_name_with_idx, "%s_%u", avl_name, avl_idx) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError building an AVL tree name with index");
		return NULL;
	}

	return avl_name_with_idx;
}


char* obi_get_full_path_of_avl_file_name(OBIDMS_p dms, const char* avl_name, int avl_idx)
{
	char* complete_avl_name;
	char* full_path;
	char* avl_file_name;

	if (avl_idx >= 0)
	{
		complete_avl_name = obi_build_avl_name_with_idx(avl_name, avl_idx);
		if (complete_avl_name == NULL)
			return NULL;
	}
	else
	{
		complete_avl_name = (char*) malloc((strlen(avl_name)+1)*sizeof(char));
		if (complete_avl_name == NULL)
		{
    		obi_set_errno(OBI_MALLOC_ERROR);
    		obidebug(1, "\nError allocating memory for an AVL name");
			return NULL;
		}
		strcpy(complete_avl_name, avl_name);
	}

	avl_file_name = build_avl_file_name(complete_avl_name);
	if (avl_file_name == NULL)
	{
		free(complete_avl_name);
		return NULL;
	}

	full_path = get_full_path_of_avl_dir(dms, avl_name);
	if (full_path == NULL)
	{
		free(complete_avl_name);
		free(avl_file_name);
		return NULL;
	}

	strcat(full_path, "/");
	strcat(full_path, avl_file_name);

	free(complete_avl_name);

	return full_path;
}


char* obi_get_full_path_of_avl_data_file_name(OBIDMS_p dms, const char* avl_name, int avl_idx)
{
	char* complete_avl_name;
	char* full_path;
	char* avl_data_file_name;

	if (avl_idx >= 0)
	{
		complete_avl_name = obi_build_avl_name_with_idx(avl_name, avl_idx);
		if (complete_avl_name == NULL)
			return NULL;
	}
	else
	{
		complete_avl_name = (char*) malloc((strlen(avl_name)+1)*sizeof(char));
		if (complete_avl_name == NULL)
		{
    		obi_set_errno(OBI_MALLOC_ERROR);
    		obidebug(1, "\nError allocating memory for an AVL name");
			return NULL;
		}
		strcpy(complete_avl_name, avl_name);
	}

	avl_data_file_name = build_avl_data_file_name(complete_avl_name);
	if (avl_data_file_name == NULL)
	{
		free(complete_avl_name);
		return NULL;
	}

	full_path = get_full_path_of_avl_dir(dms, avl_name);
	if (full_path == NULL)
	{
		free(complete_avl_name);
		free(avl_data_file_name);
		return NULL;
	}

	strcat(full_path, "/");
	strcat(full_path, avl_data_file_name);

	free(complete_avl_name);

	return full_path;
}


1355 1356
int obi_avl_exists(OBIDMS_p dms, const char* avl_name)
{
1357
	struct stat buffer;
1358
	char* 	    avl_dir_path;
1359
	int 	    check_dir;
1360

Celine Mercier's avatar
Celine Mercier committed
1361
	// Build the AVL tree file path
1362
	avl_dir_path = get_full_path_of_avl_dir(dms, avl_name);
1363
	if (avl_dir_path == NULL)
1364 1365
		return -1;

1366
	check_dir = stat(avl_dir_path, &buffer);
1367

1368
	free(avl_dir_path);
1369 1370 1371 1372 1373 1374 1375 1376

	if (check_dir == 0)
		return 1;
	else
		return 0;
}


1377
OBIDMS_avl_p obi_create_avl(OBIDMS_p dms, const char* avl_name, int avl_idx)
1378
{
1379 1380
	char* 				complete_avl_name;
	char*				avl_dir_name;
1381 1382 1383 1384 1385 1386 1387
	char* 				avl_file_name;
	char* 				avl_data_file_name;
	size_t				header_size;
	size_t				data_size;
	size_t				file_size;
	int					avl_file_descriptor;
	int					avl_data_file_descriptor;
1388
	int					avl_dir_fd;
1389 1390
	OBIDMS_avl_data_p   avl_data;
	OBIDMS_avl_p		avl;
1391 1392 1393 1394 1395 1396 1397
 	DIR*				directory;
    struct stat 		buffer;
    int 				check_dir;

	// Get complete name of AVL if index
	if (avl_idx >= 0)
	{
1398
		complete_avl_name = obi_build_avl_name_with_idx(avl_name, avl_idx);
1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
		if (complete_avl_name == NULL)
			return NULL;
	}
	else
	{
		complete_avl_name = (char*) malloc((strlen(avl_name)+1)*sizeof(char));
		if (complete_avl_name == NULL)
		{
    		obi_set_errno(OBI_MALLOC_ERROR);
    		obidebug(1, "\nError allocating memory for an AVL name");
			return NULL;
		}
		strcpy(complete_avl_name, avl_name);
	}

	// Create that AVL's directory if needed
	avl_dir_name = get_full_path_of_avl_dir(dms, avl_name);
	if (avl_dir_name == NULL)
		return NULL;
	// Check if the AVL's directory already exists
	check_dir = stat(avl_dir_name, &buffer);
	// Create that AVL's directory if it doesn't already exist
   	if (check_dir < 0)
   	{
Celine Mercier's avatar
Celine Mercier committed
1423
		if (mkdirat(dms->indexer_dir_fd, avl_dir_name, 00777) < 0)
1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451
		{
			obi_set_errno(OBI_AVL_ERROR);
			obidebug(1, "\nError creating an AVL directory");
			if (avl_idx >= 0)
				free(complete_avl_name);
			free(avl_dir_name);
			return NULL;
		}
   	}
    // Open the AVL directory
    directory = opendir(avl_dir_name);
	if (directory == NULL)
	{
		obidebug(1, "\nError opening an AVL directory");
		if (avl_idx >= 0)
			free(complete_avl_name);
		free(avl_dir_name);
		return NULL;
	}
	free(avl_dir_name);
    avl_dir_fd = dirfd(directory);
	if (avl_dir_fd < 0)
	{
		obidebug(1, "\nError getting an AVL directory file descriptor");
		if (avl_idx >= 0)
			free(complete_avl_name);
		return NULL;
	}
1452 1453 1454 1455

	// Create the data file

		// Build file name
1456
	avl_data_file_name = build_avl_data_file_name(complete_avl_name);
1457
	if (avl_data_file_name == NULL)
1458 1459 1460
	{
		if (avl_idx >= 0)
			free(complete_avl_name);
1461
		return NULL;
1462
	}
1463 1464

		// Create file
1465
	avl_data_file_descriptor = openat(avl_dir_fd, avl_data_file_name, O_RDWR | O_CREAT | O_EXCL, 0777);
1466 1467 1468
	if (avl_data_file_descriptor < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1469
		obidebug(1, "\nError creating an AVL tree data file");
1470 1471
		if (avl_idx >= 0)
			free(complete_avl_name);
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
		free(avl_data_file_name);
		return NULL;
	}
	free(avl_data_file_name);

		// Calculate the size needed
	header_size = get_avl_data_header_size();
	data_size   = get_initial_avl_data_size();
	file_size   = header_size + data_size;

1482
		// Truncate the AVL tree data file to the right size
1483 1484 1485
	if (ftruncate(avl_data_file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1486
		obidebug(1, "\nError truncating an AVL tree data file to the right size");
1487 1488
		if (avl_idx >= 0)
			free(complete_avl_name);
1489 1490 1491 1492
		close(avl_data_file_descriptor);
		return NULL;
	}

1493
		// Allocate the memory for the AVL tree data structure
1494 1495 1496
	avl_data = (OBIDMS_avl_data_p) malloc(sizeof(OBIDMS_avl_data_t));
	if (avl_data == NULL)
	{
1497
		obi_set_errno(OBI_MALLOC_ERROR);
1498
		obidebug(1, "\nError allocating the memory for the AVL tree data structure");
1499 1500
		if (avl_idx >= 0)
			free(complete_avl_name);
1501 1502 1503 1504
		close(avl_data_file_descriptor);
		return NULL;
	}

1505
		// Fill the AVL tree data structure
1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
	avl_data->header = mmap(NULL,
			                header_size,
			                PROT_READ | PROT_WRITE,
			                MAP_SHARED,
			                avl_data_file_descriptor,
			                0
			               );
	if (avl_data->header == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1516
		obidebug(1, "\nError mmapping the header of an AVL tree data file");
1517 1518
		if (avl_idx >= 0)
			free(complete_avl_name);
1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
		close(avl_data_file_descriptor);
		free(avl_data);
		return NULL;
	}

	avl_data->data   = mmap(NULL,
			                data_size,
			                PROT_READ | PROT_WRITE,
			                MAP_SHARED,
			                avl_data_file_descriptor,
			                header_size
			               );
	if (avl_data->data == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1534
		obidebug(1, "\nError mmapping the data of an AVL tree data file");
1535 1536
		if (avl_idx >= 0)
			free(complete_avl_name);
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547
		munmap(avl_data->header, header_size);
		close(avl_data_file_descriptor);
		free(avl_data);
		return NULL;
	}

	(avl_data->header)->header_size 	 = header_size;
	(avl_data->header)->data_size_max	 = data_size;
	(avl_data->header)->data_size_used   = 0;
	(avl_data->header)->nb_items	 	 = 0;
	(avl_data->header)->creation_date    = time(NULL);
1548 1549 1550
	strcpy((avl_data->header)->avl_name, complete_avl_name);

	avl_data->data_fd = avl_data_file_descriptor;
1551 1552 1553 1554 1555

	// Initialize all bits to 0
	memset(avl_data->data, 0, (avl_data->header)->data_size_max);


1556
	// Create the AVL tree file
1557 1558

		// Build file name
1559
	avl_file_name = build_avl_file_name(complete_avl_name);
1560 1561
	if (avl_file_name == NULL)
	{
1562
		close_avl_data(avl_data, true);
1563 1564 1565 1566 1567 1568 1569 1570 1571
		return NULL;
	}

		// Calculate the size needed
	header_size = get_avl_header_size();
	data_size = get_initial_avl_size();
	file_size = header_size + data_size;

		// Create file
1572
	avl_file_descriptor = openat(avl_dir_fd, avl_file_name, O_RDWR | O_CREAT | O_EXCL, 0777);
1573 1574 1575
	if (avl_file_descriptor < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1576
		obidebug(1, "\nError creating an AVL tree file");
1577 1578
		if (avl_idx >= 0)
			free(complete_avl_name);
1579
		close_avl_data(avl_data, true);
1580 1581 1582 1583 1584
		free(avl_file_name);
		return NULL;
	}
	free(avl_file_name);

1585
		// Truncate the AVL tree file to the right size
1586 1587 1588
	if (ftruncate(avl_file_descriptor, file_size) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1589
		obidebug(1, "\nError truncating an AVL tree file to the right size");
1590 1591
		if (avl_idx >= 0)
			free(complete_avl_name);
1592
		close_avl_data(avl_data, true);
1593 1594 1595 1596
		close(avl_file_descriptor);
		return NULL;
	}

1597
		// Allocate the memory for the AVL tree structure
1598 1599 1600
	avl = (OBIDMS_avl_p) malloc(sizeof(OBIDMS_avl_t));
	if (avl == NULL)
	{
1601
		obi_set_errno(OBI_MALLOC_ERROR);
1602
		obidebug(1, "\nError allocating the memory for the AVL tree structure");
1603 1604
		if (avl_idx >= 0)
			free(complete_avl_name);
1605
		close_avl_data(avl_data, true);
1606 1607 1608 1609
		close(avl_file_descriptor);
		return NULL;
	}

1610
		// Fill the AVL tree structure
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620
	avl->header = mmap(NULL,
			           header_size,
			           PROT_READ | PROT_WRITE,
			           MAP_SHARED,
			           avl_file_descriptor,
			           0
			          );
	if (avl->header == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1621
		obidebug(1, "\nError mmapping the header of an AVL tree file");
1622 1623
		if (avl_idx >= 0)
			free(complete_avl_name);
1624
		close_avl_data(avl_data, true);
1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
		close(avl_file_descriptor);
		free(avl);
		return NULL;
	}

	avl->tree  = mmap(NULL,
			          data_size,
			          PROT_READ | PROT_WRITE,
			          MAP_SHARED,
			          avl_file_descriptor,
			          header_size
			         );
	if (avl->tree == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1640
		obidebug(1, "\nError mmapping the data of an AVL tree file");
1641 1642
		if (avl_idx >= 0)
			free(complete_avl_name);
1643
		close_avl_data(avl_data, true);
1644 1645 1646 1647 1648 1649
		munmap(avl->header, header_size);
		close(avl_file_descriptor);
		free(avl);
		return NULL;
	}

1650 1651 1652
	avl->dms       = dms;
	avl->data      = avl_data;
	avl->avl_fd    = avl_file_descriptor;
1653 1654 1655 1656

	(avl->header)->header_size 	 = header_size;
	(avl->header)->avl_size 	 = data_size;
	(avl->header)->nb_items	 	 = 0;
1657
	(avl->header)->nb_items_max	 = (index_t) floor(((double) (avl->header)->avl_size) / ((double) sizeof(AVL_node_t)));
1658
	(avl->header)->root_idx      = -1;
1659
	(avl->header)->creation_date = time(NULL);
1660
	strcpy((avl->header)->avl_name, complete_avl_name);
1661 1662

	// Bloom filter
1663
	bloom_init(&((avl->header)->bloom_filter), MAX_NODE_COUNT_PER_AVL);
1664

1665
	free(complete_avl_name);
1666

1667 1668 1669 1670 1671 1672 1673
	if (closedir(directory) < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError closing an AVL directory");
		return NULL;
	}

1674 1675 1676 1677
	return avl;
}


1678
OBIDMS_avl_p obi_open_avl(OBIDMS_p dms, const char* avl_name, int avl_idx)
1679 1680
{
	char* 	 			avl_file_name;
1681 1682
	char*				complete_avl_name;
	char*				avl_dir_name;
1683
	char* 				avl_data_file_name;
1684
	DIR*				directory;
1685 1686 1687 1688 1689 1690
	size_t				header_size;
	int					avl_file_descriptor;
	int					avl_data_file_descriptor;
	int					avl_dir_file_descriptor;
	OBIDMS_avl_data_p   avl_data;
	OBIDMS_avl_p		avl;
1691 1692 1693 1694

	// Get complete name of AVL if index
	if (avl_idx >= 0)
	{
1695
		complete_avl_name = obi_build_avl_name_with_idx(avl_name, avl_idx);
1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709
		if (complete_avl_name == NULL)
			return NULL;
	}
	else
	{
		complete_avl_name = (char*) malloc((strlen(avl_name)+1)*sizeof(char));
		if (complete_avl_name == NULL)
		{
    		obi_set_errno(OBI_MALLOC_ERROR);
    		obidebug(1, "\nError allocating memory for an AVL name");
			return NULL;
		}
		strcpy(complete_avl_name, avl_name);
	}
1710

1711 1712 1713 1714
    // Open the AVL directory
	avl_dir_name = get_full_path_of_avl_dir(dms, avl_name);
	if (avl_dir_name == NULL)
	{
1715
		free(complete_avl_name);
1716 1717 1718 1719 1720 1721 1722
		return NULL;
	}
    directory = opendir(avl_dir_name);
	if (directory == NULL)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError opening an AVL directory");
1723
		free(complete_avl_name);
1724 1725 1726 1727 1728 1729 1730 1731 1732
		free(avl_dir_name);
		return NULL;
	}
	free(avl_dir_name);
	avl_dir_file_descriptor = dirfd(directory);
	if (avl_dir_file_descriptor < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
		obidebug(1, "\nError getting the file descriptor of an AVL directory");
1733
		free(complete_avl_name);
1734 1735
		return NULL;
	}
1736

1737 1738

	// Open the data file
1739 1740

		// Build file name
1741
	avl_data_file_name = build_avl_data_file_name(complete_avl_name);
1742 1743 1744 1745
	if (avl_data_file_name == NULL)
		return NULL;

		// Open file
1746
	avl_data_file_descriptor = openat(avl_dir_file_descriptor, avl_data_file_name, O_RDWR, 0777);
1747 1748 1749
	if (avl_data_file_descriptor < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1750
		obidebug(1, "\nError opening an AVL tree data file");
1751
		free(complete_avl_name);
1752 1753 1754 1755 1756
		free(avl_data_file_name);
		return NULL;
	}
	free(avl_data_file_name);

1757
		// Allocate the memory for the AVL tree data structure
1758 1759 1760
	avl_data = (OBIDMS_avl_data_p) malloc(sizeof(OBIDMS_avl_data_t));
	if (avl_data == NULL)
	{
1761
		obi_set_errno(OBI_MALLOC_ERROR);
1762
		obidebug(1, "\nError allocating the memory for the AVL tree data structure");
1763
		free(complete_avl_name);
1764 1765 1766 1767 1768 1769 1770 1771
		close(avl_data_file_descriptor);
		return NULL;
	}

	// Read the header size
	if (read(avl_data_file_descriptor, &header_size, sizeof(size_t)) < ((ssize_t) sizeof(size_t)))
	{
		obi_set_errno(OBI_AVL_ERROR);
1772
		obidebug(1, "\nError reading the header size to open an AVL tree data file");
1773
		free(complete_avl_name);
1774 1775 1776 1777 1778 1779 1780
		close(avl_data_file_descriptor);
		return NULL;
	}

		// Fill the avl data structure
	avl_data->header = mmap(NULL,
			                header_size,
1781
							PROT_READ | PROT_WRITE,
1782 1783 1784 1785 1786 1787 1788
			                MAP_SHARED,
			                avl_data_file_descriptor,
			                0
			               );
	if (avl_data->header == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1789
		obidebug(1, "\nError mmapping the header of an AVL tree data file");
1790
		free(complete_avl_name);
1791 1792 1793 1794 1795 1796 1797
		close(avl_data_file_descriptor);
		free(avl_data);
		return NULL;
	}

	avl_data->data   = mmap(NULL,
			                (avl_data->header)->data_size_max,
1798
			                PROT_READ,
1799 1800 1801 1802 1803 1804 1805
			                MAP_SHARED,
			                avl_data_file_descriptor,
			                header_size
			               );
	if (avl_data->data == MAP_FAILED)
	{
		obi_set_errno(OBI_AVL_ERROR);
1806
		obidebug(1, "\nError mmapping the data of an AVL tree data file");
1807
		free(complete_avl_name);
1808 1809 1810 1811 1812 1813
		munmap(avl_data->header, header_size);
		close(avl_data_file_descriptor);
		free(avl_data);
		return NULL;
	}

1814
	avl_data->data_fd = avl_data_file_descriptor;
1815 1816


1817
	// Open the AVL tree file
1818 1819

		// Build file name
1820
	avl_file_name = build_avl_file_name(complete_avl_name);
1821 1822
	if (avl_file_name == NULL)
	{
1823
		close_avl_data(avl_data, false);
1824 1825 1826 1827
		return NULL;
	}

		// Open file
1828
	avl_file_descriptor = openat(avl_dir_file_descriptor, avl_file_name, O_RDWR, 0777);
1829 1830 1831
	if (avl_file_descriptor < 0)
	{
		obi_set_errno(OBI_AVL_ERROR);
1832
		obidebug(1, "\nError opening an AVL tree file");
1833 1834
		free(complete_avl_name);
		close_avl_data(avl_data, false);
1835 1836 1837 1838 1839
		free(avl_file_name);
		return NULL;
	}
	free(avl_file_name);

1840
		// Allocate the memory for the AVL tree structure
1841 1842 1843
	avl = (OBIDMS_avl_p) malloc(sizeof(OBIDMS_avl_t));
	if (avl == NULL)
	{
1844
		obi_set_errno(OBI_MALLOC_ERROR);
1845
		obidebug(1, "\nError allocating the memory for the AVL tree structure");
1846 1847
		free(complete_avl_name);
		close_avl_data(avl_data, false);
1848 1849 1850 1851 1852 1853 1854 1855
		close(avl_file_descriptor);
		return NULL;
	}

		// Read the header size
	if (read(avl_file_descriptor, &header_size, sizeof(size_t)) < ((ssize_t) sizeof(size_t)))
	{
		obi_set_errno(OBI_AVL_ERROR);
1856
		obidebug(1, "\nError reading the header size to open an AVL tree");
1857 1858
		free(complete_avl_name);
		close_avl_data(avl_data, false);