adseqcompr.cxx File Reference

#include <arb_progress.h>
#include <arb_file.h>
#include <arb_misc.h>
#include <arb_diff.h>
#include "ad_cb.h"
#include "gb_key.h"
#include "TreeNode.h"
#include <climits>

Include dependency graph for adseqcompr.cxx:

Go to the source code of this file.

Classes
struct	CompressionRoot
class	FINAL_TYPE< SEQTYPE >
	xml element More...
struct	Consensus
struct	Sequence
struct	MasterSequence

Macros
#define	MAX_SEQUENCE_PER_MASTER   50
#define	MAX_NUMBER   0x7fffffff

Functions
static Consensus *	g_b_new_Consensus (long len)
static void	g_b_delete_Consensus (Consensus *gcon)
static void	g_b_Consensus_add (Consensus *gcon, unsigned char *seq, long seq_len)
static char *	g_b_Consensus_get_sequence (Consensus *gcon)
static int	g_b_count_leafs (CompressionTree *node)
static void	g_b_put_sequences_in_container (CompressionTree *ctree, Sequence *seqs, MasterSequence **masters, Consensus *gcon)
static void	g_b_create_master (CompressionTree *node, Sequence *seqs, MasterSequence **masters, int my_master, const char *ali_name, long seq_len, arb_progress &progress)
static void	subtract_sons_from_tree (CompressionTree *node, int subtract)
static int	set_masters_with_sons (CompressionTree *node, int wantedSons, long *mcount)
static int	maxCompressionSteps (CompressionTree *node)
static int	init_indices_and_count_sons (CompressionTree *node, long *scount, const char *ali_name)
static void	distribute_masters (CompressionTree *tree, long *mcount, int *max_masters)
	STATIC_ASSERT (MAX_NUMBER<=INT_MAX)
int	g_b_read_number2 (const unsigned char *&s)
void	g_b_put_number2 (int i, unsigned char *&s)
static char *	gb_compress_seq_by_master (const char *master, size_t master_len, int master_index, GBQUARK q, const char *seq, size_t seq_len, size_t *memsize, int old_flag)
static char *	gb_compress_sequence_by_master (GBDATA *gbd, const char *master, size_t master_len, int master_index, GBQUARK q, const char *seq, size_t seq_len, size_t *memsize)
static GB_ERROR	compress_sequence_tree (GBCONTAINER *gb_main, CompressionTree *tree, const char *ali_name)
GB_ERROR	GBT_compress_sequence_tree2 (GBDATA *gbd, const char *tree_name, const char *ali_name)
static char *	g_b_uncompress_single_sequence_by_master (const char *s, const char *master, size_t size, size_t *new_size)
char *	gb_uncompress_by_sequence (GBDATA *gbd, const char *ss, size_t size, GB_ERROR *error, size_t *new_size)

Macro Definition Documentation

#define MAX_SEQUENCE_PER_MASTER   50

Definition at line 23 of file adseqcompr.cxx.

Referenced by compress_sequence_tree(), distribute_masters(), g_b_Consensus_add(), and set_masters_with_sons().

#define MAX_NUMBER   0x7fffffff

Definition at line 324 of file adseqcompr.cxx.

Referenced by g_b_put_number2(), and g_b_read_number2().

Function Documentation

static Consensus* g_b_new_Consensus ( long  len )

static

Definition at line 81 of file adseqcompr.cxx.

References Consensus::con, and Consensus::len.

Referenced by g_b_create_master().

static void g_b_delete_Consensus ( Consensus *  gcon )

static

Definition at line 94 of file adseqcompr.cxx.

References Consensus::con.

Referenced by g_b_create_master().

static void g_b_Consensus_add ( Consensus *  gcon, unsigned char *  seq, long  seq_len  )

static

Definition at line 100 of file adseqcompr.cxx.

References Consensus::con, gb_assert, GB_RUNLENGTH_SIZE, Consensus::len, li, MAX_SEQUENCE_PER_MASTER, s, seq, and Consensus::used.

Referenced by g_b_put_sequences_in_container().

static char* g_b_Consensus_get_sequence ( Consensus *  gcon )

static

Definition at line 147 of file adseqcompr.cxx.

References Consensus::con, Consensus::len, max, s, seq, and Consensus::used.

Referenced by g_b_create_master().

static int g_b_count_leafs ( CompressionTree *  node )

static

Definition at line 173 of file adseqcompr.cxx.

References NULp.

Referenced by compress_sequence_tree().

static void g_b_put_sequences_in_container ( CompressionTree *  ctree, Sequence *  seqs, MasterSequence **  masters, Consensus *  gcon  )

static

Definition at line 179 of file adseqcompr.cxx.

References g_b_Consensus_add(), MasterSequence::gb_mas, GB_read_char_pntr(), GB_read_string_count(), and Sequence::gb_seq.

Referenced by g_b_create_master().

static void g_b_create_master ( CompressionTree *  node, Sequence *  seqs, MasterSequence **  masters, int  my_master, const char *  ali_name, long  seq_len, arb_progress &  progress  )

static

Definition at line 198 of file adseqcompr.cxx.

References arb_progress::aborted(), GBDATA::as_entry(), g_b_Consensus_get_sequence(), g_b_delete_Consensus(), g_b_new_Consensus(), g_b_put_sequences_in_container(), Sequence::gb_seq, GB_write_security_write(), GB_write_string(), GBT_find_sequence(), Sequence::master, and MasterSequence::master.

Referenced by compress_sequence_tree().

static void subtract_sons_from_tree ( CompressionTree *  node, int  subtract  )

static

Definition at line 239 of file adseqcompr.cxx.

Referenced by set_masters_with_sons().

static int set_masters_with_sons ( CompressionTree *  node, int  wantedSons, long *  mcount  )

static

Definition at line 246 of file adseqcompr.cxx.

References gb_assert, MAX_SEQUENCE_PER_MASTER, and subtract_sons_from_tree().

Referenced by distribute_masters().

static int maxCompressionSteps ( CompressionTree *  node )

static

Definition at line 271 of file adseqcompr.cxx.

References GBS_global_string_copy().

Referenced by distribute_masters().

static int init_indices_and_count_sons ( CompressionTree *  node, long *  scount, const char *  ali_name  )

static

Definition at line 289 of file adseqcompr.cxx.

References GBT_find_sequence().

Referenced by compress_sequence_tree().

static void distribute_masters ( CompressionTree *  tree, long *  mcount, int *  max_masters  )

static

Definition at line 310 of file adseqcompr.cxx.

References gb_assert, MAX_SEQUENCE_PER_MASTER, maxCompressionSteps(), and set_masters_with_sons().

Referenced by compress_sequence_tree().

STATIC_ASSERT

(

MAX_NUMBER<=

INT_MAX

)

int g_b_read_number2 ( const unsigned char *&  s )

inline

Definition at line 327 of file adseqcompr.cxx.

References gb_assert, MAX_NUMBER, and result.

Referenced by gb_uncompress_by_sequence().

void g_b_put_number2 ( int  i, unsigned char *&  s  )

inline

Definition at line 359 of file adseqcompr.cxx.

References gb_assert, and MAX_NUMBER.

Referenced by gb_compress_seq_by_master().

static char* gb_compress_seq_by_master ( const char *  master, size_t  master_len, int  master_index, GBQUARK  q, const char *  seq, size_t  seq_len, size_t *  memsize, int  old_flag  )

static

Definition at line 522 of file adseqcompr.cxx.

References buffer, g_b_put_number2(), gb_compress_equal_bytes_2(), GB_COMPRESSION_SEQUENCE, and GB_give_other_buffer().

Referenced by gb_compress_sequence_by_master().

static char* gb_compress_sequence_by_master ( GBDATA *  gbd, const char *  master, size_t  master_len, int  master_index, GBQUARK  q, const char *  seq, size_t  seq_len, size_t *  memsize  )

static

Definition at line 572 of file adseqcompr.cxx.

References gb_compress_data(), gb_compress_seq_by_master(), GB_COMPRESSION_DICTIONARY, GB_COMPRESSION_LAST, GB_COMPRESSION_RUNLENGTH, and GB_COMPRESSION_SORTBYTES.

Referenced by compress_sequence_tree().

static GB_ERROR compress_sequence_tree ( GBCONTAINER *  gb_main, CompressionTree *  tree, const char *  ali_name  )

static

Definition at line 581 of file adseqcompr.cxx.

References arb_assert, ARB_calloc(), ARB_strdup(), GBDATA::as_container(), GBDATA::as_entry(), distribute_masters(), arb_progress::done(), error(), arb_progress::error_if_aborted(), g_b_count_leafs(), g_b_create_master(), gb_assert, GB_await_error(), GB_clear_error(), gb_compress_sequence_by_master(), gb_create(), GB_CREATE_CONTAINER, gb_create_container(), GB_delete(), GB_entry(), GB_export_error(), GB_FIND, gb_find_or_create_quark(), GB_information(), GB_internal_error(), GB_MAIN(), MasterSequence::gb_mas, gb_Key::gb_master_ali, GB_nextEntry(), GB_read_clock(), GB_read_memuse(), gb_read_nr(), GB_read_string(), GB_read_string_count(), GB_search(), gb_search(), Sequence::gb_seq, gb_species_data, GB_STRING, GB_SYSTEM_FOLDER, GB_warning(), GB_warningf(), gb_write_compressed_pntr(), GB_write_security_delete(), GB_write_string(), GBS_global_string(), GBS_global_string_copy(), GBS_readable_size(), GBT_find_sequence(), GBT_first_species_rel_species_data(), GBT_get_alignment_len(), GBT_get_species_data(), GBT_link_tree(), GBT_next_species(), GBT_write_tree(), arb_progress::inc_and_check_user_abort(), init_indices_and_count_sons(), SmartPtr< T, C >::isSet(), GB_MAIN_TYPE::keys, Sequence::master, MasterSequence::master, MAX_SEQUENCE_PER_MASTER, NULp, seq, and warning().

Referenced by GBT_compress_sequence_tree2().

GB_ERROR GBT_compress_sequence_tree2	(	GBDATA *	gbd,
		const char *	tree_name,
		const char *	ali_name
	)

Definition at line 889 of file adseqcompr.cxx.

References ASSERT_NO_ERROR, GB_transaction::close(), compress_sequence_tree(), destroy(), DOWNCAST, error(), GB_await_error(), GB_disable_quicksave(), GB_get_requested_undo_type(), GB_internal_error(), gb_main, GB_MAIN(), GB_request_undo_type(), GB_UNDO_KILL, GBT_link_tree(), GBT_name_of_largest_tree(), GBT_read_tree(), GB_MAIN_TYPE::get_transaction_level(), NULp, GB_transaction::ok(), GB_MAIN_TYPE::root_container, and ta.

Referenced by ARB_main(), and NT_database_optimization().

static char* g_b_uncompress_single_sequence_by_master ( const char *  s, const char *  master, size_t  size, size_t *  new_size  )

static

Definition at line 959 of file adseqcompr.cxx.

References buffer, gb_assert, GB_give_other_buffer(), and GB_internal_error().

Referenced by gb_uncompress_by_sequence().

char* gb_uncompress_by_sequence	(	GBDATA *	gbd,
		const char *	ss,
		size_t	size,
		GB_ERROR *	error,
		size_t *	new_size
	)

Definition at line 1016 of file adseqcompr.cxx.

References g_b_read_number2(), g_b_uncompress_single_sequence_by_master(), gb_assert, GB_await_error(), GB_check_out_buffer(), GB_FATHER(), gb_find_by_nr(), gb_get_main_during_cb(), gb_load_single_key_data(), GB_MAIN_TYPE::gb_main(), GB_MAIN(), gb_Key::gb_master_ali, GB_read_char_pntr(), GB_read_string_count(), GB_MAIN_TYPE::keys, NULp, and s.

Referenced by GB_is_dictionary_compressed(), and gb_uncompress_data().