int core_map_add_value(struct core_map *self, void *key, void *value)
{
void *bucket;
int value_size;
#ifdef CORE_MAP_ALIGNMENT_ENABLED
key = core_map_pad_key(self, key);
#endif
bucket = core_map_get(self, key);
/* it's already there...
*/
if (bucket != NULL) {
return 0;
}
bucket = core_map_add(self, key);
#ifdef CORE_MAP_ALIGNMENT_ENABLED
value_size = self->original_value_size;
#else
value_size = core_map_get_value_size(self);
#endif
core_memory_copy(bucket, value, value_size);
return 1;
}
int thorium_balancer_select_worker_script_round_robin(struct thorium_balancer *self, int script)
{
int worker;
int worker_count;
int next_worker;
int *bucket;
bucket = core_map_get(&self->current_script_workers, &script);
worker_count = thorium_worker_pool_worker_count(self->pool);
/* First time for this script
*/
if (bucket == NULL) {
bucket = core_map_add(&self->current_script_workers, &script);
*bucket = self->first_worker;
self->first_worker = (self->first_worker + 1) % worker_count;
}
worker = *bucket;
next_worker = (worker + 1) % worker_count;
*bucket = next_worker;
return worker;
}
void biosal_input_controller_receive(struct thorium_actor *actor, struct thorium_message *message)
{
int tag;
int count;
char *file;
void *buffer;
struct biosal_input_controller *controller;
struct biosal_input_controller *concrete_actor;
int destination;
int script;
int stream;
char *local_file;
int i;
int name;
int source;
int destination_index;
struct thorium_message new_message;
int error;
int stream_index;
int64_t entries;
int64_t *bucket;
int *int_bucket;
int spawner;
int command_name;
int stream_name;
int consumer;
int consumer_index;
int *bucket_for_requests;
char *new_buffer;
int new_count;
int file_index;
struct core_vector mega_blocks;
struct core_vector_iterator vector_iterator;
struct biosal_mega_block *mega_block;
struct core_vector *vector_bucket;
struct core_vector block_counts;
uint64_t block_entries;
int mega_block_index;
uint64_t offset;
struct biosal_mega_block *block;
int acquaintance_index;
if (thorium_actor_take_action(actor, message)) {
return;
}
thorium_message_get_all(message, &tag, &count, &buffer, &source);
name = thorium_actor_name(actor);
controller = (struct biosal_input_controller *)thorium_actor_concrete_actor(actor);
concrete_actor = controller;
if (tag == ACTION_START) {
core_vector_init(&concrete_actor->spawners, 0);
core_vector_unpack(&concrete_actor->spawners, buffer);
core_vector_resize(&concrete_actor->stores_per_spawner,
core_vector_size(&concrete_actor->spawners));
for (i = 0; i < core_vector_size(&concrete_actor->spawners); i++) {
int_bucket = (int *)core_vector_at(&concrete_actor->stores_per_spawner, i);
*int_bucket = 0;
spawner = core_vector_at_as_int(&concrete_actor->spawners, i);
core_queue_enqueue(&concrete_actor->unprepared_spawners, &spawner);
}
concrete_actor->state = BIOSAL_INPUT_CONTROLLER_STATE_PREPARE_SPAWNERS;
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL
printf("DEBUG preparing first spawner\n");
#endif
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_CONTROLLER_PREPARE_SPAWNERS);
/*
thorium_dispatcher_print(thorium_actor_dispatcher(actor));
*/
} else if (tag == ACTION_ADD_FILE) {
file = (char *)buffer;
local_file = core_memory_allocate(strlen(file) + 1, MEMORY_CONTROLLER);
strcpy(local_file, file);
printf("controller %d ACTION_ADD_FILE %s\n",
thorium_actor_name(actor),
local_file);
core_vector_push_back(&concrete_actor->files, &local_file);
bucket = core_vector_at(&concrete_actor->files, core_vector_size(&concrete_actor->files) - 1);
local_file = *(char **)bucket;
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG11 ACTION_ADD_FILE %s %p bucket %p index %d\n",
local_file, local_file, (void *)bucket, core_vector_size(&concrete_actor->files) - 1);
#endif
thorium_actor_send_reply_empty(actor, ACTION_ADD_FILE_REPLY);
} else if (tag == ACTION_SPAWN_REPLY) {
if (concrete_actor->state == BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_STORES) {
biosal_input_controller_add_store(actor, message);
return;
} else if (concrete_actor->state == BIOSAL_INPUT_CONTROLLER_STATE_PREPARE_SPAWNERS) {
concrete_actor->ready_spawners++;
thorium_message_unpack_int(message, 0, &name);
thorium_actor_send_empty(actor, name, ACTION_ASK_TO_STOP);
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_CONTROLLER_PREPARE_SPAWNERS);
if (concrete_actor->ready_spawners == (int)core_vector_size(&concrete_actor->spawners)) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG all spawners are prepared\n");
#endif
thorium_actor_send_to_supervisor_empty(actor, ACTION_START_REPLY);
}
return;
} else if (concrete_actor->state == BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_PARTITIONER) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG received spawn reply, state is spawn_partitioner\n");
#endif
thorium_message_unpack_int(message, 0, &concrete_actor->partitioner);
/* configure the partitioner
*/
destination = concrete_actor->partitioner;
thorium_actor_send_int(actor, destination,
ACTION_SEQUENCE_PARTITIONER_SET_BLOCK_SIZE,
concrete_actor->block_size);
thorium_actor_send_int(actor, destination,
ACTION_SEQUENCE_PARTITIONER_SET_ACTOR_COUNT,
core_vector_size(&concrete_actor->consumers));
core_vector_init(&block_counts, sizeof(uint64_t));
for (i = 0; i < core_vector_size(&concrete_actor->mega_block_vector); i++) {
block = (struct biosal_mega_block *)core_vector_at(&concrete_actor->mega_block_vector, i);
block_entries = biosal_mega_block_get_entries(block);
core_vector_push_back_uint64_t(&block_counts, block_entries);
}
new_count = core_vector_pack_size(&block_counts);
new_buffer = thorium_actor_allocate(actor, new_count);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG packed counts, %d bytes\n", count);
#endif
core_vector_pack(&block_counts, new_buffer);
thorium_message_init(&new_message, ACTION_SEQUENCE_PARTITIONER_SET_ENTRY_VECTOR,
new_count, new_buffer);
thorium_actor_send(actor, destination, &new_message);
core_vector_destroy(&block_counts);
return;
} else if (concrete_actor->state == BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_READING_STREAMS) {
thorium_message_unpack_int(message, 0, &stream);
stream_index = stream;
mega_block_index = core_vector_size(&concrete_actor->reading_streams);
core_vector_push_back_int(&concrete_actor->reading_streams, stream_index);
core_vector_push_back_int(&concrete_actor->partition_commands, -1);
core_vector_push_back_int(&concrete_actor->stream_consumers, -1);
stream_index = core_vector_size(&concrete_actor->reading_streams) - 1;
mega_block = (struct biosal_mega_block *)core_vector_at(&concrete_actor->mega_block_vector,
mega_block_index);
offset = biosal_mega_block_get_offset(mega_block);
core_map_add_value(&concrete_actor->assigned_blocks,
&stream_index, &mega_block_index);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_READING_STREAMS
printf("DEBUG setting offset to %" PRIu64 " for stream/%d\n",
offset, stream);
#endif
thorium_actor_send_uint64_t(actor, stream, ACTION_INPUT_STREAM_SET_START_OFFSET, offset);
return;
}
stream = *(int *)buffer;
file_index = core_vector_size(&concrete_actor->counting_streams);
local_file = *(char **)core_vector_at(&concrete_actor->files, file_index);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_READING_STREAMS
printf("DEBUG actor %d receives stream %d from spawner %d for file %s\n",
name, stream, source,
local_file);
#endif
core_vector_push_back(&concrete_actor->counting_streams, &stream);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_READING_STREAMS
printf("asking stream/%d to open %s\n", stream, local_file);
#endif
thorium_message_init(&new_message, ACTION_INPUT_OPEN, strlen(local_file) + 1, local_file);
#ifdef DEBUG_ISSUE_594
thorium_message_print(&new_message);
printf("SEND Buffer %s\n", local_file);
#endif
thorium_actor_send(actor, stream, &new_message);
thorium_message_destroy(&new_message);
if (core_vector_size(&concrete_actor->counting_streams) != core_vector_size(&concrete_actor->files)) {
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_SPAWN);
}
#ifdef DEBUG_ISSUE_594
printf("EXIT Buffer %s\n", local_file);
#endif
} else if (tag == ACTION_INPUT_OPEN_REPLY) {
if (concrete_actor->state == BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_READING_STREAMS) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_READING_STREAMS
printf("DEBUG receives open.reply for reading stream/%d\n", source);
#endif
concrete_actor->opened_streams++;
if (concrete_actor->opened_streams == core_vector_size(&concrete_actor->mega_block_vector)) {
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_CONTROLLER_CREATE_STORES);
}
return;
}
concrete_actor->opened_streams++;
stream = source;
thorium_message_unpack_int(message, 0, &error);
if (error == BIOSAL_INPUT_ERROR_NO_ERROR) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG actor %d asks %d ACTION_INPUT_COUNT_IN_PARALLEL\n", name, stream);
#endif
thorium_actor_send_vector(actor, stream, ACTION_INPUT_COUNT_IN_PARALLEL,
&concrete_actor->spawners);
} else {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG actor %d received error %d from %d\n", name, error, stream);
#endif
concrete_actor->counted++;
}
/* if all streams failed, notice supervisor */
if (concrete_actor->counted == core_vector_size(&concrete_actor->files)) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
#endif
printf("DEBUG %d: Error all streams failed.\n",
thorium_actor_name(actor));
thorium_actor_send_to_supervisor_empty(actor, ACTION_INPUT_DISTRIBUTE_REPLY);
}
/*
if (concrete_actor->opened_streams == core_vector_size(&concrete_actor->files)) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG controller %d sends ACTION_INPUT_DISTRIBUTE_REPLY to supervisor %d [%d/%d]\n",
name, thorium_actor_supervisor(actor),
concrete_actor->opened_streams, core_vector_size(&concrete_actor->files));
#endif
}
*/
} else if (tag == ACTION_INPUT_COUNT_PROGRESS) {
stream_index = core_vector_index_of(&concrete_actor->counting_streams, &source);
local_file = core_vector_at_as_char_pointer(&concrete_actor->files, stream_index);
thorium_message_unpack_int64_t(message, 0, &entries);
bucket = (int64_t *)core_vector_at(&concrete_actor->counts, stream_index);
printf("controller/%d receives progress from stream/%d file %s %" PRIu64 " entries so far\n",
name, source, local_file, entries);
*bucket = entries;
} else if (tag == ACTION_INPUT_COUNT_IN_PARALLEL_REPLY) {
stream_index = core_vector_index_of(&concrete_actor->counting_streams, &source);
local_file = core_vector_at_as_char_pointer(&concrete_actor->files, stream_index);
core_vector_init(&mega_blocks, 0);
core_vector_unpack(&mega_blocks, buffer);
printf("DEBUG receive mega blocks from %d\n", source);
/*
* Update the file index for every mega block.
*/
core_vector_iterator_init(&vector_iterator, &mega_blocks);
bucket = (int64_t*)core_vector_at(&concrete_actor->counts, stream_index);
(*bucket) = 0;
while (core_vector_iterator_has_next(&vector_iterator)) {
core_vector_iterator_next(&vector_iterator, (void **)&mega_block);
printf("SETTING setting file to %d for mega block\n", stream_index);
biosal_mega_block_set_file(mega_block, stream_index);
entries = biosal_mega_block_get_entries_from_start(mega_block);
printf("Cataloging %d ENTRIES\n", (int)entries);
(*bucket) = entries;
biosal_mega_block_print(mega_block);
}
core_vector_iterator_destroy(&vector_iterator);
vector_bucket = (struct core_vector *)core_map_add(&concrete_actor->mega_blocks, &stream_index);
core_vector_init_copy(vector_bucket, &mega_blocks);
core_vector_destroy(&mega_blocks);
concrete_actor->counted++;
printf("controller/%d received from stream/%d for file %s %" PRIu64 " entries (final) %d/%d\n",
name, source, local_file, entries,
concrete_actor->counted, (int)core_vector_size(&concrete_actor->files));
thorium_actor_send_reply_empty(actor, ACTION_INPUT_CLOSE);
/* continue work here, tell supervisor about it */
if (concrete_actor->counted == core_vector_size(&concrete_actor->files)) {
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_CONTROLLER_SPAWN_READING_STREAMS);
}
} else if (tag == ACTION_INPUT_DISTRIBUTE) {
core_timer_start(&concrete_actor->input_timer);
core_timer_start(&concrete_actor->counting_timer);
/* for each file, spawn a stream to count */
/* no files, return immediately
*/
if (core_vector_size(&concrete_actor->files) == 0) {
printf("Error: no file to distribute...\n");
thorium_actor_send_reply_empty(actor, ACTION_INPUT_DISTRIBUTE_REPLY);
return;
}
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG actor %d receives ACTION_INPUT_DISTRIBUTE\n", name);
#endif
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG send ACTION_INPUT_SPAWN to self\n");
#endif
thorium_actor_send_to_self_empty(actor, ACTION_INPUT_SPAWN);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG resizing counts to %d\n", core_vector_size(&concrete_actor->files));
#endif
core_vector_resize(&concrete_actor->counts, core_vector_size(&concrete_actor->files));
for (i = 0; i < core_vector_size(&concrete_actor->counts); i++) {
bucket = (int64_t*)core_vector_at(&concrete_actor->counts, i);
*bucket = 0;
}
} else if (tag == ACTION_INPUT_SPAWN && source == name) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG ACTION_INPUT_SPAWN\n");
#endif
script = SCRIPT_INPUT_STREAM;
concrete_actor->state = BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_STREAMS;
/* the next file name to send is the current number of streams */
i = core_vector_size(&concrete_actor->counting_streams);
destination_index = i % core_vector_size(&concrete_actor->spawners);
destination = *(int *)core_vector_at(&concrete_actor->spawners, destination_index);
thorium_message_init(message, ACTION_SPAWN, sizeof(script), &script);
thorium_actor_send(actor, destination, message);
bucket = core_vector_at(&concrete_actor->files, i);
local_file = *(char **)core_vector_at(&concrete_actor->files, i);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
printf("DEBUG890 local_file %p bucket %p index %d\n", local_file, (void *)bucket,
i);
#endif
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG actor %d spawns a stream for file %d/%d via spawner %d\n",
name, i, core_vector_size(&concrete_actor->files), destination);
#endif
/* also, spawn 4 stores on each node */
} else if (tag == ACTION_ASK_TO_STOP && ( source == thorium_actor_supervisor(actor)
|| source == thorium_actor_name(actor))) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_LEVEL_2
#endif
/* stop streams
*/
for (i = 0; i < core_vector_size(&concrete_actor->counting_streams); i++) {
stream = *(int *)core_vector_at(&concrete_actor->counting_streams, i);
thorium_actor_send_empty(actor, stream, ACTION_ASK_TO_STOP);
}
for (i = 0; i < core_vector_size(&concrete_actor->reading_streams); i++) {
stream = *(int *)core_vector_at(&concrete_actor->reading_streams, i);
thorium_actor_send_empty(actor, stream, ACTION_ASK_TO_STOP);
}
#if 0
/* stop data stores
*/
for (i = 0; i < core_vector_size(&concrete_actor->consumers); i++) {
store = core_vector_at_as_int(&concrete_actor->consumers, i);
thorium_actor_send_empty(actor, store, ACTION_ASK_TO_STOP);
}
#endif
/* stop partitioner
*/
if (concrete_actor->partitioner != THORIUM_ACTOR_NOBODY) {
thorium_actor_send_empty(actor,
concrete_actor->partitioner,
ACTION_ASK_TO_STOP);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG controller %d sends ACTION_ASK_TO_STOP_REPLY to %d\n",
thorium_actor_name(actor),
thorium_message_source(message));
#endif
}
thorium_actor_send_reply_empty(actor, ACTION_ASK_TO_STOP_REPLY);
/* stop self
*/
thorium_actor_send_to_self_empty(actor, ACTION_STOP);
thorium_actor_ask_to_stop(actor, message);
printf("DEBUG controller %d dies\n", name);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
#endif
} else if (tag == ACTION_INPUT_CONTROLLER_CREATE_PARTITION && source == name) {
spawner = *(int *)core_vector_at(&concrete_actor->spawners,
core_vector_size(&concrete_actor->spawners) / 2);
thorium_actor_send_int(actor, spawner, ACTION_SPAWN,
SCRIPT_SEQUENCE_PARTITIONER);
concrete_actor->state = BIOSAL_INPUT_CONTROLLER_STATE_SPAWN_PARTITIONER;
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG input controller %d spawns a partitioner via spawner %d\n",
name, spawner);
#endif
} else if (tag == ACTION_SEQUENCE_PARTITIONER_COMMAND_IS_READY) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG controller receives ACTION_SEQUENCE_PARTITIONER_COMMAND_IS_READY, asks for command\n");
#endif
thorium_actor_send_reply_empty(actor, ACTION_SEQUENCE_PARTITIONER_GET_COMMAND);
} else if (tag == ACTION_SEQUENCE_PARTITIONER_GET_COMMAND_REPLY) {
biosal_input_controller_receive_command(actor, message);
} else if (tag == ACTION_SEQUENCE_PARTITIONER_FINISHED) {
thorium_actor_send_empty(actor,
concrete_actor->partitioner,
ACTION_ASK_TO_STOP);
biosal_input_controller_verify_requests(actor, message);
} else if (tag == ACTION_SEQUENCE_PARTITIONER_PROVIDE_STORE_ENTRY_COUNTS) {
biosal_input_controller_receive_store_entry_counts(actor, message);
} else if (tag == ACTION_RESERVE_REPLY) {
concrete_actor->ready_consumers++;
printf("DEBUG marker ACTION_RESERVE_REPLY %d/%d\n",
concrete_actor->ready_consumers,
(int)core_vector_size(&concrete_actor->consumers));
if (concrete_actor->ready_consumers == core_vector_size(&concrete_actor->consumers)) {
concrete_actor->ready_consumers = 0;
printf("DEBUG all consumers are ready\n");
thorium_actor_send_empty(actor,
concrete_actor->partitioner,
ACTION_SEQUENCE_PARTITIONER_PROVIDE_STORE_ENTRY_COUNTS_REPLY);
}
} else if (tag == ACTION_INPUT_PUSH_SEQUENCES_READY) {
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG biosal_input_controller_receive received ACTION_INPUT_PUSH_SEQUENCES_READY\n");
#endif
stream_name = source;
acquaintance_index = stream_name;
stream_index = core_vector_index_of(&concrete_actor->reading_streams, &acquaintance_index);
command_name = *(int *)core_vector_at(&concrete_actor->partition_commands,
stream_index);
thorium_actor_send_int(actor,
concrete_actor->partitioner,
ACTION_SEQUENCE_PARTITIONER_GET_COMMAND_REPLY_REPLY,
command_name);
} else if (tag == ACTION_INPUT_PUSH_SEQUENCES_REPLY) {
stream_name = source;
thorium_message_unpack_int(message, 0, &consumer);
consumer_index = core_vector_index_of(&concrete_actor->consumers,
&consumer);
bucket_for_requests = (int *)core_vector_at(&concrete_actor->consumer_active_requests, consumer_index);
(*bucket_for_requests)--;
biosal_input_controller_verify_requests(actor, message);
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG_CONSUMERS
printf("DEBUG consumer # %d has %d active requests\n",
consumer_index, *bucket_for_requests);
#endif
} else if (tag == ACTION_SET_CONSUMERS) {
core_vector_init(&concrete_actor->consumers, 0);
core_vector_unpack(&concrete_actor->consumers, buffer);
printf("controller %d receives %d consumers\n",
thorium_actor_name(actor),
(int)core_vector_size(&concrete_actor->consumers));
for (i = 0; i < core_vector_size(&concrete_actor->consumers); i++) {
core_vector_push_back_int(&concrete_actor->consumer_active_requests, 0);
}
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
core_vector_print_int(&concrete_actor->consumers);
printf("\n");
#endif
thorium_actor_send_reply_empty(actor, ACTION_SET_CONSUMERS_REPLY);
} else if (tag == ACTION_SET_BLOCK_SIZE) {
thorium_message_unpack_int(message, 0, &concrete_actor->block_size);
thorium_actor_send_reply_empty(actor, ACTION_SET_BLOCK_SIZE_REPLY);
} else if (tag == ACTION_SEQUENCE_STORE_READY) {
concrete_actor->filled_consumers++;
#ifdef BIOSAL_INPUT_CONTROLLER_DEBUG
printf("DEBUG ACTION_SEQUENCE_STORE_READY %d/%d\n", concrete_actor->filled_consumers,
(int)core_vector_size(&concrete_actor->consumers));
#endif
if (concrete_actor->filled_consumers == core_vector_size(&concrete_actor->consumers)) {
concrete_actor->filled_consumers = 0;
printf("DEBUG: all consumers are filled, sending ACTION_INPUT_DISTRIBUTE_REPLY\n");
core_timer_stop(&concrete_actor->input_timer);
core_timer_stop(&concrete_actor->distribution_timer);
core_timer_print_with_description(&concrete_actor->distribution_timer,
"Load input / Distribute input data");
core_timer_print_with_description(&concrete_actor->input_timer,
"Load input");
thorium_actor_send_to_supervisor_empty(actor, ACTION_INPUT_DISTRIBUTE_REPLY);
}
}
}
void biosal_assembly_graph_store_push_kmer_block(struct thorium_actor *self, struct thorium_message *message)
{
struct core_memory_pool *ephemeral_memory;
struct biosal_dna_kmer_frequency_block block;
struct biosal_assembly_vertex *bucket;
void *packed_kmer;
struct core_map_iterator iterator;
struct biosal_assembly_graph_store *concrete_self;
/*int tag;*/
void *key;
struct core_map *kmers;
struct biosal_dna_kmer kmer;
void *buffer;
int count;
struct biosal_dna_kmer encoded_kmer;
char *raw_kmer;
int period;
struct biosal_dna_kmer *kmer_pointer;
int *frequency;
ephemeral_memory = thorium_actor_get_ephemeral_memory(self);
concrete_self = thorium_actor_concrete_actor(self);
/*tag = thorium_message_action(message);*/
buffer = thorium_message_buffer(message);
count = thorium_message_count(message);
/*
* Handler for PUSH_DATA
*/
biosal_dna_kmer_frequency_block_init(&block, concrete_self->kmer_length,
ephemeral_memory, &concrete_self->transport_codec, 0);
biosal_dna_kmer_frequency_block_unpack(&block, buffer, ephemeral_memory,
&concrete_self->transport_codec);
key = core_memory_pool_allocate(ephemeral_memory, concrete_self->key_length_in_bytes);
kmers = biosal_dna_kmer_frequency_block_kmers(&block);
core_map_iterator_init(&iterator, kmers);
period = 2500000;
raw_kmer = core_memory_pool_allocate(thorium_actor_get_ephemeral_memory(self),
concrete_self->kmer_length + 1);
if (!concrete_self->printed_vertex_size) {
printf("DEBUG VERTEX DELIVERY %d bytes\n", count);
concrete_self->printed_vertex_size = 1;
}
while (core_map_iterator_has_next(&iterator)) {
/*
* add kmers to store
*/
core_map_iterator_next(&iterator, (void **)&packed_kmer, (void **)&frequency);
/* Store the kmer in 2 bit encoding
*/
biosal_dna_kmer_init_empty(&kmer);
biosal_dna_kmer_unpack(&kmer, packed_kmer, concrete_self->kmer_length,
ephemeral_memory,
&concrete_self->transport_codec);
kmer_pointer = &kmer;
if (concrete_self->codec_are_different) {
/*
* Get a copy of the sequence
*/
biosal_dna_kmer_get_sequence(kmer_pointer, raw_kmer, concrete_self->kmer_length,
&concrete_self->transport_codec);
biosal_dna_kmer_init(&encoded_kmer, raw_kmer, &concrete_self->storage_codec,
thorium_actor_get_ephemeral_memory(self));
kmer_pointer = &encoded_kmer;
}
biosal_dna_kmer_pack_store_key(kmer_pointer, key,
concrete_self->kmer_length, &concrete_self->storage_codec,
thorium_actor_get_ephemeral_memory(self));
#ifdef BIOSAL_DEBUG_ISSUE_540
if (strcmp(raw_kmer, "AGCTGGTAGTCATCACCAGACTGGAACAG") == 0
|| strcmp(raw_kmer, "CGCGATCTGTTGCTGGGCCTAACGTGGTA") == 0
|| strcmp(raw_kmer, "TACCACGTTAGGCCCAGCAACAGATCGCG") == 0) {
printf("Examine store key for %s\n", raw_kmer);
core_debugger_examine(key, concrete_self->key_length_in_bytes);
}
#endif
bucket = core_map_get(&concrete_self->table, key);
if (bucket == NULL) {
/* This is the first time that this kmer is seen.
*/
bucket = core_map_add(&concrete_self->table, key);
biosal_assembly_vertex_init(bucket);
#if 0
printf("DEBUG303 ADD_KEY");
biosal_dna_kmer_print(&encoded_kmer, concrete_self->kmer_length,
&concrete_self->storage_codec, ephemeral_memory);
#endif
}
if (concrete_self->codec_are_different) {
biosal_dna_kmer_destroy(&encoded_kmer,
thorium_actor_get_ephemeral_memory(self));
}
biosal_dna_kmer_destroy(&kmer, ephemeral_memory);
biosal_assembly_vertex_increase_coverage_depth(bucket, *frequency);
if (concrete_self->received >= concrete_self->last_received + period) {
printf("%s/%d received %" PRIu64 " kmers so far,"
" store has %" PRIu64 " canonical kmers, %" PRIu64 " kmers\n",
thorium_actor_script_name(self),
thorium_actor_name(self), concrete_self->received,
core_map_size(&concrete_self->table),
2 * core_map_size(&concrete_self->table));
concrete_self->last_received = concrete_self->received;
}
concrete_self->received += *frequency;
}
core_memory_pool_free(ephemeral_memory, key);
core_memory_pool_free(ephemeral_memory, raw_kmer);
core_map_iterator_destroy(&iterator);
biosal_dna_kmer_frequency_block_destroy(&block, thorium_actor_get_ephemeral_memory(self));
thorium_actor_send_reply_empty(self, ACTION_PUSH_KMER_BLOCK_REPLY);
}
void biosal_assembly_graph_store_yield_reply(struct thorium_actor *self, struct thorium_message *message)
{
struct biosal_dna_kmer kmer;
void *key;
struct biosal_assembly_vertex *value;
int coverage;
int customer;
uint64_t *count;
int new_count;
void *new_buffer;
struct thorium_message new_message;
struct core_memory_pool *ephemeral_memory;
struct biosal_assembly_graph_store *concrete_self;
int i;
int max;
ephemeral_memory = thorium_actor_get_ephemeral_memory(self);
concrete_self = thorium_actor_concrete_actor(self);
customer = concrete_self->customer;
#if 0
printf("YIELD REPLY\n");
#endif
i = 0;
max = 1024;
key = NULL;
value = NULL;
while (i < max
&& core_map_iterator_has_next(&concrete_self->iterator)) {
core_map_iterator_next(&concrete_self->iterator, (void **)&key, (void **)&value);
biosal_dna_kmer_init_empty(&kmer);
biosal_dna_kmer_unpack(&kmer, key, concrete_self->kmer_length,
ephemeral_memory,
&concrete_self->storage_codec);
coverage = biosal_assembly_vertex_coverage_depth(value);
count = (uint64_t *)core_map_get(&concrete_self->coverage_distribution, &coverage);
if (count == NULL) {
count = (uint64_t *)core_map_add(&concrete_self->coverage_distribution, &coverage);
(*count) = 0;
}
/* increment for the lowest kmer (canonical) */
(*count)++;
biosal_dna_kmer_destroy(&kmer, ephemeral_memory);
++i;
}
/* yield again if the iterator is not at the end
*/
if (core_map_iterator_has_next(&concrete_self->iterator)) {
#if 0
printf("yield ! %d\n", i);
#endif
thorium_actor_send_to_self_empty(self, ACTION_YIELD);
return;
}
/*
printf("ready...\n");
*/
core_map_iterator_destroy(&concrete_self->iterator);
new_count = core_map_pack_size(&concrete_self->coverage_distribution);
new_buffer = thorium_actor_allocate(self, new_count);
core_map_pack(&concrete_self->coverage_distribution, new_buffer);
printf("SENDING %s/%d sends map to %d, %d bytes / %d entries\n",
thorium_actor_script_name(self),
thorium_actor_name(self),
customer, new_count,
(int)core_map_size(&concrete_self->coverage_distribution));
thorium_message_init(&new_message, ACTION_PUSH_DATA, new_count, new_buffer);
thorium_actor_send(self, customer, &new_message);
thorium_message_destroy(&new_message);
core_map_destroy(&concrete_self->coverage_distribution);
thorium_actor_send_empty(self, concrete_self->source,
ACTION_PUSH_DATA_REPLY);
}
void biosal_sequence_partitioner_receive(struct thorium_actor *actor, struct thorium_message *message)
{
int tag;
int source;
int count;
void *buffer;
int bytes;
struct biosal_sequence_partitioner *concrete_actor;
struct biosal_partition_command command;
struct thorium_message response;
int command_number;
struct biosal_partition_command *active_command;
int stream_index;
struct biosal_partition_command *command_bucket;
int i;
thorium_message_get_all(message, &tag, &count, &buffer, &source);
concrete_actor = (struct biosal_sequence_partitioner *)thorium_actor_concrete_actor(actor);
if (tag == ACTION_SEQUENCE_PARTITIONER_SET_BLOCK_SIZE) {
thorium_message_unpack_int(message, 0, &concrete_actor->block_size);
thorium_actor_send_reply_empty(actor, ACTION_SEQUENCE_PARTITIONER_SET_BLOCK_SIZE_REPLY);
biosal_sequence_partitioner_verify(actor);
/*
printf("DEBUG biosal_sequence_partitioner_receive received block size\n");
*/
} else if (tag == ACTION_SEQUENCE_PARTITIONER_SET_ENTRY_VECTOR) {
/*
printf("DEBUG biosal_sequence_partitioner_receive unpacking vector, %d bytes\n",
count);
*/
core_vector_init(&concrete_actor->stream_entries, 0);
core_vector_unpack(&concrete_actor->stream_entries, buffer);
/*
printf("DEBUG after unpack\n");
*/
thorium_actor_send_reply_empty(actor, ACTION_SEQUENCE_PARTITIONER_SET_ENTRY_VECTOR_REPLY);
/*
printf("DEBUG biosal_sequence_partitioner_receive received received entry vector\n");
*/
biosal_sequence_partitioner_verify(actor);
} else if (tag == ACTION_SEQUENCE_PARTITIONER_SET_ACTOR_COUNT) {
thorium_message_unpack_int(message, 0, &concrete_actor->store_count);
thorium_actor_send_reply_empty(actor, ACTION_SEQUENCE_PARTITIONER_SET_ACTOR_COUNT_REPLY);
biosal_sequence_partitioner_verify(actor);
/*
printf("DEBUG biosal_sequence_partitioner_receive received received store count\n");
*/
} else if (tag == ACTION_SEQUENCE_PARTITIONER_GET_COMMAND) {
if (core_queue_dequeue(&concrete_actor->available_commands, &command)) {
bytes = biosal_partition_command_pack_size(&command);
/*
printf("DEBUG partitioner has command, packing %d bytes!\n", bytes);
*/
buffer = thorium_actor_allocate(actor, bytes);
biosal_partition_command_pack(&command, buffer);
thorium_message_init(&response, ACTION_SEQUENCE_PARTITIONER_GET_COMMAND_REPLY,
bytes, buffer);
thorium_actor_send_reply(actor, &response);
/* store the active command
*/
command_number = biosal_partition_command_name(&command);
command_bucket = (struct biosal_partition_command *)core_map_add(&concrete_actor->active_commands,
&command_number);
*command_bucket = command;
/* there may be other command available too !
*/
}
} else if (tag == ACTION_SEQUENCE_PARTITIONER_GET_COMMAND_REPLY_REPLY) {
/*
* take the name of the command, find it in the active
* command, generate a new command, and send ACTION_SEQUENCE_PARTITIONER_COMMAND_IS_READY
* as a reply
*/
thorium_message_unpack_int(message, 0, &command_number);
active_command = core_map_get(&concrete_actor->active_commands,
&command_number);
if (active_command == NULL) {
return;
}
stream_index = biosal_partition_command_stream_index(active_command);
active_command = NULL;
core_map_delete(&concrete_actor->active_commands,
&command_number);
biosal_sequence_partitioner_generate_command(actor, stream_index);
if (core_map_size(&concrete_actor->active_commands) == 0
&& core_queue_size(&concrete_actor->available_commands) == 0) {
thorium_actor_send_reply_empty(actor, ACTION_SEQUENCE_PARTITIONER_FINISHED);
}
} else if (tag == ACTION_ASK_TO_STOP
&& source == thorium_actor_supervisor(actor)) {
#ifdef BIOSAL_SEQUENCE_PARTITIONER_DEBUG
printf("DEBUG biosal_sequence_partitioner_receive ACTION_ASK_TO_STOP\n");
#endif
thorium_actor_send_to_self_empty(actor,
ACTION_STOP);
} else if (tag == ACTION_SEQUENCE_PARTITIONER_PROVIDE_STORE_ENTRY_COUNTS_REPLY) {
/* generate commands
*/
for (i = 0; i < core_vector_size(&concrete_actor->stream_entries); i++) {
biosal_sequence_partitioner_generate_command(actor, i);
}
}
}
int main(int argc, char **argv)
{
BEGIN_TESTS();
{
struct core_map big_map;
int kmer_length = 43;
struct biosal_dna_kmer kmer;
int count;
int run_test;
int coverage;
void *key;
int key_length;
int *bucket;
int i;
struct biosal_dna_codec codec;
struct core_memory_pool memory;
core_memory_pool_init(&memory, 1048576, -1);
biosal_dna_codec_init(&codec);
run_test = 1;
count = 100000000;
printf("STRESS TEST\n");
biosal_dna_kmer_init_mock(&kmer, kmer_length, &codec, &memory);
key_length = biosal_dna_kmer_pack_size(&kmer, kmer_length, &codec);
biosal_dna_kmer_destroy(&kmer, &memory);
core_map_init(&big_map, key_length, sizeof(coverage));
key = core_memory_allocate(key_length, -1);
i = 0;
while (i < count && run_test) {
biosal_dna_kmer_init_random(&kmer, kmer_length, &codec, &memory);
biosal_dna_kmer_pack_store_key(&kmer, key, kmer_length, &codec, &memory);
bucket = core_map_add(&big_map, key);
coverage = 99;
(*bucket) = coverage;
biosal_dna_kmer_destroy(&kmer, &memory);
if (i % 100000 == 0) {
printf("ADD %d/%d %" PRIu64 "\n", i, count,
core_map_size(&big_map));
}
i++;
}
core_map_destroy(&big_map);
core_memory_free(key, -1);
biosal_dna_codec_destroy(&codec);
core_memory_pool_destroy(&memory);
}
END_TESTS();
return 0;
}
void biosal_coverage_distribution_receive(struct thorium_actor *self, struct thorium_message *message)
{
int tag;
struct core_map map;
struct core_map_iterator iterator;
int *coverage_from_message;
uint64_t *count_from_message;
uint64_t *frequency;
int count;
void *buffer;
struct biosal_coverage_distribution *concrete_actor;
int name;
int source;
struct core_memory_pool *ephemeral_memory;
ephemeral_memory = thorium_actor_get_ephemeral_memory(self);
name = thorium_actor_name(self);
source = thorium_message_source(message);
concrete_actor = (struct biosal_coverage_distribution *)thorium_actor_concrete_actor(self);
tag = thorium_message_action(message);
count = thorium_message_count(message);
buffer = thorium_message_buffer(message);
if (tag == ACTION_PUSH_DATA) {
core_map_init(&map, 0, 0);
core_map_set_memory_pool(&map, ephemeral_memory);
core_map_unpack(&map, buffer);
core_map_iterator_init(&iterator, &map);
while (core_map_iterator_has_next(&iterator)) {
core_map_iterator_next(&iterator, (void **)&coverage_from_message,
(void **)&count_from_message);
#ifdef BIOSAL_COVERAGE_DISTRIBUTION_DEBUG
thorium_actor_log(self, "DEBUG DATA %d %d\n", (int)*coverage_from_message, (int)*count_from_message);
#endif
frequency = core_map_get(&concrete_actor->distribution, coverage_from_message);
if (frequency == NULL) {
frequency = core_map_add(&concrete_actor->distribution, coverage_from_message);
(*frequency) = 0;
}
(*frequency) += (*count_from_message);
}
core_map_iterator_destroy(&iterator);
thorium_actor_send_reply_empty(self, ACTION_PUSH_DATA_REPLY);
concrete_actor->actual++;
thorium_actor_log(self, "distribution/%d receives coverage data from producer/%d, %d entries / %d bytes %d/%d\n",
name, source, (int)core_map_size(&map), count,
concrete_actor->actual, concrete_actor->expected);
if (concrete_actor->expected != 0 && concrete_actor->expected == concrete_actor->actual) {
thorium_actor_log(self, "received everything %d/%d\n", concrete_actor->actual, concrete_actor->expected);
biosal_coverage_distribution_write_distribution(self);
thorium_actor_send_empty(self, concrete_actor->source,
ACTION_NOTIFY);
}
core_map_destroy(&map);
} else if (tag == ACTION_ASK_TO_STOP) {
biosal_coverage_distribution_ask_to_stop(self, message);
} else if (tag == ACTION_SET_EXPECTED_MESSAGE_COUNT) {
concrete_actor->source = source;
thorium_message_unpack_int(message, 0, &concrete_actor->expected);
thorium_actor_log(self, "distribution %d expects %d messages\n",
thorium_actor_name(self),
concrete_actor->expected);
thorium_actor_send_reply_empty(self, ACTION_SET_EXPECTED_MESSAGE_COUNT_REPLY);
}
}
