/*
* Basically, this actor does this:
* - spawn visitors
* - let them visit stuff
* - kill them.
* - spawn walkers
* - let them walk
* - kill the walkers
* - return OK
*/
void biosal_unitig_manager_receive(struct thorium_actor *self, struct thorium_message *message)
{
struct biosal_unitig_manager *concrete_self;
int tag;
void *buffer;
int spawner;
int expected;
int script;
int actor_count;
int source;
struct core_string file_name;
char *directory;
int argc;
char **argv;
char *path;
tag = thorium_message_action(message);
source = thorium_message_source(message);
buffer = thorium_message_buffer(message);
concrete_self = (struct biosal_unitig_manager *)thorium_actor_concrete_actor(self);
if (tag == ACTION_START) {
core_vector_unpack(&concrete_self->spawners, buffer);
spawner = thorium_actor_get_random_spawner(self, &concrete_self->spawners);
concrete_self->state = STATE_SPAWN_WRITER;
thorium_actor_send_int(self, spawner, ACTION_SPAWN, SCRIPT_WRITER_PROCESS);
} else if (tag == ACTION_SPAWN_REPLY
&& concrete_self->state == STATE_SPAWN_WRITER) {
thorium_message_unpack_int(message, 0, &concrete_self->writer_process);
/*
* open the file now.
*/
argc = thorium_actor_argc(self);
argv = thorium_actor_argv(self);
directory = core_command_get_output_directory(argc, argv);
core_string_init(&file_name, directory);
core_string_append(&file_name, "/");
core_string_append(&file_name, "unitigs.fasta");
path = core_string_get(&file_name);
thorium_actor_send_buffer(self, concrete_self->writer_process,
ACTION_OPEN, strlen(path) + 1, path);
core_string_destroy(&file_name);
} else if (tag == ACTION_OPEN_REPLY
&& source == concrete_self->writer_process) {
/*
* Spawn visitors.
*/
concrete_self->state = STATE_VISITORS;
thorium_actor_send_to_self_empty(self, ACTION_PING);
} else if (tag == ACTION_PING) {
spawner = thorium_actor_get_random_spawner(self, &concrete_self->spawners);
thorium_actor_send_int(self, spawner, ACTION_SPAWN, SCRIPT_MANAGER);
} else if (tag == ACTION_SPAWN_REPLY) {
thorium_message_unpack_int(message, 0, &concrete_self->manager);
script = SCRIPT_UNITIG_VISITOR;
if (concrete_self->state == STATE_WALKERS) {
script = SCRIPT_UNITIG_WALKER;
}
thorium_actor_send_int(self, concrete_self->manager, ACTION_MANAGER_SET_SCRIPT,
script);
} else if (tag == ACTION_ASK_TO_STOP) {
thorium_actor_send_empty(self, concrete_self->writer_process,
ACTION_ASK_TO_STOP);
thorium_actor_send_empty(self, concrete_self->manager,
ACTION_ASK_TO_STOP);
thorium_actor_send_to_self_empty(self, ACTION_STOP);
thorium_actor_send_reply_empty(self, ACTION_ASK_TO_STOP_REPLY);
} else if (tag == ACTION_MANAGER_SET_SCRIPT_REPLY) {
actor_count = UNITIG_VISITOR_COUNT_PER_WORKER;
if (concrete_self->state == STATE_WALKERS)
actor_count = UNITIG_WALKER_COUNT_PER_WORKER;
thorium_actor_send_reply_int(self, ACTION_MANAGER_SET_ACTORS_PER_WORKER,
actor_count);
} else if (tag == ACTION_MANAGER_SET_ACTORS_PER_WORKER_REPLY) {
thorium_actor_send_reply_vector(self, ACTION_START,
&concrete_self->spawners);
} else if (tag == ACTION_START_REPLY
&& concrete_self->state == STATE_VISITORS
&& core_vector_size(&concrete_self->visitors) == 0) {
core_vector_unpack(&concrete_self->visitors, buffer);
printf("DEBUG the system has %d visitors\n",
(int)core_vector_size(&concrete_self->visitors));
thorium_actor_send_to_supervisor_empty(self, ACTION_START_REPLY);
} else if (tag == ACTION_START_REPLY
&& concrete_self->state == STATE_WALKERS
&& core_vector_size(&concrete_self->walkers) == 0) {
core_vector_unpack(&concrete_self->walkers, buffer);
printf("DEBUG the system has %d walkers\n",
(int)core_vector_size(&concrete_self->walkers));
core_timer_start(&concrete_self->timer);
concrete_self->completed = 0;
thorium_actor_send_range_int(self, &concrete_self->walkers,
ACTION_SET_CONSUMER, concrete_self->writer_process);
thorium_actor_send_range_vector(self, &concrete_self->walkers,
ACTION_START, &concrete_self->graph_stores);
} else if (tag == ACTION_SET_PRODUCERS) {
core_vector_unpack(&concrete_self->graph_stores, buffer);
core_timer_start(&concrete_self->timer);
concrete_self->completed = 0;
thorium_actor_send_range_vector(self, &concrete_self->visitors,
ACTION_START, &concrete_self->graph_stores);
} else if (tag == ACTION_START_REPLY && concrete_self->state == STATE_VISITORS) {
++concrete_self->completed;
expected = core_vector_size(&concrete_self->visitors);
if (concrete_self->completed % UNITIG_VISITOR_COUNT_PER_WORKER == 0
|| concrete_self->completed == expected) {
printf("PROGRESS unitig visitors %d/%d\n",
concrete_self->completed,
expected);
}
if (concrete_self->completed == expected) {
core_timer_stop(&concrete_self->timer);
core_timer_print_with_description(&concrete_self->timer, "Visit vertices for unitigs");
/*
* Stop the visitor manager and all visitors too.
*/
thorium_actor_send_empty(self, concrete_self->manager, ACTION_ASK_TO_STOP);
/*
* Reset graph stores.
*/
thorium_actor_send_range_empty(self, &concrete_self->graph_stores,
ACTION_RESET);
concrete_self->completed = 0;
}
} else if (tag == ACTION_RESET_REPLY) {
++concrete_self->completed;
expected = core_vector_size(&concrete_self->graph_stores);
if (concrete_self->completed == expected) {
concrete_self->completed = 0;
concrete_self->state = STATE_WALKERS;
/*
* Go back at the beginning.
*/
thorium_actor_send_to_self_empty(self, ACTION_PING);
}
} else if (tag == ACTION_START_REPLY && concrete_self->state == STATE_WALKERS) {
++concrete_self->completed;
expected = core_vector_size(&concrete_self->walkers);
if (concrete_self->completed % UNITIG_WALKER_COUNT_PER_WORKER == 0
|| concrete_self->completed == expected) {
printf("PROGRESS unitig walkers %d/%d\n",
concrete_self->completed,
expected);
}
if (concrete_self->completed == expected) {
core_timer_stop(&concrete_self->timer);
core_timer_print_with_description(&concrete_self->timer, "Walk for unitigs");
thorium_actor_send_to_supervisor_empty(self, ACTION_SET_PRODUCERS_REPLY);
}
}
}
void biosal_input_stream_init(struct thorium_actor *actor)
{
struct biosal_input_stream *input;
struct biosal_input_stream *concrete_self;
input = (struct biosal_input_stream *)thorium_actor_concrete_actor(actor);
concrete_self = input;
concrete_self->proxy_ready = 0;
concrete_self->buffer_for_sequence = NULL;
concrete_self->maximum_sequence_length = 0;
concrete_self->open = 0;
concrete_self->error = BIOSAL_INPUT_ERROR_NO_ERROR;
concrete_self->file_name = NULL;
biosal_dna_codec_init(&concrete_self->codec);
if (biosal_dna_codec_must_use_two_bit_encoding(&concrete_self->codec,
thorium_actor_get_node_count(actor))) {
biosal_dna_codec_enable_two_bit_encoding(&concrete_self->codec);
}
/*concrete_self->mega_block_size = 2097152*/
/*
* This is the mega block size in number of sequences.
*/
concrete_self->mega_block_size = 2097152;
concrete_self->granularity = 1024;
concrete_self->last_offset = 0;
concrete_self->last_entries = 0;
concrete_self->starting_offset = 0;
/*
* Use a large ending offset.
*/
concrete_self->ending_offset = 0;
--concrete_self->ending_offset;
core_vector_init(&concrete_self->mega_blocks, sizeof(struct biosal_mega_block));
thorium_actor_add_action(actor, ACTION_INPUT_STREAM_SET_START_OFFSET, biosal_input_stream_set_start_offset);
thorium_actor_add_action(actor, ACTION_INPUT_STREAM_SET_END_OFFSET, biosal_input_stream_set_end_offset);
#ifdef ENABLE_PARALLEL_COUNT
thorium_actor_add_action(actor, ACTION_INPUT_COUNT_IN_PARALLEL, biosal_input_stream_count_in_parallel);
thorium_actor_add_action(actor, ACTION_INPUT_COUNT_REPLY, biosal_input_stream_count_reply);
#else
thorium_actor_add_action(actor, ACTION_INPUT_COUNT_IN_PARALLEL, biosal_input_stream_count_in_parallel_mock);
thorium_actor_add_action(actor, ACTION_INPUT_COUNT_REPLY, biosal_input_stream_count_reply_mock);
#endif
concrete_self->count_customer = THORIUM_ACTOR_NOBODY;
#if 0
core_string_init(&concrete_self->file_for_parallel_counting, NULL);
#endif
/*
* Parallel counting.
*/
concrete_self->total_entries = 0;
/*
* Disable parallel counting.
*/
concrete_self->finished_parallel_stream_count = 0;
core_vector_init(&concrete_self->spawners, sizeof(int));
core_vector_init(&concrete_self->parallel_streams, sizeof(int));
core_vector_init(&concrete_self->start_offsets, sizeof(uint64_t));
core_vector_init(&concrete_self->end_offsets, sizeof(uint64_t));
core_vector_init(&concrete_self->parallel_mega_blocks, sizeof(struct core_vector));
printf("%s/%d is now online on node %d\n",
thorium_actor_script_name(actor),
thorium_actor_name(actor),
thorium_actor_node_name(actor));
}
void biosal_coverage_distribution_write_distribution(struct thorium_actor *self)
{
struct core_map_iterator iterator;
int *coverage;
uint64_t *canonical_frequency;
uint64_t frequency;
struct biosal_coverage_distribution *concrete_actor;
struct core_vector coverage_values;
struct core_vector_iterator vector_iterator;
struct core_buffered_file_writer descriptor;
struct core_buffered_file_writer descriptor_canonical;
struct core_string file_name;
struct core_string canonical_file_name;
int argc;
char **argv;
int name;
char *directory_name;
name = thorium_actor_name(self);
argc = thorium_actor_argc(self);
argv = thorium_actor_argv(self);
directory_name = biosal_command_get_output_directory(argc, argv);
/* Create the directory if it does not exist
*/
if (!core_directory_verify_existence(directory_name)) {
core_directory_create(directory_name);
}
core_string_init(&file_name, "");
core_string_append(&file_name, directory_name);
core_string_append(&file_name, "/");
core_string_append(&file_name, BIOSAL_COVERAGE_DISTRIBUTION_DEFAULT_OUTPUT_FILE);
core_string_init(&canonical_file_name, "");
core_string_append(&canonical_file_name, directory_name);
core_string_append(&canonical_file_name, "/");
core_string_append(&canonical_file_name, BIOSAL_COVERAGE_DISTRIBUTION_DEFAULT_OUTPUT_FILE_CANONICAL);
core_buffered_file_writer_init(&descriptor, core_string_get(&file_name));
core_buffered_file_writer_init(&descriptor_canonical, core_string_get(&canonical_file_name));
concrete_actor = (struct biosal_coverage_distribution *)thorium_actor_concrete_actor(self);
core_vector_init(&coverage_values, sizeof(int));
core_map_iterator_init(&iterator, &concrete_actor->distribution);
#ifdef BIOSAL_COVERAGE_DISTRIBUTION_DEBUG
thorium_actor_log(self, "map size %d\n", (int)core_map_size(&concrete_actor->distribution));
#endif
while (core_map_iterator_has_next(&iterator)) {
core_map_iterator_next(&iterator, (void **)&coverage, (void **)&canonical_frequency);
#ifdef BIOSAL_COVERAGE_DISTRIBUTION_DEBUG
thorium_actor_log(self, "DEBUG COVERAGE %d FREQUENCY %" PRIu64 "\n", *coverage, *frequency);
#endif
core_vector_push_back(&coverage_values, coverage);
}
core_map_iterator_destroy(&iterator);
core_vector_sort_int(&coverage_values);
#ifdef BIOSAL_COVERAGE_DISTRIBUTION_DEBUG
thorium_actor_log(self, "after sort ");
core_vector_print_int(&coverage_values);
thorium_actor_log(self, "\n");
#endif
core_vector_iterator_init(&vector_iterator, &coverage_values);
#if 0
core_buffered_file_writer_printf(&descriptor_canonical, "Coverage\tFrequency\n");
#endif
core_buffered_file_writer_printf(&descriptor, "Coverage\tFrequency\n");
#ifdef BIOSAL_COVERAGE_DISTRIBUTION_DEBUG
#endif
while (core_vector_iterator_has_next(&vector_iterator)) {
core_vector_iterator_next(&vector_iterator, (void **)&coverage);
canonical_frequency = (uint64_t *)core_map_get(&concrete_actor->distribution, coverage);
frequency = 2 * *canonical_frequency;
core_buffered_file_writer_printf(&descriptor_canonical, "%d %" PRIu64 "\n",
*coverage,
*canonical_frequency);
core_buffered_file_writer_printf(&descriptor, "%d\t%" PRIu64 "\n",
*coverage,
frequency);
}
core_vector_destroy(&coverage_values);
core_vector_iterator_destroy(&vector_iterator);
thorium_actor_log(self, "distribution %d wrote %s\n", name, core_string_get(&file_name));
thorium_actor_log(self, "distribution %d wrote %s\n", name, core_string_get(&canonical_file_name));
core_buffered_file_writer_destroy(&descriptor);
core_buffered_file_writer_destroy(&descriptor_canonical);
core_string_destroy(&file_name);
core_string_destroy(&canonical_file_name);
}