function start_process($resource_id, $call_multiple_instance)
{
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = 1;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
// Divide the big list of ids into small files
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
self::divide_text_file(10000);
//orig value 10000 debug
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
}
}
Functions::process_work_list($this);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
// Combine all XML files.
Functions::combine_all_eol_resource_xmls($resource_id, $this->TEMP_FILE_PATH . "temp_DiscoverLife_batch_*.xml");
// Set to force harvest
if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml")) {
$GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=" . ResourceStatus::force_harvest()->id . " WHERE id=" . $resource_id);
}
// Delete temp files
Functions::delete_temp_files($this->TEMP_FILE_PATH . "batch_", "txt");
Functions::delete_temp_files($this->TEMP_FILE_PATH . "temp_DiscoverLife_" . "batch_", "xml");
}
}
function start_process($resource_id, $call_multiple_instance)
{
self::$TEMP_FILE_PATH = DOC_ROOT . "/update_resources/connectors/files/DiscoverLife/";
self::$WORK_LIST = DOC_ROOT . "/update_resources/connectors/files/DiscoverLife/work_list.txt";
self::$WORK_IN_PROGRESS_LIST = DOC_ROOT . "/update_resources/connectors/files/DiscoverLife/work_in_progress_list.txt";
self::$INITIAL_PROCESS_STATUS = DOC_ROOT . "/update_resources/connectors/files/DiscoverLife/initial_process_status.txt";
self::$TEXT_FILE_FOR_DL = DOC_ROOT . "/update_resources/connectors/files/DiscoverLife/names_without_pages_in_eol.txt";
//report back to DiscoverLife
if (!trim(Functions::get_a_task(self::$WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task(self::$INITIAL_PROCESS_STATUS))) {
// Divide the big list of ids into small files
Functions::add_a_task("Initial process start", self::$INITIAL_PROCESS_STATUS);
self::divide_text_file(10000);
//orig value 10000
Functions::delete_a_task("Initial process start", self::$INITIAL_PROCESS_STATUS);
}
}
// Run multiple instances, for DiscoverLife ideally a total of 2
while (true) {
$task = Functions::get_a_task(self::$WORK_LIST);
//get a task to work on
if ($task) {
print "\n Process this: {$task}";
Functions::delete_a_task($task, self::$WORK_LIST);
Functions::add_a_task($task, self::$WORK_IN_PROGRESS_LIST);
$task = str_ireplace("\n", "", $task);
//remove carriage return got from text file
if ($call_multiple_instance) {
Functions::run_another_connector_instance($resource_id, 1);
//call 1 other instance for a total of 2 instances running
$call_multiple_instance = 0;
}
self::get_all_taxa($task);
print "\n Task {$task} is done. \n";
Functions::delete_a_task("{$task}\n", self::$WORK_IN_PROGRESS_LIST);
//remove a task from task list
} else {
print "\n\n [{$task}] Work list done --- " . date('Y-m-d h:i:s a', time()) . "\n";
break;
}
}
if (!($task = trim(Functions::get_a_task(self::$WORK_IN_PROGRESS_LIST)))) {
// Combine all XML files.
self::combine_all_xmls($resource_id);
// Set to force harvest
if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml")) {
$GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=" . ResourceStatus::insert('Force Harvest') . " WHERE id=" . $resource_id);
}
// Delete temp files
self::delete_temp_files(self::$TEMP_FILE_PATH . "batch_", "txt");
self::delete_temp_files(CONTENT_RESOURCE_LOCAL_PATH . "DiscoverLife/temp_DiscoverLife_" . "batch_", "xml");
}
}
function start_process($resource_id, $call_multiple_instance)
{
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = 1;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
// Divide the big list of ids into small files
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
$batch = Functions::create_work_list_from_master_file($this->MASTER_LIST, 5000, $this->TEMP_FILE_PATH, "batch_", $this->WORK_LIST);
//debug orig value 5000
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
}
}
Functions::process_work_list($this, $batch);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
// Combine all XML files.
Functions::combine_all_eol_resource_xmls($resource_id, $this->TEMP_FILE_PATH . "temp_Bolds_batch_*.xml");
// Delete temp files
Functions::delete_temp_files($this->TEMP_FILE_PATH . "batch_", "txt");
Functions::delete_temp_files($this->TEMP_FILE_PATH . "temp_Bolds_" . "batch_", "xml");
}
}
function start_process($resource_id, $call_multiple_instance, $connectors_to_run = 1)
{
require_library('connectors/BOLDSysAPI');
$this->func = new BOLDSysAPI();
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = $connectors_to_run;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
// Divide the big list of ids into small files
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
$this->func->create_master_list();
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
}
}
Functions::process_work_list($this);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
$this->archive_builder->finalize(true);
// Set to force harvest
Functions::set_resource_status_to_force_harvest($resource_id);
// Delete temp files
Functions::delete_temp_files($this->TEMP_FILE_PATH . "sl_batch_", "txt");
}
}
function start_process($resource_id, $call_multiple_instance, $connectors_to_run = 1)
{
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = $connectors_to_run;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
// Divide the big list of ids into small files
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
self::create_master_list();
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
}
}
Functions::process_work_list($this);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
// Combine all XML files.
Functions::combine_all_eol_resource_xmls($resource_id, $this->TEMP_FILE_PATH . "sl_batch_*.xml");
// Set to force harvest
Functions::set_resource_status_to_force_harvest($resource_id);
// Delete temp files
Functions::delete_temp_files($this->TEMP_FILE_PATH . "sl_batch_", "txt");
Functions::delete_temp_files($this->TEMP_FILE_PATH . "sl_batch_", "xml");
//debug Don't delete it if you want to check subsets of the resource XML.
}
}
function start_process($resource_id, $call_multiple_instance, $connectors_to_run = 1)
{
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = $connectors_to_run;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
// this will prepare a list of all species id; 13 mins. execution
self::build_id_list();
// divides the big list of ids into small files
self::divide_text_file(10000);
//debug orig 10000, for testing use 5
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
//remove a task from task list
}
}
Functions::process_work_list($this);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
// step 3: this should only run when all of instances of step 2 are done
sleep(10);
//debug orig 10
Functions::combine_all_eol_resource_xmls($resource_id, $this->TEMP_FILE_PATH . "temp_tropicos_batch_*.xml");
Functions::delete_temp_files($this->TEMP_FILE_PATH . "temp_tropicos_batch_", "xml");
//debug comment this line if u want to have a source for checking encoding probs in the XML
Functions::delete_temp_files($this->TEMP_FILE_PATH . "batch_", "txt");
Functions::set_resource_status_to_force_harvest($resource_id);
}
}
function start_process($resource_id, $call_multiple_instance)
{
$this->resource_id = $resource_id;
$this->call_multiple_instance = $call_multiple_instance;
$this->connectors_to_run = 1;
if (!trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST))) {
if (!trim(Functions::get_a_task($this->INITIAL_PROCESS_STATUS))) {
Functions::add_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
// step 1: divides the big list of ids into small files
$ids = self::get_id_list();
self::divide_text_file(10000, $ids);
//debug original value 10000
Functions::delete_a_task("Initial process start", $this->INITIAL_PROCESS_STATUS);
//removes a task from task list
}
}
Functions::process_work_list($this);
if (!($task = trim(Functions::get_a_task($this->WORK_IN_PROGRESS_LIST)))) {
// step 3: Combine all XML files. This only runs when all of instances of step 2 are done
self::combine_all_xmls($resource_id);
// set to force harvest
if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml")) {
$GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=" . ResourceStatus::force_harvest()->id . " WHERE id=" . $resource_id);
}
// delete temp files
Functions::delete_temp_files($this->TEMP_FILE_PATH . "batch_", "txt");
Functions::delete_temp_files($this->TEMP_FILE_PATH . "temp_worms_" . "batch_", "xml");
}
self::save_bad_ids_to_txt();
}
public static function process_work_list($class, $batch = null)
{
while (true) {
$task = Functions::get_a_task($class->WORK_LIST);
//get task to work on
if ($task) {
print "\n Process this: {$task}";
Functions::delete_a_task($task, $class->WORK_LIST);
//remove a task from task list
Functions::add_a_task($task, $class->WORK_IN_PROGRESS_LIST);
print "{$task} \n";
$task = str_ireplace("\n", "", $task);
//remove carriage return got from text file
if ($class->call_multiple_instance) {
Functions::run_another_connector_instance($class->resource_id, $class->connectors_to_run);
$class->call_multiple_instance = 0;
}
$class->get_all_taxa($task, $class->TEMP_FILE_PATH, array($batch, $task));
//main connector body
print "\n Task {$task} is done. \n";
Functions::delete_a_task("{$task}\n", $class->WORK_IN_PROGRESS_LIST);
//remove a task from task list
} else {
print "\n\n [{$task}] Work list done or list hasn't been created yet " . date('Y-m-d h:i:s a', time());
break;
}
}
}
