include_once dirname(__FILE__) . "/../../config/environment.php"; require_library('connectors/I3InteractiveAPI'); $timestart = time_elapsed(); $params["resource_id"] = 332; $params["process occurrence"] = false; $params["dwca_file"] = "http://localhost/cp/3IInteractive/DwCArchive_Cicadellinae.zip"; $params["dwca_file"] = "http://dmitriev.speciesfile.org/Export/DwCArchive_Cicadellinae.zip"; $func = new I3InteractiveAPI($params); $func->get_all_taxa(); if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . "_working/taxon.tab") > 1000) { if (is_dir(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"])) { recursive_rmdir(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . "_previous"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"], CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . "_previous"); } Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . "_working", CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"]); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . "_working.tar.gz", CONTENT_RESOURCE_LOCAL_PATH . $params["resource_id"] . ".tar.gz"); Functions::set_resource_status_to_force_harvest($params["resource_id"]); Functions::count_resource_tab_files($params["resource_id"]); if ($undefined_uris = Functions::get_undefined_uris_from_resource($params["resource_id"])) { print_r($undefined_uris); } echo "\nUndefined URIs: " . count($undefined_uris) . "\n"; require_library('connectors/DWCADiagnoseAPI'); $func = new DWCADiagnoseAPI(); $func->check_unique_ids($params["resource_id"]); } $elapsed_time_sec = time_elapsed() - $timestart; echo "\n\n"; echo "\n elapsed time = " . $elapsed_time_sec / 60 . " minutes"; echo "\n elapsed time = " . $elapsed_time_sec / 60 / 60 . " hours"; echo "\n Done processing.\n";
<?php namespace php_active_record; /* NCBI Bio Projects partner provides an XML dump estimated execution time: */ include_once dirname(__FILE__) . "/../../config/environment.php"; require_library('connectors/NCBIProjectsAPI'); $timestart = time_elapsed(); $resource_id = 173; $func = new NCBIProjectsAPI($resource_id); $func->get_all_taxa(); if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_working/taxon.tab") > 1000) { if (is_dir(CONTENT_RESOURCE_LOCAL_PATH . $resource_id)) { recursive_rmdir(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id, CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous"); } Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_working", CONTENT_RESOURCE_LOCAL_PATH . $resource_id); Functions::set_resource_status_to_force_harvest($resource_id); } $elapsed_time_sec = time_elapsed() - $timestart; echo "\n\n elapsed time = " . $elapsed_time_sec / 60 . " minutes"; echo "\n elapsed time = " . $elapsed_time_sec / 60 / 60 . " hours"; echo "\n Done processing.\n";
public static function finalize_dwca_resource($resource_id, $big_file = false) { if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_working/taxon.tab") > 200) { if (is_dir(CONTENT_RESOURCE_LOCAL_PATH . $resource_id)) { recursive_rmdir(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id, CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous"); } Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_working", CONTENT_RESOURCE_LOCAL_PATH . $resource_id); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_working.tar.gz", CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".tar.gz"); Functions::set_resource_status_to_force_harvest($resource_id); Functions::count_resource_tab_files($resource_id); if (!$big_file) { if ($undefined_uris = Functions::get_undefined_uris_from_resource($resource_id)) { print_r($undefined_uris); } echo "\nUndefined URIs: " . count($undefined_uris) . "\n"; require_library('connectors/DWCADiagnoseAPI'); $func = new DWCADiagnoseAPI(); $func->check_unique_ids($resource_id); } } }
// .3 seconds wait time include_once dirname(__FILE__) . "/../../config/environment.php"; $resource_id = 15; if (!Functions::can_this_connector_run($resource_id)) { return; } require_library('FlickrAPI'); $GLOBALS['ENV_DEBUG'] = false; $auth_token = NULL; if (FlickrAPI::valid_auth_token(FLICKR_AUTH_TOKEN)) { $auth_token = FLICKR_AUTH_TOKEN; } // create new _temp file if (!($resource_file = Functions::file_open(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_temp.xml", "w+"))) { return; } // start the resource file with the XML header fwrite($resource_file, \SchemaDocument::xml_header()); // query Flickr and write results to file FlickrAPI::get_all_eol_photos($auth_token, $resource_file); // write the resource footer fwrite($resource_file, \SchemaDocument::xml_footer()); fclose($resource_file); // cache the previous version and make this new version the current version @unlink(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous.xml"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml", CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_previous.xml"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . "_temp.xml", CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml"); // set Flickr to force harvest if (filesize(CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml") > 600) { Functions::set_resource_status_to_force_harvest($resource_id); }
<?php namespace php_active_record; /* connector for Learning + Education Group Partner provides RSS feed. estimated execution time: just a few seconds */ include_once dirname(__FILE__) . "/../../config/environment.php"; $timestart = time_elapsed(); require_library('connectors/LearningEducationAPI'); $taxa = LearningEducationAPI::get_all_taxa(); $xml = \SchemaDocument::get_taxon_xml($taxa); $resource_path = CONTENT_RESOURCE_LOCAL_PATH . "257_temp.xml"; if (!($OUT = fopen($resource_path, "w+"))) { debug(__CLASS__ . ":" . __LINE__ . ": Couldn't open file: " . $resource_path); return; } fwrite($OUT, $xml); fclose($OUT); if (filesize($resource_path) > 600) { Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "257.xml", CONTENT_RESOURCE_LOCAL_PATH . "257_previous.xml"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "257_temp.xml", CONTENT_RESOURCE_LOCAL_PATH . "257.xml"); $GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=" . ResourceStatus::find_or_create_by_translated_label('Force Harvest')->id . " WHERE id=257"); } $elapsed_time_sec = time_elapsed() - $timestart; echo "\n"; echo "elapsed time = {$elapsed_time_sec} seconds \n"; echo "elapsed time = " . $elapsed_time_sec / 60 . " minutes \n"; echo "\n\n Done processing.";
include_once dirname(__FILE__) . "/../../config/environment.php"; $timestart = time_elapsed(); require_library('connectors/ConabioAPI'); $resource_id = 106; $func = new ConabioAPI(); $func->combine_all_xmls($resource_id); $resource_path = CONTENT_RESOURCE_LOCAL_PATH . $resource_id . ".xml"; // /* working well - replaces Class='Insecta' to 'Reptilia' if Order=='Squamata' --- WEB-5509 require_library('ResourceDataObjectElementsSetting'); $func = new ResourceDataObjectElementsSetting($resource_id, $resource_path); $xml = file_get_contents($resource_path); $xml = $func->replace_taxon_element_value_with_condition("dwc:Class", "Insecta", "Reptilia", $xml, "dwc:Order", "Squamata"); $func->save_resource_document($xml); // */ // start - this will get Tamborines videos from Vimeo and append it with the main resource 106.xml (DATA-1592) Functions::file_rename($resource_path, CONTENT_RESOURCE_LOCAL_PATH . "temp_vimeo_to_tamborine1.xml"); get_videos_from_vimeo(); Functions::combine_all_eol_resource_xmls($resource_id, CONTENT_RESOURCE_LOCAL_PATH . "temp_vimeo_to_tamborine*.xml"); unlink(CONTENT_RESOURCE_LOCAL_PATH . "temp_vimeo_to_tamborine1.xml"); unlink(CONTENT_RESOURCE_LOCAL_PATH . "temp_vimeo_to_tamborine2.xml"); // end if (filesize($resource_path) > 1000) { Functions::set_resource_status_to_force_harvest($resource_id); Functions::gzip_resource_xml($resource_id); } $elapsed_time_sec = time_elapsed() - $timestart; echo "\n"; echo "elapsed time = " . $elapsed_time_sec / 60 . " minutes \n"; echo "elapsed time = " . $elapsed_time_sec / 60 / 60 . " hours \n"; function get_videos_from_vimeo() {
<?php namespace php_active_record; include_once dirname(__FILE__) . "/../../config/environment.php"; require_library('connectors/PaleobiologyAPI'); $paleobiology_connector = new PaleobiologyAPI(); $paleobiology_connector->get_all_taxa(); if (filesize(CONTENT_RESOURCE_LOCAL_PATH . "paleobiology_working/taxon.tab") > 1000) { if (is_dir(CONTENT_RESOURCE_LOCAL_PATH . "paleobiology")) { recursive_rmdir(CONTENT_RESOURCE_LOCAL_PATH . "paleobiology_previous"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "paleobiology", CONTENT_RESOURCE_LOCAL_PATH . "paleobiology_previous"); } Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "paleobiology_working", CONTENT_RESOURCE_LOCAL_PATH . "paleobiology"); #$GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=".ResourceStatus::find_or_create_by_translated_label('Force Harvest')->id." WHERE id=6"); }
public static function move_up_if_only_directory($directory_path) { $files = read_dir($directory_path); $only_file = null; foreach ($files as $file) { // there can be only one only file, must not start with a dot if ($only_file) { return; } if (substr($file, 0, 1) == ".") { continue; } $only_file = $directory_path . "/" . $file; } if (is_dir($only_file)) { Functions::file_rename($only_file, $directory_path . "_swap"); rmdir($directory_path); if (copy($directory_path . "_swap", $directory_path)) { unlink($directory_path . "_swap"); } } }
<?php namespace php_active_record; include_once dirname(__FILE__) . "/../../config/environment.php"; require_library('connectors/BiopixAPI'); $biopix_connector = new BiopixAPI(); $biopix_connector->get_all_taxa(); if (filesize(CONTENT_RESOURCE_LOCAL_PATH . "31_working/taxon.tab") > 1000 && filesize(CONTENT_RESOURCE_LOCAL_PATH . "31_working/media_resource.tab") > 1000) { if (is_dir(CONTENT_RESOURCE_LOCAL_PATH . "31")) { recursive_rmdir(CONTENT_RESOURCE_LOCAL_PATH . "31_previous"); Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "31", CONTENT_RESOURCE_LOCAL_PATH . "31_previous"); } Functions::file_rename(CONTENT_RESOURCE_LOCAL_PATH . "31_working", CONTENT_RESOURCE_LOCAL_PATH . "31"); $GLOBALS['db_connection']->update("UPDATE resources SET resource_status_id=" . ResourceStatus::find_or_create_by_translated_label('Force Harvest')->id . " WHERE id=31"); }