<?php
namespace CKAN\Manager;
use CKAN\OrganizationList;
/**
* http://www.data.gov/app/themes/roots-nextdatagov/assets/Json/fed_agency.json
*/
define('ORGANIZATION_TO_EXPORT', 'Environmental Protection Agency');
require_once dirname(dirname(__DIR__)) . '/inc/common.php';
/**
* Get organization terms, including all children, as Array
*/
$OrgList = new OrganizationList(AGENCIES_LIST_URL);
$termsArray = $OrgList->getTreeArrayFor(ORGANIZATION_TO_EXPORT);
/**
* sometimes there is no parent term (ex. Department of Labor)
*/
if (!defined('PARENT_TERM')) {
define('PARENT_TERM', '_');
}
/**
* Create results dir for logs and json results
*/
$results_dir = CKANMNGR_RESULTS_DIR . date('/Ymd-His') . '_EXPORT_' . PARENT_TERM;
mkdir($results_dir);
/**
* Search for packages by terms found
*/
/**
* Production
/**
* Make it TRUE, if you want datasets to be marked as PRIVATE
* LIST_ONLY must be true
*/
define('MARK_PRIVATE', true);
/**
* Rename adding __legacy to the end of dataset name (url will be changed too)
* LIST_ONLY must be true
*/
define('RENAME_TO_LEGACY', true);
echo "Tagging " . ORGANIZATION_TO_TAG . PHP_EOL;
require_once dirname(dirname(__DIR__)) . '/inc/common.php';
/**
* Get organization terms, including all children, as Array
*/
$OrgList = new OrganizationList(AGENCIES_LIST_URL);
$termsArray = $OrgList->getTreeArrayFor(ORGANIZATION_TO_TAG);
/**
* sometimes there is no parent term (ex. Department of Labor)
*/
if (!defined('PARENT_TERM')) {
die('PARENT_TERM not found');
}
/**
* Create results dir for logs
*/
$results_dir = CKANMNGR_RESULTS_DIR . date('/Ymd-His') . '_LEGACY_' . PARENT_TERM;
mkdir($results_dir);
/**
* Adding Legacy dms tag
*/
}
// no anchors please
list($dataset, ) = explode('#', basename(trim($row['0'])));
if (!$dataset) {
continue;
}
// double trouble check
if (strpos($row['0'], '://')) {
if (!strpos($row['0'], '/dataset/')) {
if (strpos($row['0'], 'dataset?q=')) {
parse_str(parse_url($row['0'], PHP_URL_QUERY), $query_array);
if (isset($query_array['q'])) {
$query = $query_array['q'];
if (isset($query_array['organization'])) {
$org = $query_array['organization'];
$organizationList = new OrganizationList();
$org = $organizationList->getTreeArrayFor($organizationList->getNameFor($org));
if (!is_array($org) || !sizeof($org)) {
continue;
}
$org = join(' OR ', array_keys($org));
// var_dump($organizationList->getTreeArrayFor($organizationList->getNameFor($org)));
// continue;
$query = "{$query} AND organization:({$org})";
// echo $query.PHP_EOL;
}
$packages = $CkanManager->tryPackageSearch($query, '', 200);
$CkanManager->say(sizeof($packages) . " found searching: {$query},API SEARCH");
file_put_contents($results_dir . '/' . $basename . '_tags.log.csv', sizeof($packages) . " found searching: {$query},API SEARCH" . PHP_EOL, FILE_APPEND | LOCK_EX);
// print $query_array['q'];
if (!sizeof($packages)) {
/**
* @param string $organizationName
*
* @throws \Exception
* @return mixed
*/
private function findOrganization($organizationName)
{
static $OrgList;
if (!$OrgList) {
$OrgList = new OrganizationList(AGENCIES_LIST_URL);
}
return $OrgList->getTermFor($organizationName);
}
<?php
namespace CKAN\Manager;
use CKAN\OrganizationList;
/**
* @author Alex Perfilov
* @date 5/23/14
*
*/
require_once dirname(__DIR__) . '/inc/common.php';
$start = isset($argv[1]) ? trim($argv[1]) : false;
$limit = isset($argv[2]) ? intval($argv[2]) : 1;
/**
* Create results dir for logs
*/
$results_dir = CKANMNGR_RESULTS_DIR . date('/Ymd-His') . '_PRIVATE_DATASETS_' . $start ?: '';
mkdir($results_dir);
$CkanManager = new CkanManager(CKAN_API_URL, CKAN_API_KEY);
//$CkanManager = new CkanManager(CKAN_STAGING_API_URL, CKAN_STAGING_API_KEY);
//$CkanManager = new CkanManager(CKAN_DEV_API_URL, CKAN_DEV_API_KEY);
define('ERROR_REPORTING', E_ALL);
/**
* Get organization terms, including all children, as Array
*/
$OrgList = new OrganizationList(AGENCIES_LIST_URL);
$termsArray = $OrgList->getTreeArray();
$CkanManager->resultsDir = $results_dir;
$CkanManager->getPrivateList($termsArray, $start, $limit);
// show running time on finish
timer();