function count_attributes($dir, $single_file = FALSE)
{
$no_activity_dates = array();
$activities_with_at_least_one = array();
$no_activities = array();
$found_hierarchies = array();
$activities_with_attribute = array();
$activity_by = array();
$document_links = array();
$result_element = array();
$conditions = array();
$participating_org_accountable = array();
$participating_org_implementing = array();
$budget = array();
$identifiers = array();
$transaction_type_commitment = array();
$transaction_type_disbursement = array();
$transaction_type_expenditure = array();
$no_disbursements = $no_incoming_funds = $no_tracable_transactions = array();
$activities_with_sector = array();
$most_recent = array();
$activities_with_location = array();
$activities_with_coordinates = array();
$activities_with_adminstrative = array();
$activities_sector_assumed_dac = array();
$activities_sector_declared_dac = array();
$activies_in_country_lang = array();
$i = 0;
//used to count bad id's
if ($handle = opendir($dir)) {
//echo "Directory handle: $handle\n";
//echo "Files:\n";
/* This is the correct way to loop over the directory. */
while (false !== ($file = readdir($handle))) {
if ($file != "." && $file != "..") {
//ignore these system files
//echo $file . PHP_EOL;
if ($single_file && $file != $single_file) {
//skip all files except the one we want if set/requested.Handy to test just one file in a directory
continue;
}
//load the xml SAFELY
/* Some safety against XML Injection attack
* see: http://phpsecurity.readthedocs.org/en/latest/Injection-Attacks.html
*
* Attempt a quickie detection of DOCTYPE - discard if it is present (cos it shouldn't be!)
*/
$xml = file_get_contents($dir . $file);
$collapsedXML = preg_replace("/[[:space:]]/", '', $xml);
//echo $collapsedXML;
if (preg_match("/<!DOCTYPE/i", $collapsedXML)) {
//throw new InvalidArgumentException(
// 'Invalid XML: Detected use of illegal DOCTYPE'
// );
//echo "fail";
return FALSE;
}
$loadEntities = libxml_disable_entity_loader(true);
$dom = new DOMDocument();
$dom->loadXML($xml);
foreach ($dom->childNodes as $child) {
if ($child->nodeType === XML_DOCUMENT_TYPE_NODE) {
throw new Exception\ValueException('Invalid XML: Detected use of illegal DOCTYPE');
libxml_disable_entity_loader($loadEntities);
return FALSE;
}
}
libxml_disable_entity_loader($loadEntities);
if ($xml = simplexml_import_dom($dom)) {
//print_r($xml);
if (!xml_child_exists($xml, "//iati-organisation")) {
//exclude organisation files
$activities = $xml->{"iati-activity"};
//print_r($attributes); die;
foreach ($activities as $activity) {
$hierarchy = (string) $activity->attributes()->hierarchy;
if ($hierarchy && $hierarchy != NULL) {
$hierarchy = (string) $activity->attributes()->hierarchy;
} else {
$hierarchy = 0;
}
$found_hierarchies[] = $hierarchy;
if (!isset($no_activities[$hierarchy])) {
$no_activities[$hierarchy] = 0;
}
$no_activities[$hierarchy]++;
//Set up some more counters:
if (!isset($no_disbursements[$hierarchy])) {
$no_disbursements[$hierarchy] = 0;
}
if (!isset($no_incoming_funds[$hierarchy])) {
$no_incoming_funds[$hierarchy] = 0;
}
if (!isset($no_tracable_transactions[$hierarchy])) {
$no_tracable_transactions[$hierarchy] = 0;
}
//Elements check
//is <document-link>,<conditions>,<result> present
if (count($activity->{"document-link"}) > 0) {
$document_links[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
if (count($activity->conditions) > 0) {
$conditions[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
if (count($activity->result) > 0) {
$result_element[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
//More elements
//Participating Organisation (Implementing)
$participating_orgs = $activity->{"participating-org"};
foreach ($participating_orgs as $participating_org) {
//echo (string)$activity->{"participating-org"}->attributes()->role;
if ((string) $participating_org->attributes()->role == "Implementing") {
//echo "yes";
$participating_org_implementing[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
//Participating Organisation (Accountable)
if ((string) $participating_org->attributes()->role == "Accountable") {
$participating_org_accountable[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
}
//Budget/Planned Disbursement
if (count($activity->budget) > 0 || count($activity->{"planned-disbursement"}) > 0) {
$budget[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
//Unique Identifier check
//Suck up all activity identifiers - check they start with the reporting org string
//We count by storing the activity id in an array
//if there is no identifier then set a dummy one to dump it into the 'bad' pile
if (!isset($activity->{'iati-identifier'})) {
$iati_identifier = "noIdentifierGiven" . $i;
$i++;
} else {
$iati_identifier = (string) $activity->{'iati-identifier'};
}
if (isset($activity->{'reporting-org'}->attributes()->ref)) {
$reporting_org_ref = (string) $activity->{'reporting-org'}->attributes()->ref;
//echo $reporting_org_ref . PHP_EOL;
//echo $iati_identifier . PHP_EOL;
if (strpos($reporting_org_ref, $iati_identifier) == 0) {
//echo "yes";
$identifiers[$hierarchy]["good"][] = $iati_identifier;
} else {
//echo "no";
$identifiers[$hierarchy]["bad"][] = $iati_identifier;
}
} else {
$identifiers[$hierarchy]["bad"][] = $iati_identifier;
}
//Financial transaction (Commitment)
$transactions = $activity->transaction;
//if (count($transactions) == 0) {
// echo $id;
//die;
//}
if (isset($transactions) && count($transactions) > 0) {
//something not quite right here
//Loop through each of the elements
foreach ($transactions as $transaction) {
//print_r($transaction);
//Counts number of elements of this type in this activity
//$no_transactions[$hierarchy]++;
//$transaction_date = (string)$transaction->{'transaction-date'}->attributes()->{'iso-date'};
if (isset($transaction->{'transaction-type'})) {
$transaction_type = (string) $transaction->{'transaction-type'}->attributes()->{'code'};
if ($transaction_type == "C") {
$transaction_type_commitment[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
if ($transaction_type == "D") {
$transaction_type_disbursement[$hierarchy][] = (string) $activity->{'iati-identifier'};
//Count the number of disbursements at this level
$no_disbursements[$hierarchy]++;
//now test it and count the passes
if (isset($transaction->{"receiver-org"})) {
//We have a provider-org = pass!
$no_tracable_transactions[$hierarchy]++;
}
//$no_disbursements = $no_incoming_funds = $no_tracable_transactions = array();
}
if ($transaction_type == "IF") {
//Count the number of IFs at this level
$no_incoming_funds[$hierarchy]++;
if (isset($transaction->{"provider-org"})) {
//We have a provider-org = pass!
$no_tracable_transactions[$hierarchy]++;
}
}
if ($transaction_type == "E") {
$transaction_type_expenditure[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
}
//if code attribute exists
}
}
//Going to need a count of disbursements and of IF transactions
//Then need to test each against a set of criteria
/*if ($transaction_type == NULL) {
$transaction_type = "Missing";
echo "missing";
}
if ($transaction_type !="D") {
echo $id;
//die;
}*/
//Locations
//We can have more than one location, but they should add up to 100%
$locations = $activity->location;
//if (!isset($activities_with_location[$hierarchy])) {
// $activities_with_location[$hierarchy] = 0;
//}
if (isset($locations) && count($locations) > 0) {
$activities_with_location[$hierarchy][] = (string) $activity->{'iati-identifier'};
foreach ($locations as $location) {
if (isset($location->coordinates)) {
$activities_with_coordinates[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
if (isset($location->administrative)) {
if (isset($location->administrative->attributes()->adm1)) {
$adm1 = string($location->administrative->attributes()->adm1);
}
if (isset($location->administrative->attributes()->adm2)) {
$adm2 = string($location->administrative->attributes()->adm2);
}
if (isset($adm1) && len($adm1) > 0 || isset($adm2) && len($adm2) > 0) {
$activities_with_adminstrative[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
}
}
}
//Sector
$sectors = $activity->sector;
if (isset($sectors) && count($sectors) > 0) {
//$activities_with_sector[$hierarchy][] = (string)$activity->{'iati-identifier'};
foreach ($sectors as $sector) {
if (!isset($sector->attributes()->vocabulary)) {
$activities_sector_assumed_dac[$hierarchy][] = (string) $activity->{'iati-identifier'};
} elseif ((string) $sector->attributes()->vocabulary == "DAC") {
//echo "DAC";
$activities_sector_declared_dac[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
}
}
//Last-updated-datetime
$last_updated = $activity->attributes()->{'last-updated-datetime'};
$last_updated = strtotime($last_updated);
if (!isset($most_recent[$hierarchy])) {
$most_recent[$hierarchy] = 0;
}
if ($last_updated > $most_recent[$hierarchy]) {
$most_recent[$hierarchy] = $last_updated;
}
//Activity dates
$activity_dates = $activity->{"activity-date"};
//if (count($activity_dates) > 0) {
//if ($activity_dates !=NULL) {
// $activities_with_at_least_one[$hierarchy]++;
//}
foreach ($activity_dates as $activity_date) {
//$attributes = array("end-actual","end-planned","start-actual","start-planned");
// $no_activity_dates[$hierarchy]++;
//foreach($attributes as $attribute) {
$type = (string) $activity_date->attributes()->type;
if ($type == "start-actual" || $type == "start-planned") {
$type = "start";
}
if ($type == "end-actual" || $type == "end-planned") {
$type = "end";
}
//$date = (string)$activity_date->attributes()->{'iso-date'};
//Special Case for DFID
//$date = (string)$activity_date;
//echo $date; die;
// $unix_time = strtotime($date);
//if ($unix_time) {
// $year = date("Y",strtotime($date));
//} else {
// $year = 0; //we could not parse the date, so store the year as 0
//// }
//$activity_by[$year][$hierarchy][$type]++;
$activities_with_attribute[$hierarchy][$type][] = (string) $activity->{'iati-identifier'};
//Languages
// if($hierarchy == 2) {
$title_langs = $country_langs = $description_langs = $all_langs = array();
//Reset each of these each run through
//Find default language of the activity
$default_lang = (string) $activity->attributes('http://www.w3.org/XML/1998/namespace')->{'lang'};
//echo $default_lang;
//Find recipient countries for this activity:
$recipient_countries = $activity->{"recipient-country"};
foreach ($recipient_countries as $country) {
$code = (string) $country->attributes()->code;
//Look up default language for this code:
$country_langs[] = look_up_lang($code);
}
//print_r($country_langs);
//Find all the different languages used on the title element
$titles = $activity->title;
foreach ($titles as $title) {
//create an array of all declared languages on titles
$title_lang = (string) $title->attributes('http://www.w3.org/XML/1998/namespace')->{'lang'};
if ($title_lang == NULL) {
$title_langs[] = $default_lang;
} else {
$title_langs[] = $title_lang;
}
$title_lang = "";
}
//Find all the different languages used on the description element
$descriptions = $activity->description;
foreach ($descriptions as $description) {
//create an array of all declared languages on titles
$description_lang = (string) $description->attributes('http://www.w3.org/XML/1998/namespace')->{'lang'};
if ($description_lang == NULL) {
$description_langs[] = $default_lang;
} else {
$description_langs[] = $description_lang;
}
$description_lang = "";
}
//print_r($title_langs);
//die;
//Merge these arrays
$all_langs = array_merge($description_langs, $title_langs);
$all_langs = array_unique($all_langs);
//Loop through the country languiages and see if they are found on either the title or description
foreach ($country_langs as $lang) {
if (in_array($lang, $all_langs)) {
$activies_in_country_lang[$hierarchy][] = (string) $activity->{'iati-identifier'};
}
}
//$description_lang = (string)$activity->description->attributes('http://www.w3.org/XML/1998/namespace')->{'lang'};
// }
}
}
//end foreach
}
//end if not organisation file
}
//end if xml is created
}
// end if file is not a system file
}
//end while
closedir($handle);
}
//if (isset($types)) {
//echo "no_activities" . PHP_EOL;
//print_r($no_activities);
//echo "activities_with_at_least_one" . PHP_EOL;
//print_r($activities_with_at_least_one);
//echo "no_activity_dates" . PHP_EOL;
//print_r($no_activity_dates);
//echo "activity_by_year" . PHP_EOL;
ksort($activity_by);
//print_r($activity_by);
//echo "activities_with_attribute" . PHP_EOL;
//print_r($activities_with_attribute);
//foreach($types as $attribute_name=>$attribute) {
/// echo $attribute_name;
//foreach($attribute as $hierarchy=>$values) {
// echo $hierarchy;
// print_r(array_count_values($values));
// }
// }
//echo count($participating_org_implementing[0]); die;
$found_hierarchies = array_unique($found_hierarchies);
sort($found_hierarchies);
//die;
return array("no-activities" => $no_activities, "activities_with_at_least_one" => $activities_with_at_least_one, "no_activity_dates" => $no_activity_dates, "activity_by_year" => $activity_by, "hierarchies" => array_unique($found_hierarchies), "activities_with_attribute" => $activities_with_attribute, "document_links" => $document_links, "result_element" => $result_element, "conditions" => $conditions, "participating_org_accountable" => $participating_org_accountable, "participating_org_implementing" => $participating_org_implementing, "budget" => $budget, "identifiers" => $identifiers, "transaction_type_commitment" => $transaction_type_commitment, "transaction_type_disbursement" => $transaction_type_disbursement, "transaction_type_expenditure" => $transaction_type_expenditure, "no_disbursements" => $no_disbursements, "no_tracable_transactions" => $no_tracable_transactions, "no_incoming_funds" => $no_incoming_funds, "activities_with_location" => $activities_with_location, "activities_with_coordinates" => $activities_with_coordinates, "activities_with_adminstrative" => $activities_with_adminstrative, "activities_sector_assumed_dac" => $activities_sector_assumed_dac, "activities_sector_declared_dac" => $activities_sector_declared_dac, "most_recent" => $most_recent, "activies_in_country_lang" => $activies_in_country_lang);
//} else {
// return FALSE;
//}
}
if ($encoding != FALSE) {
$basic['DetectEncoding'] = $encoding;
} else {
$basic['DetectEncoding'] = "Encoding: Not detected";
}
//Activty or Organisation specific tests
if (xml_child_exists($xml, "//iati-activity")) {
//ignore organisation files
$checking_activity_file = true;
$basic['activities'] = count($xml->xpath("//iati-activity"));
//$generated = $xml->attributes()->{'generated-datetime'};
//$version = $xml->attributes()->version;
//$activities = count($xml->xpath("//iati-activity"));
$hierarchies = $xml->xpath("//@hierarchy");
$basic['hierarchies'] = get_values($hierarchies, "int");
} elseif (xml_child_exists($xml, "//iati-organisation")) {
$checking_organisation_file = true;
$org_identifier = $xml->xpath("//iati-identifier");
$basic['org_iati_identifier'] = (string) $org_identifier[0];
//print_r($xml->xpath("//name")); die;
$org_name = $xml->xpath("//name");
//a simplexml object
$name = (string) $org_name[0];
$basic['org_name'] = $name;
//$basic['org_name'] = $basic['org_name']->0;
$org_ref = $xml->xpath("//reporting-org/@ref");
$basic['org_reporting_org_ref'] = (string) $org_ref[0];
$basic['org_recipient_country_budget'] = count($xml->xpath("//recipient-country-budget"));
$basic['org_recipient_org_budget'] = count($xml->xpath("//recipient-org-budget"));
$basic['org_total_budget'] = count($xml->xpath("//total-budget"));
$basic['org_document_link'] = count($xml->xpath("//document-link"));
