Split GTFS files into one file per transport operator

splitGtfs.php

#!/usr/bin/env php
<?php

const GTFS_PATH = '/tmp/gtfs/';
const TARGET_DIR  = './splitGtfs/';

@mkdir(GTFS_PATH);
@mkdir(TARGET_DIR);

$zip = new ZipArchive;
$res = $zip->open($argv[1]);
if ($res === TRUE) {
  $zip->extractTo(GTFS_PATH);
  $zip->close();
} else {
	die("Failed to get GTFS");
}

$gtfs_agency = deserializeCSV(GTFS_PATH . 'agency.txt');
$gtfs_calendar = deserializeCSV(GTFS_PATH . 'calendar.txt');
$gtfs_calendar_dates = deserializeCSV(GTFS_PATH . 'calendar_dates.txt');
$gtfs_feed_info = deserializeCSV(GTFS_PATH . 'feed_info.txt');
$gtfs_routes = deserializeCSV(GTFS_PATH . 'routes.txt');
$gtfs_stops = deserializeCSV(GTFS_PATH . 'stops.txt');
$gtfs_stop_times = deserializeCSV(GTFS_PATH . 'stop_times.txt');
$gtfs_transfers = deserializeCSV(GTFS_PATH . 'transfers.txt');
$gtfs_trips = deserializeCSV(GTFS_PATH . 'trips.txt');

foreach ($gtfs_agency as $agencyId => $agency) {
	$part_feed_info = $gtfs_feed_info;
	$part_agency = [$agency];

	echo PHP_EOL . 'parsing... ' .  $agency['agency_name'] . PHP_EOL;

    $part_routes = [];
    $used_routes = [];
	foreach ($gtfs_routes as $key => $route) {
		if ($route['agency_id'] != $agency['agency_id']){
			continue;
		}
        
        $part_routes[] = $route;
        $used_routes[$route['route_id']] = $route['route_id'];
	}

	echo 'Kept ' .  count($part_routes) . ' out of ' .  count($gtfs_routes) . ' routes' . PHP_EOL;

	if(count($part_routes) == 0) {
		echo 'No routes, skipped!' . PHP_EOL;
		continue;
	}

	$part_trips = [];
	$used_services = [];
	$used_stops = [];
	$used_trips = [];
	foreach ($gtfs_trips as $key => $trip) {
		if (!array_key_exists($trip['route_id'], $used_routes)){
			continue;
		}
		$part_trips[] = $trip;
		$used_trips[] = $trip['trip_id'];
		$used_services[$trip['service_id']] = $trip['service_id'];
		$used_trips[$trip['trip_id']] = $trip['trip_id'];
	}

	echo $agency['agency_name'] . ' has ' . count($part_trips) . ' trips' . PHP_EOL;

	$part_stop_times = [];
	foreach ($gtfs_stop_times as $key => $stop_time) {
		if (!array_key_exists($stop_time['trip_id'], $used_trips)){
			continue;
		}
		$part_stop_times [] = $stop_time;
		$used_stops[$stop_time['stop_id']] = $stop_time['stop_id'];
	}
	echo $agency['agency_name'] . ' has ' . count($part_stop_times) . ' stop times' . PHP_EOL;


	$part_calendar = [];
	foreach ($gtfs_calendar as $key => $calendar) {
		if (!array_key_exists($calendar['service_id'], $used_services)){
			continue;
		}
		$part_calendar [] = $calendar;
	}

	$part_calendar_dates = [];
	foreach ($gtfs_calendar_dates as $key => $calendar_date) {
		if (!in_array($calendar_date['service_id'], $used_services)){
			continue;
		}
		$part_calendar_dates [] = $calendar_date;
	}

	$part_stops = [];
	foreach ($gtfs_stops as $key => $stop) {
		if (!in_array($stop['stop_id'], $used_stops)){
			continue;
		}
		$part_stops [] = $stop;
	}

	echo $agency['agency_name'] . ' has ' . count($part_stops) . ' stops' . PHP_EOL;
	
	$part_transfers = [];
	foreach ($gtfs_transfers as $key => $transfer) {
		if (!array_key_exists($transfer['from_stop_id'], $used_stops) && !array_key_exists($transfer['to_stop_id'], $used_stops)){
			continue;
		}
		$part_transfers [] = $transfer;
	}

	echo 'exporting... ' .  $agency['agency_name'] . PHP_EOL;
	@mkdir(TARGET_DIR . $agency['agency_id'] . '/');
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/agency.txt', $part_agency);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/calendar.txt', $part_calendar);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/calendar_dates.txt', $part_calendar_dates);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/feed_info.txt', $part_feed_info);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/routes.txt', $part_routes);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/stops.txt', $part_stops);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/stop_times.txt', $part_stop_times);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/transfers.txt', $part_transfers);
	serializeCSV(TARGET_DIR . $agency['agency_id'] . '/trips.txt', $part_trips);

	$zip = new ZipArchive;
	echo 'Saving ' . TARGET_DIR . $agency['agency_id'] . '.zip...' . PHP_EOL;

	if ($zip->open(TARGET_DIR . $agency['agency_id'] . '.zip', ZipArchive::CREATE) === TRUE)
	{

		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/agency.txt','agency.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/calendar.txt','calendar.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/calendar_dates.txt','calendar_dates.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/feed_info.txt','feed_info.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/routes.txt','routes.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/stops.txt','stops.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/stop_times.txt','stop_times.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/transfers.txt','transfers.txt');
		$zip->addFile(TARGET_DIR . $agency['agency_id'] . '/trips.txt','trips.txt');

	    // All files are added, so close the zip file.
	    $zip->close();
	}
}


/**
 * Serialize data to a CSV file
 *
 * @param $headers array The headers which should be written
 * @param $station array The data as an associative array (header => value) to serialize to CSV
 * @return string CSV representation of the data
 */
function serializeCSV($path, $data)
{
    if ($data === []) {
        return;
    }
	$headers = array_keys($data[0]);

	echo 'serializing... ' .  $path . ' with ' . count($data). ' rows' . PHP_EOL;
  
        $file = fopen($path, 'w');
        fputcsv($file, $headers);

	foreach ($data as $key => $dataRow) {
            fputcsv($file, serializeCSVLine($headers, $dataRow));
	}
        fclose($file);
}

/**
 * Serialize data to a CSV row
 *
 * @param $headers array The headers which should be written
 * @param $station array The data as an associative array (header => value) to serialize to CSV
 * @return string[] the row containing the data
 */
function serializeCSVLine($headers, $station): array
{
    $row = [];
    // Loop over all headers
    for ($i = 0; $i < count($headers); $i++) {
        // Which value we are appending
        $header = $headers[$i];
        // Add key if it exists, otherwise leave empty
        if (key_exists($header, $station)) {
            $row[] = $station[$header];
        }
    }
    return $row;
}

/**
 * Load a CSV file and store it in an associative array with incremental keys
 * Each line is stored as an associative array using column headers as key and the fields as value.
 *
 * @param $csvPath string File path leading to the CSV file
 * @return array the deserialized data
 */
function deserializeCSV($csvPath): array
{
    // Open the GTFS stops file and read it into an associative array
    $fileReadHandle = fopen($csvPath, 'r');
    if (!$fileReadHandle) {
        die($csvPath . ' could not be opened!');
    } else {
    	echo 'Deserializing ' . $csvPath . PHP_EOL;
    }
    // Read the original headers
    $headers = fgetcsv($fileReadHandle);
    $entries = [];
    // Go through all rows
    while (($entry = fgetcsv($fileReadHandle)) !== false) {
        $entries[] = array_combine($headers, $entry);
    }
    return $entries;
}


?>