programs2csv.php

<?php

// written by AJ Slaghuis 2013
define('DS', DIRECTORY_SEPARATOR);
$basepath = './program_info';
$fields = array();
$fields["prid"] = "prid";
$data = array();

function validate_end($file) {
    $arr = array(");\n//e", ");\n//ep",");\n//epc");
    foreach ($arr as $str) {
        if (substr($file, - strlen($str) ) == $str){         
            return strlen($str);
        }
    }
    return -1;
}

function validate_begin($file) {
    $str = 'parseMetadata(';
    if (substr($file, 0, strlen($str)) != $str) {
        return -1;
    }
    return strlen($str);
}

@unlink('output.csv');
$entries = scandir($basepath); //sorted alphabetically

foreach ($entries as $entry) {
    if ($entry == '.' || $entry == '..' || is_dir($entry) || substr($entry, -5) != '.json')
        continue; //  
    $prid = substr($entry, 0, -5);
    if ($prid == null)
        continue;

    $file = file_get_contents($basepath . DS . $entry);

    $end_trim = validate_end($file);
    $begin_trim = validate_begin($file);

    if ($end_trim < 1 || $begin_trim < 1) {
          if (strcmp($file, "parseMetadata({\"STATUS\":\"ERROR\",\"VERSION\":\"1.9.3\",\"code\":440,\"error\":\"De opgevraagde aflevering is niet gevonden.\"});") != 0) {
            echo("$entry FILE FORMAT ERROR \n");
            echo($file . "\n\n");
            die(-1);
        }// niet gevonden
        continue;
    }

    $fields["prid"] = "prid";
    $data[$prid]["prid"] = $prid;

    $file_sub = substr((substr($file, $begin_trim)), 0, - ( $end_trim ));

//    if (strlen($file_sub) > 100)
//        continue;
//    echo $file;
//    die();

    $json = json_decode($file_sub);


    foreach ($json as $key => $value) {
        if ($key == 'images' || $key == 'pubopties' || $key == 'sitestat')
            continue;
        if (is_object($value)) {
            foreach ($value as $subkey => $subvalue) {
                $fields["$key" . "_" . "$subkey"] = "$key" . "_" . "$subkey";
                $data[$prid]["$key" . "_" . "$subkey"] = str_replace(array("\n", "\r", "\t"), '', html_entity_decode($subvalue)
                );
            }
        } else if (is_array($value)) {
            for ($i = 0; $i < count($value); $i++) {
                foreach ($value[$i] as $vkey => $vvalue) {
                    $fields[$key . "_" . $vkey . "_" . $vvalue]
                            = $key . "_" . $vkey . "_" . $vvalue;
                    $data[$prid][$key . "_" . $vkey . "_" . $vvalue]
                            = str_replace(array("\n", "\r", "\t"), ''
                            , html_entity_decode($vvalue));
                }
            }
        } else {// key value pair
            $fields[$key] = $key;
            $data[$prid][$key] = str_replace(array("\n", "\r", "\t")
                    , '', html_entity_decode($value));
        }
    }
}
$outputheader = "";
$outputheader .= implode("\t ", $fields);
$outputheader.= "\n";
file_put_contents('output.csv', $outputheader);

foreach ($data as $item) {
    $outputrow = '';
    foreach ($fields as $field) {
        if (isset($item[$field])) {
            $outputrow .= $item[$field];
        } else {
            $outputrow .= ""; //\"\"";
        }
        $outputrow .= "\t ";
    }
    $outputrow = substr($outputrow, 0, -2);
    $outputrow .= "\n";
    file_put_contents('output.csv', $outputrow, FILE_APPEND | LOCK_EX);
}
?>