legacy-links-php-export.php

legacy-links-php-export.php

<?php

//$hostname = "";     // the address of the MySQL server
//$username = "";    // your username
//$password = "";     //your password
//$databaseName = "";   // name of the database

$cn = mysqli_connect($hostname, $username, $password, $databaseName) or die("Cannot connect to database.");

//Array
//(
//    [0] => 6978
//    [id] => 6978
//    [1] => 2013-05-01 18:41:56
//    [dateadded] => 2013-05-01 18:41:56
//    [2] => http://oembed.com/
//    [url] => http://oembed.com/
//    [3] => oEmbed
//[urltext] => oEmbed
//[4] => oEmbed is a format for allowing an embedded representation of a URL on third party sites. The simple API allows a website to display embedded content (such as photos or videos) when a user posts a link to that resource, without having to parse the resource directly.
//[accompanyingtext] => oEmbed is a format for allowing an embedded representation of a URL on third party sites. The simple API allows a website to display embedded content (such as photos or videos) when a user posts a link to that resource, without having to parse the resource directly.
//[5] =>
//    [via] =>
//    [6] =>
//    [category] =>
//    [7] => standards,oembed,multimedia,programming,webdev,json,link,api,web services,video,embed
//[tags] => standards,oembed,multimedia,programming,webdev,json,link,api,web services,video,embed
//[8] => 1
//    [liveonsite] => 1
//    [9] =>
//    [hash] =>
//    [10] =>
//    [meta] =>
//    [11] =>
//    [toread] =>
//)


/**
 * @param $row
 * @return void
 */
function rowToStructuredData($row) {
    $id = 'sidebarlinks.' . $row['id'];
    $tags = explode(",", $row['tags']);
    // trim any empty tags
    $tags = array_filter($tags, 'trim');
    $date_added = $row['dateadded'];
    // format as iso 8601
    $date_added = date('c', strtotime($date_added));
    $url_text = $row['urltext'];
    $accompanying_text = $row['accompanyingtext'];
    $via = $row['via'];
    $category = $row['category'];
    $live_on_site = $row['liveonsite'];
    $url = $row['url'];
    $hash = $row['hash'];
    $meta = $row['meta'];
    $to_read = $row['toread'];

    $nonsense = '';
    if (trim($hash) != '') {
        $nonsense .= 'hash:' . $hash . "|";
    }
    if (trim($meta) != '') {
        $nonsense .= 'meta:' . $meta . "|";
    }

      $private = !$live_on_site;


    if ($to_read) {
        $nonsense .= 'to_read:' . json_encode($to_read) . "|";
    }

    // trim final pipe from nonsense

    if (trim($via) != '') {
        // if it begins with http, add it to nonsense
        if (substr($via, 0, 4) == 'http') {
            $nonsense .= 'via:' . $via . "|";
            $domain_name = parse_url($via, PHP_URL_HOST);
            // if it has a www. at beginning, strip it
            if (substr($domain_name, 0, 4) == 'www.') {
                $domain_name = substr($domain_name, 4);
            }
            $viaTag = 'via:' . $domain_name;
            $tags[] = $viaTag;
        } else {
            $viaTag = 'via:' . $via;
            $tags[] = $viaTag;
        }
    }

    // add the domain as a tag
    $domain_name = parse_url($url, PHP_URL_HOST);
    // if it has a www. at beginning, strip it
    if (substr($domain_name, 0, 4) == 'www.') {
        $domain_name = substr($domain_name, 4);
    }
    $domainTag = 'domain:' . $domain_name;
    $tags[] = $domainTag;




    $nonsense = rtrim($nonsense, '|');



    return [
        'id' => $id,
        'url' => $url,
        'tags' => $tags,
        'date_added' => $date_added,
        'title' => $url_text,
        'description' => $accompanying_text,
        'private' => $private,
        'via' => $via,
//        'category' => $category, NOT USED
//        'live_on_site' => $live_on_site,
//        'hash' => $hash,
//        'meta' => $meta,
//        'to_read' => $to_read,
        'extra_derived' => $nonsense,
    ];
}


//$requested_id = (int)$_GET['id'];

$query = "SELECT 
    * 
FROM `sidebarlinks` 
order by id asc limit 30000";

// min 3
// max 28109


// get by column name
$result_mode = MYSQLI_ASSOC;

$result = mysqli_query($cn, $query, $result_mode);

// verbose error
if (!$result) {
    die('Invalid query: ' . mysqli_error());
}
// json header
//header('Content-Type: application/json');

while ($row = mysqli_fetch_array($result)) {

    $id = $row['id'];
    // format filename like 00003.json
    $filename = str_pad($id, 5, '0', STR_PAD_LEFT) . '.json';
    $data = rowToStructuredData($row);
    $json = json_encode($data, JSON_PRETTY_PRINT);
    file_put_contents($filename, $json);
    print $filename . "\n";
//    print json_encode(rowToStructuredData($row));

}

// 3 to 28109