Metropolitan Museum of Art Public Images Downloader

getMetImages.php

<?php

/*
 * Metropolitan Museum of Art Public Images Downloader
 *  (c) 2017/02/17(Fri) yoya@awm.jp
 * Usage) php getMetImages.php MetObjects.csv
 */
$file = new SplFileObject($argv[1]);
$file->setFlags(SplFileObject::READ_CSV | SplFileObject::READ_AHEAD | SplFileObject::SKIP_EMPTY | SplFileObject::DROP_NEW_LINE);
$idx = 0;
foreach ($file as $record) {
    if (is_null($record[0])) {
        continue;
    }
    foreach ($record as $i => $value) {
        if (substr($value, 0, 3) === "\xef\xbb\xbf") { // <U+FEFF>
            $value = substr($value, 3);
        }
        $record[$i] = trim($value);
    }
    if ($idx === 0) {
        $keys = $record;
    } else {
        echo "$idx".PHP_EOL;
        $new_record = [];
        foreach ($record as $i => $value) {
            $new_record[$keys[$i]] = $value;
        }
        $record = $new_record;
        /*
         * main routine
         */
        $isPublicDomain = $record["Is Public Domain"];
        if ($isPublicDomain !== "False") {
            if ($isPublicDomain !== "True") {
                $errMesg = "Unknown Is Public Domain flag:".$isPublicDomain;
                trigger_error($errMesg, E_USER_ERROR);
            }
            $linkResource = $record["Link Resource"];
            echo $linkResource.PHP_EOL;
            $html = file_get_contents($linkResource);
            file_put_contents("$idx.html", $html);
            if (preg_match("(http://.+/original/.*\.jpg)", $html, $matches) ||
                preg_match("(http://.+/original/.*\.png)", $html, $matches)) {
                $originalUrl = $matches[0];
                echo $originalUrl.PHP_EOL;
                `wget $originalUrl`;
            }
        }
    }
    $idx ++;
}