[PHP] 特定ディレクトリ内の全 *.html ファイルの中身を相対URLに変更

replace-relative-uri.php

<?php
$dir = './path/';
$org_url = 'http://example.com/';

function scan_file($dir, $file_type = '{*.html,*.css,*.js,*.gif,*.png,*.jpg,*.zip,*.xml}') {
    $list = $tmp = array();
    foreach(glob($dir . '*/', GLOB_ONLYDIR) as $child_dir) {
        if ($tmp = scan_file($child_dir, $file_type)) {
            $list = array_merge($list, $tmp);
        }
    }

    foreach(glob($dir . $file_type, GLOB_BRACE) as $image) {
        $list[] = $image;
    }

    return $list;
}

function replace_relative_URI($content, $home_url) {
    $parsed   = parse_url($home_url);
    $replace  = $parsed['scheme'] . '://' . $parsed['host'];
    if (isset($parsed['port']))
        $replace .= ':'.$parsed['port'];

    $pattern  = array(
        '# (href|src|action)="'.preg_quote($replace).'([^"]*)"#ism',
        "# (href|src|action)='".preg_quote($replace)."([^']*)'#ism",
    );
    $content  = preg_replace($pattern, ' $1="$2"', $content);

    $pattern  = '#<(meta [^>]*property=[\'"]og:[^\'"]*[\'"] [^>]*content=|link [^>]*rel=[\'"]canonical[\'"] [^>]*href=|link [^>]*rel=[\'"]shortlink[\'"] [^>]*href=|data-href=|data-url=)[\'"](/[^\'"]*)[\'"]([^>]*)>#uism';
    $content = preg_replace($pattern, '<$1"'.$replace.'$2"$3>', $content);


    return $content;
}


$files = scan_file($dir, '{*.html}');
$info = new FInfo(FILEINFO_MIME_TYPE);

foreach ($files as $filename) {
    $filetype = $info->file($filename);

    if ( $filetype == 'text/plain') {
        if (preg_match('/\.css$/i', $filename))
            $filetype = 'text/css';
        else if (preg_match('/\.js$/i', $filename))
            $filetype = 'application/x-javascript';
        else if (preg_match('/\.html?$/i', $filename))
            $filetype = 'text/html';
        else if (preg_match('/\.xml$/i', $filename))
            $filetype = 'application/xml';
    }
    echo $filetype . ':' . str_replace($dir, '', $filename) . "\n";

    $file_content = replace_relative_URI(file_get_contents($filename), $org_url);
    rename( $filename, $filename . '.org');
    file_put_contents( $filename, $file_content );
}