php 爬小说站

原始站点总是弹窗，弹窗就弹窗吧，还有声音，声音还竟是哪种的

虽说可以用chrome屏蔽掉，也可写hosts，但是换个站点还的重新弄，怪麻烦的

<?php
$base = 'http://xx/oo/';
$start = 'xx.shtml';

$content_grep = '/    (.*)<br \\/>/';
$next_grep = '/href=\\'(\\d+\\.shtml)\\'><b>下一页/';

$next = $start;
$file_name = 'out.txt';

while($next) {
    echo 'getting ' . $next . PHP_EOL;
    $result = file_get_contents($base . $next);

    preg_match_all($content_grep, $result, $match);
    $isTitle = true;
    $content = "";
    foreach($match[1] as $line) {
        if($isTitle) {
            $content = $line . PHP_EOL . PHP_EOL;
            $isTitle = false;
        } else {
            $content .= '        ' . $line . PHP_EOL . PHP_EOL;
        }
    }

    $file = fopen($file_name, 'a');
    echo 'write length: ' . strlen($content) . PHP_EOL;
    fwrite($file, $content);
    fclose($file);

    echo '.';

    preg_match($next_grep, $result, $match);
    $next = $match[1];
}
//该片段来自于http://outofmemory.cn