爬小说站

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

$content_grep = '/&nbsp;&nbsp;&nbsp;&nbsp;(.*)<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];
}