spider/lib/Parse.php

<?php
namespace Spider\Lib;
use Dever;
use Spider\Lib\Doc\Dom;
class Parse
{
	private $url = '';
	private $host = '';
	private $log;
	private $doc = array();
	private $data = array();

	public function __construct($url, $project, $rule, $param, $col, $set, $push)
	{
		$doc = Doc::getInstance($url, $rule);
		$doc->log(new Log($project));
		$data = $doc->get($param);
		if ($data) {
			if (!is_array($data) && !is_object($data)) {
				$state = Dever::json_decode($data);
				if ($state) {
					$data = $state;
				}
			}
			if ($data) {
				if (is_array($data)) {
					$domain = parse_url($url);
					$host = $domain['scheme'] . '://' . $domain['host'] . '/';

					foreach ($data as $k => $v) {
						if (is_string($v) && !strstr($v, 'http')) {
							$v = $host . ltrim($v, '/');
						}
						$this->data[$k] = $this->load($doc, $k, $v, $col, $set, $push, $project, $v);
					}
				} else {
					$this->data = $this->load($doc, 0, $data, $col, $set, $push, $project, $url);
				}
			}
		}
		$doc->saveLog();
	}

	public function get()
	{
		return $this->data;
	}

	private function load($doc, $index, $data, $col, $set, $push, $project, $source)
	{
		if (!$col) {
			if (Dever::input('test') == 1) {
				$doc->outLog();
				echo 'error';die;
			}
			return false;
		}
		$result = $table = array();

		if (isset($col[1])) {
			$data = $doc->init($data);
			$this->getCol($doc, $col[1], $data, $result, $table, $source);
		}

		if (isset($col[2])) {
			$data = $doc->getCur();
			$this->getCol($doc, $col[2], $data, $result, $table, $source);
		}

		if ($set) {
			foreach ($set as $v) {
				$value = $this->set($index, $v, $project);
				$result[$v['key']] = $value;
				if (Dever::input('test') == 1) {
					$table[$v['name']] = $value;
				}
			}
		}

		if ($push) {
			$result['test'] = Dever::input('test');
			$this->push($push, $result, $project);
		}

		if (Dever::input('test') == 1) {
			$doc->outLog();
			echo Dever::table($table);die;
		}

		$this->update($result, $project, $source);
		return $result;
	}

	private function getCol($doc, $col, $data, &$result, &$table, $source = false)
	{
		foreach ($col as $v) {
			$callback = false;
			if (strpos($v['key'], '.') !== false) {
				$temp = explode('.', $v['key']);
				$v['key'] = $temp[1];
				$callback = $temp[0];
			}

			if ($source && strstr($v['collect_rule'], '{link}')) {
				$v['collect_rule'] = str_replace('{link}', $source, $v['collect_rule']);
			}

			$value = $doc->rule($data, $col, $v);
			if ($value == 'error') {
				break;
			}
			if ($callback) {
				if (function_exists($callback)) {
					$value = $callback($value);
				} else {
					$value = Dever::{$callback}($value);
				}
			}

			if ($v['local'] == 1) {
				$this->res = $v['res_key'];
				$value = $this->local($value, $v['type']);
			} elseif ($v['type'] == 2) {
				if (is_string($value) && strstr($value, '[')) {
					$temp = Dever::json_decode($value);
					if ($temp) {
						$value = implode(',', $temp);
					}
				}
			}

			if ($v['collect_filter_link'] == 1) {
				$value = $this->filter($value);
			}

			if ($value) {
				$result[$v['key']] = $value;
				if (Dever::input('test') == 1) {
					$table[$v['name']] = $value;
				}
			}
		}
	}

	private function push($push, $data, $project)
	{
		$push = Dever::split($push);
		$data['project_id'] = $project;
		foreach ($push as $k => $v) {
			if (strstr($v, 'http')) {
				Dever::curl($v, $data, 'post');
			} else {
				Dever::load($v, $data);
			}
		}
	}

	private function set($index, $data, $project)
	{
		if ($data['type'] == 1) {
			return $data['value'];
		} elseif ($data['type'] == 2) {
			$old = 0;
			$info = Dever::db('spider/data')->getOne(array('pid' => $project));
			if($info) {
				$value = json_decode($info['value'], true);
				if (isset($value[$data['key']])) {
					$old = $value[$data['key']];
				}
			}
			return $data['value'] + $index + $old;
		} elseif ($data['type'] == 3) {
			$eval = '$value = ' . $data['value'] . ';';
			eval($eval);
			return $value;
		} elseif ($data['type'] == 4) {
			$temp = Dever::split($data['value']);
			return mt_rand($temp[0], $temp[1]);
		} elseif ($data['type'] == 5) {
			$temp = Dever::split($data['value']);
			$temp[0] = Dever::maketime($temp[0]);
			$temp[1] = Dever::maketime($temp[1]);
			return mt_rand($temp[0], $temp[1]);
		}
	}

	private function filter($content)
	{
		$rule = '<(a).+href="(.*?)"(.*?)>(.*?)<\/a>';
		$content = preg_replace_callback('/' . $rule . '/i', array($this, 'filter_replace'), $content);
		return $content;
	}

	private function filter_replace($result)
	{
		if (isset($result[4]) && $result[4]) {
			return $result[4];
		}
	}

	private function local($content, $type = 1)
	{
		if ($type == 1) {
			$doc = Dom::init($content);
			$pic = Dom::find($doc, '$("img").each().attr("src")');
			if ($pic) {
				$content = $this->local_replace($pic, $content);
			}
			$video = Dom::find($doc, '$("video").each().attr("src")');
			if ($video) {
				$content = $this->local_replace($video, $content);
			}
			$audio = Dom::find($doc, '$("audio").each().attr("src")');
			if ($audio) {
				$content = $this->local_replace($audio, $content);
			}
			/*
			$rule = '<(img|video|audio).+src=\"?(.+\.(jpg|gif|bmp|bnp|png))\"?.+>';
			$content = preg_replace_callback('/' . $rule . '/i', array($this, 'local_rule_replace'), $content);
			*/
		} else {
			$content = $this->copy($content);
		}

		return $content;
	}

	private function local_replace($file, $content)
	{
		if (is_string($file) && strstr($file, '[')) {
			$file = Dever::json_decode($file);
		}
		if (is_array($file)) {
			foreach ($file as $k => $v) {
				$content = $this->local_replace($v, $content);
			}
		} else {
			$result = $this->copy($file);
			if ($result) {
				$content = str_replace($file, $result, $content);
			}
		}
		return $content;
	}

	private function local_rule_replace($result)
	{
		if (isset($result[2]) && $result[2]) {
			$file = $this->copy($result[2]);
			if ($file) {
				$result[0] = str_replace($result[2], $file, $result[0]);
				return $result[0];
			}
		}
	}

	private function copy($file)
	{
		if (is_string($file) && strstr($file, '[')) {
			$temp = Dever::json_decode($file);
			if ($temp) {
				$file = array();
				foreach($temp as $k => $v) {
					$f = $this->copy($v);
					if ($f) {
						$file[] = $f;
					}
				}
				$file = implode(',', $file);
				return $file;
			}
		}
		if (strstr($file, ',')) {
			$temp = explode(',', $file);
			$file = array();
			foreach($temp as $k => $v) {
				$f = $this->copy($v);
				if ($f) {
					$file[] = $f;
				}
			}
			$file = implode(',', $file);
			return $file;
		}
		if (!strstr($file, 'http') && strstr($file, '//')) {
			$file = 'https:' . $file;
		}
		$data = Dever::load('upload/save.copy?file=' . $file . '&key=' . $this->res . '&state=1');
		if (isset($data['status']) && $data['status'] == -1) {
			return '';
		} elseif (isset($data['url'])) {
			return $data['url'];
		} else {
			return '';
		}
	}

	private function update($data, $project, $source)
	{
		if (!$data) {
			return;
		}
		$param['pid'] = $project;
		$param['source'] = $source;
		$info = Dever::db('spider/data')->one($param);
		$param['value'] = json_encode($data, JSON_UNESCAPED_UNICODE);
		if ($info) {
			$update = $param;
			$id = $update['where_id'] = $info['id'];
			Dever::db('spider/data')->update($update);
		} else {
			$update = $param;
			$id = Dever::db('spider/data')->insert($update);
		}
		echo $id;
		echo "\r\n";
	}
}