/**
* loopnet.com需要美国ip才能访问，所以使用代理服务器
* 免费代理：http://www.xicidaili.com/wt/
* @param $url
* @return mixed
* @throws FunFangException
*/
public static function parseLoopnetProperty($url)
{
if (!$url) {
return null;
}
$dataSourceId = 1000;
$html = new \simple_html_dom();
$httpClient = new GuzzleHttp\Client();
try {
$content = $httpClient->get($url, ['proxy' => 'tcp://108.61.218.68:8888', 'timeout' => 30])->getBody()->getContents();
} catch (GuzzleHttp\Exception\RequestException $e) {
throw new FunFangException("访问 {$url} 时出错！" . $e->getMessage());
}
$html->load($content);
//对该页面进行简单的检查，判断是否有DataId
$dataIdWrapper = $html->find('.property-timestamp', 0);
if (!$dataIdWrapper) {
throw new FunFangException('您输入的URL可能不正确，解析失败！');
}
$dataId = $dataIdWrapper->find('td', 0)->plaintext;
$dataId = str_replace(' ', '', $dataId);
$dataId = str_replace('ListingID:', '', $dataId);
if (!$dataId) {
throw new FunFangException('未找到房源ID！');
}
//先从数据库中查找
if ($dataId) {
$record = Property::where('DataSourceId', $dataSourceId)->where('DataId', $dataId)->first();
if ($record) {
return $record;
}
}
//如果数据库中没有记录，从html中解析并保存到数据库
//只采集特定类型房源
$record = new Property();
$record->DataSourceId = $dataSourceId;
$record->DataId = trim($dataId);
$record->ReferenceUrl = $url;
$propertyTds = $html->find('.property-data', 0)->find('td');
for ($i = 0; $i < count($propertyTds); $i = $i + 2) {
$dataType = trim($propertyTds[$i]->plaintext);
$dataValue = trim($propertyTds[$i + 1]->plaintext);
if (StringUtil::equalsIgnoreCase('Price', $dataType)) {
//价格
$record->ListPrice = str_replace(',', '', str_replace('$', '', $dataValue));
} else {
if (StringUtil::equalsIgnoreCase('Property Type', $dataType)) {
//房产类型
//指定采集的房源类型
$specTypes = ['Multifamily', 'Office', 'Industrial', 'Land', 'Residential Income'];
if (!in_array($dataValue, $specTypes)) {
throw new FunFangException('暂不采集该类型的房源！');
}
$propertyTypeEnum = Util::getPropertyTypeEnum($dataValue);
if ($propertyTypeEnum) {
$record->PropertyType = $propertyTypeEnum['name'];
$record->PropertyTypeEnum = $propertyTypeEnum['value'];
}
} else {
if (StringUtil::equalsIgnoreCase('Lot Size', $dataType)) {
//土地面积
$lotSqFt = str_ireplace('AC', '', $dataValue);
$lotSqFt = str_replace(' ', '', $lotSqFt);
$lotSqFt = $lotSqFt * 43560;
$record->LotSqFt = $lotSqFt;
} else {
if (StringUtil::equalsIgnoreCase('Building Size', $dataType)) {
//建筑面积
$structureSqFt = str_ireplace('SF', '', $dataValue);
$structureSqFt = str_ireplace(' ', '', $structureSqFt);
$structureSqFt = str_ireplace(',', '', $structureSqFt);
$record->StructureSqFt = $structureSqFt;
}
}
}
}
}
$addressStr = $html->find('.basic-info', 0)->find('h1', 0)->plaintext;
$addressStr = str_replace('·', '', $addressStr);
$addressRes = $httpClient->get('http://maps.googleapis.com/maps/api/geocode/json?address=' . $addressStr);
$addressJson = json_decode($addressRes->getBody()->getContents());
if ($addressJson->status == 'OK') {
$addressObj = GoogleGeoHelper::getAddress($addressJson->results);
}
if ($addressObj) {
$record->State = $addressObj->state;
$record->County = $addressObj->county;
$record->City = $addressObj->city;
$record->Address = "{$addressObj->streetNumber} {$addressObj->street}";
$record->PostalCode = $addressObj->postalCode;
//$record->Location = DB::raw("GeomFromText('POINT($addressObj->lng $addressObj->lat)')");
$record->Location = "{$addressObj->lng},{$addressObj->lat}";
}
$record->Description = $html->find('.description', 0)->find('.row', 1)->plaintext;
//描述
$photos = $html->find('.carousel-wrapper', 0)->find('.carousel-inner', 0)->find('img');
//照片
$photoUrls = [];
foreach ($photos as $photo) {
if ($photo->src) {
array_push($photoUrls, $photo->src);
} else {
array_push($photoUrls, $photo->getAttribute('lazy-src'));
}
}
$record->PhotoUrls = implode(',', $photoUrls);
$record->save();
$html->clear();
$newRecord = Property::find($record->Id);
return $newRecord;
}
/**
* 通过MLSID或第三方网站URL获取房源
* @param Request $request
* @return \Illuminate\Contracts\View\Factory|\Illuminate\View\View|string
* @throws \Exception
*/
public function fetch(Request $request)
{
$this->validate($request, ['type' => 'required', 'q' => 'required']);
//搜索内容
$q = trim($request->q);
//判断搜索类型
if ($request->type == 0) {
//MLSID
$model = Property::where('MLSNumber', $q)->first();
if ($model) {
//return view('property._detail', ['record' => $record]);
$view = view('property._detail', ['record' => $model]);
return response()->json(['model' => $model, 'view' => $view]);
}
throw new Exception('未找到！');
} else {
if ($request->type == 1) {
//第三方网站URL
$domain = UrlUtil::getMainDomain($q);
if (StringUtil::equalsIgnoreCase($domain, 'loopnet.com')) {
$model = Util::parseLoopnetProperty($q);
$view = view('property._detail', ['model' => $model])->render();
return response()->json(['model' => $model, 'view' => $view]);
} else {
if (StringUtil::equalsIgnoreCase($domain, 'newhomesource.com')) {
$model = Util::parseNewhomesourceProperty($q);
$view = view('property._detail', ['model' => $model])->render();
return response()->json(['model' => $model, 'view' => $view]);
} else {
throw new Exception('您输入的网站URL暂不支持！');
}
}
} else {
throw new Exception('搜索类型错误！');
}
}
}
