public static function saveParseOlxAuto()
{
foreach (self::$urls as $id => $auto) {
foreach (parent::$cities as $city) {
$xml = file_get_contents('http://olx.ua/transport/' . $auto . '/' . $city . '/rss/');
$movies = new \SimpleXMLElement($xml, LIBXML_NOCDATA);
unset($xml);
foreach ($movies->channel->item as $item) {
if (parent::productUnique(trim((string) $item->link))) {
preg_match("/Цена:\\s\\d(.*?)\\sгрн/", $item->description, $m);
if (isset($m[0])) {
preg_match_all("/\\d/", str_replace(" ", "", $m[0]), $prices);
foreach ($prices as $price) {
$model = new Items();
$model->product = trim((string) $item->title);
$model->price = implode('', $price);
$model->url = trim((string) $item->link);
$model->store = self::STORE;
$model->phone = parent::getPhoneNumber(trim((string) $item->link));
$model->subcategory_id = $id;
$model->options = '{"b/u":"1","city":"' . $city . '"}';
$model->save();
}
}
}
}
}
}
}
public static function processLink($link, $category_id, $optionType)
{
$page = self::getPage($link);
$document = new \DOMDocument();
try {
$document->loadHTML($page);
} catch (\Exception $e) {
return;
}
self::logFile('file.log', $page);
$products = self::findByDoc($document, '.item');
echo "founded {$products->length} elements in {$link} \n";
if (!$products->length) {
return;
}
foreach ($products as $product) {
/**
* @var $product \DOMNode
*/
$a = self::findByDoc($document, 'h4 a', $product)->item(0);
$href = $a ? $a->getAttribute('href') . "\n" : null;
try {
if ($href && self::productUnique('http://foxmart.ua' . $href)) {
$name = trim(explode('.', $a->textContent)[0]);
$name = self::decode(trim(explode('+', $name)[0]));
$price = self::findByDoc($document, '.price_prod', $product)->item(0)->textContent;
$price = preg_replace('/[^\\d]+/', '', $price);
$diagonal = self::findByDoc($document, '.product_view', $product)->item(0)->textContent;
preg_match('/����: ([\\d\\.]+)/', $diagonal, $matchesDia);
$product = new Items();
$product->url = 'http://foxmart.ua' . $href;
$product->store = self::STORE;
$product->price = $price;
$product->product = $name;
$product->phone = self::PHONE;
$product->subcategory_id = $category_id;
// $product->options = '{' .
// '"type": "' . $optionType . '",' .
// (isset($matchesDia[1]) ? ('"display": "'. $matchesDia[1] .'"') : '' ) .
// '}';
$product->options = json_encode(array_merge(['type' => $optionType], isset($matchesDia[1]) ? ['display' => $matchesDia[1]] : []));
$product->save();
}
} catch (\Exception $e) {
echo $e->getTraceAsString();
}
}
if (preg_match('/\\/([\\d]+).html/', $link, $matchCurrent)) {
$current = $matchCurrent[1];
self::processLink(preg_replace('/\\/[\\d]+.html/', '/' . ($current + 1) . '.html', $link), $category_id, $optionType);
} else {
self::processLink(preg_replace('/.html/', '/2.html', $link), $category_id, $optionType);
}
}
public static function processLink($link, $category_id, $optionType)
{
$page = self::getPage($link);
$document = new \DOMDocument();
try {
$document->loadHTML($page);
} catch (\Exception $e) {
return;
}
$products = self::findByDoc($document, '.item');
echo "founded {$products->length} elements in {$link} \n";
if (!$products->length) {
return;
}
foreach ($products as $product) {
/**
* @var $product \DOMNode
*/
$a = self::findByDoc($document, 'a.product-name', $product)->item(0);
$href = $a ? $a->getAttribute('href') . "\n" : null;
try {
if ($href && self::productUnique($href)) {
$name = trim(explode('.', $a->textContent)[0]);
$name = self::decode(trim(explode('+', $name)[0]));
$price = self::findByDoc($document, '.price .sum', $product)->item(0)->textContent;
$price = preg_replace('/[^\\d]+/', '', $price);
$diagonal = self::findByDoc($document, '.attr-container', $product)->item(0)->textContent;
preg_match('/([\\d\\.]+)"/', $diagonal, $matchesDia);
$product = new Items();
$product->url = $href;
$product->store = self::STORE;
$product->price = $price;
$product->product = $name;
$product->phone = self::PHONE;
$product->subcategory_id = $category_id;
// $product->options = '{' .
// '"type": "' . $optionType . '",' .
// (isset($matchesDia[1]) ? ('"display": "'. $matchesDia[1] .'"') : '' ) .
// '}';
$product->options = json_encode(array_merge(['type' => $optionType], isset($matchesDia[1]) ? ['display' => $matchesDia[1]] : []));
$product->save();
}
} catch (\Exception $e) {
}
}
preg_match('/\\/p-([\\d]+)/', $link, $matchCurrent);
$current = isset($matchCurrent[1]) ? $matchCurrent[1] : 1;
self::processLink(preg_replace('/\\/p-[\\d]+/', '/p-' . ($current + 1), $link), $category_id, $optionType);
}
/**
* Creates a new Items model.
* If creation is successful, the browser will be redirected to the 'view' page.
* @return mixed
*/
public function actionCreate()
{
$model = new Items();
$model_tr = new \common\models\ItemsTr();
$trips_array = ArrayHelper::map(\common\models\Trips::find()->orderBy('from')->all(), 'id', function ($model, $defaultValue) {
return $model->from . '-' . $model->to;
});
if ($model->load(Yii::$app->request->post()) && $model->save()) {
if ($_POST['ItemsTr']) {
$tr = Yii::$app->request->post('ItemsTr');
$params = [];
foreach ($tr as $lang) {
$lang['item_id'] = $model->id;
array_push($params, $lang);
}
Yii::$app->db->createCommand()->batchInsert('items_tr', ['name', 'type', 'lang', 'item_id'], $params)->execute();
}
return $this->redirect(['view', 'id' => $model->id]);
} else {
return $this->render('create', ['model' => $model, 'model_tr' => $model_tr, 'trips_array' => $trips_array]);
}
}
/**
* Парсинг каталога работы Olx
*/
public static function saveParseOlxWork()
{
foreach (self::$category_work as $id => $work) {
foreach (parent::$cities as $city) {
$xml = file_get_contents('http://olx.ua/rabota/' . $work . '/' . $city . '/rss/');
$movies = new \SimpleXMLElement($xml, LIBXML_NOCDATA);
unset($xml);
foreach ($movies->channel->item as $item) {
if (parent::productUnique(trim((string) $item->link))) {
$model = new Items();
$model->product = trim((string) $item->title);
$model->price = self::getPrice(trim((string) $item->link))[0];
$model->url = trim((string) $item->link);
$model->store = self::STORE;
$model->phone = parent::getPhoneNumber(trim((string) $item->link));
$model->subcategory_id = $id;
$model->options = '{"city":"' . $city . '"}';
$model->save();
}
}
}
}
}
/**
* @param $baseUrl
* @param $subcategory
*/
private function saveParseRst($baseUrl, $subcategory)
{
set_time_limit(0);
error_reporting(E_ALL & ~E_NOTICE);
$j = 1;
while ($j <= 1000) {
$url = $baseUrl . '&start=' . $j;
$parser = new HtmlDomParser();
$html = iconv('windows-1251', 'UTF-8//IGNORE', $parser->file_get_html($url));
$dom = $parser->str_get_html($html);
$year = null;
$fuel = null;
$price = null;
$link = null;
$product = null;
$city = null;
$phone = null;
for ($i = 0; $i < count($dom->find('div[class=rst-ocb-i]')) - 1; $i++) {
$dparser = new HtmlDomParser();
$ddom = $dparser->str_get_html($dom->find('div[class=rst-ocb-i]')[$i]->innertext);
$link = 'http://rst.ua' . $ddom->find('a.rst-ocb-i-a')[0]->href;
preg_match("/(.*)/", $ddom->find('li[class=rst-ocb-i-d-l-i]')[1]->plaintext, $year);
preg_match("/(.*)/", $ddom->find('li[class=rst-ocb-i-d-l-i]')[2]->plaintext, $fuel);
preg_match("/(.*)/", $ddom->find('li[class=rst-ocb-i-d-l-j]')[0]->plaintext, $city);
$product = $ddom->find('h3[class=rst-ocb-i-h]')[0]->plaintext;
$phone_parser = $dparser->file_get_html($link);
$phone = utf8_encode($phone_parser->find('p[class=rst-page-oldcars-item-option-block-container]')[0]->plaintext);
if (isset($phone) && !empty($phone)) {
preg_match("/\\d+/", $phone, $phone);
$phone = $phone[0];
} else {
$phone = utf8_encode($phone_parser->find('div.rst-page-oldcars-item-option-block-container td')[0]->plaintext);
}
unset($phone_parser);
$price = str_replace("'", "", $ddom->find('span[class=rst-ocb-i-d-l-i-s rst-ocb-i-d-l-i-s-p]')[0]->plaintext);
unset($dparser);
if ($this->productUnique($link)) {
preg_match("/\\((\\d+).*?\\)/", $year[0], $running);
preg_match("/\\((.*?)\\)/", $fuel[0], $transmission);
preg_match("/\\d+/", $year[0], $year);
preg_match("/\\d+/", $phone[0], $phone);
preg_match("/\\-?\\d+(\\.\\d{0,})?(.*?)\\(/", $fuel[0], $fuel);
preg_match("/(\\d+)/", $price, $price);
$city = explode(":", $city[0]);
$model = new Items();
$model->product = $product;
if (!empty($price)) {
$model->price = $price[0];
} else {
$model->options = "договорная";
$model->price = "0";
}
$model->url = $link;
$model->store = 'Rst';
$model->phone = $phone;
$model->subcategory_id = $subcategory;
$model->options .= '{"year":"' . trim($year[0]) . '","fuel":"' . trim($fuel[2]) . '","transmission":"' . trim($transmission[1]) . '","running":"' . trim($running[1]) . '","city":"' . trim($city[1]) . '","b/u":"1"}';
$model->save();
}
}
$j++;
}
}
public function saveMicrotronShit($item)
{
if (self::productUnique($item['url'])) {
$url = $item['url'];
$price = $item['price'];
$title = $item['product'];
if (isset($price) && isset($title)) {
preg_match_all("/\\d/", str_replace(" ", "", $price), $price);
$model = new Items();
$model->product = trim((string) $title);
$model->price = implode('', $price[0]);
$model->url = $url;
$model->store = 'Microtron';
$model->phone = self::PHONE;
$model->subcategory_id = 8;
$model->options = '-';
$model->save();
}
}
}
public function saveParseOlxSumki()
{
foreach (parent::$cities as $city) {
$xml = file_get_contents('http://olx.ua/elektronika/foto-video/aksessuary-dlya-foto-videokamer/sumki/' . $city . '/rss/');
$movies = new \SimpleXMLElement($xml, LIBXML_NOCDATA);
unset($xml);
foreach ($movies->channel->item as $item) {
if (self::productUnique(trim((string) $item->link))) {
preg_match("/Цена:\\s\\d(.*?)\\sгрн/", $item->description, $m);
if (isset($m[0])) {
preg_match_all("/\\d/", str_replace(" ", "", $m[0]), $prices);
foreach ($prices as $price) {
$model = new Items();
$model->product = trim((string) $item->title);
$model->price = implode('', $price);
$model->url = trim((string) $item->link);
$model->store = 'Olx';
$model->phone = parent::getPhoneNumber(trim((string) $item->link));
$model->subcategory_id = 69;
$model->options = '{"type":"sumki-photo","b/u":"1","city":"' . $city . '"}';
$model->save();
}
}
}
}
}
}
/**
* Парсинг Стиральных машин
*/
private function saveParseComfyWasher()
{
set_time_limit(0);
$file = 'http://comfy.ua/media/sitemap.xml';
$links = [];
// Формируем список урлов и сносим xml
$xml = simplexml_load_file($file, "SimpleXMLElement", LIBXML_NOCDATA);
$p_cnt = count($xml->url);
for ($i = 0; $i < $p_cnt; $i++) {
$item = $xml->url[$i];
if (preg_match("/(http\\:\\/\\/comfy\\.ua\\/stiral\\-naja\\-mashina)(.*?)(\\.html)/", $item->loc)) {
array_push($links, trim((string) $item->loc));
}
unset($item);
}
unset($xml);
unset($p_cnt);
// Проходим по списку урлов
foreach ($links as $link) {
if (self::productUnique($link)) {
$page = new AdvancedHtmlDom();
$page->load_file($link);
$url = $link;
$type_washer = null;
$max_load = null;
$not = $page->find('span[class=informerText]');
if ($not->text()) {
$price = $page->find('span[class=price-value]')[0]->text();
$title = $page->find('h1[class=product-name]')[0]->text();
for ($i = 0; $i < $page->find('li.features__item dl')->count(); $i++) {
if (stripos($page->find('li.features__item dl')[$i]->text(), 'Тип стиральной машины')) {
$type_washer = $page->find('li.features__item dl')[$i]->text();
}
if (stripos($page->find('li.features__item dl')[$i]->text(), 'Макс. загрузка')) {
$max_load = $page->find('li.features__item dl')[$i]->text();
}
if (!empty($type_washer) && !empty($max_load)) {
break;
}
}
unset($page);
if (isset($price) && isset($title) && $type_washer != null && $max_load != null) {
$type_washer = split(":", $type_washer);
$max_load = split(":", $max_load);
preg_match_all("/\\d/", str_replace(" ", "", $price), $price);
$model = new Items();
$model->product = trim((string) $title);
$model->price = str_replace(",", ".", implode('', $price[0]));
$model->url = $url;
$model->store = 'Comfy';
$model->phone = self::PHONE;
$model->subcategory_id = 4;
$model->options = '{"type":"washer","type_washer":"' . trim($type_washer[1]) . '","max_load":"' . trim($max_load[1]) . '","b/u":"0"}';
$model->save();
}
}
}
}
}
