Exemplo n.º 1
0
<?php

require_once "config/default_config.php";
includeMyFiles('urlgenerator');
$goutte = new Goutte\Client();
$baseURL = 'https://www.lashowroom.com/';
$womenManuEndpoint = 'women/manufacturers';
//*[@id="lnv_level_content_cat"]/ul
$status_code = 200;
$count = 1;
getLinkEndPoints($goutte, $baseURL . $womenManuEndpoint);
while ($status_code == 200) {
    $status_code = getLink($goutte, $baseURL . $womenManuEndpoint . $count);
    $count++;
    $sleep_time = rand(3 * 1000000, 4 * 1000000);
    echo "\tSleeping for " . number_format($sleep_time / 1000000, 2) . " sec\n";
    usleep($sleep_time);
}
function getLinkEndPoints($goutte, $url)
{
    $crawler = $goutte->request('GET', $url);
    $status_code = $goutte->getResponse()->getStatus();
    if ($status_code == 200) {
        // Plus size
        $domSelector = '//*[@id="lnv_level_content_cat"]/ul';
        $crawler->filterXPath($domSelector)->each(function ($node) {
            $domS = '//li/a';
            $node->filterXPath($domS)->each(function ($node) {
                print_r($node->attr('href'));
                echo "\n";
            });
Exemplo n.º 2
0
<?php

// Include config and initiate
include_once __DIR__ . '/../config/default_config.php';
includeMyFiles();
// Initialize goutte
$goutte = new Goutte\Client();
// Get Database
$db = new Db();
$loginUrl = 'http://www.colorfive.com/sign-in';
$crawler = $goutte->request('GET', $loginUrl);
print_r($crawler->html());
exit;
// Default Status code
$status_code = 200;
$count = 1;
$data = $db->query("SELECT url from product_urls WHERE source='numberonewholesales'");
$url = '';
foreach ($data as $value) {
    $url = $value['url'];
    $data = getProductData($goutte, $url, $db);
}
function getProductData($goutte, $url, $db)
{
    $crawler = $goutte->request('GET', $url);
    $status_code = $goutte->getResponse()->getStatus();
    $data = array();
    if ($status_code == 200) {
        $attribs = array();
        $domSelector = '//*[@class="productInfoName"]';
        $attribs['title'] = $data = $crawler->filterXPath($domSelector)->each(function ($node) {
Exemplo n.º 3
0
<?php

require_once "config/default_config.php";
includeMyFiles('master');
print_r(PROCESS_CONF);
exit;
class masterProcess
{
    private $result = NULL;
    private $connx = NULL;
    private $numRows = 0;
    private $childs = array();
    public function __construct(array $arguments = array())
    {
        if (!empty($arguments)) {
            foreach ($arguments as $property => $argument) {
                $this->{$property} = $argument;
            }
        }
    }
    // Starting up the process
    public function init()
    {
        // $this->forkProcess('test',array('test.php'));
        // $this->forkProcess('test',array('test2.php'));
        $schedules = $this->getSchedules();
        foreach ($schedules as $schedule) {
            if ($schedule['process_name'] == 'bestbuy') {
                $this->fetchBestbuyData($schedule);
            }
            if ($schedule['process_name'] == 'input_file_generator') {