asko-get-nokogiri.php

<?php

// https://raw.github.com/olamedia/nokogiri/master/nokogiri.php
require 'nokogiri.php';

$template = array
(
    "title" => "ASKO nábytek",
    "descr" => null,
    "datetime_end" => null,
    "datetime_begin" => null,
    "file_orig" => null,
    "source_url" => "http://www.asko-nabytek.cz/katalog-letak-asko"
);

function find_dates($str)
{
    preg_match_all("/[\d\s\.]+/u", $str, $dates);
    $out = array();
    foreach($dates[0] as $date)
    {
        $date = strtotime(preg_replace("/([^\d^\.]+)/", "", $date));
        if($date)
        {
            $out[] = $date;
        }
    }
    return (count($out) == 2 ? $out : FALSE);
};

function find_code($url)
{
    $ret = preg_match("/katalog.asko.cz\/([^\/]+)/", $url, $match);
    return $ret ? $match[1] : FALSE;
};

function check_content_type($url, $content_type)
{
    $headers = get_headers($url, 1);
    return $headers['Content-Type'] == $content_type;
};

// zaciname zpracovani

$data = array('urls' => array(), 'validity' => array(), 'valid_for' => array());
$html = new nokogiri(file_get_contents($template['source_url']));
$list = $html->get('.content-main .list-page');

foreach($list->get('p.ta-c') as $line)
{
    $data['urls'][] = $line['a'][0]['href'];
}

foreach($list->get('ul.list-arrow-small') as $line)
{
    $data['validity'][] = $line['li'][0]['#text'][0];
    $data['valid_for'][] = (count($line['li'][1]) == 1) ? "všechny pobočky" : "E-shop";
}

// zkontroluje pocet prvku

$items_count = count($data['urls']);
foreach($data as $key=>$vals)
{
    if(count($data[$key]) != $items_count)
    {
        throw new Exception('Nesouhlasi pocet dat');
    }
}


$out = array();
for($i=0;$i<count($data['urls']);$i++)
{
    $dates = find_dates($data['validity'][$i]);
    $code = find_code($data['urls'][$i]);
    $file_orig = $data['urls'][$i]."/includes/".str_replace("-","_", $code).".pdf";

    if(!$dates OR !$code)
    {
        throw new Exception('Spatne data');
    }

    // overime existenci PDFka
    if(!check_content_type($file_orig, "application/pdf"))
    {
        throw new Exception('PDF neexistuje: '.$file_orig);
    }

    $item = $template;
    $item['datetime_begin'] = date("Y-m-d H:i:s",$dates[0]);
    $item['datetime_end'] = date("Y-m-d H:i:s",strtotime("-1 second", $dates[1]));
    $item['file_orig'] = $file_orig;
    $item['descr'] = "Platnost letáku od ".date("j.m.Y", $dates[0])." do ".date("j.m.Y", $dates[1])." pro ".$data['valid_for'][$i];
    $out[] = $item;
}

var_dump($out);