public function web_scraping($array)
{
$ri = new resizeImage();
$pa = new postarticle();
// $get = $array['get'];
// $url = $array['url'];
// $this->print_r1( $array );
// init arrays
$get = array();
$response = array();
$img = array();
$video = array();
// init size
$size['image']['width'] = 'auto';
$size['image']['height'] = 'auto';
$size['video']['height'] = 'auto';
$size['video']['width'] = 'auto';
// init attr
$title = '';
$description = '';
$keyword = '';
// clean session
unset($_SESSION['article_image']);
unset($_SESSION['article_title']);
unset($_SESSION['article_description']);
unset($_SESSION['article_keyword']);
unset($_SESSION['article_video']);
// initialized get
if (!empty($array['get'])) {
foreach ($array['get'] as $value) {
// $data[$value] = $value;
// echo " $value <br> ";
$get[$value] = $value;
}
}
// print_r($response);
// initialize size
if (!empty($array['size'])) {
foreach ($array['size'] as $key => $value) {
// $data[$value] = $value;
$height = !empty($value['height']) ? $value['height'] : 'auto';
$width = !empty($value['width']) ? $value['width'] : 'auto';
// echo " $key width $width , height $height <br> ";
$size[$key]['height'] = $height;
$size[$key]['width'] = $width;
}
}
// initialize url
$url = !empty($array['url']) ? $array['url'] : '';
echo " url {$url} <br>";
// initialized limit
$limit = !empty($array['limit']) ? $array['limit'] : 10;
echo " limit {$limit} <br>";
// initialized general data
if (!empty($get['image'])) {
// echo $get['image'].' height = '.$size['image']['height'].' width = '.$size['image']['width'].'<br>';
}
if (!empty($get['video'])) {
// echo $get['video'].' height = '.$size['video']['height'].' width = '.$size['video']['width'].'<br>';
unset($_SESSION['article_video']);
}
// initialize curl
$counter = 0;
$images = array();
// echo " scraping <br>";
$ch = curl_init("{$url}");
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
$cl = curl_exec($ch);
$dom = new DOMDocument();
@$dom->loadHTML($cl);
$images = $dom->getElementsByTagName('img');
$videos = $dom->getElementsByTagName('iframe');
$paharagraph = $dom->getElementsByTagName('p');
$embededs = $dom->getElementsByTagName('embed');
// modal image
if (!empty($get['image'])) {
// init
$height = $size['image']['height'];
$width = $size['image']['width'];
$c = 0;
$ph = '';
$t[0]['title'] = '';
$t[1]['title'] = '';
$t[2]['title'] = '';
$t[3]['title'] = '';
// retrieve keyword
#code to retrieve keyword here
// retrieve title
$title = $pa->get_title_in_a_website($url);
if (empty($title)) {
$title1 = $dom->getElementsByTagName('h1');
$title2 = $dom->getElementsByTagName('h2');
$title3 = $dom->getElementsByTagName('h3');
$title4 = $dom->getElementsByTagName('h4');
$c = 0;
// echo "<h1> title1 </h1>";
foreach ($title1 as $text) {
if ($c == 0) {
$t[0]['title'] = $text->nodeValue . ' <br> ';
}
break;
$c++;
}
$c = 0;
// echo "<h1> title2 </h1>";
foreach ($title2 as $text) {
if ($c == 0) {
$t[1]['title'] = $text->nodeValue . ' <br> ';
}
break;
$c++;
}
// $c=0; print_r(expression)
// echo "<h1> title3 </h1>";
foreach ($title3 as $text) {
if ($c == 0) {
$t[2]['title'] = $text->nodeValue . ' <br> ';
}
break;
$c++;
}
$c = 0;
// echo "<h1> title4 </h1>";
foreach ($title4 as $text) {
if ($c == 0) {
$t[3]['title'] = $text->nodeValue . ' <br> ';
}
break;
$c++;
}
$x[0] = strlen($t[0]['title']);
$x[1] = strlen($t[1]['title']);
$x[2] = strlen($t[2]['title']);
$x[3] = strlen($t[3]['title']);
$s = $x[0];
for ($i = 1; $i < count($x); $i++) {
if ($s < $x[$i]) {
$s = $x[$i];
$title = $t[$i]['title'];
}
}
}
// retrieve description
#code to retrieved description here in the main site meta
$c = 0;
// echo "<h1> paharagraph </h1>";
foreach ($paharagraph as $p) {
$ph[$c]['description'] = $p->nodeValue . ' <br> ';
$c++;
}
// print_r($ph);
// get greatest value desc
$s = !empty($ph[0]['description']) ? strlen($ph[0]['description']) : 0;
// $description = '';
for ($i = 1; $i < count($ph); $i++) {
$len = !empty($ph[$i]['description']) ? strlen($ph[$i]['description']) : 0;
if ($s < $len) {
$s = $len;
$description = $ph[$i]['description'];
}
}
// print description
// echo " description : $description <br> ";
$_SESSION['article_title'] = $title;
$_SESSION['article_description'] = $description;
$_SESSION['article_keyword'] = $keyword;
// assign image
$c = 0;
foreach ($images as $image) {
$imgSrc = $image->getAttribute('src');
// echo " $imgSrc <br> ";
// echo " <img src='$imgSrc' style='width:$width; height:$height' /> ";
#get image
foreach ($array['accept_image'] as $url) {
# foreach for the allowed extention
if (strpos($imgSrc, $url)) {
# check if the extention allowed is exist
if ($c < $limit) {
# set limit how many modals show
$imgSrc = $pa->add_img_url_if_dont_have($imgSrc, $url);
# add url link is only direct to the folder
if ($pa->url_exists($imgSrc)) {
# check if the image exist
// $ri->load( $imgSrc );
// $img_width = $ri->getWidth();
// if ( $img_width > $width ) {
// $jpg = intval( strpos($imgSrc, '.jpg') );
// if ( $jpg > 0 ) { #only allows if jpg
$img[$c] = $imgSrc;
$c++;
// echo " pos $jpg $imgSrc <br>";
// }
// else {
// echo " not jpg pos $jpg";
// }
// }
}
}
}
}
}
$_SESSION['article_image'] = $img;
echo " after foreach ";
// print_r($_SESSION['article_image']);
}
// modal video
if (!empty($get['video'])) {
$height = $size['video']['height'];
$width = $size['video']['width'];
/*
vimeo
url
http://vube.com/trending/Aj+Silva+Music/0dPsIWYjoj?t=s
http://vube.com/trending/Andrea+Kaden/W8nM24DqUw?t=s
http://vube.com/trending/Andrea+Kaden/oFzAdU9Dqe
youtube
embeded:
//www.youtube.com/embed/1oju14nG5vo
url:
https://www.youtube.com/watch?v=1oju14nG5vo
https://www.youtube.com/watch?v=zC617kE1maU
https://www.youtube.com/watch?v=sR8rlTIU8_Y
video_id = 1oju14nG5vo
vimeo
https://vimeo.com/channels/staffpicks/102550866
https://vimeo.com/channels/staffpicks/102399221
https://vimeo.com/22506211
*/
$c = 0;
// iframe
foreach ($embededs as $video) {
$vidSrc = $video->getAttribute('src');
foreach ($array['accept_video'] as $url) {
if (strpos($vidSrc, $url)) {
echo " video src = " . $vidSrc . '<br>';
$vid[$c] = $vidSrc;
// $vid[$c]['id'] = '123123';
$c++;
}
}
}
// embeded
foreach ($videos as $video) {
$vidSrc = $video->getAttribute('src');
foreach ($array['accept_video'] as $url) {
if (strpos($vidSrc, $url)) {
echo " video src = " . $vidSrc . '<br>';
$vid[$c] = $vidSrc;
// $vid[$c]['id'] = '123123';
$c++;
}
}
}
$_SESSION['article_video'] = $vid;
}
}
require "../../../fs_folders/php_functions/connect.php";
require "../../../fs_folders/php_functions/function.php";
require "../../../fs_folders/php_functions/myclass.php";
require "../../../fs_folders/php_functions/library.php";
require "../../../fs_folders/php_functions/source.php";
$mc = new myclass();
$pa = new postarticle();
$ri = new resizeImage();
$url = $_GET['url'];
$url = str_replace(' ', '.', $url);
if (empty($url)) {
$url = 'http://freelancersfashion.blogspot.com/';
}
$bool = true;
if ($bool) {
if ($pa->url_exists($url)) {
if ($pa->with_http($url)) {
// echo "http"
/*
$content = file_get_contents($url);
$imagesArray = $pa->retrieve_images_from_url( $content , $mc , $pa , $ri );
*/
echo " url1 = {$url} <br>";
// $article_len = count($article);
echo "<hr>";
if (strpos($url, 'youtube.com')) {
// $pa->test_get_images_from_url( $url );
$video_array = $pa->video($url);
// echo "youtube";
echo '
<h1> "' . $video_array[0]['vtitle'] . '"</h1>