<?php
require 'vendor/autoload.php';
use Goutte\Client;

function strip_tags_blacklist($html, $tags) {
    $html = preg_replace('/<'. $tags .'\b[^>]*>(.*?)<\/'. $tags .'>/is', "", $html);
    return $html;
}

$url = 'https://healthyceleb.com/myra-molloy/';
if(isset($_GET['url'])){
    $url = $_GET['url'];
}
$prefix = 'hton_';
$meta = array();
$client = new Client();
$crawler = $client->request('GET', $url);


$full_name = array_shift($crawler->filter('.tablepress .odd .column-1')->extract('_text'));

if (strpos($full_name, 'Quick Info') == false) {
    echo json_encode(array('error'=>'Data not found'));exit;
}

$full_name = trim(str_replace('Quick Info', '', $full_name));
$meta['hton_background_full_name'] = $full_name;

$name = array_shift($crawler->filter('h1.entry-title')->extract('_text'));

// META FOR YOAST
$meta_desc = 'Who is '.$name.' ? View the latest Biography of '.$name.' and also find Personal Life, estimated Net Worth, Salary, Age, Career &amp; More.';
$meta['_yoast_wpseo_metadesc'] = $meta_desc;

try{
    $image = $crawler->filterXpath('//meta[@property="og:image"]')->attr('content');
}catch (Exception $e){
    $image = '';
}

// PULL META
$count = 0;
$crawler->filter('.tablepress tr td')->each(function($node) use (&$meta, &$count) {
    if($count == 1){ // Weight
        $height = convertHeight($node->text());
        if($height){
            $meta['hton_body_height'] = $height;
        }
    }else if($count == 3){ // Height
        $weight = convertWeight($node->text());
        if($weight){
            $meta['hton_body_weight'] = $weight;
        }
    }else if($count == 5){

        $date = convertDOB($node->text());
        if($date){
            $meta['hton_background_dob'] = $date;
        }
    }
    $count++;
    // $node->filter('td')->each(function($nodeChild) use(&$meta){
        // if($nodeChild->text() == ''){
            // echo 'its height';exit;
        // }
    // });
});
// $meta[$prefix.'background_full_name'] = get_meta($crawler, 'name');
// $meta[$prefix.'background_dob'] = get_meta($crawler, 'birthDate');
// $meta[$prefix.'body_gender'] = get_meta($crawler, 'gender');
// $meta[$prefix.'background_profession'] = get_meta($crawler, 'jobTitle');
// $meta[$prefix.'background_nicknames'] = get_meta($crawler, 'givenName');


$html = $crawler->filter('.td-post-content')->html();
// REPLACE h3 with h2
// $html = str_replace('<h3>','<h2>',$html);
// $html = str_replace('</h3>','</h2>',$html);

// First paragraph into H3
$start = strpos($html, '<p>');
$end = strpos($html, '</p>', $start);
$paragraph = substr($html, $start, $end-$start+4);

$paragraph_h3 = str_replace('<p>','<h3>',$paragraph);
$paragraph_h3 = str_replace('</p>','</h3>',$paragraph_h3);
$html = str_replace($paragraph,$paragraph_h3,$html);


$remove_tags_and_content = array('table','figcaption','div','iframe','font','noscript');
foreach($remove_tags_and_content as $tg){
    $html = strip_tags_and_content($html, $tg);
}

$html = preg_replace('/(<(script|style)\b[^>]*>).*?(<\/\2>)/is', "$1$3", $html);
$strip_tags = "center|style|span|ins|div|script|a|img|input|button|figure|strong|em";
$html = preg_replace('#<a.*?>.*?</a>#i', '', $html);
$html = preg_replace("#<\s*\/?(".$strip_tags.")\s*[^>]*?>#im", '', $html);

// Empty spaces
$html = preg_replace("#<p[^>]*>(\s|&nbsp;|</?\s?br\s?/?>)*</?p>#", '', $html);
// Remove empty paragraphs
$html = str_replace("<strong> </strong>","",$html);
$html = str_replace("<p></p>","",$html);
$html = str_replace("<p><br></p>","",$html);

// Remove New lines
$html = preg_replace('/[\s\t\n]{2,}/', ' ', $html);

// Remove HTML Comments
$html = preg_replace('/<!--(.|\s)*?-->/', '', $html);

$html = trim($html);


$fields['name'] = $name;
$fields['image'] = $image;
$fields['meta'] = $meta;
$fields['body'] = $html;

header('Content-Type: application/json');
echo json_encode($fields);
exit;

// Date covert
function convertDOB($datestr,$format = 'Y-m-d'){
    if(strtotime($datestr)){
        $date = date($format, strtotime($datestr));
        return $date;
    }else{
        return false;
    }
}
function convertWeight($weight){
    if (strpos($weight, 'kg') !== false) {
        return $weight;
    }else{
        return false;
    }
}

function convertHeight($height){
    if (strpos($height, 'ft') !== false) {
        return $height;
    }else{
        return false;
    }
}

function strip_tags_and_content($html,$tag){
    return preg_replace('/<'.$tag.'[^>]*>([\s\S]*?)<\/'.$tag.'[^>]*>/', '', $html);
}

function curlTest(){
    $url = 'https://healthyton.com';
    $ch = curl_init();
    $timeout = 10;
    curl_setopt($ch, CURLOPT_URL, $url);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, $timeout);
    $data = curl_exec($ch);
    curl_close($ch);
    return $data;
}
function spinbot($textToSpin){
    // echo $textToSpin;exit;
    // debug(curlTest());
    $url = 'https://api.spinbot.com';
    $header = array();
    $spinbotApiKey = '87b0d1bb9b3c420381ee141837cbc7f5';
    $header[] = "x-auth-key:$spinbotApiKey";
    $header[] = 'x-spin-cap-words:true';
    $header[] = 'x-words-to-skip:rewrit,nonExistentWordPart';
    $header[] = 'x-min-percent-change-per-sentence:any';
    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $url);
    curl_setopt($ch, CURLOPT_POST, true);
    curl_setopt($ch, CURLOPT_POSTFIELDS, $textToSpin);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($ch, CURLOPT_HEADER, true);
    curl_setopt($ch, CURLOPT_HTTPHEADER, $header);
    curl_setopt($ch, CURLOPT_VERBOSE, 1);
    curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 0);
    curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, 0);
    // echo 'curl here';
    $response = curl_exec($ch);
    // debug($response);
    curl_close($ch);
    return $response;
}


function stringClean($str)
{       
    $str = utf8_decode($str);
    $str = str_replace("&nbsp;", "", $str);
    $str = preg_replace("/\s+/", " ", $str);
    $str = trim($str);
    return $str;
}

function cleanDate($date){
    $date = trim($date);
    $date = str_replace(',','',$date);
    $date = str_replace('.','',$date);
    $date = str_replace('th','',$date);
    $date = str_replace('st','',$date);
    $date = str_replace('rd','',$date);
    $date = str_replace('<sup></sup>','',$date);

    try{
        $dt = new DateTime($date);
    }catch (Exception $e){
        debug($e);
    }

    return $dt->format('Y-m-d');
}


function putHeadersTextIntoArray($header_text) {
    $headers = array();
    foreach (explode("\r\n", $header_text) as $i => $line)
        if ($i === 0) {
            $headers['http_code'] = $line;
        } else {
            list ($key, $value) = explode(': ', $line);
            $headers[$key] = $value;
        }
    return $headers;
}

function debug($arr, $exit = true){
    echo '<pre>';
    print_r($arr);
    echo '</pre>';
    if($exit) exit;
}

function get_meta($crawler, $meta_name){
    $meta_info = array_shift($crawler->filterXpath('//td[@itemprop="'.$meta_name.'"]')->extract('_text'));
    return $meta_info;
}


function splitName($name) {
    $parts = explode(' ', $name);
    return array(
        'firstname' => array_shift($parts),
        'lastname' => array_pop($parts),
        'middlename' => join(' ', $parts)
    );
}