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

//$url = "https://www.famousbirthdays.com/people/selena-gomez.html";
//if(isset($_GET['url'])){
//    $url = $_GET['url'];
//}

$url = 'https://www.famousbirthdays.com/random';


$prefix = 'bio_';
$meta = array();
$mapping = array();
$client = new Client();
$crawler = $client->request('POST', $url);

try{
    $name = $crawler->filterXpath('//meta[@name="twitter:title"]')->attr('content');
}catch (Exception $e){
    $name = '';
}
$name = str_replace('Learn about ','',$name);

// META
$names = splitName($name);
$meta[$prefix.'first_name'] = $names['firstname'];
if($names['lastname'] != ''){
    $meta[$prefix.'last_name'] = $names['lastname'];
}

$count = 0;
$dob = '';
$birthplace = '';
$city = '';

$crawler->filter('a.btn-rank')->each(function ($node) use(&$city){
    $href = $node->attr('href');
    if (strpos($href, '/birthplace') !== false) {
        $city = array_shift($node->filter('.title')->extract('_text'));
        $city = trim(str_replace('Born in ','',$city));
    }
});
$elements = $crawler->filter('.stats .main-stats .is-flex .col-md-12')->each(function($node) use(&$count, &$dob,&$birthplace){
    if($count == 0){
        $dob = $node->text();
        $dob = str_replace('Birthday','',$dob);
    }else if($count == 1){
        $birthplace = $node->text();
        $birthplace = trim(str_replace('Birthplace','',$birthplace));
        $birthplace = trim(preg_replace('/\s\s+/', ' ', $birthplace));
    }else{
        return;
    }
    $count++;
});

$dob = cleanDate($dob);
$meta[$prefix.'dob'] = $dob;
$meta[$prefix.'fact_birth-place'] = $birthplace;
if($city != ''){
    $meta[$prefix.'city'] = $city;
}

//$name = array_shift($crawler->filter('.main-info .col-md-6 h1')->extract('_text'));
$category = array_shift($crawler->filter('.main-info .col-md-6 h1 .person-title')->extract('_text'));
$fields['category'] = trim($category);
$fields['name'] = $name;
try{
    $image = $crawler->filterXpath('//meta[@name="og:image"]')->attr('content');
}catch (Exception $e){
    $image = '';
}

// Get second image from list of image
$count = 0;
$elements = $crawler->filter('.famous-slider img')->each(function($node) use(&$count, &$image){
    if($count <= 1){
        $image = $node->attr('src');
    }else{
        return;
    }
    $count++;
});


$fields['image'] = $image;

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

$strip_tags = "figcaption";
/*$html = preg_replace("#<\s*\/?(".$strip_tags.")\s*[^>]*?>#im", '', $html);*/
$html = preg_replace('/(<(figcaption)\b[^>]*>).*?(<\/\2>)/is', "$1$3", $html);
$html = preg_replace('/<figcaption[^>]*>([\s\S]*?)<\/figcaption[^>]*>/', '', $html);
$html = preg_replace('/<div[^>]*>([\s\S]*?)<\/div[^>]*>/', '', $html);
//$html = trim(preg_replace('/\s\s+/', ' ', $html));
//$fields['body'] = $html;
$fields['meta'] = $meta;

$textToSpin = str_replace('</div>','',$html);
$response = spinbot($textToSpin);
//debug($response);
list($strResponseHeaders, $strResponseBody) = explode("\r\n\r\n", $response, 2);
//list($strResponseHeaders, $strResponseBody) = explode("\r\n\r\n", $strResponseBody, 2);
//$aHeaders = putHeadersTextIntoArray($strResponseHeaders);
$fields['body'] = $strResponseBody;


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

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


function spinbot($textToSpin){
    $url = 'https://api.spinbot.com';
    $header = array();
// Required header:
    $spinbotApiKey = '87b0d1bb9b3c420381ee141837cbc7f5';
    $header[] = "x-auth-key:$spinbotApiKey";

// optional header values
    $header[] = 'x-spin-cap-words:true';
    $header[] = 'x-words-to-skip:rewrit,nonExistentWordPart';
    $header[] = 'x-min-percent-change-per-sentence:any';
//$header[] = 'x-action:getavailablespins';

// Execute cURL request, get response
    $ch = curl_init();
    curl_setopt($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);
    $response = curl_exec($ch);
    curl_close($ch);

    return $response;
}

function cleanDate($date){
    $date = trim($date);
    for($i = 1; $i <= 12; $i++){
        $month = date('F',strtotime('2020-'.$i.'-01'));
        $date = str_replace($month,'',$date);
    }

    try{
        $dt = new DateTime($date);
    }catch (Exception $e){
        debug($e);
    }
    return $dt->format('Y-m-d');
}

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