151 lines
5.3 KiB
PHP
151 lines
5.3 KiB
PHP
<?php
|
|
include 'vendor/autoload.php';
|
|
|
|
use DiDom\Document;
|
|
|
|
set_time_limit(-1);
|
|
$settings = include 'settings.php';
|
|
|
|
$loginData = [
|
|
'email' => $settings['email'],
|
|
'password' => $settings['password'],
|
|
'op' => 'Login',
|
|
'form_build_id' => 'form-fba4b62ee04aafbf045b1d9ae019d90b',
|
|
'form_id' => 'packt_user_login_form'
|
|
];
|
|
|
|
|
|
function c($url, $post = [])
|
|
{
|
|
$cookie = "cookie.txt";
|
|
$ch = curl_init();
|
|
curl_setopt($ch, CURLOPT_URL, $url);
|
|
curl_setopt($ch, CURLOPT_POST, 1);
|
|
|
|
curl_setopt($ch, CURLOPT_POSTFIELDS,
|
|
http_build_query($post));
|
|
curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
|
|
curl_setopt($ch, CURLOPT_COOKIEJAR, $cookie);
|
|
curl_setopt($ch, CURLOPT_COOKIEFILE, $cookie);
|
|
|
|
// receive server response ...
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
|
|
|
|
$server_output = curl_exec($ch);
|
|
strlen($server_output);
|
|
|
|
// $info = curl_getinfo($ch);
|
|
|
|
|
|
curl_close($ch);
|
|
|
|
return $server_output;
|
|
|
|
|
|
}
|
|
|
|
|
|
function getBookInfo($bookUrl)
|
|
{
|
|
|
|
print_r($bookUrl . "\n");
|
|
$bookData = [];
|
|
$bookPage = new Document($bookUrl, true);
|
|
$bookData['datepublished'] = $bookPage->first('.book-top-block-info-authors time[itemprop="datePublished"]::attr(datetime)');
|
|
$bookData['numberofpages'] = $bookPage->first('span[itemprop="numberOfPages"]::text');
|
|
$bookData['isbn'] = $bookPage->first('span[itemprop="isbn"]::text');
|
|
$bookData['reviewCount'] = $bookPage->first('meta[itemprop="reviewCount"]::attr(content)');
|
|
$bookData['ratingValue'] = $bookPage->first('meta[itemprop="ratingValue"]::attr(content)');
|
|
|
|
$bookData['toc'] = [];
|
|
foreach ($bookPage->find('#book-info-toc.onlyDesktop .book-toc-chapter') as $chapter) {
|
|
if (!is_null($chapter->first('div[class*="book-toc-chapter-title"]'))) {
|
|
$tocSection = [];
|
|
$tocSection['title'] = trim($chapter->first('div[class*="book-toc-chapter-title"]')->text());
|
|
$tocSection['subchapters'] = $chapter->find('div[class*="book-toc-section-text"]::text');
|
|
$bookData['toc'][] = $tocSection;
|
|
}
|
|
}
|
|
|
|
$bookData['description'] = $bookPage->find('div.book-info-bottom-indetail-text[itemprop="description"] p::text');
|
|
$bookData['willLearn'] = $bookPage->find('div.book-info-will-learn-text li::text');
|
|
|
|
$bookData['category'] = $bookPage->first('div[data-product-id="' . $bookData['isbn'] . '"]::attr(data-product-category)');
|
|
|
|
$bookData['authors'] = [];
|
|
foreach ($bookPage->find('[itemprop="author"]') as $author) {
|
|
$authorData = [];
|
|
$authorData['name'] = $author->first('h3::text');
|
|
$authorData['bio'] = $author->find('p::text');
|
|
$bookData['authors'][] = $authorData;
|
|
}
|
|
|
|
return $bookData;
|
|
}
|
|
|
|
|
|
//$return = c('https://www.packtpub.com/', $loginData);
|
|
//$return = c('https://www.packtpub.com/account/my-ebooks');
|
|
//var_dump($return);
|
|
//var_dump($return);
|
|
//$document = new Document($return);
|
|
|
|
$document = new Document('packt.html', true);
|
|
|
|
$booksData = [];
|
|
$books = $document->find('.product-line.unseen');
|
|
//$books = array_slice($books, 2, 1);
|
|
|
|
$bookData = json_decode('books.txt', 1);
|
|
|
|
foreach ($books as $book) {
|
|
|
|
|
|
// print_r($book->html());
|
|
$bookData = [];
|
|
$bookData['title'] = str_replace(["\r\n"], '', trim($book->first('.title::text')));
|
|
$bookData['nid'] = (int)$book->attr('nid');
|
|
|
|
$bookData['pdf'] = $book->first('a[href$=pdf]::attr(href)');
|
|
$bookData['epub'] = $book->first('a[href$=epub]::attr(href)');
|
|
$bookData['mobi'] = $book->first('a[href$=mobi]::attr(href)');
|
|
$bookData['code'] = $book->first('a[href*=code_download]::attr(href)');
|
|
$bookData['isbn'] = $book->first('div[isbn]::attr(isbn)');
|
|
$bookData['img'] = $book->first('img[class*=imagecache]::attr(src)');
|
|
$bookData['url'] = $book->first('div[class*=product-thumbnail]')->first('a::attr(href)');
|
|
$bookData['info'] = getBookInfo($bookData['url']);
|
|
|
|
|
|
|
|
|
|
$db = new SQLite3('data.db');
|
|
$stmt = $db->prepare('REPLACE INTO book (nid, title, isbn, img, url, datepublished, numberofpages, reviewCount, ratingValue, category, pdf, epub, mobi, code)
|
|
VALUES (:nid, :title, :isbn, :img, :url, :datepublished, :numberofpages, :reviewCount, :ratingValue, :category, :pdf, :epub, :mobi, :code)');
|
|
$stmt->bindValue(':nid', $bookData['nid'], SQLITE3_INTEGER);
|
|
$stmt->bindValue(':title', $bookData['title'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':isbn', $bookData['isbn'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':img', $bookData['img'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':url', $bookData['url'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':datepublished', $bookData['info']['datepublished'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':numberofpages', $bookData['info']['numberofpages'], SQLITE3_INTEGER);
|
|
$stmt->bindValue(':reviewCount', $bookData['info']['reviewCount'], SQLITE3_INTEGER);
|
|
$stmt->bindValue(':ratingValue', (float)$bookData['info']['ratingValue'], SQLITE3_FLOAT);
|
|
$stmt->bindValue(':category', $bookData['info']['category'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':pdf', $bookData['pdf'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':epub', $bookData['epub'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':mobi', $bookData['mobi'], SQLITE3_TEXT);
|
|
$stmt->bindValue(':code', $bookData['code'], SQLITE3_TEXT);
|
|
|
|
$result = $stmt->execute();
|
|
|
|
|
|
$booksData[$bookData['nid']] = $bookData;
|
|
|
|
}
|
|
print_r($booksData);
|
|
|
|
file_put_contents('books.txt', json_encode($booksData));
|
|
|
|
|
|
//print_r($return);
|