packtpub-browser/packt.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 = [], $localFilePath = null)
{
    $cookie = "cookie.txt";
    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $url);

    if (!empty($post) && count($post)) {
        curl_setopt($ch, CURLOPT_POST, 1);
        curl_setopt($ch, CURLOPT_POSTFIELDS, http_build_query($post));
    }

    /**
     * this means url is a path to a file and needs to be saved localy under $localFilePath location
     */
    if (!is_null($localFilePath)) {
        $fp = fopen($localFilePath, 'w+');
        curl_setopt($ch, CURLOPT_FILE, $fp);
    } else {
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    }

    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
    curl_setopt($ch, CURLOPT_COOKIEJAR, $cookie);
    curl_setopt($ch, CURLOPT_COOKIEFILE, $cookie);

    $server_output = curl_exec($ch);
    strlen($server_output);
    curl_close($ch);
    return $server_output;
}


function resolveBookUrl($bookUrl)
{
    return 'https://www.packtpub.com/' . trim(str_replace('https://www.packtpub.com/', '', $bookUrl), '/');
}

function truncateBookUrl($bookUrl)
{
    return str_replace('https://www.packtpub.com', '', $bookUrl);
}

function getBookInfo($bookUrl)
{

    $bookUrl = resolveBookUrl($bookUrl);
    $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');
$document = new Document($return);
//
//$document = new Document('packt.html', true);

$booksData = [];
$books = $document->find('.product-line.unseen');

$dl = 0;
$db = new SQLite3('data.db');
//shuffle($books);
foreach ($books as $book) {

    $bookData = [];
    $bookData['title'] = str_replace(["\r\n"], '', trim($book->first('.title::text')));
    $bookData['nid'] = (int)$book->attr('nid');

    $bookData['pdf'] = truncateBookUrl($book->first('a[href$=pdf]::attr(href)'));
    $bookData['epub'] = truncateBookUrl($book->first('a[href$=epub]::attr(href)'));
    $bookData['mobi'] = truncateBookUrl($book->first('a[href$=mobi]::attr(href)'));
    $bookData['code'] = truncateBookUrl($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)');
    $bookData['url'] = truncateBookUrl($book->first('div[class*=product-thumbnail]')->first('a::attr(href)'));


    $stmt = $db->prepare('SELECT id FROM book WHERE nid = :nid');
    $stmt->bindValue(':nid', $bookData['nid'], SQLITE3_INTEGER);
    $result = $stmt->execute();

    $resultData = $result->fetchArray(SQLITE3_ASSOC);

    $dl += downloadBook($bookData['pdf']);
    $dl += downloadBook($bookData['epub']);
    $dl += downloadBook($bookData['mobi']);
    $dl += downloadBook($bookData['code']);


    if (!$resultData) {
        $bookData['info'] = getBookInfo($bookData['url']);
        $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();
    }
    echo $dl.' - ';
    if ($dl > 500) {
//        var_dump($dl);
        die();
    }

}

function downloadBook($url)
{

    if (strlen($url) && !fileExists(fileNameFromPath($url))) {
        $localPath = fileNameFromPath($url);
        $directory = directoryNameFromPath($url);
        $fullUrl = resolveBookUrl($url);

        echo ($fullUrl)."\n";
//        var_dump($localPath);
        if (!is_dir($directory)) {
            mkdir($directory, 0777, true);
        }
        c($fullUrl, [], $localPath);
        return 1;
    }
    return 0;
}

function fileNameFromPath($filePath)
{
    $filePathParts = explode('/', trim($filePath, '/'));
    return 'books' . DIRECTORY_SEPARATOR . $filePathParts[1] . DIRECTORY_SEPARATOR . $filePathParts[1] . '.' . (isset($filePathParts[2]) ? $filePathParts[2] : 'zip');
}

function directoryNameFromPath($filePath)
{
    $filePathParts = explode('/', trim($filePath, '/'));
    return 'books' . DIRECTORY_SEPARATOR . $filePathParts[1] . DIRECTORY_SEPARATOR;
}

function fileExists($filePath)
{
    return file_exists($filePath);
}