krzysiej/packtpub-browser
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity

Poprawianie kodu

Browse Source
This commit is contained in:
krzysiej
2018-07-13 14:59:25 +02:00
parent 42beb193e4
commit 22d0bb4f04
2 changed files with 279 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

288
packt.php
View File

@@ -15,6 +15,284 @@ $loginData = [
];
class Packt
{
private $credentials = [];
const BASE_URL = 'https://www.packtpub.com/';
private $database = null;
public function __construct($credentials)
{
$this->credentials = $credentials;
}
/**
* @return null|SQLite3
*/
public function getDatabase()
{
if (!$this->database) {
$this->database = new SQLite3('data.db');
}
return $this->database;
}
public function resolveBookUrl($bookUrl)
{
return self::BASE_URL . trim(str_replace(self::BASE_URL, '', $bookUrl), '/');
}
public function truncateBookUrl($bookUrl)
{
return str_replace('https://www.packtpub.com', '', $bookUrl);
}
public function getBookInfo($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 = [];
$authorData['name'] = trim($author->first('.book-info-bottom-author-title')->text());
$authorData['bio'] = preg_replace('#\s{2,}#', ' ', trim($author->first('.book-info-bottom-author-body')->text()));
$bookData['authors'][] = $authorData;
}
return $bookData;
}
public function saveBookChapters($data, $bookId)
{
$db = $this->getDatabase();
foreach ($data['toc'] as $chapterNumber => $chapter) {
$stmt = $db->prepare('REPLACE INTO chapter (book_id, chapter_number, parent_id, name)
VALUES (:book_id, :chapter_number, :parent_id, :name)');
$stmt->bindValue(':book_id', $bookId, SQLITE3_INTEGER);
$stmt->bindValue(':chapter_number', $chapterNumber + 1, SQLITE3_INTEGER);
$stmt->bindValue(':parent_id', null, SQLITE3_INTEGER);
$stmt->bindValue(':name', $chapter['title'], SQLITE3_TEXT);
$result = $stmt->execute();
$stmt = $db->prepare('select seq from sqlite_sequence where name="chapter"');
$result = $stmt->execute();
$lastInsertedId = $result->fetchArray(SQLITE3_ASSOC);
$lastInsertedId['seq'];
if (isset($chapter['subchapters'])) {
foreach ($chapter['subchapters'] as $subChapterNumber => $subChapterName) {
$stmt = $db->prepare('REPLACE INTO chapter (book_id, chapter_number, parent_id, name)
VALUES (:book_id, :chapter_number, :parent_id, :name)');
$stmt->bindValue(':book_id', $bookId, SQLITE3_INTEGER);
$stmt->bindValue(':chapter_number', $subChapterNumber + 1, SQLITE3_INTEGER);
$stmt->bindValue(':parent_id', $lastInsertedId['seq'], SQLITE3_INTEGER);
$stmt->bindValue(':name', $subChapterName, SQLITE3_TEXT);
$result = $stmt->execute();
}
}
}
}
public function directoryNameFromPath($filePath)
{
$filePathParts = explode('/', trim($filePath, '/'));
return 'books' . DIRECTORY_SEPARATOR . $filePathParts[1] . DIRECTORY_SEPARATOR;
}
public function downloadBook($url)
{
if (strlen($url) && !file_exists(fileNameFromPath($url))) {
$localPath = $this->fileNameFromPath($url);
$directory = $this->directoryNameFromPath($url);
$fullUrl = $this->resolveBookUrl($url);
if (!is_dir($directory)) {
mkdir($directory, 0777, true);
}
c($fullUrl, [], $localPath);
return 1;
}
return 0;
}
public function fileNameFromPath($filePath)
{
$filePathParts = explode('/', trim($filePath, '/'));
return 'books' . DIRECTORY_SEPARATOR . $filePathParts[1] . DIRECTORY_SEPARATOR . $filePathParts[1] . '.' . (isset($filePathParts[2]) ? $filePathParts[2] : 'zip');
}
public function getBooks()
{
//logowanie
c('https://www.packtpub.com/', $this->credentials);
//pobieranie listy ksiązek
$return = c('https://www.packtpub.com/account/my-ebooks');
$document = new Document($return);
$books = $document->find('.product-line.unseen');
foreach ($books as $book) {
$bookData = [];
$bookData['nid'] = (int)$book->attr('nid');
$bookData['title'] = str_replace(["\r\n"], '', trim($book->first('.title::text')));
$bookData['ebook_isbn'] = $book->first('div[isbn]::attr(isbn)');
$bookData['img'] = $book->first('img[class*=imagecache]::attr(src)');
$bookData['url'] = $this->truncateBookUrl($book->first('div[class*=product-thumbnail]')->first('a::attr(href)'));
$bookData['pdf'] = $this->truncateBookUrl($book->first('a[href$=pdf]::attr(href)'));
$bookData['epub'] = $this->truncateBookUrl($book->first('a[href$=epub]::attr(href)'));
$bookData['mobi'] = $this->truncateBookUrl($book->first('a[href$=mobi]::attr(href)'));
$bookData['code'] = $this->truncateBookUrl($book->first('a[href*=code_download]::attr(href)'));
yield $bookData;
}
}
/**
* @param int $nid
* @return bool
*/
public function isBookByNid(int $nid)
{
$stmt = $this->getDatabase()->prepare('SELECT nid FROM book WHERE nid = :nid');
$stmt->bindValue(':nid', $nid, SQLITE3_INTEGER);
$result = $stmt->execute();
return $result->fetchArray(SQLITE3_ASSOC) === false ? false : true;
}
public function saveBooks()
{
foreach ($this->getBooks() as $basicBookInfo) {
if ($this->isBookByNid($basicBookInfo['nid'])) {
$stmt = $this->getDatabase()->prepare('UPDATE book SET nid = :nid, title = :title, ebook_isbn = :ebook_isbn, img = :img, url = :url, pdf = :pdf, epub = :epub, mobi = :mobi, code = :code, upd_dt = (DATETIME(\'now\')) WHERE nid = :nid');
} else {
$stmt = $this->getDatabase()->prepare('INSERT INTO book (nid, title, isbn, img, url, pdf, epub, mobi, code)
VALUES (:nid, :title, :ebook_isbn, :img, :url, :pdf, :epub, :mobi, :code)');
}
$stmt->bindValue(':nid', $basicBookInfo['nid'], SQLITE3_INTEGER);
$stmt->bindValue(':title', $basicBookInfo['title'], SQLITE3_TEXT);
$stmt->bindValue(':ebook_isbn', $basicBookInfo['ebook_isbn'], SQLITE3_TEXT);
$stmt->bindValue(':img', $basicBookInfo['img'], SQLITE3_TEXT);
$stmt->bindValue(':url', $basicBookInfo['url'], SQLITE3_TEXT);
$stmt->bindValue(':pdf', $basicBookInfo['pdf'], SQLITE3_TEXT);
$stmt->bindValue(':epub', $basicBookInfo['epub'], SQLITE3_TEXT);
$stmt->bindValue(':mobi', $basicBookInfo['mobi'], SQLITE3_TEXT);
$stmt->bindValue(':code', $basicBookInfo['code'], SQLITE3_TEXT);
$stmt->execute();
}
}
public function saveBookDetails($bookDetails)
{
print_r($bookDetails['nid']);
if ($this->isBookByNid($bookDetails['nid'])) {
print_r($bookDetails);
echo 'xxx';
$stmt = $this->getDatabase()->prepare('UPDATE book SET datepublished = :datepublished, reviewCount = :reviewCount, ratingValue = :ratingValue, numberofpages = :numberofpages, upd_dt = (DATETIME(\'now\')) WHERE nid = :nid');
$stmt->bindValue(':datepublished', $bookDetails['datepublished'], SQLITE3_TEXT);
$stmt->bindValue(':reviewCount', $bookDetails['reviewCount'], SQLITE3_INTEGER);
$stmt->bindValue(':ratingValue', $bookDetails['ratingValue'], SQLITE3_FLOAT);
$stmt->bindValue(':numberofpages', $bookDetails['numberofpages'], SQLITE3_INTEGER);
$stmt->bindValue(':nid', $bookDetails['nid'], SQLITE3_INTEGER);
print_r($stmt->execute());
}
}
public function getBookDetails($bookUrl)
{
$bookUrl = $this->resolveBookUrl($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['paper_isbn'] = $bookPage->first('.book-info-details span[itemprop="isbn"]::text');
$bookData['ebook_isbn'] = $bookPage->first('div.digital-product[data-product-type="ebooks"]::attr(data-product-id)');
$bookData['ebook_price'] = $bookPage->first('div.digital-product[data-product-type="ebooks"]::attr(data-product-price)');
$bookData['paper_price'] = $bookPage->first('div.physical-product[data-product-type="books"]::attr(data-product-price)');
$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['title'] = $bookPage->first('h1[itemprop="name"]::text');
$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['paper_isbn'] . '"]::attr(data-product-category)');
// $bookData['price'] = $bookPage->first('div[data-product-id="' . $bookData['isbn'] . '"]::attr(data-product-price)');
// $bookData['price'] = $bookPage->first('div[data-product-id="' . $bookData['paper_isbn'] . '"][data-product-type="ebooks"]');
// $bookData['nid'] = $bookPage->first('div[data-product-id="' . $bookData['paper_isbn'] . '"]::attr(data-product-nid)');
$bookData['nid'] = $bookPage->first('div.digital-product[data-product-type="ebooks"]::attr(data-product-nid)');
$bookData['authors'] = [];
foreach ($bookPage->find('[itemprop="author"]') as $author) {
$authorData = [];
$authorData['name'] = trim($author->first('.book-info-bottom-author-title')->text());
$authorData['bio'] = preg_replace('#\s{2,}#', ' ', trim($author->first('.book-info-bottom-author-body')->text()));
$bookData['authors'][] = $authorData;
}
return $bookData;
}
}
echo '<pre>';
$packt = new Packt($loginData);
//$books = $packt->saveBooks();
//print_r($books);
echo '<pre>';
$bookDetails = $packt->getBookDetails('/web-development/dependency-injection-angularjs');
print_r($bookDetails);
//$packt->saveBookDetails($bookDetails);
die();
function c($url, $post = [], $localFilePath = null)
{
$cookie = "cookie.txt";
@@ -127,7 +405,6 @@ function saveBookChapters($data, $bookId)
}
}
}
}
@@ -176,7 +453,6 @@ while ($book = $booksResult->fetchArray(SQLITE3_ASSOC)) {
}
}
//$data = json_decode(file_get_contents('data.txt'), 1);
@@ -304,13 +580,11 @@ VALUES (:nid, :title, :isbn, :img, :url, :datepublished, :numberofpages, :revie
function downloadBook($url)
{
if (strlen($url) && !fileExists(fileNameFromPath($url))) {
if (strlen($url) && !file_exists(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);
}
@@ -332,8 +606,4 @@ function directoryNameFromPath($filePath)
return 'books' . DIRECTORY_SEPARATOR . $filePathParts[1] . DIRECTORY_SEPARATOR;
}
function fileExists($filePath)
{
return file_exists($filePath);
}