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

Fixed checking if book info is already in the database

Browse Source
This commit is contained in:
krzysiej
2018-04-25 12:37:23 +02:00
parent b3a1ebafe7
commit b496f8a531
2 changed files with 2799 additions and 2593 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4117
packt.html
View File

File diff suppressed because one or more lines are too long

23
packt.php
View File

@@ -34,8 +34,6 @@ function c($url, $post = [])
$server_output = curl_exec($ch); $server_output = curl_exec($ch);
strlen($server_output); strlen($server_output);
// $info = curl_getinfo($ch);
curl_close($ch); curl_close($ch);
@@ -50,6 +48,11 @@ function resolveBookUrl($bookUrl)
return 'https://www.packtpub.com/' . str_replace('https://www.packtpub.com/', '', $bookUrl); return 'https://www.packtpub.com/' . str_replace('https://www.packtpub.com/', '', $bookUrl);
} }
function truncateBookUrl($bookUrl)
{
return str_replace('https://www.packtpub.com', '', $bookUrl);
}
function getBookInfo($bookUrl) function getBookInfo($bookUrl)
{ {
@@ -99,32 +102,30 @@ $booksData = [];
$books = $document->find('.product-line.unseen'); $books = $document->find('.product-line.unseen');
$db = new SQLite3('data.db'); $db = new SQLite3('data.db');
foreach ($books as $book) { foreach ($books as $book) {
$bookData = []; $bookData = [];
$bookData['title'] = str_replace(["\r\n"], '', trim($book->first('.title::text'))); $bookData['title'] = str_replace(["\r\n"], '', trim($book->first('.title::text')));
$bookData['nid'] = (int)$book->attr('nid'); $bookData['nid'] = (int)$book->attr('nid');
$bookData['pdf'] = $book->first('a[href$=pdf]::attr(href)'); $bookData['pdf'] = truncateBookUrl($book->first('a[href$=pdf]::attr(href)'));
$bookData['epub'] = $book->first('a[href$=epub]::attr(href)'); $bookData['epub'] = truncateBookUrl($book->first('a[href$=epub]::attr(href)'));
$bookData['mobi'] = $book->first('a[href$=mobi]::attr(href)'); $bookData['mobi'] = truncateBookUrl($book->first('a[href$=mobi]::attr(href)'));
$bookData['code'] = $book->first('a[href*=code_download]::attr(href)'); $bookData['code'] = truncateBookUrl($book->first('a[href*=code_download]::attr(href)'));
$bookData['isbn'] = $book->first('div[isbn]::attr(isbn)'); $bookData['isbn'] = $book->first('div[isbn]::attr(isbn)');
$bookData['img'] = $book->first('img[class*=imagecache]::attr(src)'); $bookData['img'] = $book->first('img[class*=imagecache]::attr(src)');
$bookData['url'] = $book->first('div[class*=product-thumbnail]')->first('a::attr(href)'); $bookData['url'] = truncateBookUrl($book->first('div[class*=product-thumbnail]')->first('a::attr(href)'));
$stmt = $db->prepare('SELECT id FROM book WHERE nid = :nid'); $stmt = $db->prepare('SELECT id FROM book WHERE nid = :nid');
$stmt->bindValue(':nid', $bookData['nid'], SQLITE3_INTEGER); $stmt->bindValue(':nid', $bookData['nid'], SQLITE3_INTEGER);
$result = $stmt->execute(); $result = $stmt->execute();
if ($result->numColumns() > 0) { $resultData = $result->fetchArray(SQLITE3_ASSOC);
if (!$resultData) {
} else {
$bookData['info'] = getBookInfo($bookData['url']); $bookData['info'] = getBookInfo($bookData['url']);
$stmt = $db->prepare('REPLACE INTO book (nid, title, isbn, img, url, datepublished, numberofpages, reviewCount, ratingValue, category, pdf, epub, mobi, code) $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)'); VALUES (:nid, :title, :isbn, :img, :url, :datepublished, :numberofpages, :reviewCount, :ratingValue, :category, :pdf, :epub, :mobi, :code)');