Excluded data parsing to a data parser class that handles document and returns info class or array of those.

2022-05-25 19:58:15 +02:00
parent 0e3c8852ce
commit 7d70ad40cb
5 changed files with 85 additions and 42 deletions
--- a/src/DataParser.php
+++ b/src/DataParser.php
@@ -0,0 +1,57 @@
+<?php
+
+namespace Techtube\Bookinfo;
+
+use DiDom\Document;
+use DiDom\Exceptions\InvalidSelectorException;
+use Techtube\Bookinfo\Api\AbstractBookInfo;
+
+class DataParser
+{
+    /**
+     * @param Document $document
+     * @return AbstractBookInfo
+     * @throws InvalidSelectorException
+     */
+    public function singlePage(Document $document): AbstractBookInfo
+    {
+        $info = new Info();
+        $jsonInfo = json_decode($document->first('script[type="application/ld+json"]')->text());
+        $info->author = $document->first('meta[property="books:author"]')->getAttribute('content');
+        $info->isbn = $document->first('meta[property="books:isbn"]')->getAttribute('content');
+        $info->description = $document->first('meta[property="og:description"]')->getAttribute('content');
+        $info->title = trim($document->first('h1.book__title')->text());
+        $info->category = trim($document->first('.book__category')->text());
+        $info->cover_url = $document->first('meta[property="og:image"]')->getAttribute('content');
+        $info->pages = (int)$document->first('span.book__pages')?->text();
+        if (preg_match('#(.*) \(tom (\d*)\)#ism', trim($document->first('a[href*="/cykl/"]')?->text()), $series)) {
+            $info->cycle = $series[1];
+            $info->volume = $series[2];
+        }
+        $info->language = $jsonInfo->inLanguage ?? null;
+        $info->datePublished = $jsonInfo->datePublished ?? null;
+
+        return $info;
+    }
+
+    /**
+     * @param Document $document
+     * @return AbstractBookInfo[]
+     * @throws InvalidSelectorException
+     */
+    public function searchPage(Document $document): array
+    {
+        $books = $document->find('#search .authorAllBooks__single');
+
+        $booksInfo = [];
+        foreach ($books as $book) {
+            $bookInfo = new Info();
+            $bookInfo->title = trim($book->first('.authorAllBooks__singleTextTitle')->text());
+            $bookInfo->author = trim($book->first('.authorAllBooks__singleTextAuthor')->text());
+            $bookInfo->url = $book->first('button[data-book-url]')->getAttribute('data-book-url');
+            $bookInfo->cover_url = $book->first('.img-fluid')->getAttribute('src');
+            $booksInfo[] = $bookInfo;
+        }
+        return $booksInfo;
+    }
+}