Update search to website changes (#1)

Co-authored-by: Krzysztof Płaczek <krzysztofplaczek@techtube.pl>
Reviewed-on: #1

README.md

@@ -0,0 +1,50 @@
+## Installation 
+
+Place this in the composer.json.
+
+```composer 
+{
+  "repositories": [{
+    "type": "composer",
+    "url": "https://satis.techtube.pl"
+  }]
+}
+```
+Then execute to get latest version:
+
+```bash
+composer require techtube/bookinfo
+```
+
+or this to get specific version:
+
+```bash
+composer require techtube/bookinfo:"1.0.4"
+```
+
+## Usage
+
+To get information about the book by the link:
+
+```php
+<?php
+
+include_once 'vendor/autoload.php';
+
+$searcher = new \Techtube\Bookinfo\BookFinder();
+$data = $searcher->byUrl('https://lubimyczytac.pl/ksiazka/5008411/oni');
+
+print_r($data);
+```
+
+To search the book by phrase:
+```php
+<?php
+
+include_once 'vendor/autoload.php';
+
+$searcher = new \Techtube\Bookinfo\BookFinder();
+$data = $searcher->search('jack reacher');
+
+print_r($data);
+```

composer.json

@@ -13,8 +13,8 @@
     },
     "authors": [
         {
-            "name": "kplaczek",
+            "name": "Krzysztof Płaczek",
-            "email": "kplaczek@wi.ps.pl"
+            "email": "krzysztofplaczek@techtube.pl"
         }
     ]
 }

src/Api/AbstractBookInfo.php

@@ -9,11 +9,15 @@ abstract class AbstractBookInfo
     public string $isbn;
     public string $description;
     public string $title;
+    public string $originalTitle;
     public string $category;
-    public string $cover_url;
+    public array $cover_url;
     public int $pages;
     public string $cycle;
-    public int $volume;
+    public string $cycleUrl;
-    public string $language;
+    public string $volume;
+    public ?string $language;
     public string $datePublished;
+    public ?string $publisher;
+    public ?string $translator;
 }

src/BookFinder.php

@@ -7,7 +7,7 @@ use Techtube\Bookinfo\Api\AbstractBookInfo;
 
 class BookFinder
 {
-    private static $searchUrl = 'https://lubimyczytac.pl/szukaj/ksiazki?phrase=';
+    private static string $searchUrl = 'https://lubimyczytac.pl/szukaj/ksiazki?phrase=';
 
     private DataParser $parser;
 
@@ -21,14 +21,14 @@ class BookFinder
         return $this->parser->searchPage(new Document($this->getSearchUrl($phrase), true));
     }
 
-    public function byUrl($url): AbstractBookInfo
+    public function byUrl(string $url): AbstractBookInfo
     {
         return $this->parser->singlePage(new Document($url, true));
     }
 
-    public function getSearchUrl(string $phrase): string
+    private function getSearchUrl(string $phrase): string
     {
-        return self::$searchUrl . $phrase;
+        return sprintf('%s%s', self::$searchUrl, $phrase);
     }
 }
 

src/DataParser.php

@@ -4,6 +4,8 @@ namespace Techtube\Bookinfo;
 
 use DiDom\Document;
 use DiDom\Exceptions\InvalidSelectorException;
+use DiDom\Query;
+use JetBrains\PhpStorm\ArrayShape;
 use Techtube\Bookinfo\Api\AbstractBookInfo;
 
 class DataParser
@@ -17,19 +19,25 @@ class DataParser
     {
         $info = new Info();
         $jsonInfo = json_decode($document->first('script[type="application/ld+json"]')->text());
+        $info->url = $document->getDocument()->baseURI;
+        $info->publisher = $document->first('a[href*="wydawnictwo"]')?->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');
+        $info->description = trim($document->first('#book-description p')->text());
         $info->title = trim($document->first('h1.book__title')->text());
+        $info->originalTitle = trim($document->first("//*[contains(text(), 'Tytuł oryginału:')]", Query::TYPE_XPATH)?->nextSibling('dd')?->text() ?? '');
         $info->category = trim($document->first('.book__category')->text());
-        $info->cover_url = $document->first('meta[property="og:image"]')->getAttribute('content');
+        $info->cover_url = $this->generateCoverUrls(
+            $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)) {
+        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->language = $jsonInfo?->inLanguage ?? trim($document->xpath("//*[contains(text(), 'Język:')]")[0]->nextSibling('dd')->text());
-        $info->datePublished = $jsonInfo->datePublished ?? null;
+        $info->datePublished = $jsonInfo?->datePublished ?? null;
+        $info->translator = trim($document->first("//*[contains(text(), 'Tłumacz:')]", Query::TYPE_XPATH)?->nextSibling('dd')?->text() ?? '');
 
         return $info;
     }
@@ -41,17 +49,38 @@ class DataParser
      */
     public function searchPage(Document $document): array
     {
-        $books = $document->find('#search .authorAllBooks__single');
+        $books = $document->find('#ksiazki .authorAllBooks__single');
-
         $booksInfo = [];
-        foreach ($books as $book) {
+        if ($document->has('#searchksiazki')) {
-            $bookInfo = new Info();
+            foreach ($books as $book) {
-            $bookInfo->title = trim($book->first('.authorAllBooks__singleTextTitle')->text());
+                $bookInfo = new Info();
-            $bookInfo->author = trim($book->first('.authorAllBooks__singleTextAuthor')->text());
+                $bookInfo->title = trim($book->first('.authorAllBooks__singleTextTitle')->text());
-            $bookInfo->url = $book->first('button[data-book-url]')->getAttribute('data-book-url');
+                $bookInfo->author = trim($book->first('.authorAllBooks__singleTextAuthor')->text());
-            $bookInfo->cover_url = $book->first('.img-fluid')->getAttribute('data-src');
+                $bookInfo->url = $book->first('button[data-book-url]')->getAttribute('data-book-url');
-            $booksInfo[] = $bookInfo;
+                $bookInfo->cover_url = $this->generateCoverUrls($book->first('.img-fluid')->getAttribute('data-src'));
+                if (preg_match('#(.*) \(tom (.*)\)#ism', trim($book->first('a[href*="/cykl/"]')?->text() ?? ''), $series)) {
+                    $bookInfo->cycle = $series[1];
+                    $bookInfo->volume = $series[2];
+                    $bookInfo->cycleUrl = $book->first('a[href*="/cykl/"]')?->href;
+                }
+                $booksInfo[] = $bookInfo;
+            }
         }
         return $booksInfo;
     }
+
+    /**
+     * @param string $coverUrl
+     * @return array
+     */
+    #[ArrayShape(['small' => "string", 'medium' => "string", 'large' => "string"])]
+    private function generateCoverUrls(string $coverUrl): array
+    {
+        $coverUrlBase = preg_replace('(\d*?x\d*?\.jpg)', '', $coverUrl);
+        return [
+            'small' => $coverUrlBase . '70x100.jpg',
+            'medium' => $coverUrlBase . '170x243.jpg',
+            'large' => $coverUrlBase . '352x500.jpg'
+        ];
+    }
 }