From 293a776d16070a46dd1637296d8d195049cf489e Mon Sep 17 00:00:00 2001
From: krzysiej <krzysztofplaczek@techtube.pl>
Date: Fri, 27 May 2022 10:54:39 +0200
Subject: [PATCH] Updated a way of parsing a single book page to find language,
 publisher and book description. Also, url of a page is now attached to the
 book info object.

---
 src/Api/AbstractBookInfo.php |  3 ++-
 src/BookFinder.php           |  6 +++---
 src/DataParser.php           | 10 ++++++----
 3 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/src/Api/AbstractBookInfo.php b/src/Api/AbstractBookInfo.php
index cc286d7..99cecb4 100644
--- a/src/Api/AbstractBookInfo.php
+++ b/src/Api/AbstractBookInfo.php
@@ -14,6 +14,7 @@ abstract class AbstractBookInfo
     public int $pages;
     public string $cycle;
     public int $volume;
-    public string $language;
+    public ?string $language;
     public string $datePublished;
+    public ?string $publisher;
 }
\ No newline at end of file
diff --git a/src/BookFinder.php b/src/BookFinder.php
index 5529256..ee7ec90 100644
--- a/src/BookFinder.php
+++ b/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;
 
@@ -21,12 +21,12 @@ 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;
     }
diff --git a/src/DataParser.php b/src/DataParser.php
index 7cb81c3..53d92a9 100644
--- a/src/DataParser.php
+++ b/src/DataParser.php
@@ -18,21 +18,23 @@ 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());
         $info->category = trim($document->first('.book__category')->text());
         $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->datePublished = $jsonInfo->datePublished ?? null;
+        $info->language = $jsonInfo?->inLanguage ?? trim($document->xpath("//*[contains(text(), 'Język:')]")[0]->nextSibling('dd')->text());
+        $info->datePublished = $jsonInfo?->datePublished ?? null;
 
         return $info;
     }