From 1c3c85d8ff5a6d071f688ef09ca93f275b4995af Mon Sep 17 00:00:00 2001
From: Niehztog <Niehztog@users.noreply.github.com>
Date: Sun, 31 Mar 2024 18:46:07 +0200
Subject: [PATCH] [XPathBridge] Allow multiple categories (#4038)

* [XPathAbstract] allow multiple categories

* fix feed icons in two bridges

* fix warning

* fix linter errors
---
 bridges/BlizzardNewsBridge.php |  7 ++++
 bridges/NiusBridge.php         |  5 +++
 lib/XPathAbstract.php          | 74 +++++++++++++++++++++++++---------
 3 files changed, 67 insertions(+), 19 deletions(-)

diff --git a/bridges/BlizzardNewsBridge.php b/bridges/BlizzardNewsBridge.php
index 3930e0a4d1a..19c38152cc0 100644
--- a/bridges/BlizzardNewsBridge.php
+++ b/bridges/BlizzardNewsBridge.php
@@ -57,4 +57,11 @@ protected function getSourceUrl()
         }
         return 'https://news.blizzard.com/' . $locale;
     }
+
+    public function getIcon()
+    {
+        return <<<icon
+https://blznews.akamaized.net/images/favicon-cb34a003c6f2f637ee8f4f7b406f3b9b120b918c04cabec7f03a760e708977ea9689a1c638f4396def8dce7b202cd007eae91946cc3c4a578aa8b5694226cfc6.ico
+icon;
+    }
 }
diff --git a/bridges/NiusBridge.php b/bridges/NiusBridge.php
index c76b29f0809..0be6e89c2fd 100644
--- a/bridges/NiusBridge.php
+++ b/bridges/NiusBridge.php
@@ -23,6 +23,11 @@ class NiusBridge extends XPathAbstract
     const XPATH_EXPRESSION_ITEM_CATEGORIES = './/div[@class="subtitle"]/text()';
     const SETTING_FIX_ENCODING = false;
 
+    public function getIcon()
+    {
+        return 'https://www.nius.de/favicon.ico';
+    }
+
     protected function formatItemTitle($value)
     {
         return strip_tags($value);
diff --git a/lib/XPathAbstract.php b/lib/XPathAbstract.php
index 2206f79ac9d..224d8e875a1 100644
--- a/lib/XPathAbstract.php
+++ b/lib/XPathAbstract.php
@@ -439,7 +439,8 @@ public function collectData()
                 }
 
                 $isContent = $param === 'content';
-                $value = $this->getItemValueOrNodeValue($typedResult, $isContent, $isContent && !$this->getSettingUseRawItemContent());
+                $isCategories = 'categories' === $param;
+                $value = $this->getItemValueOrNodeValue($typedResult, $isContent, $isContent && !$this->getSettingUseRawItemContent(), $isCategories);
                 $item->__set($param, $this->formatParamValue($param, $value));
             }
 
@@ -459,7 +460,7 @@ public function collectData()
      */
     protected function formatParamValue($param, $value)
     {
-        $value = $this->fixEncoding($value);
+        $value = is_array($value) ? array_map([$this, 'fixEncoding'], $value) : $this->fixEncoding($value);
         switch ($param) {
             case 'title':
                 return $this->formatItemTitle($value);
@@ -572,12 +573,12 @@ protected function formatItemEnclosures($value)
      * formatted as array.
      * Can be easily overwritten for in case the values need to be transformed into something
      * else.
-     * @param string $value
+     * @param string|array $value
      * @return array
      */
     protected function formatItemCategories($value)
     {
-        return [$value];
+        return is_array($value) ? $value : [$value];
     }
 
     /**
@@ -596,22 +597,21 @@ protected function cleanMediaUrl($mediaUrl)
 
     /**
      * @param $typedResult
-     * @return string
+     * @param bool $returnXML
+     * @param bool $escapeHtml
+     * @param bool $allowMultiple
+     * @return string|array
+     * @throws Exception
      */
-    protected function getItemValueOrNodeValue($typedResult, $returnXML = false, $escapeHtml = false)
+    protected function getItemValueOrNodeValue($typedResult, $returnXML = false, $escapeHtml = false, $allowMultiple = false)
     {
-        if ($typedResult instanceof \DOMNodeList) {
+        if ($typedResult instanceof \DOMNodeList && !$allowMultiple) {
             $item = $typedResult->item(0);
-            if ($item instanceof \DOMElement) {
-                // Don't escape XML
-                if ($returnXML) {
-                    return ($item->ownerDocument ?? $item)->saveXML($item);
-                }
-                $text = $item->nodeValue;
-            } elseif ($item instanceof \DOMAttr) {
-                $text = $item->value;
-            } elseif ($item instanceof \DOMText) {
-                $text = $item->wholeText;
+            $text = $this->extractNodeListContent($item, $returnXML);
+        } elseif ($typedResult instanceof \DOMNodeList && $allowMultiple) {
+            $text = [];
+            foreach ($typedResult as $item) {
+                $text[] = $this->extractNodeListContent($item, $returnXML);
             }
         } elseif (is_string($typedResult) && strlen($typedResult) > 0) {
             $text = $typedResult;
@@ -619,10 +619,46 @@ protected function getItemValueOrNodeValue($typedResult, $returnXML = false, $es
             throw new \Exception('Unknown type of XPath expression result.');
         }
 
+        if (is_array($text)) {
+            foreach ($text as &$element) {
+                $element = $this->cleanExtractedText($element, $escapeHtml, $returnXML);
+            }
+        } else {
+            $text = $this->cleanExtractedText($text, $escapeHtml, $returnXML);
+        }
+        return $text;
+    }
+
+    /**
+     * @param $item
+     * @param $returnXML
+     * @return false|string
+     * @throws Exception
+     */
+    protected function extractNodeListContent($item, $returnXML)
+    {
+        if ($item instanceof \DOMElement) {
+            return $returnXML ? ($item->ownerDocument ?? $item)->saveXML($item) : $item->nodeValue;
+        } elseif ($item instanceof \DOMAttr) {
+            return $item->value;
+        } elseif ($item instanceof \DOMText) {
+            return $item->wholeText;
+        }
+        throw new \Exception('Unknown type of XPath expression result.');
+    }
+
+    /**
+     * @param $text
+     * @param $escapeHtml
+     * @param $returnXML
+     * @return string
+     */
+    protected function cleanExtractedText($text, $escapeHtml, $returnXML)
+    {
         $text = trim($text);
 
-        if ($escapeHtml) {
-            return htmlspecialchars($text);
+        if ($escapeHtml && !$returnXML) {
+            $text = htmlspecialchars($text);
         }
         return $text;
     }