Refactored description truncator

clash82 · Nov 28, 2024 · 6bb4b03 · 6bb4b03
1 parent e0be448
commit 6bb4b03
Show file tree

Hide file tree

Showing 11 changed files with 105 additions and 33 deletions.
diff --git a/composer.json b/composer.json
@@ -15,6 +15,7 @@
     "ext-simplexml": "*",
     "ext-json": "*",
     "ext-libxml": "*",
+    "ext-dom": "*",
     "wolfcast/browser-detection": "^2.9",
     "simplepie/simplepie": "^1.8"
   },

diff --git a/composer.lock b/composer.lock
diff --git a/config/sample.ini.dist b/config/sample.ini.dist
@@ -1,2 +1,5 @@
 name=Sample RSS channel name
 rss=https://link-to-your-working-rss-channel
+
+; remove whitespaces and HTML tags from post content (0|1, disabled by default)
+trim=0
diff --git a/src/lib/App.php b/src/lib/App.php
@@ -10,6 +10,7 @@
 use MultiRssCombiner\Renderer\PageRenderer;
 use MultiRssCombiner\Renderer\RssRenderer;
 use MultiRssCombiner\Value\Item;
+use SimplePie\SimplePie;
 use Wolfcast\BrowserDetection;
 
 class App
@@ -22,10 +23,17 @@ class App
 
     final public const APP_CHANNEL_CONFIGURATION_PATH = '/config/';
 
+    private readonly ChannelConfiguration $channels;
+
+    public function __construct()
+    {
+        $this->channels = new ChannelConfiguration(self::APP_CHANNEL_CONFIGURATION_PATH);
+    }
+
     public function buildView(bool $showDefault = true): void
     {
         $configuration = new GeneralConfiguration(self::APP_CONFIGURATION_FILE);
-        $cache = new RssCacheProvider(self::APP_RSS_CACHE_FILE, $configuration->get()->getLimit());
+        $cache = new RssCacheProvider($this->channels, self::APP_RSS_CACHE_FILE, $configuration->get()->getLimit());
         $browser = new BrowserDetection();
 
         // detect if client belongs to the one of the supported web browsers or is just an RSS reader
@@ -45,17 +53,15 @@ public function buildView(bool $showDefault = true): void
 
     public function buildCache(): void
     {
-        $channels = new ChannelConfiguration(self::APP_CHANNEL_CONFIGURATION_PATH);
-
-        $feed = new \SimplePie();
-        $feed->enable_order_by_date(true);
+        $feed = new SimplePie();
+        $feed->enable_order_by_date();
         $feed->force_feed(true);
         $feed->enable_cache(false);
 
         $items = [];
 
         // fetch RSS details
-        foreach ($channels->getAll() as $channel) {
+        foreach ($this->channels->getAll() as $channel) {
             printf('Fetching %s (%s)<br>', $channel->getName(), $channel->getUrl());
 
             $feed->set_feed_url($channel->getUrl());
@@ -84,7 +90,7 @@ public function buildCache(): void
                 }
 
                 $item = new Item(
-                    $channel->getName(),
+                    $channel,
                     $item->get_title() ?? '',
                     $description ?? '',
                     $item->get_link() ?? '',
@@ -98,7 +104,7 @@ public function buildCache(): void
         }
 
         // reorder items
-        usort($items, fn($a, $b) => $b->getPubDate() <=> $a->getPubDate());
+        usort($items, fn ($a, $b) => $b->getPubDate() <=> $a->getPubDate());
 
         // store everything in cache
         $cache = new RssCacheManager(self::APP_RSS_CACHE_FILE);

diff --git a/src/lib/Manager/RssCache.php b/src/lib/Manager/RssCache.php
@@ -27,7 +27,7 @@ public function save(): void
 
         foreach ($this->cache as $item) {
             $element = $xml->addChild('item');
-            $element->addChild('channelName', $item->getChannelName());
+            $element->addAttribute('channel', $item->getChannelId());
             $element->addChild('title', $item->getTitle());
             $element->addChild('description', $item->getDescription());
             $element->addChild('link', $item->getLink());

diff --git a/src/lib/Provider/ChannelConfiguration.php b/src/lib/Provider/ChannelConfiguration.php
@@ -22,9 +22,11 @@ public function __construct(string $path)
 
             $ini = (array)parse_ini_file($fileInfo->getRealPath());
 
-            $this->channels[] = new Channel(
+            $this->channels[$fileInfo->getFilename()] = new Channel(
+                $fileInfo->getFilename(),
                 $ini['name'],
-                $ini['url']
+                $ini['url'],
+                isset($ini['trim']) && $ini['trim']
             );
         }
     }
@@ -36,4 +38,14 @@ public function getAll(): array
     {
         return $this->channels;
     }
+
+    public function getById(string $id): Channel
+    {
+        return $this->channels[$id];
+    }
+
+    public function channelExists(string $id): bool
+    {
+        return isset($this->channels[$id]);
+    }
 }
diff --git a/src/lib/Provider/GeneralConfiguration.php b/src/lib/Provider/GeneralConfiguration.php
@@ -5,9 +5,9 @@
 use MultiRssCombiner\Exception\ConfigurationNotFoundException;
 use MultiRssCombiner\Value\General;
 
-class GeneralConfiguration implements GeneralConfigurationProvider
+readonly class GeneralConfiguration implements GeneralConfigurationProvider
 {
-    private readonly General $configuration;
+    private General $configuration;
 
     public function __construct(string $filename)
     {

diff --git a/src/lib/Provider/RssCache.php b/src/lib/Provider/RssCache.php
@@ -9,7 +9,7 @@ class RssCache implements CacheProvider
     /** @var Item[] */
     private array $cache = [];
 
-    public function __construct(string $fileName, int $limit)
+    public function __construct(ChannelConfiguration $channelConfiguration, string $fileName, int $limit)
     {
         $cacheFile = sprintf('%s%s', getcwd(), $fileName);
 
@@ -34,11 +34,16 @@ public function __construct(string $fileName, int $limit)
             }
 
             $item = $cacheContent['item'][$i];
+            $channelId = $item['@attributes']['channel'];
+
+            if (!$channelConfiguration->channelExists($channelId)) {
+                continue;
+            }
 
             $date = new \DateTime($item['pubDate']);
 
             $this->cache[] = new Item(
-                $item['channelName'],
+                $channelConfiguration->getById($channelId),
                 $item['title'],
                 $item['description'],
                 $item['link'],

diff --git a/src/lib/Value/Channel.php b/src/lib/Value/Channel.php
@@ -2,10 +2,19 @@
 
 namespace MultiRssCombiner\Value;
 
-class Channel
+readonly class Channel
 {
-    public function __construct(private readonly string $name, private readonly string $url)
+    public function __construct(
+        private string $id,
+        private string $name,
+        private string $url,
+        private bool $trimContent
+    ) {
+    }
+
+    public function getId(): string
     {
+        return $this->id;
     }
 
     public function getName(): string
@@ -17,4 +26,9 @@ public function getUrl(): string
     {
         return $this->url;
     }
+
+    public function isTrimContent(): bool
+    {
+        return $this->trimContent;
+    }
 }
diff --git a/src/lib/Value/General.php b/src/lib/Value/General.php
@@ -4,10 +4,22 @@
 
 use MultiRssCombiner\App;
 
-class General
+readonly class General
 {
-    public function __construct(private readonly string $title, private readonly string $description, private readonly string $link, private readonly string $language, private readonly string $copyright, private readonly string $icon, private readonly int $iconWidth, private readonly int $iconHeight, private readonly string $dateFormat, private readonly int $limit, private readonly string $guidPrefix, private readonly string $feedUrl)
-    {
+    public function __construct(
+        private string $title,
+        private string $description,
+        private string $link,
+        private string $language,
+        private string $copyright,
+        private string $icon,
+        private int $iconWidth,
+        private int $iconHeight,
+        private string $dateFormat,
+        private int $limit,
+        private string $guidPrefix,
+        private string $feedUrl
+    ) {
     }
 
     public function getTitle(): string

diff --git a/src/lib/Value/Item.php b/src/lib/Value/Item.php
@@ -4,15 +4,27 @@
 
 use DateTime;
 
-class Item
+readonly class Item
 {
-    public function __construct(private readonly string $channelName, private readonly string $title, private readonly string $description, private readonly string $link, private readonly string $guid, private readonly ?string $image, private readonly DateTime $pubDate)
+    public function __construct(
+        private Channel $channel,
+        private string $title,
+        private string $description,
+        private string $link,
+        private string $guid,
+        private ?string $image,
+        private DateTime $pubDate
+    ) {
+    }
+
+    public function getChannelId(): string
     {
+        return $this->channel->getId();
     }
 
     public function getChannelName(): string
     {
-        return $this->channelName;
+        return $this->channel->getName();
     }
 
     public function getTitle(): string
@@ -22,17 +34,23 @@ public function getTitle(): string
 
     public function getDescription(): string
     {
-        # remove whitespaces inside HTML tags
-        $description = preg_replace('/\s+/', ' ', $this->description);
+        $description = $this->description;
+
+        if ($this->channel->isTrimContent()) {
+            # remove whitespaces inside HTML tags
+            $description = preg_replace('/\s+/', ' ', $description);
+
+            # replace <br /> and redundant <br /> tags
+            $description = preg_replace('#<br />(\s*<br />)+#', '<br>', $description);
 
-        # replace <br /> and redundant <br /> tags
-        $description = preg_replace('#<br />(\s*<br />)+#', '<br>', $description);
+            # replace redundant <br> tags
+            $description = preg_replace('#<br>(\s*<br>)+#', '<br>', $description);
 
-        # replace redundant <br> tags
-        $description = preg_replace('#<br>(\s*<br>)+#', '<br>', $description);
+            # remove all HTML tags
+            $description = strip_tags($description);
+        }
 
-        # remove all HTML tags
-        return strip_tags($description);
+        return $description;
     }
 
     public function getLink(): string