Merge remote-tracking branch 'upstream/master'

floviolleau · Aug 8, 2023 · 6ab3a5f · 6ab3a5f
2 parents b02b970 + b86ee57
commit 6ab3a5f
Show file tree

Hide file tree

Showing 24 changed files with 305 additions and 49 deletions.
diff --git a/actions/ConnectivityAction.php b/actions/ConnectivityAction.php
@@ -37,12 +37,14 @@ public function execute(array $request)
             throw new \Exception('This action is only available in debug mode!');
         }
 
-        if (!isset($request['bridge'])) {
+        $bridgeName = $request['bridge'] ?? null;
+        if (!$bridgeName) {
             return render_template('connectivity.html.php');
         }
-
-        $bridgeClassName = $this->bridgeFactory->createBridgeClassName($request['bridge']);
-
+        $bridgeClassName = $this->bridgeFactory->createBridgeClassName($bridgeName);
+        if (!$bridgeClassName) {
+            throw new \Exception(sprintf('Bridge not found: %s', $bridgeName));
+        }
         return $this->reportBridgeConnectivity($bridgeClassName);
     }
 

diff --git a/actions/DisplayAction.php b/actions/DisplayAction.php
@@ -33,17 +33,22 @@ public function execute(array $request)
     private function createResponse(array $request)
     {
         $bridgeFactory = new BridgeFactory();
-        $bridgeClassName = $bridgeFactory->createBridgeClassName($request['bridge'] ?? '');
+        $formatFactory = new FormatFactory();
 
+        $bridgeName = $request['bridge'] ?? null;
         $format = $request['format'] ?? null;
+
+        $bridgeClassName = $bridgeFactory->createBridgeClassName($bridgeName);
+        if (!$bridgeClassName) {
+            throw new \Exception(sprintf('Bridge not found: %s', $bridgeName));
+        }
         if (!$format) {
             throw new \Exception('You must specify a format!');
         }
         if (!$bridgeFactory->isEnabled($bridgeClassName)) {
             throw new \Exception('This bridge is not whitelisted');
         }
 
-        $formatFactory = new FormatFactory();
         $format = $formatFactory->create($format);
 
         $bridge = $bridgeFactory->create($bridgeClassName);

diff --git a/actions/FrontpageAction.php b/actions/FrontpageAction.php
@@ -4,12 +4,20 @@ final class FrontpageAction implements ActionInterface
 {
     public function execute(array $request)
     {
+        $messages = [];
         $showInactive = (bool) ($request['show_inactive'] ?? null);
         $activeBridges = 0;
 
         $bridgeFactory = new BridgeFactory();
         $bridgeClassNames = $bridgeFactory->getBridgeClassNames();
 
+        foreach ($bridgeFactory->getMissingEnabledBridges() as $missingEnabledBridge) {
+            $messages[] = [
+                'body' => sprintf('Warning : Bridge "%s" not found', $missingEnabledBridge),
+                'level' => 'warning'
+            ];
+        }
+
         $formatFactory = new FormatFactory();
         $formats = $formatFactory->getFormatNames();
 
@@ -24,7 +32,7 @@ public function execute(array $request)
         }
 
         return render(__DIR__ . '/../templates/frontpage.html.php', [
-            'messages' => [],
+            'messages' => $messages,
             'admin_email' => Configuration::getConfig('admin', 'email'),
             'admin_telegram' => Configuration::getConfig('admin', 'telegram'),
             'bridges' => $body,

diff --git a/actions/SetBridgeCacheAction.php b/actions/SetBridgeCacheAction.php
@@ -23,7 +23,11 @@ public function execute(array $request)
 
         $bridgeFactory = new BridgeFactory();
 
-        $bridgeClassName = $bridgeFactory->createBridgeClassName($request['bridge'] ?? '');
+        $bridgeName = $request['bridge'] ?? null;
+        $bridgeClassName = $bridgeFactory->createBridgeClassName($bridgeName);
+        if (!$bridgeClassName) {
+            throw new \Exception(sprintf('Bridge not found: %s', $bridgeName));
+        }
 
         // whitelist control
         if (!$bridgeFactory->isEnabled($bridgeClassName)) {

diff --git a/bridges/AppleMusicBridge.php b/bridges/AppleMusicBridge.php
@@ -40,6 +40,8 @@ public function collectData()
 
         foreach ($json->results as $obj) {
             if ($obj->wrapperType === 'collection') {
+                $copyright = $obj->copyright ?? '';
+
                 $this->items[] = [
                     'title' => $obj->artistName . ' - ' . $obj->collectionName,
                     'uri' => $obj->collectionViewUrl,
@@ -49,7 +51,7 @@ public function collectData()
                     . '><img src="' . $obj->artworkUrl100 . '" /></a><br><br>'
                     . $obj->artistName . ' - ' . $obj->collectionName
                     . '<br>'
-                    . $obj->copyright,
+                    . $copyright,
                 ];
             }
         }

diff --git a/bridges/AskfmBridge.php b/bridges/AskfmBridge.php
@@ -37,7 +37,8 @@ public function collectData()
 
             $item['timestamp'] = strtotime($element->find('time', 0)->datetime);
 
-            $answer = trim($element->find('div.streamItem_content', 0)->innertext);
+            $var = $element->find('div.streamItem_content', 0);
+            $answer = trim($var->innertext ?? '');
 
             // This probably should be cleaned up, especially for YouTube embeds
             if ($visual = $element->find('div.streamItem_visual', 0)) {

diff --git a/bridges/CssSelectorBridge.php b/bridges/CssSelectorBridge.php
@@ -198,6 +198,9 @@ protected function htmlFindEntries($page, $url_selector, $url_pattern = '', $lim
             }
             if ($link->tag != 'a') {
                 $link = $link->find('a', 0);
+                if (is_null($link)) {
+                    continue;
+                }
             }
             $item['uri'] = $link->href;
             $item['title'] = $link->plaintext;
@@ -209,6 +212,10 @@ protected function htmlFindEntries($page, $url_selector, $url_pattern = '', $lim
             $link_to_item[$link->href] = $item;
         }
 
+        if (empty($link_to_item)) {
+            returnClientError('The provided URL selector matches some elements, but they do not contain links.');
+        }
+
         $links = $this->filterUrlList(array_keys($link_to_item), $url_pattern, $limit);
 
         if (empty($links)) {

diff --git a/bridges/EconomistBridge.php b/bridges/EconomistBridge.php
@@ -159,7 +159,7 @@ private function processContent($html, $elem)
             $svelte->parent->removeChild($svelte);
         }
         foreach ($elem->find('img') as $strange_img) {
-            if (!str_contains($strange_img->src, 'https://economist.com')) {
+            if (!str_contains($strange_img->src, 'economist.com')) {
                 $strange_img->src = 'https://economist.com' . $strange_img->src;
             }
         }

diff --git a/bridges/FeedReducerBridge.php b/bridges/FeedReducerBridge.php
@@ -23,8 +23,9 @@ class FeedReducerBridge extends FeedExpander
 
     public function collectData()
     {
-        if (preg_match('#^http(s?)://#i', $this->getInput('url'))) {
-            $this->collectExpandableDatas($this->getInput('url'));
+        $url = $this->getInput('url');
+        if (preg_match('#^http(s?)://#i', $url)) {
+            $this->collectExpandableDatas($url);
         } else {
             throw new Exception('URI must begin with http(s)://');
         }
@@ -35,21 +36,21 @@ public function getItems()
         $filteredItems = [];
         $intPercentage = (int)preg_replace('/[^0-9]/', '', $this->getInput('percentage'));
 
-        foreach ($this->items as $thisItem) {
+        foreach ($this->items as $item) {
             // The URL is included in the hash:
             //  - so you can change the output by adding a local-part to the URL
             //  - so items with the same URI in different feeds won't be correlated
 
             // $pseudoRandomInteger will be a 16 bit unsigned int mod 100.
             // This won't be uniformly distributed 1-100, but should be close enough.
 
-            $pseudoRandomInteger = unpack(
-                'S', // unsigned 16-bit int
-                hash('sha256', $thisItem['uri'] . '::' . $this->getInput('url'), true)
-            )[1] % 100;
+            $data = $item['uri'] . '::' . $this->getInput('url');
+            $hash = hash('sha256', $data, true);
+            // S = unsigned 16-bit int
+            $pseudoRandomInteger = unpack('S', $hash)[1] % 100;
 
             if ($pseudoRandomInteger < $intPercentage) {
-                $filteredItems[] = $thisItem;
+                $filteredItems[] = $item;
             }
         }
 

diff --git a/bridges/FilterBridge.php b/bridges/FilterBridge.php
@@ -12,7 +12,7 @@ class FilterBridge extends FeedExpander
         'url' => [
             'name' => 'Feed URL',
             'type'  => 'text',
-            'defaultValue' => 'https://lorem-rss.herokuapp.com/feed?unit=day',
+            'exampleValue' => 'https://lorem-rss.herokuapp.com/feed?unit=day',
             'required' => true,
         ],
         'filter' => [

diff --git a/bridges/FourchanBridge.php b/bridges/FourchanBridge.php
@@ -45,7 +45,8 @@ public function collectData()
             $file = $element->find('.file', 0);
 
             if (!empty($file)) {
-                $item['image'] = $element->find('.file a', 0)->href;
+                $var = $element->find('.file a', 0);
+                $item['image'] = $var->href ?? '';
                 $item['imageThumb'] = $element->find('.file img', 0)->src;
                 if (!isset($item['imageThumb']) and strpos($item['image'], '.swf') !== false) {
                     $item['imageThumb'] = 'http://i.imgur.com/eO0cxf9.jpg';

diff --git a/bridges/GettrBridge.php b/bridges/GettrBridge.php
@@ -27,9 +27,10 @@ class GettrBridge extends BridgeAbstract
 
     public function collectData()
     {
+        $user = $this->getInput('user');
         $api = sprintf(
             'https://api.gettr.com/u/user/%s/posts?offset=0&max=%s&dir=fwd&incl=posts&fp=f_uo',
-            $this->getInput('user'),
+            $user,
             min($this->getInput('limit'), 20)
         );
         $data = json_decode(getContents($api), false);

diff --git a/bridges/HeiseBridge.php b/bridges/HeiseBridge.php
@@ -103,6 +103,16 @@ class HeiseBridge extends FeedExpander
             'required' => false,
             'title' => 'Specify number of full articles to return',
             'defaultValue' => 5
+        ],
+        'sessioncookie' => [
+            'name' => 'Session Cookie',
+            'required' => false,
+            'title' => <<<'TITLE'
+                If you have a heise+ subscription,
+                you can enter your cookie (ssohls) here to
+                have heise+ articles displayed in full.
+                By default the cookie is 1 year valid.
+                TITLE,
         ]
     ]];
     const LIMIT = 5;
@@ -118,6 +128,7 @@ public function collectData()
     protected function parseItem($feedItem)
     {
         $item = parent::parseItem($feedItem);
+        $sessioncookie = $this->getInput('sessioncookie');
 
         // strip rss parameter
         $item['uri'] = explode('?', $item['uri'])[0];
@@ -128,13 +139,15 @@ protected function parseItem($feedItem)
         }
 
         // abort on heise+ articles and link to archive.ph for full-text content
-        if (str_starts_with($item['title'], 'heise+ |')) {
+        if ($sessioncookie == '' && str_starts_with($item['title'], 'heise+ |')) {
             $item['uri'] = 'https://archive.ph/?run=1&url=' . urlencode($item['uri']);
             return $item;
         }
 
         $item['uri'] .= '?seite=all';
-        $article = getSimpleHTMLDOMCached($item['uri']);
+        $article = getSimpleHTMLDOM($item['uri'], [
+            'cookie: ssohls=' . $sessioncookie
+        ]);
 
         if ($article) {
             $article = defaultLinkTo($article, $item['uri']);

diff --git a/bridges/NiusBridge.php b/bridges/NiusBridge.php
@@ -0,0 +1,32 @@
+<?php
+
+class NiusBridge extends XPathAbstract
+{
+    const NAME = 'Nius';
+    const URI = 'https://www.nius.de/news';
+    const DESCRIPTION = 'Die Stimme der Mehrheit';
+    const MAINTAINER = 'Niehztog';
+    //const PARAMETERS = array();
+    const CACHE_TIMEOUT = 3600;
+
+    const FEED_SOURCE_URL = 'https://www.nius.de/news';
+    const XPATH_EXPRESSION_ITEM = './/div[contains(@class, "compact-story") or contains(@class, "regular-story")]';
+    const XPATH_EXPRESSION_ITEM_TITLE = './/h2[@class="title"]//node()';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './/h2[@class="title"]//node()';
+    const XPATH_EXPRESSION_ITEM_URI = './/a[1]/@href';
+
+    const XPATH_EXPR_AUTHOR_PART1 = 'normalize-space(.//span[@class="author"]/text()[1])';
+    const XPATH_EXPR_AUTHOR_PART2 = 'normalize-space(.//span[@class="author"]/text()[2])';
+    const XPATH_EXPRESSION_ITEM_AUTHOR = 'substring-after(concat(' . self::XPATH_EXPR_AUTHOR_PART1 . ', " ", ' . self::XPATH_EXPR_AUTHOR_PART2 . '), " ")';
+
+    //const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/td[3]';
+    const XPATH_EXPRESSION_ITEM_ENCLOSURES = './/img[1]/@src';
+    const XPATH_EXPRESSION_ITEM_CATEGORIES = './/div[@class="subtitle"]/text()';
+    const SETTING_FIX_ENCODING = false;
+
+    protected function cleanMediaUrl($mediaUrl)
+    {
+        $result = preg_match('~https:\/\/www\.nius\.de\/_next\/image\?url=(.*)\?~', $mediaUrl, $matches);
+        return $result ? $matches[1] : $mediaUrl;
+    }
+}