From 1262cc982ce4242b7bfcf14865d312381ddfe0bb Mon Sep 17 00:00:00 2001
From: hleskien <34342248+hleskien@users.noreply.github.com>
Date: Wed, 24 Jan 2024 19:12:38 +0100
Subject: [PATCH] added new bridges (#3920)

* added new bridges

* lint

---------

Co-authored-by: Dag <me@dvikan.no>
---
 .../BundesverbandFuerFreieKammernBridge.php   | 28 ++++++
 bridges/DeutscherAeroClubBridge.php           | 28 ++++++
 bridges/EDDHPiRepsBridge.php                  | 85 +++++++++++++++++++
 bridges/EDDHPresseschauBridge.php             | 42 +++++++++
 bridges/FliegermagazinBridge.php              | 22 +++++
 bridges/LogicMastersBridge.php                | 26 ++++++
 bridges/LuftfahrtBundesAmtBridge.php          | 41 +++++++++
 bridges/LuftsportSHBridge.php                 | 26 ++++++
 8 files changed, 298 insertions(+)
 create mode 100644 bridges/BundesverbandFuerFreieKammernBridge.php
 create mode 100644 bridges/DeutscherAeroClubBridge.php
 create mode 100644 bridges/EDDHPiRepsBridge.php
 create mode 100644 bridges/EDDHPresseschauBridge.php
 create mode 100644 bridges/FliegermagazinBridge.php
 create mode 100644 bridges/LogicMastersBridge.php
 create mode 100644 bridges/LuftfahrtBundesAmtBridge.php
 create mode 100644 bridges/LuftsportSHBridge.php

diff --git a/bridges/BundesverbandFuerFreieKammernBridge.php b/bridges/BundesverbandFuerFreieKammernBridge.php
new file mode 100644
index 00000000..147f2d47
--- /dev/null
+++ b/bridges/BundesverbandFuerFreieKammernBridge.php
@@ -0,0 +1,28 @@
+<?php
+
+class BundesverbandFuerFreieKammernBridge extends XPathAbstract
+{
+    const NAME = 'Bundesverband für freie Kammern e.V.';
+    const URI = 'https://www.bffk.de/aktuelles/aktuelle-nachrichten.html';
+    const DESCRIPTION = 'Aktuelle Nachrichten';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://www.bffk.de/aktuelles/aktuelle-nachrichten.html';
+    //const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="icon"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//ul[@class="article-list"]/li';
+    const XPATH_EXPRESSION_ITEM_TITLE = './/a/text()';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './/a/text()';
+    const XPATH_EXPRESSION_ITEM_URI = './/a/@href';
+    //const XPATH_EXPRESSION_ITEM_AUTHOR = './/';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/span/i';
+    //const XPATH_EXPRESSION_ITEM_ENCLOSURES = './';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    protected function formatItemTimestamp($value)
+    {
+        $value = trim($value, '()');
+        $dti = DateTimeImmutable::createFromFormat('d.m.Y', $value);
+        $dti = $dti->setTime(0, 0, 0);
+        return $dti->getTimestamp();
+    }
+}
diff --git a/bridges/DeutscherAeroClubBridge.php b/bridges/DeutscherAeroClubBridge.php
new file mode 100644
index 00000000..de98798d
--- /dev/null
+++ b/bridges/DeutscherAeroClubBridge.php
@@ -0,0 +1,28 @@
+<?php
+
+class DeutscherAeroClubBridge extends XPathAbstract
+{
+    const NAME = 'Deutscher Aero Club';
+    const URI = 'https://www.daec.de/news/';
+    const DESCRIPTION = 'News aus Luftsport und Dachverband';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://www.daec.de/news/';
+    const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="icon"][1]/@href';
+    const XPATH_EXPRESSION_ITEM = '//div[contains(@class, "news-list-view")]/div[contains(@class, "article")]';
+    const XPATH_EXPRESSION_ITEM_TITLE = './/span[@itemprop="headline"]';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './/div[@itemprop="description"]/p';
+    const XPATH_EXPRESSION_ITEM_URI = './/div[@class="news-header"]//a/@href';
+    //const XPATH_EXPRESSION_ITEM_AUTHOR = './/';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/time/@datetime';
+    const XPATH_EXPRESSION_ITEM_ENCLOSURES = './/img/@src';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    protected function formatItemTimestamp($value)
+    {
+        $dti = DateTimeImmutable::createFromFormat('Y-m-d', $value);
+        $dti = $dti->setTime(0, 0, 0);
+        return $dti->getTimestamp();
+    }
+}
+
diff --git a/bridges/EDDHPiRepsBridge.php b/bridges/EDDHPiRepsBridge.php
new file mode 100644
index 00000000..7d5e6a0b
--- /dev/null
+++ b/bridges/EDDHPiRepsBridge.php
@@ -0,0 +1,85 @@
+<?php
+
+class EDDHPiRepsBridge extends BridgeAbstract
+{
+    const NAME = 'EDDH.de PIREPs';
+    const URI = 'https://eddh.de/info/pireps_08days.php';
+    const DESCRIPTION = 'Erfahrungen und Tipps von Piloten für Piloten: Die Einträge der letzten 8 Tage';
+    const MAINTAINER = 'hleskien';
+    //const PARAMETERS = [];
+    //const CACHE_TIMEOUT = 3600;
+
+    public function collectData()
+    {
+        $dom = getSimpleHTMLDOM(self::URI);
+        foreach ($dom->find('table table table td') as $itemnode) {
+            $texts = $this->extractTexts($itemnode->find('text, br'));
+            $timestamp = $itemnode->find('.su_dat', 0)->innertext();
+            $uri = $itemnode->find('.pir_hd a', 0)->href;
+            $this->items[] = [
+                'timestamp' => $this->formatItemTimestamp($timestamp),
+                'title' => $this->formatItemTitle($texts),
+                'uri' => $this->formatItemUri($uri),
+                'author' => $this->formatItemAuthor($texts),
+                'content' => $this->formatItemContent($texts)
+            ];
+        }
+    }
+
+    public function getIcon()
+    {
+        return 'https://eddh.de/favicon.ico';
+    }
+
+    private function extractTexts($nodes)
+    {
+        $texts = [];
+        $i = 0;
+        foreach ($nodes as $node) {
+            $text = trim($node->outertext());
+            if ($node->tag == 'br') {
+                $texts[$i++] = "\n";
+            } elseif (($node->tag == 'text') && ($text != '')) {
+                $text = iconv('Windows-1252', 'UTF-8', $text);
+                $text = str_replace('&nbsp;', '', $text);
+                $texts[$i++] = $text;
+            }
+        }
+        return $texts;
+    }
+
+    protected function formatItemAuthor($texts)
+    {
+        $pos = array_search('Name:', $texts);
+        return $texts[$pos + 1];
+    }
+
+    protected function formatItemContent($texts)
+    {
+        $pos1 = array_search('Bemerkungen:', $texts);
+        $pos2 = array_search('Bewertung:', $texts);
+        $content = '';
+        for ($i = $pos1 + 1; $i < $pos2; $i++) {
+            $content .= $texts[$i];
+        }
+        return trim($content);
+    }
+
+    protected function formatItemTitle($texts)
+    {
+        $texts[5] = ltrim($texts[5], '(');
+        return implode(' ', [$texts[1], $texts[2], $texts[3], $texts[5]]);
+    }
+
+    protected function formatItemTimestamp($value)
+    {
+        $value = str_replace('Eintrag vom', '', $value);
+        $value = trim($value);
+        return strtotime($value);
+    }
+
+    protected function formatItemUri($value)
+    {
+        return 'https://eddh.de/info/' . $value;
+    }
+}
diff --git a/bridges/EDDHPresseschauBridge.php b/bridges/EDDHPresseschauBridge.php
new file mode 100644
index 00000000..93585684
--- /dev/null
+++ b/bridges/EDDHPresseschauBridge.php
@@ -0,0 +1,42 @@
+<?php
+
+class EDDHPresseschauBridge extends XPathAbstract
+{
+    const NAME = 'EDDH.de Presseschau';
+    const URI = 'https://eddh.de/presse/presseschau.php';
+    const DESCRIPTION = 'Luftfahrt-Presseschau: Presse-Artikel aus der Luftfahrt';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://eddh.de/presse/presseschau.php';
+    //const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="icon"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//table//table[.//p[@class="pressnews"]]//td';
+    const XPATH_EXPRESSION_ITEM_TITLE = './h4';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './p[@class="pressnews"]';
+    const XPATH_EXPRESSION_ITEM_URI = './p[@class="pressnews"]/a/@href';
+    const XPATH_EXPRESSION_ITEM_AUTHOR = './p[@class="quelle"]';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './p[@class="quelle"]';
+    //const XPATH_EXPRESSION_ITEM_ENCLOSURES = './';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    public function getIcon()
+    {
+        return 'https://eddh.de/favicon.ico';
+    }
+
+    protected function formatItemAuthor($value)
+    {
+        $parts = explode('(', $value);
+        $author = trim($parts[0]);
+        return $author;
+    }
+
+    protected function formatItemTimestamp($value)
+    {
+        $parts = explode('(', $value);
+        $ws = ["\n", "\t", ' ', ')'];
+        $value = str_replace($ws, '', $parts[1]);
+        $dti = DateTimeImmutable::createFromFormat('d.m.Y', $value);
+        $dti = $dti->setTime(0, 0, 0);
+        return $dti->getTimestamp();
+    }
+}
diff --git a/bridges/FliegermagazinBridge.php b/bridges/FliegermagazinBridge.php
new file mode 100644
index 00000000..84804e99
--- /dev/null
+++ b/bridges/FliegermagazinBridge.php
@@ -0,0 +1,22 @@
+<?php
+
+class FliegermagazinBridge extends XPathAbstract
+{
+    const NAME = 'fliegermagazin';
+    const URI = 'https://www.fliegermagazin.de/news-fuer-piloten/';
+    const DESCRIPTION = 'News für Piloten';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://www.fliegermagazin.de/news-fuer-piloten/';
+    const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="shortcut icon"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//article[@data-type="post"]';
+    const XPATH_EXPRESSION_ITEM_TITLE = './/h3/a/text()';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './/h3/a/text()';
+    const XPATH_EXPRESSION_ITEM_URI = './/h3/a/@href';
+    const XPATH_EXPRESSION_ITEM_AUTHOR = './/p[@class="author-field"]';
+    // Timestamp kann nur durch Laden des Artikels herausgefunden werden
+    //const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/span/i';
+    const XPATH_EXPRESSION_ITEM_ENCLOSURES = './/img/@src';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+}
+
diff --git a/bridges/LogicMastersBridge.php b/bridges/LogicMastersBridge.php
new file mode 100644
index 00000000..86b5bd6d
--- /dev/null
+++ b/bridges/LogicMastersBridge.php
@@ -0,0 +1,26 @@
+<?php
+
+class LogicMastersBridge extends XPathAbstract
+{
+    const NAME = 'Logic Masters Deutschland e.V.';
+    const URI = 'https://logic-masters.de/';
+    const DESCRIPTION = 'Aktuelles';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://logic-masters.de/';
+    //const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="SHORTCUT ICON"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//div[@class="aktuelles_eintrag"]';
+    const XPATH_EXPRESSION_ITEM_TITLE = './div[@class="aktuelles_titel"]';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './p';
+    //const XPATH_EXPRESSION_ITEM_URI = './a/@href';
+    //const XPATH_EXPRESSION_ITEM_AUTHOR = './/';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './div[@class="aktuelles_datum"]';
+    //const XPATH_EXPRESSION_ITEM_ENCLOSURES = './';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    protected function formatItemTimestamp($value)
+    {
+        $formatter = new IntlDateFormatter('de', IntlDateFormatter::LONG, IntlDateFormatter::NONE);
+        return $formatter->parse($value);
+    }
+}
\ No newline at end of file
diff --git a/bridges/LuftfahrtBundesAmtBridge.php b/bridges/LuftfahrtBundesAmtBridge.php
new file mode 100644
index 00000000..2b0384a2
--- /dev/null
+++ b/bridges/LuftfahrtBundesAmtBridge.php
@@ -0,0 +1,41 @@
+<?php
+
+class LuftfahrtBundesAmtBridge extends XPathAbstract
+{
+    const NAME = 'Luftfahrt-Bundesamt';
+    const URI = 'https://www.lba.de/DE/Home/Nachrichten/nachrichten_node.html';
+    const DESCRIPTION = 'alle Nachrichten: Liste aller Meldungen';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://www.lba.de/DE/Home/Nachrichten/nachrichten_node.html';
+    const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="shortcut icon"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//table/tbody/tr';
+    const XPATH_EXPRESSION_ITEM_TITLE = './td[2]/a/text()';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './td[2]/a/text()';
+    const XPATH_EXPRESSION_ITEM_URI = './td[2]/a/@href';
+    //const XPATH_EXPRESSION_ITEM_AUTHOR = './/';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './td[1]';
+    //const XPATH_EXPRESSION_ITEM_ENCLOSURES = './';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    protected function provideFeedIcon(\DOMXPath $xpath)
+    {
+        return parent::provideFeedIcon($xpath) . '?__blob=normal&v=3';
+    }
+
+    protected function formatItemTimestamp($value)
+    {
+        $value = trim($value);
+        $dti = DateTimeImmutable::createFromFormat('d.m.Y', $value);
+        $dti = $dti->setTime(0, 0, 0);
+        return $dti->getTimestamp();
+    }
+
+    // remove jsession part
+    protected function formatItemUri($value)
+    {
+        $parts = explode(';', $value);
+        return $parts[0];
+    }
+}
+
diff --git a/bridges/LuftsportSHBridge.php b/bridges/LuftsportSHBridge.php
new file mode 100644
index 00000000..66914c61
--- /dev/null
+++ b/bridges/LuftsportSHBridge.php
@@ -0,0 +1,26 @@
+<?php
+
+class LuftsportSHBridge extends XPathAbstract
+{
+    const NAME = 'Luftsportverband Schleswig-Holstein';
+    const URI = 'https://www.luftsport-sh.de/start.html';
+    const DESCRIPTION = 'Aktuelles vom Luftsportverband Schleswig-Holstein e.V.';
+    const MAINTAINER = 'hleskien';
+
+    const FEED_SOURCE_URL = 'https://www.luftsport-sh.de/start.html';
+    const XPATH_EXPRESSION_FEED_ICON = './/link[@rel="icon" and @sizes="16x16"]/@href';
+    const XPATH_EXPRESSION_ITEM = '//div[contains(@class, "mod_newslist")]/div';
+    const XPATH_EXPRESSION_ITEM_TITLE = './/*[@itemprop="name"]/a/text()';
+    const XPATH_EXPRESSION_ITEM_CONTENT = './/div[@itemprop="description"]/p/text()';
+    const XPATH_EXPRESSION_ITEM_URI = './h3/a/@href';
+    //const XPATH_EXPRESSION_ITEM_AUTHOR = './/';
+    const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/time/@datetime';
+    const XPATH_EXPRESSION_ITEM_ENCLOSURES = './/img/@src';
+    //const XPATH_EXPRESSION_ITEM_CATEGORIES = './/';
+
+    protected function formatItemTimestamp($value)
+    {
+        $dti = DateTimeImmutable::createFromFormat(DateTimeInterface::ATOM, $value);
+        return $dti->getTimestamp();
+    }
+}