[FilterMore] sort

Without this, every item has an enclosure with the same basename and
different query parameters.
This is correct, but some podcast readers (gPodder) mess it up, writing
every download to the same file.

bridges/AutoPodcasterBridge.php

@@ -0,0 +1,113 @@
+<?php
+class AutoPodcasterBridge extends FeedExpander {
+    const MAINTAINER='boyska';
+    const NAME='Auto Podcaster';
+    const URI = '';
+    const CACHE_TIMEOUT = 300; // 5 minuti
+    const DESCRIPTION='Make a "multimedia" podcast out of a normal feed';
+    const PARAMETERS = array('url' => array(
+        'url' => array(
+            'name' => 'URL',
+            'required' => true
+    )));
+
+    private function archiveIsAudioFormat($formatString) {
+        return strpos($formatString, 'MP3') !== false ||
+            strpos($formatString, 'Ogg') === 0;
+    }
+
+    private function extractAudio($dom) {
+        $audios = [];
+        foreach($dom->find('audio') as $audioEl) {
+            $sources = [];
+            if($audioEl->src !== false) {
+                $sources[] = $audioEl->src;
+            }
+            foreach($audioEl->find('source') as $sourceEl) {
+                $sources[] = $sourceEl->src;
+            }
+            if($sources) {
+                $audios[$sources[0]] = ['sources' => $sources];
+            }
+        }
+        return $audios;
+    }
+    private function extractIframeArchive($dom) {
+        $audios = [];
+
+        foreach($dom->find('iframe') as $iframeEl) {
+            if(strpos($iframeEl->src, "https://archive.org/embed/") === 0) {
+                $listURL = preg_replace("/\/embed\//", "/details/", $iframeEl->src, 1) . "?output=json";
+                $baseURL = preg_replace("/\/embed\//", "/download/", $iframeEl->src, 1);
+                $list = json_decode(file_get_contents($listURL));
+                $audios = [];
+                foreach($list->files as $name =>$data) {
+                    if($data->source === 'original' &&
+                        $this->archiveIsAudioFormat($data->format)) {
+                        $audios[$baseURL . $name] = ['sources' => [$baseURL . $name]];
+                    }
+                }
+                foreach($list->files as $name =>$data) {
+                    if($data->source === 'derivative' &&
+                        $this->archiveIsAudioFormat($data->format) &&
+                        isset($audios[$baseURL . "/" . $data->original])) {
+                        $audios[$baseURL . "/" . $data->original]['sources'][] = $baseURL . $name;
+                    }
+                }
+            }
+        }
+
+        return $audios;
+    }
+
+	protected function parseItem($newItem){
+		$item = parent::parseItem($newItem);
+
+		$dom = getSimpleHTMLDOMCached($item['uri']);
+        $audios = [];
+        if ($dom !== false) {
+            /* 1st extraction method: by "audio" tag */
+            $audios = array_merge($audios, $this->extractAudio($dom));
+
+            /* 2nd extraction method: by "iframe" tag */
+            $audios = array_merge($audios, $this->extractIframeArchive($dom));
+        }
+        elseif($item['content'] !== NULL) {
+            /* 1st extraction method: by "audio" tag */
+            $audios = array_merge($audios, $this->extractAudio(str_get_html($item['content'])));
+
+            /* 2nd extraction method: by "iframe" tag */
+            $audios = array_merge($audios,
+                $this->extractIframeArchive(str_get_html($item['content'])));
+        }
+
+        if(count($audios) === 0) {
+            return null;
+        }
+        $item['enclosures'] = array_values($audios);
+        $item['enclosures'] = [];
+        foreach(array_values($audios) as $audio) {
+            $item['enclosures'][] = $audio['sources'][0];
+        }
+        return $item;
+	}
+	public function collectData(){
+		if($this->getInput('url') && substr($this->getInput('url'), 0, strlen('http')) !== 'http') {
+			// just in case someone find a way to access local files by playing with the url
+			returnClientError('The url parameter must either refer to http or https protocol.');
+		}
+        $this->collectExpandableDatas($this->getURI());
+	}
+	public function getName(){
+		if(!is_null($this->getInput('url'))) {
+			return self::NAME . ' : ' . $this->getInput('url');
+		}
+
+		return parent::getName();
+	}
+	public function getURI(){
+		return $this->getInput('url');
+    }
+
+}
+

bridges/FilterMoreBridge.php

@@ -0,0 +1,241 @@
+<?php
+
+class FilterMoreBridge extends FeedExpander {
+
+	const MAINTAINER = 'boyska';
+	const NAME = 'FilterMore';
+	const CACHE_TIMEOUT = 2;
+	const DESCRIPTION = 'Filters a feed of your choice';
+	const URI = 'https://git.lattuta.net/boyska/rss-bridge';
+
+    const PARAMETERS = [
+        [
+		'url' => array(
+			'name' => 'Feed URL',
+			'required' => true,
+		),
+		'conj_type' => array(
+			'name' => 'Conjunction type type',
+			'type' => 'list',
+			'required' => false,
+			'values' => array(
+				'All conditions must be met' => 'and',
+				'Any condition must be met' => 'or',
+			),
+			'defaultValue' => 'permit',
+		),
+
+		'title_re' => array(
+			'name' => 'Filter item title (regular expression, see php.net/pcre_match for details)',
+			'required' => false,
+            'exampleValue' => '/breaking\ news/i',
+		),
+		'body_re' => array(
+			'name' => 'Filter body (regular expression)',
+			'required' => false,
+		),
+		'author_re' => array(
+			'name' => 'Filter author (regular expression)',
+			'required' => false,
+            'exampleValue' => '/(technology|politics)/i',
+		),
+		'newer_than' => array(
+			'name' => 'Filter date: ok if newer than the value (see php.net/strtotime for details)',
+			'required' => false,
+            'exampleValue' => '-14 days',
+		),
+		'older_than' => array(
+			'name' => 'Filter date: ok if older than the value (see php.net/strtotime for details)',
+			'required' => false,
+            'exampleValue' => '-1 hour',
+		),
+
+		'has_media' => array(
+			'name' => 'Has at least 1 media inside',
+			'type' => 'checkbox',
+			'required' => false,
+			'defaultValue' => false,
+		),
+
+		'invert_filter' => array(
+			'name' => 'Invert filter result',
+			'type' => 'checkbox',
+			'required' => false,
+			'defaultValue' => false,
+		),
+
+        'sort_by' => [
+            'name' => 'Sort by',
+            'type' => 'list',
+            'required' => true,
+            'values' => [
+                "Don't sort" => 'none',
+                'Date' => 'timestamp',
+                'Title' => 'title',
+                'Random' => 'random',
+            ],
+            'defaultValue' => 'date',
+        ],
+        'sort_dir' => [
+            'name' => 'Sort direction',
+            'type' => 'list',
+            'required' => true,
+            'values' => [
+                'Ascending' => 'asc',
+                'Descending' => 'desc',
+            ],
+            'defaultValue' => 'asc',
+        ],
+
+	]];
+
+	protected function parseItem($newItem){
+		$item = parent::parseItem($newItem);
+        $item['enclosures'] = [];
+        if(isset($newItem->enclosure)) {
+            foreach($newItem->enclosure as $encl) {
+                $serialized = [];
+                foreach($encl->attributes() as $key => $value) {
+                    $serialized[$key] = (string)$value;
+                }
+                $serialized["length"] = intval($serialized["length"]);
+                $item['enclosures'][] = $serialized;
+            }
+        }
+        if(isset($newItem->link)) {
+            foreach($newItem->link as $el) {
+                if(((string)$el['rel']) !== 'enclosure') continue;
+                $serialized = [];
+                $serialized['url'] = (string)$el['href'];
+
+                $item['enclosures'][] = $serialized;
+            }
+        }
+
+        $filters = ['filterByTitle', 'filterByBody', 'filterByAuthor', 'filterByDateNewer', 'filterByDateOlder', 'filterByMedia'];
+        $results = [];
+
+        foreach($filters as $filter) {
+            $filter_res = $this->$filter($item);
+            if($filter_res === null) continue;
+            $results[] = $filter_res;
+        }
+
+        $old_enclosures = $item['enclosures'];
+        $item['enclosures'] = [];
+        foreach($old_enclosures as $e) {
+            $item['enclosures'][] = $e['url'];
+        }
+        if(count($results) === 0) {
+            return $item;
+        }
+        if($this->getConjType() === 'and') {
+            $result = !in_array(false, $results);
+        } else { // or
+            $result = in_array(true, $results);
+        }
+        if($this->getInvertResult()) {
+            $result = !$result;
+        }
+        if($result)
+            return $item;
+        else
+            return null;
+	}
+
+    protected function sortItemKey($item) {
+        $sort_by = $this->getInput('sort_by');
+        $key = $item[$sort_by];
+        return $key;
+    }
+    public function collectExpandableDatas($url, $maxItems = -1){
+        parent::collectExpandableDatas($url, $maxItems);
+        if($this->getInput('sort_by') === 'random') {
+            shuffle($this->items);
+        } elseif($this->getInput('sort_by') !== 'none') {
+            usort($this->items, function($itemA, $itemB) {
+                $valA = $this->sortItemKey($itemA);
+                $valB = $this->sortItemKey($itemB);
+                $cmp = strcmp($valA, $valB);
+                return $cmp;
+            });
+        }
+        if($this->getInput('sort_dir') === 'desc')
+            $this->items = array_reverse($this->items);
+    }
+
+    private function cmp($a, $b) {
+        if($a > $b) return 1;
+        if($a < $b) return -1;
+        return 0;
+    }
+	private function filterByFieldRegexp($field, $re){
+        if($re === "") return null;
+        if(preg_match($re, $field)) {
+            return true;
+        }
+        return false;
+	}
+	protected function filterByTitle($item){
+		$re = $this->getInput('title_re');
+        return $this->filterByFieldRegexp($item['title'], $re);
+	}
+	protected function filterByBody($item){
+		$re = $this->getInput('body_re');
+        return $this->filterByFieldRegexp($item['content'], $re);
+	}
+	protected function filterByAuthor($item){
+		$re = $this->getInput('author_re');
+        return $this->filterByFieldRegexp($item['author'], $re);
+	}
+	private function filterByDate($item, $input, $expected){
+		$val = $this->getInput($input);
+        if($val === "") return null;
+        $ts = strtotime($val);
+        if($ts === false) {
+            throw new Exception("Invalid time specification: " . $val);
+        }
+        $cmp =  $this->cmp($item['timestamp'], $ts); // 1 if newer, -1 if older
+        return $cmp === $expected;
+	}
+	protected function filterByDateNewer($item){
+        return $this->filterByDate($item, 'newer_than', 1);
+	}
+	protected function filterByDateOlder($item){
+        return $this->filterByDate($item, 'older_than', -1);
+	}
+    protected function filterByMedia($item) {
+        if(!$this->getInput('has_media')) return null;
+        if(count($item['enclosures']) > 0) return true;
+        return false;
+    }
+
+	protected function getConjType(){
+		return $this->getInput('conj_type');
+	}
+	protected function getInvertResult(){
+		return $this->getInput('invert_filter');
+	}
+
+	public function getURI(){
+		$url = $this->getInput('url');
+
+		if(empty($url)) {
+			$url = parent::getURI();
+		}
+		return $url;
+	}
+
+	public function collectData(){
+		if($this->getInput('url') && substr($this->getInput('url'), 0, strlen('http')) !== 'http') {
+			// just in case someone find a way to access local files by playing with the url
+			returnClientError('The url parameter must either refer to http or https protocol.');
+		}
+		try{
+			$this->collectExpandableDatas($this->getURI());
+		} catch (HttpException $e) {
+			$this->collectExpandableDatas($this->getURI());
+		}
+	}
+}
+

bridges/RadioRaiBridge.php

@@ -0,0 +1,62 @@
+<?php
+class RadioRaiBridge extends BridgeAbstract {
+
+	const MAINTAINER = 'boyska';
+	const NAME = 'Radio Rai';
+	const URI = 'https://www.raiplayradio.it';
+	const CACHE_TIMEOUT = 900; // 15min
+	const DESCRIPTION = 'Segui le trasmissioni radio rai con feed/podcast valido';
+	const PARAMETERS = array( array(
+		'txname' => array(
+			'name' => 'txname',
+			'required' => true
+		)
+	));
+
+    private function getFinalURL($url) {
+        $ch = curl_init($url);
+        curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+        curl_setopt($ch, CURLOPT_NOBODY, true);
+        curl_setopt($ch, CURLOPT_HEADER, true);
+        $ret = curl_exec($ch);
+        if($ret === FALSE) {
+            return null;
+        }
+        $redirect = curl_getinfo($ch, CURLINFO_REDIRECT_URL);
+        if($redirect === false) return $url;
+        return $redirect;
+    }
+
+	public function collectData(){
+		$html = getSimpleHTMLDOM($this->getURI())
+			or returnServerError('No results for this query.');
+
+		foreach($html->find('[data-mediapolis]') as $episode) {
+            // var_dump($episode);
+            
+            $title = $episode->getAttribute('data-title');
+            if($title === FALSE) { continue; }
+            $audiourl = $episode->getAttribute('data-mediapolis');
+            $item = array();
+            $item['author'] = $this->getInput('txname');
+            $item['title'] = $title;
+            $item['content'] = $episode->plaintext;
+            $item['enclosures'] = [ $this::getFinalURL($audiourl) ];
+            $item['uri'] = $this::URI . $episode->getAttribute('data-href');
+
+            $this->items[] = $item;
+		}
+	}
+
+	public function getURI(){
+        return 'https://www.raiplayradio.it/programmi/' .  $this->getInput('txname') .  '/archivio/puntate/';
+	}
+
+	public function getName(){
+        if($this->getInput('txname')) {
+            return 'Radio Rai - ' . $this->getInput('txname');
+        }
+
+		return parent::getName();
+	}
+}

docker-compose.yaml

@@ -0,0 +1,16 @@
+version: "3"
+
+services:
+    php:
+        build:
+            context: .
+        environment:
+           - 'PHP_DISPLAY_ERRORS=0'
+           - 'PHP_LOG_ERRORS=1'
+           - 'PHP_ERROR_REPORTING=E_ALL'
+           - 'PHP_REPORT_MEMLEAKS=1'
+        volumes:
+            - ./:/app/public/
+        ports:
+            - "8001:80"
+