From 75cc52a62c5f5e6ec13b54c2e06bdce035814880 Mon Sep 17 00:00:00 2001
From: ORelio <ORelio@users.noreply.github.com>
Date: Mon, 7 Jun 2021 20:11:12 +0200
Subject: [PATCH] [FilterBridge] Various improvements (#2148)

- Add option for case-insensitive regex
- Allow matching item content or author in addition to item title
- Optionally attempt to convert encoding when applying matches
---
 bridges/FilterBridge.php | 124 +++++++++++++++++++++++++--------------
 1 file changed, 81 insertions(+), 43 deletions(-)

diff --git a/bridges/FilterBridge.php b/bridges/FilterBridge.php
index 696b100f..ea4b9742 100644
--- a/bridges/FilterBridge.php
+++ b/bridges/FilterBridge.php
@@ -2,11 +2,11 @@
 
 class FilterBridge extends FeedExpander {
 
-	const MAINTAINER = 'Frenzie';
+	const MAINTAINER = 'Frenzie, ORelio';
 	const NAME = 'Filter';
 	const CACHE_TIMEOUT = 3600; // 1h
 	const DESCRIPTION = 'Filters a feed of your choice';
-	const URI = 'https://github.com/rss-bridge/rss-bridge';
+	const URI = 'https://github.com/RSS-Bridge/rss-bridge';
 
 	const PARAMETERS = array(array(
 		'url' => array(
@@ -14,7 +14,7 @@ class FilterBridge extends FeedExpander {
 			'required' => true,
 		),
 		'filter' => array(
-			'name' => 'Filter item title (regular expression)',
+			'name' => 'Filter (regular expression)',
 			'required' => false,
 		),
 		'filter_type' => array(
@@ -22,60 +22,101 @@ class FilterBridge extends FeedExpander {
 			'type' => 'list',
 			'required' => false,
 			'values' => array(
-				'Permit' => 'permit',
-				'Block' => 'block',
+				'Keep matching items' => 'permit',
+				'Hide matching items' => 'block',
 			),
 			'defaultValue' => 'permit',
 		),
-		'title_from_content' => array(
-			'name' => 'Generate title from content',
+		'case_insensitive' => array(
+			'name' => 'Case-insensitive filter',
 			'type' => 'checkbox',
 			'required' => false,
-		)
+		),
+		'fix_encoding' => array(
+			'name' => 'Attempt Latin1/UTF-8 fixes when evaluating filter',
+			'type' => 'checkbox',
+			'required' => false,
+		),
+		'target_title' => array(
+			'name' => 'Apply filter on title',
+			'type' => 'checkbox',
+			'required' => false,
+			'defaultValue' => 'checked'
+		),
+		'target_content' => array(
+			'name' => 'Apply filter on content',
+			'type' => 'checkbox',
+			'required' => false,
+		),
+		'target_author' => array(
+			'name' => 'Apply filter on author',
+			'type' => 'checkbox',
+			'required' => false,
+		),
+		'title_from_content' => array(
+			'name' => 'Generate title from content (overwrite existing title)',
+			'type' => 'checkbox',
+			'required' => false,
+		),
+		'length_limit' => array(
+			'name' => 'Max length analyzed by filter (-1: no limit)',
+			'type' => 'number',
+			'required' => false,
+			'defaultValue' => -1,
+		),
 	));
 
 	protected function parseItem($newItem){
 		$item = parent::parseItem($newItem);
 
+		// Generate title from first 50 characters of content?
 		if($this->getInput('title_from_content') && array_key_exists('content', $item)) {
-
 			$content = str_get_html($item['content']);
-
 			$pos = strpos($item['content'], ' ', 50);
-
-			$item['title'] = substr(
-				$content->plaintext,
-				0,
-				$pos
-			);
-
+			$item['title'] = substr($content->plaintext, 0, $pos);
 			if(strlen($content->plaintext) >= $pos) {
 				$item['title'] .= '...';
 			}
-
 		}
 
-		switch(true) {
-		case $this->getFilterType() === 'permit':
-			if (preg_match($this->getFilter(), $item['title'])) {
-				return $item;
-			}
-			break;
-		case $this->getFilterType() === 'block':
-			if (!preg_match($this->getFilter(), $item['title'])) {
-				return $item;
-			}
-			break;
+		// Build regular expression
+		$regex = '/' . $this->getInput('filter') . '/';
+		if($this->getInput('case_insensitive')) {
+			$regex .= 'i';
 		}
-		return null;
-	}
 
-	protected function getFilter(){
-		return '/' . $this->getInput('filter') . '/';
-	}
+		// Retrieve fields to check
+		$filter_fields = array();
+		if($this->getInput('target_title')) {
+			$filter_fields[] = $item['title'];
+		}
+		if($this->getInput('target_content')) {
+			$filter_fields[] = $item['content'];
+		}
+		if($this->getInput('target_author')) {
+			$filter_fields[] = $item['author'];
+		}
 
-	protected function getFilterType(){
-		return $this->getInput('filter_type');
+		// Apply filter on item
+		$keep_item = false;
+		$length_limit = intval($this->getInput('length_limit'));
+		foreach($filter_fields as $field) {
+			if($length_limit > 0) {
+				$field = substr($field, 0, $length_limit);
+			}
+			$keep_item |= boolval(preg_match($regex, $field));
+			if($this->getInput('fix_encoding')) {
+				$keep_item |= boolval(preg_match($regex, utf8_decode($field)));
+				$keep_item |= boolval(preg_match($regex, utf8_encode($field)));
+			}
+		}
+
+		// Reverse result? (keep everything but matching items)
+		if($this->getInput('filter_type') === 'block') {
+			$keep_item = !$keep_item;
+		}
+
+		return $keep_item ? $item : null;
 	}
 
 	public function getURI(){
@@ -84,18 +125,15 @@ class FilterBridge extends FeedExpander {
 		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
+		if($this->getInput('url') && substr($this->getInput('url'), 0, 4) !== 'http') {
+			// just in case someone finds 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 (Exception $e) {
-			$this->collectExpandableDatas($this->getURI());
-		}
+		$this->collectExpandableDatas($this->getURI());
 	}
 }