From d1c224b48384ade65db45b8d12b5e8ddbdcda64f Mon Sep 17 00:00:00 2001
From: boyska <boyska@riseup.net>
Date: Fri, 15 Oct 2021 23:34:51 +0200
Subject: [PATCH] [FilterMore] NEW bridge for advanced filtering

---
 bridges/FilterMoreBridge.php | 157 +++++++++++++++++++++++++++++++++++
 1 file changed, 157 insertions(+)
 create mode 100644 bridges/FilterMoreBridge.php

diff --git a/bridges/FilterMoreBridge.php b/bridges/FilterMoreBridge.php
new file mode 100644
index 0000000..aa11251
--- /dev/null
+++ b/bridges/FilterMoreBridge.php
@@ -0,0 +1,157 @@
+<?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 = array(array(
+		'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',
+		),
+
+		'invert_filter' => array(
+			'name' => 'Invert filter result',
+			'type' => 'checkbox',
+			'required' => false,
+			'defaultValue' => false,
+		),
+	));
+
+	protected function parseItem($newItem){
+		$item = parent::parseItem($newItem);
+
+        $filters = ['filterByTitle', 'filterByBody', 'filterByAuthor', 'filterByDateNewer', 'filterByDateOlder'];
+        $results = [];
+
+        foreach($filters as $filter) {
+            $filter_res = $this->$filter($item);
+            if($filter_res === null) continue;
+            $results[] = $filter_res;
+        }
+        if(count($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 $item;
+        else
+            return null;
+	}
+
+    private function cmp($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 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());
+		}
+	}
+}
+