From 60dd22d3b3916f5113954fc1472b1658c3c4245f Mon Sep 17 00:00:00 2001
From: Alexandre Alapetite <alexandre@alapetite.fr>
Date: Tue, 22 Oct 2024 16:00:07 +0200
Subject: Allow regex parentheses (#6926)

* Allow regex parentheses
While waiting for a new better search parser, auto-escape parentheses in regex expressions to allow them like in `/^(ab|cd)/`

* Allow escaped parenthesis in regex

* A couple more tests
---
 app/Models/BooleanSearch.php | 15 +++++++++++++++
 app/Models/Search.php        |  1 +
 2 files changed, 16 insertions(+)

(limited to 'app')

diff --git a/app/Models/BooleanSearch.php b/app/Models/BooleanSearch.php
index 10c2321fb..749dbdfa5 100644
--- a/app/Models/BooleanSearch.php
+++ b/app/Models/BooleanSearch.php
@@ -37,6 +37,7 @@ class FreshRSS_BooleanSearch {
 		if ($level === 0) {
 			$input = $this->parseUserQueryNames($input, $allowUserQueries);
 			$input = $this->parseUserQueryIds($input, $allowUserQueries);
+			$input = self::escapeRegexParentheses($input);
 			$input = trim($input);
 		}
 
@@ -132,6 +133,20 @@ class FreshRSS_BooleanSearch {
 		return $input;
 	}
 
+	/**
+	 * Temporarily escape parentheses used in regex expressions.
+	 */
+	public static function escapeRegexParentheses(string $input): string {
+		return preg_replace_callback('#(?<=[\\s(:!-]|^)(?<![\\\\])/.*?(?<!\\\\)/[im]*#',
+			fn(array $matches): string => str_replace(['(', ')'], ['\\u0028', '\\u0029'], $matches[0]),
+			$input
+		) ?? '';
+	}
+
+	public static function unescapeRegexParentheses(string $input): string {
+		return str_replace(['\\u0028', '\\u0029'], ['(', ')'], $input);
+	}
+
 	/**
 	 * Example: 'ab cd OR ef OR "gh ij"' becomes '(ab cd) OR (ef) OR ("gh ij")'
 	 */
diff --git a/app/Models/Search.php b/app/Models/Search.php
index 0e2f5da6f..45fa742be 100644
--- a/app/Models/Search.php
+++ b/app/Models/Search.php
@@ -94,6 +94,7 @@ class FreshRSS_Search {
 	public function __construct(string $input) {
 		$input = self::cleanSearch($input);
 		$input = self::unescape($input);
+		$input = FreshRSS_BooleanSearch::unescapeRegexParentheses($input);
 		$this->raw_input = $input;
 
 		$input = $this->parseNotEntryIds($input);
-- 
cgit v1.2.3