From 5cc42b18b0e6ced675af446d0d38fd3a354a25f9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Florian=20M=C3=BCllner?= Date: Tue, 27 Feb 2018 13:20:02 +0100 Subject: [PATCH] utils: Simplify URL regex to only support one layer of parentheses The author of the original URL-matching regex warns[0] that the pattern may cause certain regex engines to lock up with certain input, namely patterns that contain parentheses. It turns out SpiderMonkey is affected, but rather than switching to the author's improved version (that is still crazy), sim- plify the pattern a bit by removing support for nested parentheses in URLs. Even a single pair of parentheses is extremely rare, so this is unlikely to make a noticeable difference (other than not locking up SpiderMonkey of course) ... [0] http://daringfireball.net/2010/07/improved_regex_for_matching_urls --- js/misc/util.js | 2 +- tests/unit/url.js | 6 +++++- 2 files changed, 6 insertions(+), 2 deletions(-) diff --git a/js/misc/util.js b/js/misc/util.js index 1d73d2076..38b67e891 100644 --- a/js/misc/util.js +++ b/js/misc/util.js @@ -17,7 +17,7 @@ const Params = imports.misc.params; var SCROLL_TIME = 0.1; // http://daringfireball.net/2010/07/improved_regex_for_matching_urls -const _balancedParens = '\\((?:[^\\s()<>]+|(?:\\(?:[^\\s()<>]+\\)))*\\)'; +const _balancedParens = '\\([^\\s()<>]+\\)'; const _leadingJunk = '[\\s`(\\[{\'\\"<\u00AB\u201C\u2018]'; const _notTrailingJunk = '[^\\s`!()\\[\\]{};:\'\\".,<>?\u00AB\u00BB\u201C\u201D\u2018\u2019]'; diff --git a/tests/unit/url.js b/tests/unit/url.js index 251725ca6..158316bb4 100644 --- a/tests/unit/url.js +++ b/tests/unit/url.js @@ -38,6 +38,10 @@ const tests = [ output: [ { url: 'http://www.gnome.org:99/port', pos: 10 } ] }, { input: 'This is an ftp://www.gnome.org/ test.', output: [ { url: 'ftp://www.gnome.org/', pos: 11 } ] }, + { input: 'https://www.gnome.org/(some_url,_with_very_unusual_characters)', + output: [ { url: 'https://www.gnome.org/(some_url,_with_very_unusual_characters)', pos: 0 } ] }, + { input: 'https://www.gnome.org/(some_url_with_unbalanced_parenthesis', + output: [ { url: 'https://www.gnome.org/', pos: 0 } ] }, { input: 'Visit http://www.gnome.org/ and http://developer.gnome.org', output: [ { url: 'http://www.gnome.org/', pos: 6 }, @@ -68,4 +72,4 @@ for (let i = 0; i < tests.length; i++) { JsUnit.assertEquals('Test ' + i + ', match ' + j + ' position', match[j].pos, tests[i].output[j].pos); } -} \ No newline at end of file +}