From 7d1b9ca5627bbc89231a51c8c98e55c9c9786b2f Mon Sep 17 00:00:00 2001
From: Aidan Woods <aidantwoods@gmail.com>
Date: Mon, 11 Oct 2021 20:09:57 +0100
Subject: [PATCH] Better autolinks

This doesn't follow gfm spec yet, work lifted from my code over
at
https://github.com/Parsemd/Parsemd-PHP/blob/8505e2737e43f89c09fc9597889e56a6f8e114b9/src/Parsers/GitHubFlavor/Inlines/AutoLink.php

Fixes #505
Fixes #717
---
 src/Components/Inlines/Url.php  | 34 ++++++++++++++++++++++++++-------
 tests/data/url_autolinking.html |  8 +++++++-
 tests/data/url_autolinking.md   | 14 +++++++++++++-
 3 files changed, 47 insertions(+), 9 deletions(-)

diff --git a/src/Components/Inlines/Url.php b/src/Components/Inlines/Url.php
index 19f818e..8a84492 100644
--- a/src/Components/Inlines/Url.php
+++ b/src/Components/Inlines/Url.php
@@ -14,6 +14,9 @@ final class Url implements BacktrackingInline
 {
     use WidthTrait;
 
+    private const ABSOLUTE_URI = '[a-z][a-z0-9+.-]{1,31}:[^\s[:cntrl:]<>]*';
+    private const NO_TRAILING_PUNCT = '(?<![?!.,:*_~])';
+
     /** @var string */
     private $url;
 
@@ -38,14 +41,31 @@ final class Url implements BacktrackingInline
      */
     public static function build(Excerpt $Excerpt, State $State)
     {
-        if (\preg_match(
-            '/(?<=^|\s|[*_~(])https?+:[\/]{2}[^\s<]+\b\/*+/ui',
-            $Excerpt->context(),
-            $matches,
-            \PREG_OFFSET_CAPTURE
-        )) {
+        // this needs some work to follow spec
+        if (
+            \preg_match(
+                '/'.self::ABSOLUTE_URI.self::NO_TRAILING_PUNCT.'/iu',
+                $Excerpt->context(),
+                $matches,
+                \PREG_OFFSET_CAPTURE
+            )
+        ) {
             /** @var array{0: array{string, int}} $matches */
-            return new self($matches[0][0], \intval($matches[0][1]));
+            $url = $matches[0][0];
+            $position = \intval($matches[0][1]);
+
+            if (\preg_match('/[)]++$/', $url, $matches)) {
+                $trailingParens = \strlen($matches[0]);
+
+                $openingParens = \substr_count($url, '(');
+                $closingParens = \substr_count($url, ')');
+
+                if ($closingParens > $openingParens) {
+                    $url = \substr($url, 0, -\min($trailingParens, $closingParens - $openingParens));
+                }
+            }
+
+            return new self($url, $position);
         }
 
         return null;
diff --git a/tests/data/url_autolinking.html b/tests/data/url_autolinking.html
index 8f937db..334355c 100644
--- a/tests/data/url_autolinking.html
+++ b/tests/data/url_autolinking.html
@@ -1,3 +1,9 @@
 <p>an autolink <a href="http://example.com">http://example.com</a></p>
 <p>inside of parentheses (<a href="http://example.com">http://example.com</a>)</p>
-<p>trailing slash <a href="http://example.com/">http://example.com/</a> and <a href="http://example.com/path/">http://example.com/path/</a></p>
\ No newline at end of file
+<p><a href="http://www.google.com/search?q=Markup+(business)">http://www.google.com/search?q=Markup+(business)</a></p>
+<p><a href="http://www.google.com/search?q=Markup+(business)">http://www.google.com/search?q=Markup+(business)</a>))</p>
+<p>(<a href="http://www.google.com/search?q=Markup+(business)">http://www.google.com/search?q=Markup+(business)</a>)</p>
+<p>(<a href="http://www.google.com/search?q=Markup+(business)">http://www.google.com/search?q=Markup+(business)</a></p>
+<p>trailing slash <a href="http://example.com/">http://example.com/</a> and <a href="http://example.com/path/">http://example.com/path/</a></p>
+<p>trailing paren  <a href="https://www.owasp.org/index.php/Cross-site_Scripting_(XSS)">https://www.owasp.org/index.php/Cross-site_Scripting_(XSS)</a></p>
+<p>complex link <a href="http://elk.canda.biz/app/kibana#/discover?_g=()&amp;_a=(columns:!(_source),index:&#039;deve-*&#039;,interval:auto,query:(query_string:(analyze_wildcard:!t,query:&#039;*&#039;)),sort:!(&#039;@timestamp&#039;,desc))">http://elk.canda.biz/app/kibana#/discover?_g=()&amp;_a=(columns:!(_source),index:'deve-*',interval:auto,query:(query_string:(analyze_wildcard:!t,query:'*')),sort:!('@timestamp',desc))</a></p>
\ No newline at end of file
diff --git a/tests/data/url_autolinking.md b/tests/data/url_autolinking.md
index 09b467b..825f17e 100644
--- a/tests/data/url_autolinking.md
+++ b/tests/data/url_autolinking.md
@@ -2,4 +2,16 @@ an autolink http://example.com
 
 inside of parentheses (http://example.com)
 
-trailing slash http://example.com/ and http://example.com/path/
\ No newline at end of file
+http://www.google.com/search?q=Markup+(business)
+
+http://www.google.com/search?q=Markup+(business)))
+
+(http://www.google.com/search?q=Markup+(business))
+
+(http://www.google.com/search?q=Markup+(business)
+
+trailing slash http://example.com/ and http://example.com/path/
+
+trailing paren  https://www.owasp.org/index.php/Cross-site_Scripting_(XSS)
+
+complex link http://elk.canda.biz/app/kibana#/discover?_g=()&_a=(columns:!(_source),index:'deve-*',interval:auto,query:(query_string:(analyze_wildcard:!t,query:'*')),sort:!('@timestamp',desc))
\ No newline at end of file