From bed389b5ec5c09c67475e093e23325ecefbfb3fd Mon Sep 17 00:00:00 2001
From: Terentev Aleksey <terentev555@gmail.com>
Date: Tue, 29 Aug 2023 17:13:06 +0500
Subject: [PATCH] feat: sanitize links in alt of img tag

https://jira.railsc.ru/browse/PC4-29908
---
 lib/string_tools.rb       |  9 +++++++++
 spec/string_tools_spec.rb | 12 ++++++++++++
 2 files changed, 21 insertions(+)

diff --git a/lib/string_tools.rb b/lib/string_tools.rb
index 30291da..175d981 100644
--- a/lib/string_tools.rb
+++ b/lib/string_tools.rb
@@ -199,6 +199,7 @@ def call(env)
           normalize_link node, 'href'
         when 'img'
           normalize_link node, 'src'
+          remove_links node, 'alt'
         end
       end
 
@@ -210,6 +211,14 @@ def normalize_link(node, attr_name)
       rescue Addressable::URI::InvalidURIError
         node.swap node.children
       end
+
+      def remove_links(node, attr_name)
+        return unless node[attr_name]
+
+        node[attr_name] = node[attr_name].gsub(URI::DEFAULT_PARSER.make_regexp, '').squish
+
+        node.remove_attribute(attr_name) if node[attr_name].empty?
+      end
     end
 
     class IframeNormalizer
diff --git a/spec/string_tools_spec.rb b/spec/string_tools_spec.rb
index 1d6b7f2..ff239fc 100644
--- a/spec/string_tools_spec.rb
+++ b/spec/string_tools_spec.rb
@@ -80,6 +80,18 @@
       expect(sanitized_string).to eq('<style type="text/css">body{color: red;}</style>')
     end
 
+    it 'removes links in alt attribute of img tag' do
+      origin_str = '<img scr="http://test.test" alt="http://test.test test https://test.test alt">'
+      sanitized_string = described_class.sanitize(origin_str, 'img' => %w(scr alt))
+      expect(sanitized_string).to eq('<img scr="http://test.test" alt="test alt">')
+    end
+
+    it 'removes alt attribute of img tag if empty value' do
+      origin_str = '<img scr="http://test.test" alt="http://test.test">'
+      sanitized_string = described_class.sanitize(origin_str, 'img' => %w(scr alt))
+      expect(sanitized_string).to eq('<img scr="http://test.test">')
+    end
+
     context 'multiple invocations of the method' do
       it 'does not mess up default config' do
         origin_str = '<p style="text-align: center;" title="foobar"></p>'