Fix poor srcset parsing

lib/html_proofer/element.rb

@@ -82,10 +82,25 @@ def multiple_srcsets?
       !blank?(srcset) && srcset.split(",").size > 1
     end
 
+    # From https://github.com/sindresorhus/srcset/blob/f7c48acd7facf18e94dec47e6b96e84e0f0e69dc/index.js#LL1-L16C71
+    # This regex represents a loose rule of an “image candidate string”; see https://html.spec.whatwg.org/multipage/images.html#srcset-attribute
+    # An “image candidate string” roughly consists of the following:
+    # 1. Zero or more whitespace characters.
+    # 2. A non-empty URL that does not start or end with `,`.
+    # 3. Zero or more whitespace characters.
+    # 4. An optional “descriptor” that starts with a whitespace character.
+    # 5. Zero or more whitespace characters.
+    # 6. Each image candidate string is separated by a `,`.
+    # We intentionally implement a loose rule here so that we can perform more aggressive error handling and reporting in the below code.
+
+    IMAGE_CANDIDATE_REGEX = /\s*([^,]\S*[^,](?:\s+[^,]+)?)\s*(?:,|$)/
+
     def srcsets
       return nil if blank?(srcset)
 
-      srcset.split(",").map(&:strip)
+      srcset.split(IMAGE_CANDIDATE_REGEX).select.with_index do |_part, idx|
+        idx.odd?
+      end.map(&:strip)
     end
 
     def multiple_sizes?

lib/html_proofer/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module HTMLProofer
-  VERSION = "5.0.2"
+  VERSION = "5.0.3"
 end

spec/html-proofer/fixtures/images/base64_srcset.html

@@ -0,0 +1 @@
+<source media="(min-width:650px)" srcset="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAACEAAAAhCAYAAABX5MJvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAK8AAACvABQqw0mAAAABZ0RVh0Q3JlYXRpb24gVGltZQAwMi8yMi8xMq6ePc8AAAAgdEVYdFNvZnR3YXJlAE1hY3JvbWVkaWEgRmlyZXdvcmtzIE1Yu5EqJAAABOhJREFUeJzFVltIVGsU">

spec/html-proofer/images_spec.rb

@@ -236,6 +236,12 @@
     expect(proofer.failed_checks.first.description).to(eq("internal image /uploads/150-marie-lloyd.jpg does not exist"))
   end
 
+  it "supports base64 srcsets" do
+    relative_images = File.join(FIXTURES_DIR, "images", "base64_srcset.html")
+    proofer = run_proofer(relative_images, :file)
+    expect(proofer.failed_checks).to(eq([]))
+  end
+
   it "works for images with a swapped data attribute src" do
     custom_data_src_check = "#{FIXTURES_DIR}/images/data_src_attribute.html"
     proofer = run_proofer(custom_data_src_check, :file, swap_attributes: { "img" => [["data-src", "src"]] })