Add explicit CDATA guard cleanup

lib/nokolexbor/node.rb

@@ -22,6 +22,32 @@ class Node
 
     LOOKS_LIKE_XPATH = %r{^(\./|/|\.\.|\.$)}
 
+    CDATA_GUARD_PATTERNS = [
+      %r{\A(?<before>[[:space:]]*)<!\[CDATA\[(?<text>.*)\]\]>(?<after>[[:space:]]*)\z}m,
+      %r{\A(?<before>[[:space:]]*)//[[:blank:]]*<!\[CDATA\[(?<text>.*)//[[:blank:]]*\]\]>(?<after>[[:space:]]*)\z}m,
+      %r{\A(?<before>[[:space:]]*)/\*[[:blank:]]*<!\[CDATA\[[[:blank:]]*\*/(?<text>.*)/\*[[:blank:]]*\]\]>[[:blank:]]*\*/(?<after>[[:space:]]*)\z}m,
+      %r{\A(?<before>[[:space:]]*)/\*[[:blank:]]*<!\[CDATA\[[[:blank:]]*/\*[[:blank:]]*\*/(?<text>.*)/\*[[:blank:]]*\]\]>[[:blank:]]*/\*[[:blank:]]*\*/(?<after>[[:space:]]*)\z}m,
+    ].freeze
+    private_constant :CDATA_GUARD_PATTERNS
+
+    # Return this node's text with one complete legacy CDATA guard removed.
+    #
+    # This does not modify the node or change {#content}. Incomplete, mismatched,
+    # and embedded guards are returned unchanged. Whitespace is preserved; use
+    # +node.unwrap_cdata_text.strip+ to remove surrounding whitespace as well.
+    #
+    # @return [String]
+    def unwrap_cdata_text
+      raw_text = content
+
+      CDATA_GUARD_PATTERNS.each do |pattern|
+        match = pattern.match(raw_text)
+        return match[:before] + match[:text] + match[:after] if match
+      end
+
+      raw_text
+    end
+
     # @return true if this is a {Comment}
     def comment?
       type == COMMENT_NODE

spec/node_spec.rb

@@ -47,6 +47,67 @@
     _{ node.content = 1 }.must_raise TypeError
   end
 
+  describe 'unwrap_cdata_text' do
+    def script_with(content)
+      Nokolexbor::HTML("<script>#{content}</script>").at_css('script')
+    end
+
+    it 'removes recognized complete CDATA guards and preserves whitespace' do
+      wrapped_text = {
+        "<![CDATA[\n  payload\n]]>" => "\n  payload\n",
+        "//<![CDATA[\n  payload\n//]]>" => "\n  payload\n",
+        "/* <![CDATA[ */\n  payload\n/* ]]> */" => "\n  payload\n",
+        "/*<![CDATA[*/\n  payload\n/*]]>*/" => "\n  payload\n",
+        "/*<![CDATA[/* */\n  payload\n/*]]>/* */" => "\n  payload\n",
+      }
+
+      wrapped_text.each do |text, expected|
+        _(script_with(text).unwrap_cdata_text).must_equal expected
+      end
+    end
+
+    it 'preserves whitespace from the node content' do
+      node = script_with('')
+      node.content = " \t<![CDATA[payload]]>\r\n"
+
+      _(node.unwrap_cdata_text).must_equal " \tpayload\r\n"
+    end
+
+    it 'composes with String#strip for generic surrounding whitespace cleanup' do
+      text_values = [
+        "  payload  ",
+        "<![CDATA[\n  payload\n]]>",
+        "  /* <![CDATA[ */\n  payload\n/* ]]> */  ",
+      ]
+
+      text_values.each do |text|
+        _(script_with(text).unwrap_cdata_text.strip).must_equal 'payload'
+      end
+    end
+
+    it 'does not change the node content' do
+      node = script_with('<![CDATA[payload]]>')
+
+      _(node.unwrap_cdata_text).must_equal 'payload'
+      _(node.content).must_equal '<![CDATA[payload]]>'
+    end
+
+    it 'preserves text without a complete matching outer wrapper' do
+      text_values = [
+        "  payload  ",
+        '<![CDATA[payload',
+        'payload]]>',
+        '//<![CDATA[payload]]>',
+        'before <![CDATA[payload]]> after',
+        '/* <![CDATA[ */payload/*]]>/* */',
+      ]
+
+      text_values.each do |text|
+        _(script_with(text).unwrap_cdata_text).must_equal text
+      end
+    end
+  end
+
   describe 'attr' do
     before do
       @doc = Nokolexbor::HTML('<div class="a" checked></div>')