Parse Unicode surrogate pairs in Encoding::JS.unescape (closes #519).

lib/ronin/support/encoding/js.rb

@@ -188,6 +188,11 @@ def self.unescape(data)
           until scanner.eos?
             unescaped << if (backslash_escape = scanner.scan(/\\[btnfr'"\\]/))
                            BACKSLASHED_CHARS[backslash_escape]
+                         elsif (surrogate_pair = scanner.scan(/\\u[dD][890abAB][0-9a-fA-F]{2}\\u[dD][cdefCDEF][0-9a-fA-F]{2}/))
+                           hi = surrogate_pair[2..6].to_i(16)
+                           lo = surrogate_pair[8..12].to_i(16)
+
+                           (0x1_0000 + ((hi - 0xd800) * 0x400) + (lo - 0xdc00))
                          elsif (unicode_escape = scanner.scan(/[\\%]u[0-9a-fA-F]{1,4}/))
                            unicode_escape[2..].to_i(16)
                          elsif (hex_escape     = scanner.scan(/[\\%][0-9a-fA-F]{1,2}/))

spec/encoding/js_spec.rb

@@ -1,6 +1,8 @@
 require 'spec_helper'
 require 'ronin/support/encoding/js'
 
+require 'json'
+
 describe Ronin::Support::Encoding::JS do
   describe ".escape_byte" do
     context "when given a byte that maps to a special character" do
@@ -97,6 +99,12 @@
       expect(subject.unescape(js_unicode)).to eq(data)
     end
 
+    it "must unescape Unicode surrogate pair characters" do
+      expect(subject.unescape("\\uD83D\\uDE80")).to eq(
+        JSON.parse("\"\\uD83D\\uDE80\"")
+      )
+    end
+
     it "must unescape JavaScript hex characters" do
       expect(subject.unescape(js_hex)).to eq(data)
     end