util.rb in right_develop-2.1.5

- old
+ new

@@ -1,5 +1,9 @@
+if RUBY_VERSION =~ /^1\.8/
+  require 'iconv'
+end
+
 module RightDevelop::CI
   module Util
     module_function
 
     # Regular expression used to determine which characters of a string are allowed
@@ -12,10 +16,15 @@
     JAVA_PACKAGE_SEPARATOR = '.'
 
     # Replacement codepoint that looks a bit like a period
     JAVE_PACKAGE_SEPARATOR_HOMOGLYPH = '&#xb7;'
 
+    # Regular expression that matches characters that need to be escaped inside CDATA
+    # c.f. http://www.w3.org/TR/xml11/#charsets
+    # RestrictedChar ::= [#x1-#x8] | [#xB-#xC] | [#xE-#x1F] | [#x7F-#x84] | [#x86-#x9F]
+    INVALID_CDATA_CHARACTER = Regexp.new '[\x01-\x08\x0b-\x0c\x0e-\x1f\x7f-\x84\x86-\x9f]', nil, 'n' # Ruby 1.8-2.1 compatible
+
     # Make a string suitable for parsing by Jenkins JUnit display plugin by escaping any non-valid
     # Java class name characters as an XML entity. This prevents Jenkins from interpreting "hi1.2"
     # as a package-and-class name.
     #
     # @param [String] name
@@ -33,8 +42,30 @@
           result << "&#x#{chr};"
         end
       end
 
       result
+    end
+
+    # Strip invalid UTF-8 sequences from a string and entity-escape any character that can't legally
+    # appear inside XML CDATA. If test output contains weird data, we could end up generating
+    # invalid JUnit XML which will choke Java. Preserve the purity of essence of our precious XML
+    # fluids!
+    #
+    # @return [String] the input with all invalid UTF-8 replaced by the empty string
+    # @param [String] untrusted a string (of any encoding) that might contain invalid UTF-8 sequences
+    def purify(untrusted)
+      # First pass: strip bad UTF-8 characters
+      if RUBY_VERSION =~ /^1\.8/
+        iconv = Iconv.new('UTF-8//IGNORE', 'UTF-8')
+        result = iconv.iconv(untrusted)
+      else
+        result = untrusted.force_encoding(Encoding::BINARY).encode('UTF-8', :undef=>:replace, :replace=>'')
+      end
+
+      # Second pass: entity escape characters that can't appear in XML CDATA.
+      result.gsub(INVALID_CDATA_CHARACTER) do |ch|
+        "&#x%s;" % [ch.unpack('H*').first]
+      end
     end
   end
 end
\ No newline at end of file