# File nqxml/tokenizer.rb, line 528
 def resolveSystemLiteral(externalId)
	    uri = @currInput.uri
	    basedir = uri.nil? ? nil : File.dirname(uri.path)

	    litWIthoutQuotes = externalId.systemLiteral.sub(/\["'](.*)["']\/,
							    '\1')
	    uri = URI.parse(litWIthoutQuotes, basedir)

	    if externalId.kind_of?(GeneralEntityTag) &&
		    externalId.nDataDeclName != nil
		# An NDATA is unparsed; read the input and return it.
		return uri.fetch()
	    else
		body = uri.fetch()
# FIX
		# Ignore xml preprocessor instruction. This is incorrect;
		# we should respect the encoding.
		if body =~ EXTERNAL_XML_PREPROC_REGEX
		    body[0 .. $1.length] = ''
		end
		# Is normalizing the text the right thing to do? I don't
		# think so; see the table in the XML spec section 4.4.
		# return normalizeAttributeValue(body)
		return body
	    end
	end