lib/marc4j4r/reader.rb in marc4j4r-0.2.4 vs lib/marc4j4r/reader.rb in marc4j4r-0.9.0

@@ -11,36 +11,32 @@
   end  
 end
 
 
 module MARC4J4R
-  # Add some sugar to the MarcReader interface
-  #
-  # Adjust the interface so that a #new call to any implementations that
-  # implement it can take a java.io.InputStream, ruby IO obejct, or String
-  # (that will be interpreted as a filename) without complaining.
-  #
-  # The mechanism -- running module_eval on a string-representation of the 
-  # new method in each of the hard-coded implementations of MarcReader
-  # (MarcStreamReader,MarcPermissiveStreamReader,MarcXmlReader) -- is ugly
-  # and deeply unsettling.
-  #
-  # @author Bill Dueber
-
+ 
   # First, add Enumerable to the interface  
   Java::org.marc4j.MarcReader.module_eval("include Enumerable")
   
-  
   class Reader
-      
+
+    ENCODINGS = ['UTF-8', 'ISO-8859-1', 'MARC-8']
+    ENCODING_ALIASES = {:utf8 => 'UTF-8', :marc8 => 'MARC-8', :iso => 'ISO-8859-1'}
+    attr_reader :handle  
+
   # Get a marc reader of the appropriate type
   # @param [String, IO, java.io.InputStream] input The IO stream (or filename) from which you want to read
   # @param [:strictmarc, :permissivemarc, :marcxml] The type of MARC reader you want. 
+  # @param [:utf8, :iso, :marc8, 'UTF-8', 'ISO-8859-1', 'MARC-8'] An explicit encoding
   # @return [MarcReader] A MarcReader object with the syntactic sugar added in this file (e.g, each)
   #
   # @example Get a strict binary MARC reader for the file 'test.mrc'
   #  reader = MARC4J4R::Reader.new('test.mrc')
+  #  reader = MARC4J4R::Reader.new('test.mrc', :strictmarc) # same thing; :strictmarc is the default
+  # 
+  # @example Get a strict binary MARC reader for the file 'test.mrc', force input to be treated as utf-8
+  #  reader = MARC4J4R::Reader.new('test.mrc', :strictmarc, :utf8)
   #
   # @example Get a permissive binary MARC reader
   #  reader = MARC4J4R::Reader.new('test.mrc', :permissivemarc)
   #
   # @example Get a reader for an xml file
@@ -49,17 +45,23 @@
   # @example Get a reader based on an existing IO object
   #   require 'open-uri'
   #   infile = open('http://my.machine.com/test.mrc')
   #   reader = MARC4J4R::Reader.new(infile)
 
-    attr_reader :handle  
-    def self.new(input, type = :strictmarc)
+    def self.new(input, type = :strictmarc, encoding = nil)
+      if encoding
+        encoding = ENCODING_ALIASES[encoding] if ENCODING_ALIASES[encoding]
+        unless ENCODINGS.include? encoding
+          raise ArgumentError, "Encoding must be in [#{ENCODINGS.map {|x| '"' + x + '"'}.join(', ')}], not \"#{encoding}\""
+        end
+      end
       @handle = IOConvert.byteinstream(input)
       case type
       when :strictmarc then
-        return Java::org.marc4j.MarcStreamReader.new(@handle)
+        return Java::org.marc4j.MarcStreamReader.new(@handle, encoding)
       when :permissivemarc then
-        return Java::org.marc4j.MarcPermissiveStreamReader.new(@handle, true, true)
+        encoding ||= 'BESTGUESS'
+        return Java::org.marc4j.MarcPermissiveStreamReader.new(@handle, true, true, encoding)
       when :marcxml then
         return Java::org.marc4j.MarcXmlReader.new(@handle)
       when :alephsequential then
         return MARC4J4R::AlephSequentialReader.new(@handle)
       else
\ No newline at end of file