#!/usr/bin/ruby -w
# Test of character encoding when moving from ruby-zoom to ruby-marc via XML.
# First, we see the record as ruby-zoom puts it.
# Then the record if we give the xml method no arguments.
# Finally if we ask yaz to convert from marc8 to utf8


require 'rubygems'
require 'zoom'
require 'UniversalDetector' #gem install chardet
require 'marc'
require 'stringio'

zservers = [ ['142.51.8.7', 2200, 'unicorn']] #dbs tells me they have no utf8 records


zservers.each do |zserver|
    puts zserver
    conn = ZOOM::Connection.new
    conn.preferred_record_syntax = 'USMARC'
    conn.set_option('charset', 'UTF-8')
    conn.connect(zserver[0], zserver[1])
    conn.database_name = zserver[2]
    rset = conn.search('@attr 1=4 "Revue philosophique de la France et de"')
    
    #how ruby-zoom sees it
    puts "the record how ruby-zoom sees it"
    puts rset[0]
    charset =  UniversalDetector.chardet(rset[0].xml) 
    puts charset.inspect
    puts
    puts "-----------------------------------"
    puts
    
    #if we give the xml method no values 
    record_no_conversion = MARC::XMLReader.new(StringIO.new(string=rset[0].xml))
    record_no_conversion = record_no_conversion.to_a
    puts record_no_conversion[0]
    marc_charset = UniversalDetector.chardet(record_no_conversion[0].to_s) 
    puts marc_charset.inspect
    puts
    puts "-----------------------------------"
    puts
    
    record_utf8 = MARC::XMLReader.new(StringIO.new(string=rset[0].xml('MARC8','UTF8')))
    record_utf8 = record_utf8.to_a
    puts record_utf8[0]
    marc_charset = UniversalDetector.chardet(record_utf8[0].to_s) 
    puts marc_charset.inspect
end