oai 0.0.6 → 0.0.7

data/README

@@ -76,5 +76,5 @@ Where x.y.z is the version of the gem that was generated.
 
 == AUTHORS
 
-- Ed Summers <ehs@pobox>
+- Ed Summers <ehs@pobox.com>
 - William Groppe <will.groppe@gmail.com>

data/Rakefile

@@ -1,4 +1,4 @@
-RUBY_OAI_VERSION = '0.0.6'
+RUBY_OAI_VERSION = '0.0.7'
 
 require 'rubygems'
 require 'rake'
@@ -25,8 +25,6 @@ spec = Gem::Specification.new do |s|
     s.bindir = 'bin'
     s.executables = 'oai'
 
-    s.add_dependency('activesupport', '>=1.3.1')
-    s.add_dependency('chronic', '>=0.0.3')
     s.add_dependency('builder', '>=2.0.0')
     
     s.files = %w(README Rakefile) +

data/lib/oai/client.rb

@@ -155,7 +155,7 @@ module OAI
     def do_request(verb, opts = nil)
       # fire off the request and return appropriate DOM object
       uri = build_uri(verb, opts)
-      xml = get(uri)
+      xml = strip_invalid_utf_8_chars(get(uri))
       if @parser == 'libxml' 
         # remove default namespace for oai-pmh since libxml
         # isn't able to use our xpaths to get at them 
@@ -176,15 +176,16 @@ module OAI
     
     def encode(value)
       return CGI.escape(value) unless value.respond_to?(:strftime)
-      if value.respond_to?(:to_time) # Usually a DateTime or Time
-        value.to_time.utc.xmlschema
+      if value.kind_of?(DateTime)
+        Time.parse(value.asctime).utc.xmlschema
+      elsif value.kind_of?(Time)
+        value.utc.xmlschema
       else # Assume something date like
         value.strftime('%Y-%m-%d')
       end
     end
 
     def load_document(xml)
-      retried = false
       case @parser
       when 'libxml'
         begin
@@ -192,28 +193,13 @@ module OAI
           parser.string = xml
           return parser.parse
         rescue XML::Parser::ParseError => e
-          if retried
-            raise OAI::Exception, 'response not well formed XML: '+e, caller
-          end
-          ic = Iconv.new('UTF-8//IGNORE', 'UTF-8')
-          xml2 = ic.iconv(xml << ' ')[0..-2]
-          puts "equal? #{xml == xml2}"
-          retried = true
-          retry
+          raise OAI::Exception, 'response not well formed XML: '+e, caller
         end
       when 'rexml'
         begin
           return REXML::Document.new(xml)
         rescue REXML::ParseException => e
-          if retried
-            puts xml
-            raise OAI::Exception, 'response not well formed XML: '+e, caller
-          end
-          puts "RETRYING"
-          ic = Iconv.new('UTF-8//IGNORE', 'UTF-8')
-          xml = ic.iconv(xml << ' ')[0..-2]
-          retried = true
-          retry
+          raise OAI::Exception, 'response not well formed XML: '+e.message, caller
         end
       end
     end
@@ -288,13 +274,27 @@ module OAI
     def parse_date(value)
       return value if value.respond_to?(:strftime)
       
-      # Oddly Chronic doesn't parse an UTC encoded datetime.  
-      # Luckily Time does
-      dt = Chronic.parse(value) || Time.parse(value)
-      raise OAI::ArgumentError.new unless dt
-      
-      dt.utc
+      Date.parse(value) # This will raise an exception for badly formatted dates
+      Time.parse(value).utc # Sadly, this will not
+    rescue
+      raise OAI::ArgumentError.new 
     end
     
+    # Strip out invalid UTF-8 characters.  Regex from the W3C, inverted.
+    # http://www.w3.org/International/questions/qa-forms-utf-8.en.php
+    #
+    # Regex is from WebCollab: 
+    #   http://webcollab.sourceforge.net/unicode.html
+    def strip_invalid_utf_8_chars(xml)
+      simple_bytes = xml.gsub(/[\x00-\x08\x10\x0B\x0C\x0E-\x19\x7F]
+                             | [\x00-\x7F][\x80-\xBF]+
+                             | ([\xC0\xC1]|[\xF0-\xFF])[\x80-\xBF]*
+                             | [\xC2-\xDF]((?![\x80-\xBF])|[\x80-\xBF]{2,})
+                             | [\xE0-\xEF](([\x80-\xBF](?![\x80-\xBF]))
+                             | (?![\x80-\xBF]{2})|[\x80-\xBF]{3,})/x, '?')
+      simple_bytes.gsub(/\xE0[\x80-\x9F][\x80-\xBF]
+                       | \xED[\xA0-\xBF][\x80-\xBF]/,'?')
+    end
+        
   end
 end

data/lib/oai/harvester.rb

@@ -9,6 +9,7 @@ require 'readline'
 require 'chronic'
 require 'socket'
 
+require 'oai/client'
 require 'oai/harvester/config'
 require 'oai/harvester/harvest'
 require 'oai/harvester/logging'

data/lib/oai/harvester/shell.rb

@@ -47,8 +47,8 @@ module OAI
             end
           rescue 
             puts "Not a recognized command, or bad options.  Type 'help' for clues."
-            #puts $!
-            #puts $!.backtrace.join("\n")
+            puts $!
+            puts $!.backtrace.join("\n")
           end
         end
       end

data/lib/oai/provider.rb

@@ -1,6 +1,6 @@
-require 'active_support'
+require 'rexml/document'
+require 'singleton'
 require 'builder'
-require 'chronic'
 
 if not defined?(OAI::Const::VERBS)
   require 'oai/exception'
@@ -266,8 +266,8 @@ module OAI::Provider
       Response::GetRecord.new(self.class, options).to_xml
     end
     
-    #  xml_response = process_verb('ListRecords', :from => 'October', 
-    #    :until => 'November') # thanks Chronic!
+    #  xml_response = process_verb('ListRecords', :from => 'October 1, 2005', 
+    #    :until => 'November 1, 2005') 
     #
     # If you are implementing a web interface using process_request is the
     # preferred way.

data/lib/oai/provider/metadata_format.rb

@@ -1,3 +1,5 @@
+require 'singleton'
+
 module OAI::Provider::Metadata
   # == Metadata Base Class
   #
@@ -54,10 +56,9 @@ module OAI::Provider::Metadata
     def value_for(field, record, map)
       method = map[field] ? map[field].to_s : field.to_s
       
-      methods = record.public_methods(false)
-      if methods.include?(method.pluralize)
-        record.send method.pluralize
-      elsif methods.include?(method)
+      if record.respond_to?(pluralize(method))
+        record.send pluralize(method)
+      elsif record.respond_to?(method)
         record.send method
       else
         []
@@ -68,6 +69,47 @@ module OAI::Provider::Metadata
     def header_specification
       raise NotImplementedError.new
     end
+    
+    # Shamelessly lifted form ActiveSupport.  Thanks Rails community!
+    def pluralize(word)
+      # Use ActiveSupports pluralization if it's available.
+      return word.pluralize if word.respond_to?(:pluralize)
+      
+      # Otherwise use our own simple pluralization rules.
+      result = word.to_s.dup
+      
+      # Uncountable words
+      return result if %w(equipment information rice money species series fish sheep).include?(result)
+      
+      # Irregular words
+      { 'person' => 'people', 'man' => 'men', 'child' => 'children', 'sex' => 'sexes', 
+        'move' => 'moves', 'cow' => 'kine' }.each { |k,v| return v if word == k }
+      
+      rules.each { |(rule, replacement)| break if result.gsub!(rule, replacement) }
+      result
+    end
+    
+    def rules
+      [
+        [/$/, 's'],
+        [/s$/i, 's'],
+        [/(ax|test)is$/i, '\1es'],
+        [/(octop|vir)us$/i, '\1i'],
+        [/(alias|status)$/i, '\1es'],
+        [/(bu)s$/i, '\1ses'],
+        [/(buffal|tomat)o$/i, '\1oes'],
+        [/([ti])um$/i, '\1a'],
+        [/sis$/i, 'ses'],
+        [/(?:([^f])fe|([lr])f)$/i, '\1\2ves'],
+        [/(hive)$/i, '\1s'],
+        [/([^aeiouy]|qu)y$/i, '\1ies'],
+        [/(x|ch|ss|sh)$/i, '\1es'],
+        [/(matr|vert|ind)(?:ix|ex)$/i, '\1ices'],
+        [/([m|l])ouse$/i, '\1ice'],
+        [/^(ox)$/i, '\1en'],
+        [/(quiz)$/i, '\1zes']
+      ]
+    end
 
   end
   

data/lib/oai/provider/model/activerecord_caching_wrapper.rb

@@ -103,7 +103,6 @@ module OAI::Provider
       end
       
       oaitoken = OaiToken.find_by_token(token.to_s)
-      
       raise ResumptionTokenException.new unless oaitoken
 
       PartialResult.new(

data/lib/oai/provider/model/activerecord_wrapper.rb

@@ -20,8 +20,8 @@ module OAI::Provider
       @limit = options.delete(:limit)
       
       unless options.empty?
-        raise ArgumentException.new(
-          "Unsupported options [#{options.join(', ')}]"
+        raise ArgumentError.new(
+          "Unsupported options [#{options.keys.join(', ')}]"
         )
       end
     end

data/lib/oai/provider/response.rb

@@ -94,12 +94,10 @@ module OAI
     def parse_date(value)
       return value if value.respond_to?(:strftime)
       
-      # Oddly Chronic doesn't parse an UTC encoded datetime.  
-      # Luckily Time does
-      dt = Chronic.parse(value) || Time.parse(value)
-      raise OAI::ArgumentError.new unless dt
-      
-      dt.utc
+      Date.parse(value) # This will raise an exception for badly formatted dates
+      Time.parse(value).utc # Sadly, this will not
+    rescue
+      raise OAI::ArgumentError.new 
     end
     
     def internalize(hash = {})

data/lib/oai/provider/response/identify.rb

@@ -8,9 +8,11 @@ module OAI::Provider::Response
           r.repositoryName provider.name
           r.baseURL provider.url
           r.protocolVersion 2.0
-          provider.email.each do |address|
-            r.adminEmail address
-          end if provider.email
+          if provider.email and provider.email.respond_to?(:each)
+            provider.email.each { |address| r.adminEmail address }
+          else
+            r.adminEmail provider.email.to_s
+          end
           r.earliestDatestamp provider.model.earliest
           r.deletedRecord provider.delete_support.to_s
           r.granularity provider.granularity

data/lib/oai/xpath.rb

@@ -5,7 +5,7 @@ module OAI
     def xpath_all(doc, path)
       case parser_type(doc)
       when 'libxml'
-        return doc.find(path)
+        return doc.find(path).to_a if doc.find(path)
       when 'rexml'
         return REXML::XPath.match(doc, path)
       end

data/test/activerecord_provider/helpers/set_provider.rb

@@ -32,5 +32,5 @@ class ARSetProvider < OAI::Provider::Base
   repository_name 'ActiveRecord Set Based Provider'
   repository_url 'http://localhost'
   record_prefix = 'oai:test'
-  source_model SetModel.new(DCField)
+  source_model SetModel.new(DCField, :timestamp_field => 'date')
 end

data/test/activerecord_provider/tc_ar_provider.rb

@@ -13,7 +13,8 @@ class ActiveRecordProviderTest < Test::Unit::TestCase
   end
   
   def test_metadata_formats_for_record
-    assert_nothing_raised { REXML::Document.new(@provider.list_metadata_formats(:identifier => 'oai:test/1')) }
+    record_id = DCField.find(:first).id
+    assert_nothing_raised { REXML::Document.new(@provider.list_metadata_formats(:identifier => "oai:test/#{record_id}")) }
     doc =  REXML::Document.new(@provider.list_metadata_formats)
     assert doc.elements['/OAI-PMH/ListMetadataFormats/metadataFormat/metadataPrefix'].text == 'oai_dc'
   end
@@ -31,25 +32,29 @@ class ActiveRecordProviderTest < Test::Unit::TestCase
   end
 
   def test_get_record
-    assert_nothing_raised { REXML::Document.new(@provider.get_record(:identifier => 'oai:test/1')) }
-    doc = REXML::Document.new(@provider.get_record(:identifier => 'oai:test/1'))
-    assert_equal 'oai:test/1', doc.elements['OAI-PMH/GetRecord/record/header/identifier'].text
+    record_id = DCField.find(:first).id
+    assert_nothing_raised { REXML::Document.new(@provider.get_record(:identifier => "oai:test/#{record_id}")) }
+    doc = REXML::Document.new(@provider.get_record(:identifier => "#{record_id}"))
+    assert_equal "oai:test/#{record_id}", doc.elements['OAI-PMH/GetRecord/record/header/identifier'].text
   end
   
   def test_deleted
-    DCField.update(5, :deleted => true)
-    doc = REXML::Document.new(@provider.get_record(:identifier => 'oai:test/5'))
-    assert_equal 'oai:test/5', doc.elements['OAI-PMH/GetRecord/record/header/identifier'].text
+    record = DCField.find(:first)
+    record.deleted = true;
+    record.save
+    doc = REXML::Document.new(@provider.get_record(:identifier => "oai:test/#{record.id}"))
+    assert_equal "oai:test/#{record.id}", doc.elements['OAI-PMH/GetRecord/record/header/identifier'].text
     assert_equal 'deleted', doc.elements['OAI-PMH/GetRecord/record/header'].attributes["status"]
   end
   
   def test_from
-    DCField.update_all(['updated_at = ?', Chronic.parse("January 1 2005")],
-      "id < 90")
-    DCField.update_all(['updated_at = ?', Chronic.parse("June 1 2005")],
-      "id < 10")
+    first_id = DCField.find(:first, :order => "id asc").id
+    DCField.update_all(['updated_at = ?', Time.parse("January 1 2005")],
+      "id < #{first_id + 90}")
+    DCField.update_all(['updated_at = ?', Time.parse("June 1 2005")],
+      "id < #{first_id + 10}")
     
-    from_param = Chronic.parse("January 1 2006")
+    from_param = Time.parse("January 1 2006")
     
     doc = REXML::Document.new(
       @provider.list_records(:from => from_param)
@@ -58,31 +63,33 @@ class ActiveRecordProviderTest < Test::Unit::TestCase
       doc.elements['OAI-PMH/ListRecords'].size
 
     doc = REXML::Document.new(
-      @provider.list_records(:from => Chronic.parse("May 30 2005"))
+      @provider.list_records(:from => Time.parse("May 30 2005"))
       )
     assert_equal 20, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end
   
   def test_until
-    DCField.update_all(['updated_at = ?', Chronic.parse("June 1 2005")],
-      "id < 10")
+    first_id = DCField.find(:first, :order => "id asc").id
+    DCField.update_all(['updated_at = ?', Time.parse("June 1 2005")],
+      "id < #{first_id + 10}")
 
     doc = REXML::Document.new(
-      @provider.list_records(:until => Chronic.parse("June 1 2005"))
+      @provider.list_records(:until => Time.parse("June 1 2005"))
       )
-    assert_equal 9, doc.elements['OAI-PMH/ListRecords'].to_a.size
+    assert_equal 10, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end
   
   def test_from_and_until
-    DCField.update_all(['updated_at = ?', Chronic.parse("June 1 2005")])
-    DCField.update_all(['updated_at = ?', Chronic.parse("June 15 2005")],
-      "id < 50")
-    DCField.update_all(['updated_at = ?', Chronic.parse("June 30 2005")],
-      "id < 10")
+    first_id = DCField.find(:first, :order => "id asc").id
+    DCField.update_all(['updated_at = ?', Time.parse("June 1 2005")])
+    DCField.update_all(['updated_at = ?', Time.parse("June 15 2005")],
+      "id < #{first_id + 50}")
+    DCField.update_all(['updated_at = ?', Time.parse("June 30 2005")],
+      "id < #{first_id + 10}")
 
     doc = REXML::Document.new(
-      @provider.list_records(:from => Chronic.parse("June 3 2005"),
-        :until => Chronic.parse("June 16 2005"))
+      @provider.list_records(:from => Time.parse("June 3 2005"),
+        :until => Time.parse("June 16 2005"))
       )
     assert_equal 40, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end

data/test/activerecord_provider/tc_ar_sets_provider.rb

@@ -25,7 +25,8 @@ class ActiveRecordSetProviderTest < Test::Unit::TestCase
   end
   
   def test_record_with_multiple_sets
-    assert_equal 2, DCField.find(32).sets.size
+    record = DCSet.find(:first, :conditions => "spec = 'C'").dc_fields.first
+    assert_equal 2, record.sets.size
   end
 
   def setup
@@ -43,24 +44,29 @@ class ActiveRecordSetProviderTest < Test::Unit::TestCase
   def define_sets
     set_a = DCSet.create(:name => "Set A", :spec => "A")
     set_b = DCSet.create(:name => "Set B", :spec => "B")
-    set_c = DCSet.create(:name => "Set B", :spec => "B")
+    set_c = DCSet.create(:name => "Set C", :spec => "C")
     set_ab = DCSet.create(:name => "Set A:B", :spec => "A:B")
   
-    DCField.find([1,2,3,4,5,6,7,8,9,10]).each do |record|
+    next_id = 0
+    DCField.find(:all, :limit => 10, :order => "id asc").each do |record|
       set_a.dc_fields << record
+      next_id = record.id
     end
     
-    DCField.find([11,12,13,14,15,16,17,18,19,20]).each do |record|
+    DCField.find(:all, :limit => 10, :order => "id asc", :conditions => "id > #{next_id}").each do |record|
       set_b.dc_fields << record
+      next_id = record.id
     end
 
-    DCField.find([21,22,23,24,25,26,27,28,29,30]).each do |record|
+    DCField.find(:all, :limit => 10, :order => "id asc", :conditions => "id > #{next_id}").each do |record|
       set_ab.dc_fields << record
+      next_id = record.id
     end
   
-    DCField.find([31,32,33,34,35,36,37,38,39,40]).each do |record|
+    DCField.find(:all, :limit => 10, :order => "id asc", :conditions => "id > #{next_id}").each do |record|
       set_a.dc_fields << record
       set_c.dc_fields << record
+      next_id = record.id
     end
   end
 end

data/test/activerecord_provider/tc_caching_paging_provider.rb

@@ -5,40 +5,41 @@ class CachingPagingProviderTest < Test::Unit::TestCase
   
   def test_full_harvest
     doc = Document.new(@provider.list_records)
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].size
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].size
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].size
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_nil doc.elements["/OAI-PMH/resumptionToken"]
+    assert_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
     assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
   end
   
   def test_from_and_until
-    DCField.update_all(['updated_at = ?', Chronic.parse("September 15 2005")],
-      "id <= 25")
-    DCField.update_all(['updated_at = ?', Chronic.parse("November 1 2005")],
-      "id <= 50 and id > 25")
-    
+    first_id = DCField.find(:first, :order => "id asc").id
+    DCField.update_all(['updated_at = ?', Time.parse("September 15 2005")],
+      "id <= #{first_id + 25}")
+    DCField.update_all(['updated_at = ?', Time.parse("November 1 2005")],
+      "id < #{first_id + 50} and id > #{first_id + 25}")
+
     # Should return 50 records broken into 2 groups of 25.
     doc = Document.new(
       @provider.list_records(
-        :from => Chronic.parse("September 1 2005"),
-        :until => Chronic.parse("November 30 2005"))
+        :from => Time.parse("September 1 2005"),
+        :until => Time.parse("November 30 2005"))
       )
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].size
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
     doc = Document.new(@provider.list_records(:resumption_token => token))
     assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].size
-    assert_nil doc.elements["/OAI-PMH/resumptionToken"]
+    assert_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
   end
 
   def setup

data/test/activerecord_provider/tc_simple_paging_provider.rb

@@ -5,42 +5,43 @@ class SimpleResumptionProviderTest < Test::Unit::TestCase
   
   def test_full_harvest
     doc = Document.new(@provider.list_records)
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].to_a.size
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].to_a.size
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].to_a.size
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].to_a.size
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
-    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].to_a.size
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].to_a.size
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_nil doc.elements["/OAI-PMH/resumptionToken"]
+    assert_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
     assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].to_a.size
   end
   
   def test_from_and_until
-    DCField.update_all(['updated_at = ?', Chronic.parse("September 15 2005")],
-      "id < 26")
-    DCField.update_all(['updated_at = ?', Chronic.parse("November 1 2005")],
-      "id < 51 and id > 25")
+    first_id = DCField.find(:first, :order => "id asc").id
+    DCField.update_all(['updated_at = ?', Time.parse("September 15 2005")],
+      "id < #{first_id + 25}")
+    DCField.update_all(['updated_at = ?', Time.parse("November 1 2005")],
+      "id <= #{first_id + 50} and id > #{first_id + 25}")
     
-    total = DCField.count(:id, :conditions => ["updated_at >= ? AND updated_at <= ?", Chronic.parse("September 1 2005"), Chronic.parse("November 30 2005")])
+    total = DCField.count(:id, :conditions => ["updated_at >= ? AND updated_at <= ?", Time.parse("September 1 2005"), Time.parse("November 30 2005")])
 
     # Should return 50 records broken into 2 groups of 25.
     doc = Document.new(
       @provider.list_records(
-        :from => Chronic.parse("September 1 2005"),
-        :until => Chronic.parse("November 30 2005"))
+        :from => Time.parse("September 1 2005"),
+        :until => Time.parse("November 30 2005"))
       )
-    assert_equal total/2, doc.elements["/OAI-PMH/ListRecords"].to_a.size
-    assert_not_nil doc.elements["/OAI-PMH/resumptionToken"]
-    token = doc.elements["/OAI-PMH/resumptionToken"].text
+    assert_equal 26, doc.elements["/OAI-PMH/ListRecords"].to_a.size
+    assert_not_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
+    token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text
     doc = Document.new(@provider.list_records(:resumption_token => token))
-    assert_equal total/2, doc.elements["/OAI-PMH/ListRecords"].to_a.size
-    assert_nil doc.elements["/OAI-PMH/resumptionToken"]
+    assert_equal 25, doc.elements["/OAI-PMH/ListRecords"].to_a.size
+    assert_nil doc.elements["/OAI-PMH/ListRecords/resumptionToken"]
   end
 
   def setup

data/test/client/tc_exception.rb

@@ -2,9 +2,9 @@ require 'test_helper'
 
 class ExceptionTest < Test::Unit::TestCase
 
-  def test_http_error
-    client = OAI::Client.new 'http://www.example.com'
-    assert_raises(OAI::Exception) { client.identify }
+  def test_not_found
+    client = OAI::Client.new 'http://www.google.com'
+    assert_raises(ArgumentError) { client.identify }
   end
 
   def test_xml_error

data/test/client/tc_list_records.rb

@@ -1,6 +1,7 @@
 require 'test_helper'
 
 class GetRecordsTest < Test::Unit::TestCase
+
   def test_get_records
     client = OAI::Client.new 'http://localhost:3333/oai'
     response = client.list_records 

data/test/client/tc_low_resolution_dates.rb

@@ -0,0 +1,14 @@
+require 'test_helper'
+
+class LowResolutionDatesTest < Test::Unit::TestCase
+
+  def test_low_res_date_parsing
+    client = OAI::Client.new 'http://caltechcstr.library.caltech.edu/perl/oai2' 
+
+    date = Date.new 2003, 1, 1
+    
+    # get a list of identifier headers
+    assert_nothing_raised { client.list_identifiers :from => date } 
+  end
+  
+end

data/test/client/tc_utf8_escaping.rb

@@ -0,0 +1,11 @@
+require 'test_helper'
+
+class UTF8Test < Test::Unit::TestCase
+  
+  def test_escaping_invalid_utf_8_characters
+    client = OAI::Client.new 'http://localhost:3333/oai' #, :parser => 'libxml'
+    invalid_utf_8 = [2, 3, 4, 104, 5, 101, 6, 108, 66897, 108, 66535, 111, 1114112, 33, 55234123, 33].pack("U*")
+    assert_equal("hello!!", client.send(:strip_invalid_utf_8_chars, invalid_utf_8).gsub(/\?/, ''))
+  end
+
+end

data/test/client/tc_xpath.rb

@@ -13,7 +13,7 @@ class XpathTest < Test::Unit::TestCase
   def test_libxml
     begin 
       require 'xml/libxml'
-    rescue
+    rescue LoadError
       # libxml not available so nothing to test!
       return
     end
@@ -23,7 +23,4 @@ class XpathTest < Test::Unit::TestCase
     assert_equal xpath(doc, './/foobar'), nil
   end
 
-end
-
-__END__
-
+end

data/test/provider/models.rb

@@ -143,9 +143,9 @@ class SimpleModel < TestModel
     set_two.spec = "A:B"
     set_two.description = "This is test set two."
 
-    generate_records(5, Chronic.parse("oct 5 2002"), set_one)
-    generate_records(1, Chronic.parse("nov 5 2002"), [set_two], true)
-    generate_records(4, Chronic.parse("nov 5 2002"), [set_two])
+    generate_records(5, Time.parse("2002-10-5"), set_one)
+    generate_records(1, Time.parse("2002-11-5"), [set_two], true)
+    generate_records(4, Time.parse("2002-11-5"), [set_two])
   end
 
 end
@@ -154,11 +154,11 @@ class BigModel < TestModel
   
   def initialize(limit = nil)
     super(limit)
-    generate_records(100, Chronic.parse("October 2 2000"))
-    generate_records(100, Chronic.parse("November 2 2000"))
-    generate_records(100, Chronic.parse("December 2 2000"))
-    generate_records(100, Chronic.parse("January 2 2001"))
-    generate_records(100, Chronic.parse("February 2 2001"))
+    generate_records(100, Time.parse("October 2 2000"))
+    generate_records(100, Time.parse("November 2 2000"))
+    generate_records(100, Time.parse("December 2 2000"))
+    generate_records(100, Time.parse("January 2 2001"))
+    generate_records(100, Time.parse("February 2 2001"))
   end
   
 end
@@ -172,7 +172,7 @@ class MappedModel < TestModel
     set_one.spec = "A"
     set_one.description = "This is test set one."
 
-    generate_records(5, Chronic.parse("dec 1 2006"), set_one)
+    generate_records(5, Time.parse("dec 1 2006"), set_one)
   end
   
   def map_oai_dc
@@ -216,14 +216,14 @@ class ComplexModel < TestModel
     set_three_four.spec = "Three:Four"
     set_three_four.description = "This is combination set of Three and Four."
 
-    generate_records(250, Chronic.parse("May 2 1998"), [set_one, set_one_two])
-    generate_records(50, Chronic.parse("June 2 1998"), [set_one, set_one_two], true)
-    generate_records(50, Chronic.parse("October 10 1998"), [set_three, set_three_four], true)
-    generate_records(250, Chronic.parse("July 2 2002"), [set_two, set_one_two])
+    generate_records(250, Time.parse("May 2 1998"), [set_one, set_one_two])
+    generate_records(50, Time.parse("June 2 1998"), [set_one, set_one_two], true)
+    generate_records(50, Time.parse("October 10 1998"), [set_three, set_three_four], true)
+    generate_records(250, Time.parse("July 2 2002"), [set_two, set_one_two])
     
-    generate_records(250, Chronic.parse("September 15 2004"), [set_three, set_three_four])
-    generate_records(50, Chronic.parse("October 10 2004"), [set_three, set_three_four], true)
-    generate_records(250, Chronic.parse("December 25 2005"), [set_four, set_three_four])
+    generate_records(250, Time.parse("September 15 2004"), [set_three, set_three_four])
+    generate_records(50, Time.parse("October 10 2004"), [set_three, set_three_four], true)
+    generate_records(250, Time.parse("December 25 2005"), [set_four, set_three_four])
   end
   
 end

data/test/provider/tc_exceptions.rb

@@ -1,7 +1,8 @@
 require 'test_helper'
 
 class ProviderExceptions < Test::Unit::TestCase
-
+  include Singleton
+  
   def setup
     @provider = ComplexProvider.new
   end
@@ -51,8 +52,8 @@ class ProviderExceptions < Test::Unit::TestCase
   
   def test_no_records_match_dates_that_are_out_of_range
     assert_raise(OAI::NoMatchException) do
-      @provider.list_records(:from => Chronic.parse("November 2 2000"), 
-                             :until => Chronic.parse("November 1 2000"))
+      @provider.list_records(:from => Time.parse("November 2 2000"), 
+                             :until => Time.parse("November 1 2000"))
     end
   end
   

data/test/provider/tc_functional_tokens.rb

@@ -23,8 +23,8 @@ class ResumptionTokenFunctionalTest < Test::Unit::TestCase
     assert_nothing_raised { Document.new(@provider.list_records) }
     doc = Document.new(
       @provider.list_records(
-        :from => Chronic.parse("September 1 2004"),
-        :until => Chronic.parse("November 30 2004"))
+        :from => Time.parse("September 1 2004"),
+        :until => Time.parse("November 30 2004"))
       )
     assert_equal 101, doc.elements["/OAI-PMH/ListRecords"].to_a.size
     token = doc.elements["/OAI-PMH/ListRecords/resumptionToken"].text

data/test/provider/tc_provider.rb

@@ -33,12 +33,12 @@ class OaiTest < Test::Unit::TestCase
   def test_from
     assert_nothing_raised { REXML::Document.new(@big_provider.list_records) }
     doc = REXML::Document.new(
-      @big_provider.list_records(:from => Chronic.parse("February 1 2001"))
+      @big_provider.list_records(:from => Time.parse("February 1 2001"))
       )
     assert_equal 100, doc.elements['OAI-PMH/ListRecords'].to_a.size
 
     doc = REXML::Document.new(
-      @big_provider.list_records(:from => Chronic.parse("January 1 2001"))
+      @big_provider.list_records(:from => Time.parse("January 1 2001"))
       )
     assert_equal 200, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end
@@ -46,7 +46,7 @@ class OaiTest < Test::Unit::TestCase
   def test_until
     assert_nothing_raised { REXML::Document.new(@big_provider.list_records) }
     doc = REXML::Document.new(
-      @big_provider.list_records(:until => Chronic.parse("November 1 2000"))
+      @big_provider.list_records(:until => Time.parse("November 1 2000"))
       )
     assert_equal 100, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end
@@ -54,14 +54,14 @@ class OaiTest < Test::Unit::TestCase
   def test_from_and_until
     assert_nothing_raised { REXML::Document.new(@big_provider.list_records) }
     doc = REXML::Document.new(
-      @big_provider.list_records(:from => Chronic.parse("November 1 2000"),
-        :until => Chronic.parse("November 30 2000"))
+      @big_provider.list_records(:from => Time.parse("November 1 2000"),
+        :until => Time.parse("November 30 2000"))
       )
     assert_equal 100, doc.elements['OAI-PMH/ListRecords'].to_a.size
 
     doc = REXML::Document.new(
-      @big_provider.list_records(:from => Chronic.parse("December 1 2000"),
-      :until => Chronic.parse("December 31 2000"))
+      @big_provider.list_records(:from => Time.parse("December 1 2000"),
+      :until => Time.parse("December 31 2000"))
       )
     assert_equal 100, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end

metadata

@@ -1,10 +1,10 @@
 --- !ruby/object:Gem::Specification 
-rubygems_version: 0.9.0
+rubygems_version: 0.9.4
 specification_version: 1
 name: oai
 version: !ruby/object:Gem::Version 
-  version: 0.0.6
-date: 2007-06-05 00:00:00 -04:00
+  version: 0.0.7
+date: 2008-01-14 00:00:00 -05:00
 summary: A ruby library for working with the Open Archive Initiative Protocol for Metadata Harvesting (OAI-PMH)
 require_paths: 
 - lib
@@ -33,30 +33,30 @@ files:
 - Rakefile
 - bin/oai
 - test/activerecord_provider
-- test/client
-- test/provider
-- test/test.xml
 - test/activerecord_provider/config
-- test/activerecord_provider/database
-- test/activerecord_provider/fixtures
-- test/activerecord_provider/helpers
-- test/activerecord_provider/models
-- test/activerecord_provider/tc_ar_provider.rb
-- test/activerecord_provider/tc_ar_sets_provider.rb
-- test/activerecord_provider/tc_caching_paging_provider.rb
-- test/activerecord_provider/tc_simple_paging_provider.rb
-- test/activerecord_provider/test_helper.rb
 - test/activerecord_provider/config/connection.rb
 - test/activerecord_provider/config/database.yml
+- test/activerecord_provider/database
 - test/activerecord_provider/database/ar_migration.rb
 - test/activerecord_provider/database/oaipmhtest
+- test/activerecord_provider/fixtures
 - test/activerecord_provider/fixtures/dc.yml
+- test/activerecord_provider/helpers
 - test/activerecord_provider/helpers/providers.rb
 - test/activerecord_provider/helpers/set_provider.rb
+- test/activerecord_provider/models
 - test/activerecord_provider/models/dc_field.rb
 - test/activerecord_provider/models/dc_set.rb
 - test/activerecord_provider/models/oai_token.rb
+- test/activerecord_provider/tc_ar_provider.rb
+- test/activerecord_provider/tc_ar_sets_provider.rb
+- test/activerecord_provider/tc_caching_paging_provider.rb
+- test/activerecord_provider/tc_simple_paging_provider.rb
+- test/activerecord_provider/test_helper.rb
+- test/client
 - test/client/helpers
+- test/client/helpers/provider.rb
+- test/client/helpers/test_wrapper.rb
 - test/client/tc_exception.rb
 - test/client/tc_get_record.rb
 - test/client/tc_identify.rb
@@ -65,10 +65,11 @@ files:
 - test/client/tc_list_metadata_formats.rb
 - test/client/tc_list_records.rb
 - test/client/tc_list_sets.rb
+- test/client/tc_low_resolution_dates.rb
+- test/client/tc_utf8_escaping.rb
 - test/client/tc_xpath.rb
 - test/client/test_helper.rb
-- test/client/helpers/provider.rb
-- test/client/helpers/test_wrapper.rb
+- test/provider
 - test/provider/models.rb
 - test/provider/tc_exceptions.rb
 - test/provider/tc_functional_tokens.rb
@@ -76,19 +77,9 @@ files:
 - test/provider/tc_resumption_tokens.rb
 - test/provider/tc_simple_provider.rb
 - test/provider/test_helper.rb
+- test/test.xml
 - lib/oai
-- lib/oai.rb
-- lib/test.rb
 - lib/oai/client
-- lib/oai/client.rb
-- lib/oai/constants.rb
-- lib/oai/exception.rb
-- lib/oai/harvester
-- lib/oai/harvester.rb
-- lib/oai/provider
-- lib/oai/provider.rb
-- lib/oai/set.rb
-- lib/oai/xpath.rb
 - lib/oai/client/get_record.rb
 - lib/oai/client/header.rb
 - lib/oai/client/identify.rb
@@ -99,23 +90,27 @@ files:
 - lib/oai/client/metadata_format.rb
 - lib/oai/client/record.rb
 - lib/oai/client/response.rb
+- lib/oai/client.rb
+- lib/oai/constants.rb
+- lib/oai/exception.rb
+- lib/oai/harvester
 - lib/oai/harvester/config.rb
 - lib/oai/harvester/harvest.rb
 - lib/oai/harvester/logging.rb
 - lib/oai/harvester/mailer.rb
 - lib/oai/harvester/shell.rb
+- lib/oai/harvester.rb
+- lib/oai/provider
 - lib/oai/provider/extensions
 - lib/oai/provider/metadata_format
+- lib/oai/provider/metadata_format/oai_dc.rb
 - lib/oai/provider/metadata_format.rb
 - lib/oai/provider/model
+- lib/oai/provider/model/activerecord_caching_wrapper.rb
+- lib/oai/provider/model/activerecord_wrapper.rb
 - lib/oai/provider/model.rb
 - lib/oai/provider/partial_result.rb
 - lib/oai/provider/response
-- lib/oai/provider/response.rb
-- lib/oai/provider/resumption_token.rb
-- lib/oai/provider/metadata_format/oai_dc.rb
-- lib/oai/provider/model/activerecord_caching_wrapper.rb
-- lib/oai/provider/model/activerecord_wrapper.rb
 - lib/oai/provider/response/error.rb
 - lib/oai/provider/response/get_record.rb
 - lib/oai/provider/response/identify.rb
@@ -124,6 +119,13 @@ files:
 - lib/oai/provider/response/list_records.rb
 - lib/oai/provider/response/list_sets.rb
 - lib/oai/provider/response/record_response.rb
+- lib/oai/provider/response.rb
+- lib/oai/provider/resumption_token.rb
+- lib/oai/provider.rb
+- lib/oai/set.rb
+- lib/oai/xpath.rb
+- lib/oai.rb
+- lib/test.rb
 - examples/models/file_model.rb
 - examples/providers/dublin_core.rb
 test_files: []
@@ -139,24 +141,6 @@ extensions: []
 requirements: []
 
 dependencies: 
-- !ruby/object:Gem::Dependency 
-  name: activesupport
-  version_requirement: 
-  version_requirements: !ruby/object:Gem::Version::Requirement 
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        version: 1.3.1
-    version: 
-- !ruby/object:Gem::Dependency 
-  name: chronic
-  version_requirement: 
-  version_requirements: !ruby/object:Gem::Version::Requirement 
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        version: 0.0.3
-    version: 
 - !ruby/object:Gem::Dependency 
   name: builder
   version_requirement: