Merge pull request #76 from code4lib/configured_id_column

Add configurable `identifier_field` analagous to existing `timestamp_field`.

Author: jrochkind

lib/oai/provider.rb

@@ -155,8 +155,10 @@
 #
 # Valid options include:
 #
-# * `timestamp_field` - Specifies the model field to use as the update
+# * `timestamp_field` - Specifies the model field/method to use as the update
 #   filter.  Defaults to `updated_at`.
+# * `identifier_field` -- specifies the model field/method to use to get value to use
+#    as oai identifier (method return value should not include prefix)
 # * `limit` -           Maximum number of records to return in each page/set.
 #   Defaults to 100, set to `nil` for all records in one page. Otherwise
 #   the wrapper will paginate the result via resumption tokens.

lib/oai/provider/model.rb

@@ -1,21 +1,21 @@
 module OAI::Provider
   # = OAI::Provider::Model
   #
-  # Model implementers should subclass OAI::Provider::Model and override 
+  # Model implementers should subclass OAI::Provider::Model and override
   # Model#earliest, Model#latest, and Model#find.  Optionally Model#sets and
-  # Model#deleted? can be used to support sets and record deletions.  It 
-  # is also the responsibility of the model implementer to account for 
-  # resumption tokens if support is required.  Models that don't support 
-  # resumption tokens should raise an exception if a limit is requested     
+  # Model#deleted? can be used to support sets and record deletions.  It
+  # is also the responsibility of the model implementer to account for
+  # resumption tokens if support is required.  Models that don't support
+  # resumption tokens should raise an exception if a limit is requested
   # during initialization.
   #
   # earliest - should return the earliest update time in the repository.
   # latest - should return the most recent update time in the repository.
   # sets - should return an array of sets supported by the repository.
   # deleted? - individual records returned should respond true or false
   # when sent the deleted? message.
-  # available_formats - if overridden, individual records should return an 
-  # array of prefixes for all formats in which that record is available, 
+  # available_formats - if overridden, individual records should return an
+  # array of prefixes for all formats in which that record is available,
   # if other than ["oai_dc"]
   # about - if overridden, should return a String or Array of XML Strings to
   # insert into the OAI Record <about> chunks.
@@ -28,29 +28,29 @@ module OAI::Provider
   # There are several helper models for dealing with resumption tokens please
   # see the ResumptionToken class for more details.
   #
-
   class Model
-    attr_reader :timestamp_field
-    
-    def initialize(limit = nil, timestamp_field = 'updated_at')
+    attr_reader :timestamp_field, :identifier_field
+
+    def initialize(limit = nil, timestamp_field = 'updated_at', identifier_field = 'id')
       @limit = limit
+      @identifier_field = identifier_field
       @timestamp_field = timestamp_field
     end
 
     # should return the earliest timestamp available from this model.
     def earliest
       raise NotImplementedError.new
     end
-    
+
     # should return the latest timestamp available from this model.
     def latest
       raise NotImplementedError.new
     end
-    
+
     def sets
       nil
     end
-  
+
     # find is the core method of a model, it returns records from the model
     # bases on the parameters passed in.
     #
@@ -61,12 +61,12 @@ def sets
     # * :from => earliest timestamp to be included in the results
     # * :until => latest timestamp to be included in the results
     # * :set => the set from which to retrieve the results
-    # * :metadata_prefix => type of metadata requested (this may be useful if 
+    # * :metadata_prefix => type of metadata requested (this may be useful if
     #                       not all records are available in all formats)
     def find(selector, options={})
       raise NotImplementedError.new
     end
-    
+
     def deleted?
       false
     end
@@ -76,5 +76,5 @@ def about record
       nil
     end
   end
-  
+
 end

lib/oai/provider/model/activerecord_wrapper.rb

@@ -10,11 +10,18 @@ module OAI::Provider
   #
   class ActiveRecordWrapper < Model
 
-    attr_reader :model, :timestamp_field
-
+    attr_reader :model, :timestamp_field, :identifier_field
+
+    # If custom 'timestamp_field' is used, be aware this will be an ActiveRecord
+    # attribute that we will limit on, so perhaps should be indexe appropriately.
+    #
+    # If custom `identifier_field` is used, be aware this will be an ActiveRecord
+    # attribute that we will sort on, and use in WHERE clauses with `=` as well as
+    # greater than/less than, so should be indexed appropriately.
     def initialize(model, options={})
       @model = model
       @timestamp_field = options.delete(:timestamp_field) || 'updated_at'
+      @identifier_field = options.delete(:identifier_field) || model.primary_key || "id"
       @limit = options.delete(:limit) || 100
 
       unless options.empty?
@@ -54,7 +61,7 @@ def find(selector, options={})
           find_scope.where(conditions)
         end
       else
-        find_scope.where(conditions).find(selector)
+        find_scope.where(conditions).find_by!(identifier_field => selector)
       end
     end
 
@@ -129,7 +136,7 @@ def next_set(find_scope, token_string)
       else # end of result set
         find_scope.where(token_conditions(token))
           .limit(@limit)
-          .order("#{model.primary_key} asc")
+          .order("#{identifier_field} asc")
       end
     end
 
@@ -138,9 +145,9 @@ def next_set(find_scope, token_string)
     def select_partial(find_scope, token)
       records = find_scope.where(token_conditions(token))
         .limit(@limit)
-        .order("#{model.primary_key} asc")
+        .order("#{identifier_field} asc")
       raise OAI::ResumptionTokenException.new unless records
-      offset = records.last.send(model.primary_key.to_sym)
+      offset = records.last.send(identifier_field)
 
       PartialResult.new(records, token.next(offset))
     end
@@ -157,7 +164,7 @@ def token_conditions(token)
 
       return sql if 0 == last
       # Now add last id constraint
-      sql.first << " AND #{model.primary_key} > :id"
+      sql.first << " AND #{identifier_field} > :id"
       sql.last[:id] = last
 
       return sql

lib/oai/provider/response/record_response.rb

@@ -50,7 +50,7 @@ def about_for(record)
 
     # Namespace syntax suggested in http://www.openarchives.org/OAI/2.0/guidelines-oai-identifier.htm
     def identifier_for(record)
-      "#{provider.prefix}:#{record.id}"
+      "#{provider.prefix}:#{record.send( provider.model.identifier_field )}"
     end
 
     def timestamp_for(record)

test/activerecord_provider/helpers/providers.rb

@@ -12,6 +12,13 @@ class ARProvider < OAI::Provider::Base
   source_model ActiveRecordWrapper.new(DCField)
 end
 
+class ARProviderCustomIdentifierField < OAI::Provider::Base
+  repository_name 'ActiveRecord Based Provider'
+  repository_url 'http://localhost'
+  record_prefix 'oai:test'
+  source_model ActiveRecordWrapper.new(DCField, identifier_field: "source")
+end
+
 class ARProviderWithScope < OAI::Provider::Base
   DATE_LESS_THAN_RESTRICTION = Time.parse("2007-03-12 19:30:22 UTC")
 

test/activerecord_provider/tc_ar_provider.rb

@@ -40,6 +40,18 @@ def test_list_records_scope
     assert_equal expected_count, doc.elements['OAI-PMH/ListRecords'].to_a.size
   end
 
+
+  def test_get_record_alternate_identifier_column
+    @provider = ARProviderCustomIdentifierField.new
+
+    record_id = DCField.first.send(@provider.class.model.identifier_field)
+
+    doc = REXML::Document.new(@provider.get_record(
+        :identifier => "oai:test:#{record_id}", :metadata_prefix => 'oai_dc'))
+
+    assert_equal "oai:test:#{record_id}", doc.elements['OAI-PMH/GetRecord/record/header/identifier'].text
+  end
+
   def test_list_identifiers
     assert_nothing_raised { REXML::Document.new(@provider.list_identifiers) }
     doc = REXML::Document.new(@provider.list_identifiers)