picky 1.5.1 → 1.5.2

data/lib/picky/sources/db.rb

@@ -40,6 +40,8 @@ module Sources
     def create_database_adapter # :nodoc:
       # TODO Do not use ActiveRecord directly.
       #
+      # TODO Use set_table_name etc.
+      #
       adapter_class = Class.new ActiveRecord::Base
       adapter_class.abstract_class = true
       adapter_class
@@ -76,56 +78,58 @@ module Sources
     #
     # Uses CREATE TABLE AS with the given SELECT statement to create a snapshot of the data.
     #
-    def take_snapshot type
+    def take_snapshot index
       connect_backend
       
-      origin = snapshot_table_name type
-      
+      origin = snapshot_table_name index
       on_database = database.connection
       
-      on_database.execute "DROP TABLE IF EXISTS #{origin}"
+      # Drop the table if it exists.
+      #
+      on_database.drop_table origin if on_database.table_exists?(origin)
+      
+      # The adapters currently do not support this.
+      #
       on_database.execute "CREATE TABLE #{origin} AS #{select_statement}"
-      # TODO Use rename_column ASAP.
+      
+      # Add a column that Picky uses to traverse the table's entries.
       #
-      if on_database.adapter_name == "PostgreSQL"
-        on_database.execute "ALTER TABLE #{origin} ADD COLUMN #{@@traversal_id} SERIAL PRIMARY KEY"
-      else
-        on_database.execute "ALTER TABLE #{origin} ADD COLUMN #{@@traversal_id} INTEGER NOT NULL PRIMARY KEY AUTO_INCREMENT"
-      end
+      on_database.add_column origin, @@traversal_id, :primary_key, :null => :false
       
-      # Execute any special queries this type needs executed.
+      # Execute any special queries this index needs executed.
       #
-      on_database.execute type.after_indexing if type.after_indexing
+      on_database.execute index.after_indexing if index.after_indexing
     end
     
     # Counts all the entries that are used for the index.
     #
-    def count type # :nodoc:
+    def count index
       connect_backend
       
-      database.connection.select_value("SELECT COUNT(#{@@traversal_id}) FROM #{snapshot_table_name(type)}").to_i
+      database.connection.select_value("SELECT COUNT(#{@@traversal_id}) FROM #{snapshot_table_name(index)}").to_i
     end
     
+    # The name of the snapshot table created by Picky.
     #
-    #
-    def snapshot_table_name type # :nodoc:
-      "#{type.name}_type_index"
+    def snapshot_table_name index
+      "picky_#{index.name}_index"
     end
     
     # Harvests the data to index in chunks.
     #
-    def harvest type, category, &block
+    def harvest index, category, &block
       connect_backend
       
-      (0..count(type)).step(chunksize) do |offset|
-        get_data type, category, offset, &block
+      (0..count(index)).step(chunksize) do |offset|
+        get_data index, category, offset, &block
       end
     end
     
     # Gets the data from the backend.
     #
-    def get_data type, category, offset, &block # :nodoc:
-      select_statement = harvest_statement_with_offset(type, category, offset)
+    def get_data index, category, offset, &block # :nodoc:
+      
+      select_statement = harvest_statement_with_offset index, category, offset
       
       # TODO Rewrite ASAP.
       #
@@ -134,25 +138,19 @@ module Sources
         text_key = category.from.to_s
         database.connection.execute(select_statement).each do |hash|
           id, text = hash.values_at id_key, text_key
-          next unless text
-          text.force_encoding 'utf-8' # TODO Still needed? Or move to backend?
-          yield id, text
+          yield id, text if text
         end
       else
         database.connection.execute(select_statement).each do |id, text|
-          next unless text
-          text.force_encoding 'utf-8' # TODO Still needed? Or move to backend?
-          yield id, text
+          yield id, text if text
         end
       end
     end
     
     # Builds a harvest statement for getting data to index.
     #
-    # TODO Use the adapter for this.
-    #
-    def harvest_statement_with_offset type, category, offset # :nodoc:
-      statement = harvest_statement type, category
+    def harvest_statement_with_offset index, category, offset
+      statement = harvest_statement index, category
       
       statement += statement.include?('WHERE') ? ' AND' : ' WHERE'
       
@@ -161,15 +159,13 @@ module Sources
     
     # The harvest statement used to pull data from the snapshot table.
     #
-    def harvest_statement type, category # :nodoc:
-      "SELECT id, #{category.from} FROM #{snapshot_table_name(type)} st"
+    def harvest_statement index, category
+      "SELECT id, #{category.from} FROM #{snapshot_table_name(index)} st"
     end
     
     # The amount of records that are loaded each chunk.
     #
-    def chunksize # :nodoc:
-      # TODO Make parametrizable.
-      #
+    def chunksize
       25_000
     end
     

data/spec/lib/sources/db_spec.rb

@@ -30,7 +30,7 @@ describe Sources::DB do
 
           @connection.should_receive(:execute).
                       once.
-                      with('SELECT id, some_category FROM some_type_type_index st WHERE st.__picky_id > some_offset LIMIT 25000').
+                      with('SELECT id, some_category FROM picky_some_type_index st WHERE st.__picky_id > some_offset LIMIT 25000').
                       and_return []
 
           @source.get_data @type, @category, :some_offset
@@ -40,7 +40,7 @@ describe Sources::DB do
         it 'yields to the caller' do
           @connection.should_receive(:execute).
                       any_number_of_times.
-                      with('SELECT id, some_category FROM some_type_type_index st WHERE st.__picky_id > some_offset LIMIT 25000').
+                      with('SELECT id, some_category FROM picky_some_type_index st WHERE st.__picky_id > some_offset LIMIT 25000').
                       and_return [[1, 'text']]
 
           @source.get_data @type, @category, :some_offset do |id, text|
@@ -86,7 +86,7 @@ describe Sources::DB do
     end
     it "should get the id count" do
       result = stub(:result, :to_i => 12_345)
-      @connection.should_receive(:select_value).once.with("SELECT COUNT(__picky_id) FROM some_type_name_type_index")
+      @connection.should_receive(:select_value).once.with("SELECT COUNT(__picky_id) FROM picky_some_type_name_index")
       
       @source.count @type
     end

metadata

@@ -2,7 +2,7 @@
 name: picky
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 1.5.1
+  version: 1.5.2
 platform: ruby
 authors: 
 - Florian Hanke