picky 4.6.3 → 4.6.4

data/lib/picky/generators/partial.rb

@@ -0,0 +1,27 @@
+module Picky
+
+  module Generators
+    
+    module Partial
+      extend Helpers::Identification
+      
+      def self.from thing, index_name = nil, category_name = nil
+        return Default unless thing
+
+        if thing.respond_to? :each_partial
+          thing
+        else
+          raise <<-ERROR
+partial options #{identifier_for(index_name, category_name)}should be either
+* for example a Partial::Substring.new(from: m, to: n), Partial::Postfix.new(from: n), Partial::Infix.new(min: m, max: n) etc.
+or
+* an object that responds to #each_partial(str_or_sym) and yields each partial
+ERROR
+        end
+      end
+    
+    end
+    
+  end
+
+end

data/lib/picky/generators/similarity.rb

@@ -0,0 +1,27 @@
+module Picky
+  
+  module Generators
+
+    module Similarity
+      extend Helpers::Identification
+
+      def self.from thing, index_name = nil, category_name = nil
+        return Default unless thing
+
+        if thing.respond_to?(:encode) && thing.respond_to?(:prioritize)
+          thing
+        else
+          raise <<-ERROR
+similarity options #{identifier_for(index_name, category_name)}should be either
+* for example a Similarity::Phonetic.new(n), Similarity::Metaphone.new(n), Similarity::DoubleMetaphone.new(n) etc.
+or
+* an object that responds to #encode(text) => encoded_text and #prioritize(array_of_encoded, encoded)
+ERROR
+        end
+      end
+
+    end
+
+  end
+  
+end

data/lib/picky/generators/weights.rb

@@ -0,0 +1,32 @@
+module Picky
+
+  module Generators
+
+    module Weights
+      extend Helpers::Identification
+      
+      # Factory method to return a fitting
+      # weight handling thing for the given thing.
+      #  
+      def self.from thing, index_name = nil, category_name = nil
+        return Default unless thing
+
+        if thing.respond_to? :weight_for
+          thing
+        elsif thing.respond_to? :to_int
+          Logarithmic.new thing
+        else
+          raise <<-ERROR
+weight options #{identifier_for(index_name, category_name)}should be either
+* for example a Weights::Logarithmic.new, Weights::Constant.new(int = 0), Weights::Dynamic.new(&block) etc.
+or
+* an object that responds to #weight_for(amount_of_ids_for_token) => float
+ERROR
+        end
+      end
+
+    end
+    
+  end
+  
+end

data/lib/picky/helpers/identification.rb

@@ -0,0 +1,18 @@
+module Picky
+
+  module Helpers
+    
+    module Identification
+      
+      def identifier_for index_name = nil, category_name = nil
+        specifics = ""
+        specifics << index_name.to_s if index_name
+        specifics << ":#{category_name}" if category_name
+        specifics = "for #{specifics} " unless specifics.empty?
+      end
+    
+    end
+    
+  end
+
+end

data/lib/picky/helpers/indexing.rb

@@ -7,10 +7,26 @@ module Picky
 
       include Measuring
       
+      # Runs the block and logs a few infos regarding the time it took.
+      #
       def timed_indexing scheduler, &block
         Picky.logger.info "Picky is indexing using #{scheduler.fork? ? 'multiple processes' : 'a single process'}: "
         Picky.logger.info " Done in #{timed(&block).round}s.\n"
       end
+      
+      # Indexing works the same way, always:
+      #  * Prepare the scheduler.
+      #  * Cache the scheduler.
+      #
+      def index scheduler = Scheduler.new
+        timed_indexing scheduler do
+          prepare scheduler
+          scheduler.finish
+
+          cache scheduler
+          scheduler.finish
+        end
+      end
 
     end
   end

data/lib/picky/index.rb

@@ -178,6 +178,12 @@ module Picky
 
       new_category
     end
+    
+    # The directory used by this index.
+    #
+    def directory
+      @directory ||= ::File.join(PICKY_ROOT, 'index', PICKY_ENVIRONMENT, name.to_s)
+    end
 
     # Make this category range searchable with a fixed range. If you need other
     # ranges, define another category with a different range value.

data/lib/picky/index_indexing.rb

@@ -3,10 +3,6 @@ module Picky
   #
   #
   class Index
-
-    include API::Tokenizer
-    include API::Source
-
     include Helpers::Indexing
 
     # Delegators for indexing.
@@ -20,19 +16,7 @@ module Picky
     # Parameters are the exact same as for indexing.
     #
     def indexing options = {}
-      @tokenizer = extract_tokenizer options
-    end
-
-    #
-    #
-    def index scheduler = Scheduler.new
-      timed_indexing scheduler do
-        prepare scheduler
-        scheduler.finish
-
-        cache scheduler
-        scheduler.finish
-      end
+      @tokenizer = Tokenizer.from options
     end
 
     # Calling prepare on an index will call prepare
@@ -46,9 +30,7 @@ module Picky
         check_source_empty
         prepare_in_parallel scheduler
       else
-        with_data_snapshot do
-          categories.prepare scheduler
-        end
+        with_data_snapshot { categories.prepare scheduler }
       end
     end
 
@@ -98,7 +80,13 @@ module Picky
     #
     def source some_source = nil, &block
       some_source ||= block
-      some_source ? (@source = extract_source(some_source)) : unblock_source
+      some_source ? (@source = Source.from(some_source, false, name)) : unblock_source
+    end
+    # Get the actual source if it is wrapped in a time
+    # capsule, i.e. a block/lambda.
+    #
+    def unblock_source
+      @source.respond_to?(:call) ? @source.call : @source
     end
 
     # Define a key_format on the index.

data/lib/picky/indexes_indexing.rb

@@ -5,14 +5,17 @@ module Picky
   class Indexes
 
     extend Helpers::Indexing
+    include Helpers::Indexing
 
     instance_delegate :clear,
                       :tokenizer
 
-    each_delegate :clear,
+    each_delegate :cache,
+                  :clear,
+                  :prepare,
                   :to => :indexes
 
-    #
+    # Overrides index from the helper.
     #
     def self.index scheduler = Scheduler.new
       timed_indexing scheduler do
@@ -20,18 +23,6 @@ module Picky
       end
     end
 
-    #
-    #
-    def index scheduler = Scheduler.new
-      indexes.each { |index| index.prepare scheduler }
-      scheduler.finish
-
-      # timed_exclaim "Tokenizing finished, generating data for indexes from tokenized data."
-
-      indexes.each { |index| index.cache scheduler }
-      scheduler.finish
-    end
-
     #
     #
     def tokenizer