picky 4.10.0 → 4.11.0

data/lib/picky/backends/memory/json.rb

@@ -33,12 +33,15 @@ module Picky
 
         # Dump JSON into the cache file.
         #
-        # TODO Ask MultiJson people to add IO option:
-        # MultiJson.encode(object, out_file)
+        # TODO Add IO option:
+        # MultiJson.encode(object, io: out_file)
         #
         def dump_json internal
           ::File.open(cache_path, 'w') do |out_file|
-            # MultiJson.encode internal, out_file
+            # If using Yajl, this will stream write to out_file.
+            # Note: But it fails on oj.
+            #
+            # MultiJson.dump internal, [out_file]
             out_file.write MultiJson.encode internal
           end
         end

data/lib/picky/categories.rb

@@ -29,10 +29,22 @@ module Picky
       @categories    = []
       @category_hash = {}
     end
+    
+    # Updates the qualifier ("qualifier:searchterm") mapping.
+    #
+    # Example:
+    #   You dynamically add a new category to an index.
+    #   To add the qualifiers to a search, you call this
+    #   method.
+    #
+    def mapper
+      @mapper ||= QualifierMapper.new self
+    end
 
     # Add the given category to the list of categories.
     #
     def << category
+      @mapper = nil # TODO Move. (Resets category mapping)
       categories << category
       category_hash[category.name] = category
     end

data/lib/picky/categories_indexed.rb

@@ -61,8 +61,10 @@ module Picky
     #
     # Note: Once I thought this was called too often. But it is not (18.01.2011).
     #
+    # TODO Called too often?
+    #
     def possible_categories token
-      token.predefined_categories || categories
+      token.predefined_categories(mapper) || categories
     end
 
   end

data/lib/picky/extensions/array.rb

@@ -2,21 +2,11 @@
 #
 class Array
 
-  # Cluster-uniqs equal neighborly elements.
-  #
-  # Returns a copy.
-  #
-  def clustered_uniq
-    self.inject([]) do |result, element|
-      result << element if element != result.last
-      result
-    end
-  end
   # Around 10% faster than the above.
   #
   # Returns a copy.
   #
-  def clustered_uniq_fast
+  def clustered_uniq
     result = []
     self.inject(nil) do |last, element|
       if last == element

data/lib/picky/extensions/class.rb

@@ -2,9 +2,11 @@
 #
 class Class
 
-  def instance_delegate *methods
-    methods.each do |method|
-      module_eval("def self.#{method}(*args, &block)\nself.instance.__send__(#{method.inspect}, *args, &block)\nend\n", "(__DELEGATION__)", 1)
+  def instance_delegate *method_names
+    method_names.each do |method_name|
+      module_eval(<<-DELEGATION, "(__DELEGATION__)", 1)
+def self.#{method_name}(*args, &block)\n  self.instance.#{method_name}(*args, &block)\nend
+DELEGATION
     end
   end
 

data/lib/picky/extensions/object.rb

@@ -9,6 +9,8 @@ class Object
 
   # Just outputs the given text to the logger.
   #
+  # Note: stubbed in spec_helper.rb
+  #
   def exclaim text
     Picky.logger.info text
     Picky.logger.flush
@@ -17,7 +19,17 @@ class Object
   # Puts a text that informs the user of a missing gem.
   #
   def warn_gem_missing gem_name, message
-    Picky.logger.warn "#{gem_name} gem missing!\nTo use #{message}, you need to:\n  1. Add the following line to Gemfile:\n     gem '#{gem_name}'\n     or\n     require '#{gem_name}'\n     for example at the top of your app.rb file.\n  2. Then, run:\n     bundle update\n"
+    Picky.logger.warn <<-WARNING
+Warning: #{gem_name} gem missing!
+To use #{message}, you need to:
+  1. Add the following line to Gemfile:
+     gem '#{gem_name}'
+     or
+     require '#{gem_name}'
+     for example at the top of your app.rb file.
+  2. Then, run:
+     bundle update
+WARNING
   end
 
   # Indents each line by <tt>amount=2</tt> spaces.

data/lib/picky/generators/similarity/phonetic.rb

@@ -20,8 +20,6 @@ module Picky
         #
         def initialize amount = 3
           check_gem
-
-          raise "In Picky 2.0+, the Similarity::Phonetic has been renamed to Similarity::DoubleMetaphone. Please use that one. Thanks!" if self.class == Phonetic
           @amount = amount
         end
 

data/lib/picky/index.rb

@@ -140,6 +140,23 @@ module Picky
         @backend ||= Backends::Memory.new
       end
     end
+    
+    # Ignore the categories with these qualifiers.
+    #
+    # Example:
+    #   search = Search.new(index1, index2, index3) do
+    #     ignore :name, :first_name
+    #   end
+    #
+    # Cleans up / optimizes after being called.
+    #
+    # TODO This needs to move to the index.
+    #
+    def ignore *qualifiers
+      @ignored_categories ||= []
+      @ignored_categories += qualifiers.map { |qualifier| @mapper.map qualifier }.compact
+      @ignored_categories.uniq!
+    end
 
     # SYMBOLS.
     #
@@ -179,6 +196,22 @@ module Picky
       new_category
     end
     
+    # Restrict categories to the given ones.
+    #
+    # Functionally equivalent as if indexes didn't
+    # have the categories at all.
+    #
+    # Note: Probably only makes sense when an index
+    # is used in multiple searches. If not, why even
+    # have the categories?
+    #
+    # TODO Redesign.
+    #
+    def only *qualifiers
+      raise "Sorry, Picky::Search#only has been removed in version."
+      # @mapper.restrict_to *qualifiers
+    end
+    
     # The directory used by this index.
     #
     # Note: Used @directory ||=, but needs to be dynamic.

data/lib/picky/index_facets.rb

@@ -15,12 +15,16 @@ module Picky
       text_ids = self[category_identifier].exact.inverted
       no_counts = options[:counts] == false
       minimal_counts = options[:at_least]
-      text_ids.inject(no_counts ? [] : {}) do |result, (text, ids)|
-        size = ids.size
-        next result if minimal_counts && size < minimal_counts
-        if no_counts
+      
+      if no_counts
+        text_ids.inject([]) do |result, (text, ids)|
+          next result if minimal_counts && ids.size < minimal_counts
           result << text
-        else
+        end
+      else
+        text_ids.inject({}) do |result, (text, ids)|
+          size = ids.size
+          next result if minimal_counts && size < minimal_counts
           result[text] = size; result
         end
       end

data/lib/picky/loader.rb

@@ -170,8 +170,6 @@ module Picky
         load_relative 'query/allocation',
                       'query/allocations'
 
-        load_relative 'query/qualifier_category_mapper'
-
         load_relative 'query/boosts'
 
         load_relative 'query/indexes',
@@ -219,6 +217,8 @@ module Picky
       end
       
       def load_inner_api
+        load_relative 'qualifier_mapper'
+        
         load_relative 'category',
                       'category_indexed',
                       'category_indexing',
@@ -235,7 +235,7 @@ module Picky
                       'indexes_indexed',
                       'indexes_indexing',
                       'indexes_convenience'
-
+        
         load_relative 'index',
                       'index_indexed',
                       'index_indexing',
@@ -261,14 +261,12 @@ module Picky
       
       # Loads the user interface parts.
       #
-      # TODO Move tokenizer etc.?
-      #
       def load_user_interface
         load_api
         load_logging
         load_relative 'source'
         load_relative 'tokenizer'
-        load_relative 'rack/harakiri'
+        # load_relative 'rack/harakiri' # Needs to be explicitly loaded/required.
         load_relative 'character_substituters/west_european'
         load_generators
         load_inner_api
@@ -276,7 +274,6 @@ module Picky
         load_search
         load_interfaces
         load_relative 'scheduler'
-        load_relative 'migrations/from_30_to_31' # TODO Remove.
       end
 
       # Loads the framework.

data/lib/picky/pool.rb

@@ -12,7 +12,8 @@ module Picky
       # Note: If you need to run two consecutive queries,
       # this can't be used.
       #
-      # TODO Also install calling release_all after each query.
+      # Note: You need to call Picky::Pool.release_all after each query
+      # (or after a few queries).
       #
       def install
         Query::Token.extend self
@@ -76,19 +77,21 @@ module Picky
         # (And removes it from the used pool)
         #
         def release instance
-           @__free__ << instance
+          @__free__ << instance
            
-           # Note: This is relatively fast as there are often only
-           # few instances in the used pool.
-           @__used__.delete instance 
+          # Note: This is relatively fast as there are often only
+          # few instances in the used pool.
+          #
+          @__used__.delete instance 
         end
       
-        # After you have called release all, you can't
+        # After you have called release all, you can't externally
         # use any reference that has formerly been obtained
         # anymore.
         #
         def release_all
-          @__free__ += @__used__
+          @__used__.each { |used| @__free__ << used } # +0 Array per release_all
+          # @__used__ += @__free__ # +1 Array per release_all
           @__used__.clear
         end
       

data/lib/picky/qualifier_mapper.rb

@@ -0,0 +1,51 @@
+module Picky
+
+  # Collection class for qualifiers.
+  #
+  class QualifierMapper
+
+    attr_reader :mapping
+      
+    #
+    #
+    def initialize categories
+      @mapping = {}
+      categories.each { |category| add category }
+    end
+
+    #
+    #
+    def add category
+      category.qualifiers.each do |qualifier|
+        sym_qualifier = qualifier.intern
+        Picky.logger.warn %Q{Warning: Qualifier "#{qualifier}" already mapped to category #{mapping[sym_qualifier].identifier} (ambiguous qualifier mapping).} if mapping.has_key? sym_qualifier
+        mapping[sym_qualifier] = category
+      end
+    end
+
+    # Normalizes the given qualifier.
+    #
+    # Returns nil if it is not allowed, the normalized qualifier if it is.
+    #
+    def map qualifier
+      return nil if qualifier.empty?
+
+      mapping[qualifier.intern]
+    end
+      
+    # Restricts the given categories.
+    #
+    def restrict user_qualified
+      if @restricted
+        user_qualified ? @restricted & user_qualified : @restricted 
+      else
+        user_qualified
+      end
+    end
+    def restrict_to *qualifiers
+      @restricted = qualifiers.map { |qualifier| map qualifier }.compact
+    end
+
+  end
+
+end

data/lib/picky/query/boosts.rb

@@ -44,11 +44,11 @@ module Picky
       #   [:name, :height, :color] returns +3, but
       #   [:name, :height, :street] returns -1.
       #
-      # Note: Use Array#clustered_uniq_fast to make
+      # Note: Use Array#clustered_uniq to make
       #       [:a, :a, :b, :a] => [:a, :b, :a]
       #
       def boost_for_categories names
-        @boosts[names.clustered_uniq_fast] || 0
+        @boosts[names.clustered_uniq] || 0
       end
 
       # API.

data/lib/picky/query/indexes.rb

@@ -29,19 +29,6 @@ module Picky
         IndexesCheck.check_backends indexes
 
         @indexes = indexes
-        
-        remap_qualifiers
-      end
-      
-      # Updates the qualifier ("qualifier:searchterm") mapping.
-      #
-      # Example:
-      #   You dynamically add a new category to an index.
-      #   To add the qualifiers to a search, you call this
-      #   method.
-      #
-      def remap_qualifiers
-        @mapper = QualifierCategoryMapper.new @indexes # TODO Move into search?
       end
 
       # Ignore the categories with these qualifiers.
@@ -51,26 +38,13 @@ module Picky
       #     ignore :name, :first_name
       #   end
       #
-      # Cleans up / optimizes after being called.
+      # Note: Cleans up / optimizes after being called.
       #
       def ignore *qualifiers
         @ignored_categories ||= []
         @ignored_categories += qualifiers.map { |qualifier| @mapper.map qualifier }.compact
         @ignored_categories.uniq!
       end
-      
-      # Restrict categories to the given ones.
-      #
-      # Functionally equivalent as if indexes didn't
-      # have the categories at all.
-      #
-      # Note: Probably only makes sense when an index
-      # is used in multiple searches. If not, why even
-      # have the categories?
-      #
-      def only *qualifiers
-        @mapper.restrict_to *qualifiers
-      end
 
       # Returns a number of prepared (sorted, reduced etc.) allocations for the given tokens.
       #
@@ -105,8 +79,6 @@ module Picky
       # Returns a number of possible allocations for the given tokens.
       #
       def allocations_for tokens
-        tokens.categorize @mapper
-
         Allocations.new allocations_ary_for(tokens)
       end
       def allocations_ary_for tokens

data/lib/picky/query/token.rb

@@ -13,7 +13,7 @@ module Picky
 
       attr_reader :text, :original
       attr_writer :similar
-      attr_accessor :predefined_categories
+      attr_writer :predefined_categories
       
       delegate :blank?, :to => :@text
       
@@ -22,6 +22,8 @@ module Picky
       # Note:
       # Use this if you do not want a normalized token.
       #
+      # TODO Throw away @predefined_categories?
+      #
       def initialize text, original = nil, categories = nil
         @text     = text
         @original = original
@@ -59,8 +61,10 @@ module Picky
       # Note: If this is not done, there is no mapping.
       # Note: predefined is an Array of mapped categories.
       #
-      def categorize mapper
-        @predefined_categories ||= extract_predefined mapper
+      # TODO Do we really need to set the predefined categories on the token?
+      #
+      def predefined_categories mapper
+        @predefined_categories || extract_predefined(mapper)
       end
       def extract_predefined mapper
         user_qualified = categorize_with mapper, @qualifiers

data/lib/picky/query/tokens.rb

@@ -63,12 +63,6 @@ module Picky
         @tokens.last.partial = true unless empty?
       end
 
-      #
-      #
-      def categorize mapper
-        @tokens.each { |token| token.categorize mapper }
-      end
-
       #
       #
       def originals

data/lib/picky/search_facets.rb

@@ -16,6 +16,8 @@ module Picky
     #   search.facets :name, filter: 'surname:peter', more_than: 0
     #
     def facets category_identifier, options = {}
+      # TODO Make it work. How should it work with multiple indexes?
+      #
       raise "#{__method__} cannot be used on searches with more than 1 index yet. Sorry!" if indexes.size > 1
       index = indexes.first
       
@@ -27,9 +29,21 @@ module Picky
       #
       return counts unless filter_query = options[:filter]
       
+      # Pre-tokenize query token category.
+      #
+      predefined_categories = [index[category_identifier]]
+      
+      # Pre-tokenize key token – replace text below.
+      # Note: The original is not important.
+      #
+      # TODO Don't use predefined.
+      #
+      key_token = Query::Token.new '', nil, predefined_categories
+      
       # Pre-tokenize filter for reuse.
       #
-      tokenized_filter = tokenized filter_query, false
+      tokenized_filter_query = tokenized filter_query, false
+      tokenized_filter_query.tokens.push key_token
       
       # Extract options.
       #
@@ -38,15 +52,37 @@ module Picky
       
       # Get actual counts.
       #
-      counts.inject(no_counts ? [] : {}) do |result, (key, _)|
-        tokenized_query = tokenized "#{category_identifier}:#{key}", false
-        total = search_with(tokenized_filter + tokenized_query, 0, 0).total
+      if no_counts
+        facets_without_counts counts, minimal_counts, tokenized_filter_query, key_token.text
+      else
+        facets_with_counts    counts, minimal_counts, tokenized_filter_query, key_token.text
+      end
+    end
+    def facets_without_counts counts, minimal_counts, tokenized_filter_query, last_token_text
+      counts.inject([]) do |result, (key, _)|
+        # Replace only the key token text because that
+        # is the only information that changes in between
+        # queries.
+        #
+        last_token_text.replace key
+        total = search_with(tokenized_filter_query, 0, 0).total
+        
+        next result unless total >= minimal_counts
+        result << key
+      end
+    end
+    def facets_with_counts counts, minimal_counts, tokenized_filter_query, last_token_text
+      counts.inject({}) do |result, (key, _)|
+        # Replace only the key token text because that
+        # is the only information that changes in between
+        # queries.
+        #
+        last_token_text.replace key
+        total = search_with(tokenized_filter_query, 0, 0).total
+        
         next result unless total >= minimal_counts
-        if no_counts
-          result << key
-        else
-          result[key] = total; result
-        end
+        result[key] = total
+        result
       end
     end
 

data/lib/picky/sinatra.rb

@@ -2,6 +2,8 @@
 #
 require_relative 'sinatra/index_actions'
 
+# TODO Remove for 5.0.
+#
 module Picky
 
   # This Module is used to install delegator methods

data/lib/picky/tokenizer.rb

@@ -93,27 +93,37 @@ Case sensitive?     #{@case_sensitive ? "Yes." : "-"}
 
     # Splitting.
     #
-    # We allow Strings and Regexps.
+    # We allow Strings, Regexps, and things that respond to #split.
+    #
     # Note: We do not test against to_str since symbols do not work with String#split.
     #
-    def splits_text_on regexp_or_string
-      raise ArgumentError.new "#{__method__} takes a Regexp or String as argument, not a #{regexp_or_string.class}." unless Regexp === regexp_or_string || String === regexp_or_string
-      @splits_text_on = regexp_or_string
-    end
-    def split text
-      text.split @splits_text_on
+    def splits_text_on thing
+      raise ArgumentError.new "#{__method__} takes a Regexp or String or a thing that responds to #split as argument, not a #{thing.class}." unless Regexp === thing || thing.respond_to?(:split)
+      @splits_text_on = thing
+      if thing.respond_to? :split
+        def split text
+          @splits_text_on.split text
+        end
+      else
+        def split text
+          text.split @splits_text_on
+        end
+      end
     end
 
     # Normalizing.
     #
     # We only allow arrays.
     #
+    # TODO 5.0 Rename to normalizes(config)
+    # TODO 5.0 Rename to normalize(text)
+    #
     def normalizes_words regexp_replaces
-      raise ArgumentError.new "#{__method__} takes an Array of replaces as argument, not a #{regexp_replaces.class}." unless regexp_replaces.respond_to?(:to_ary)
+      raise ArgumentError.new "#{__method__} takes an Array of replaces as argument, not a #{regexp_replaces.class}." unless regexp_replaces.respond_to?(:to_ary) || regexp_replaces.respond_to?(:normalize_with_patterns)
       @normalizes_words_regexp_replaces = regexp_replaces
     end
     def normalize_with_patterns text
-      return text unless @normalizes_words_regexp_replaces
+      return text unless @normalizes_words_regexp_replaces # TODO Remove.
 
       @normalizes_words_regexp_replaces.each do |regex, replace|
         # This should be sufficient
@@ -196,8 +206,6 @@ Case sensitive?     #{@case_sensitive ? "Yes." : "-"}
         send method_name, value unless value.nil?
       end
     rescue NoMethodError => e
-      # TODO Print out valid options.
-      #
       raise <<-ERROR
 The option "#{e.name}" is not a valid option for a Picky tokenizer.
 Please see https://github.com/floere/picky/wiki/Indexing-configuration for valid options.

data/lib/picky.rb

@@ -18,6 +18,8 @@ module Picky
 
   # External libraries.
   #
+  # TODO Remove active support.
+  #
   require 'active_support/core_ext/module/delegation'
   require 'active_support/core_ext/logger'
   require 'active_support/core_ext/object/blank'
@@ -69,7 +71,7 @@ module Picky
   #  * Loggers::Silent
   #  * Loggers::Concise (default)
   #  * Loggers::Verbose
-  #  
+  #
   self.logger = Loggers::Default
 
 end