picky 1.4.1 → 1.4.2

data/lib/picky/cacher/partial/none.rb

@@ -1,31 +0,0 @@
-module Cacher
-
-  module Partial
-
-    # Does not generate a partial index.
-    #
-    class None < Strategy
-
-      # Returns an empty index.
-      #
-      def generate_from index
-        {}
-      end
-      
-      # Returns if this strategy's generated file is saved.
-      #
-      def saved?
-        false
-      end
-      
-      # Do not use the partial bundle for getting ids and weights.
-      #
-      def use_exact_for_partial?
-        true
-      end
-
-    end
-
-  end
-
-end

data/lib/picky/cacher/partial/strategy.rb

@@ -1,21 +0,0 @@
-module Cacher
-  module Partial
-    # Superclass for partial strategies.
-    #
-    class Strategy < Cacher::Strategy
-      
-      # Defines whether to use the exact bundle
-      # instead of the partial one.
-      #
-      # Default is @false@.
-      #
-      # For example: 
-      #  Partial::None.new # Uses the exact index instead of the partial one.
-      #
-      def use_exact_for_partial?
-        false
-      end
-      
-    end
-  end
-end

data/lib/picky/cacher/partial/substring.rb

@@ -1,118 +0,0 @@
-module Cacher
-
-  module Partial
-    
-    # Generates the right substrings for use in the substring strategy.
-    #
-    class SubstringGenerator
-      
-      attr_reader :from, :to
-      
-      def initialize from, to
-        @from, @to = from, to
-        
-        if @to.zero?
-          def each_subtoken token, &block
-            token.each_subtoken @from, &block
-          end
-        else
-          def each_subtoken token, &block
-            token[0..@to].intern.each_subtoken @from, &block
-          end
-        end
-        
-      end
-      
-    end
-    
-    # The subtoken partial strategy.
-    #
-    # If given "florian"
-    # it will index "floria", "flori", "flor", "flo", "fl", "f"
-    # (Depending on what the given from value is, the example is with option from: 1)
-    #
-    class Substring < Strategy
-      
-      # The from option signifies where in the symbol it
-      # will start in generating the subtokens.
-      #
-      # Examples:
-      #
-      # With :hello, and to: -1 (default)
-      # * from: 1 # => [:hello, :hell, :hel, :he, :h]
-      # * from: 4 # => [:hello, :hell]
-      #
-      # With :hello, and to: -2
-      # * from: 1 # => [:hell, :hel, :he, :h]
-      # * from: 4 # => [:hell]
-      #
-      def initialize options = {}
-        from     = options[:from] || 1
-        to = options[:to] || -1
-        @generator = SubstringGenerator.new from, to
-      end
-      
-      # Delegator to generator#from.
-      #
-      def from
-        @generator.from
-      end
-      
-      # Delegator to generator#to.
-      #
-      def to
-        @generator.to
-      end
-      
-      # Generates a partial index from the given index.
-      #
-      def generate_from index
-        result = {}
-        
-        # Generate for each key token the subtokens.
-        #
-        i = 0
-        index.each_key do |token|
-          i += 1
-          if i == 5000
-            timed_exclaim "Generating partial tokens for token #{token}. This appears every 5000 tokens."
-            i = 0
-          end
-          generate_for token, index, result
-        end
-        
-        # Remove duplicate ids.
-        #
-        # THINK If it is unique for a subtoken, it is
-        #       unique for all derived longer tokens.
-        #
-        result.each_value &:uniq!
-        
-        result
-      end
-      
-      private
-        
-        # To each shortened token of :test
-        # :test, :tes, :te, :t
-        # add all ids of :test
-        #
-        # "token" here means just text.
-        #
-        # THINK Could be improved by appending the aforegoing ids?
-        #
-        def generate_for token, index, result
-          @generator.each_subtoken(token) do |subtoken|
-            if result[subtoken]
-              result[subtoken] += index[token] # unique
-            else
-              result[subtoken] = index[token].dup
-            end
-          end
-        end
-        
-    end
-    
-  end
-  
-end

data/lib/picky/cacher/partial_generator.rb

@@ -1,15 +0,0 @@
-module Cacher
-
-  # The partial generator uses a subtoken(downto:1) generator as default.
-  #
-  class PartialGenerator < Generator
-    
-    # Generate a partial index based on the given index.
-    #
-    def generate strategy = Partial::Substring.new(from: 1)
-      strategy.generate_from self.index
-    end
-
-  end
-  
-end

data/lib/picky/cacher/similarity/default.rb

@@ -1,7 +0,0 @@
-module Cacher
-  module Similarity
-    # Default is no similarity.
-    #
-    Default = None.new
-  end
-end

data/lib/picky/cacher/similarity/double_levenshtone.rb

@@ -1,77 +0,0 @@
-# encoding: utf-8
-#
-module Cacher
-
-  module Similarity
-
-    # DoubleLevensthone means that it's a combination of
-    # * DoubleMetaphone
-    # and
-    # * Levenshtein
-    # :)
-    #
-    class DoubleLevenshtone < Strategy
-
-      attr_reader :amount
-
-      #
-      #
-      def initialize amount = 10
-        @amount = amount
-      end
-
-      # Encodes the given symbol.
-      #
-      # Returns a symbol.
-      #
-      def encoded sym
-        codes = Text::Metaphone.double_metaphone sym.to_s
-        codes.first.to_sym unless codes.empty?
-      end
-
-      # Generates an index for the given index (in exact index style).
-      #
-      # In the following form:
-      # [:meier, :mueller, :peter, :pater] => { MR: [:meier], MLR: [:mueller], PTR: [:peter, :pater] }
-      #
-      def generate_from index
-        hash = hashify index.keys
-        sort hash
-      end
-
-      private
-
-        # Sorts the index values in place.
-        #
-        def sort index
-          index.each_pair.each do |code, ary|
-            ary.sort_by_levenshtein! code
-            ary.slice! amount, ary.size # size is not perfectly correct, but anyway
-          end
-          index
-        end
-
-        # Hashifies a list of symbols.
-        #
-        # Where:
-        # { encoded_sym => [syms] }
-        #
-        def hashify list
-          list.inject({}) do |total, element|
-            if code = encoded(element)
-              total[code] ||= []
-              total[code] << element
-            end
-            total
-          end
-        end
-
-    end
-    
-    # ... aka Phonetic.
-    #
-    Phonetic = DoubleLevenshtone
-
-  end
-
-end

data/lib/picky/cacher/similarity/none.rb

@@ -1,31 +0,0 @@
-module Cacher
-
-  module Similarity
-
-    # Similarity strategy that does nothing.
-    #
-    class None < Strategy
-
-      # Does not encode text. Just returns nil.
-      #
-      def encoded text
-        nil
-      end
-
-      # Returns an empty index.
-      #
-      def generate_from index
-        {}
-      end
-      
-      # Returns if this strategy's generated file is saved.
-      #
-      def saved?
-        false
-      end
-
-    end
-
-  end
-
-end

data/lib/picky/cacher/similarity/strategy.rb

@@ -1,9 +0,0 @@
-module Cacher
-  module Similarity
-    # Base class for all similarity strategies.
-    #
-    class Strategy < Cacher::Strategy
-      
-    end
-  end
-end

data/lib/picky/cacher/similarity_generator.rb

@@ -1,15 +0,0 @@
-module Cacher
-
-  # Uses no similarity as default.
-  #
-  class SimilarityGenerator < Generator
-
-    # Generate a similarity index based on the given index.
-    #
-    def generate strategy = Similarity::None.new
-      strategy.generate_from self.index
-    end
-
-  end
-
-end

data/lib/picky/cacher/strategy.rb

@@ -1,12 +0,0 @@
-module Cacher
-  class Strategy
-    
-    # By default, all caches are saved in a
-    # storage (like a file).
-    #
-    def saved?
-      true
-    end
-    
-  end
-end

data/lib/picky/cacher/weights/default.rb

@@ -1,7 +0,0 @@
-module Cacher
-  module Weights
-    # Default is Logarithmic.
-    #
-    Default = Logarithmic.new
-  end
-end

data/lib/picky/cacher/weights/logarithmic.rb

@@ -1,39 +0,0 @@
-module Cacher
-
-  module Weights
-
-    # Uses a logarithmic weight.
-    # If for a key k we have x ids, the weight is:
-    # w(x): log(x)
-    # Special case: If x < 1, then we use 0.
-    #
-    class Logarithmic < Strategy
-
-      # Generates a partial index from the given index.
-      #
-      def generate_from index
-        index.inject({}) do |hash, text_ids|
-          text, ids = *text_ids
-          weight = weight_for ids.size
-          hash[text] ||= weight.round(2) if weight
-          hash
-        end
-      end
-
-      # Sets the weight value.
-      #
-      # If the size is 0 or one, we would get -Infinity or 0.0.
-      # Thus we do not set a value if there is just one. The default, dynamically, is 0.
-      #
-      # BUT: We need the value, even if 0. To designate that there is a weight!
-      #
-      def weight_for amount
-        return 0 if amount < 1
-        Math.log amount
-      end
-
-    end
-
-  end
-
-end

data/lib/picky/cacher/weights/strategy.rb

@@ -1,9 +0,0 @@
-module Cacher
-  module Weights
-    # Superclass for weighing strategies.
-    #
-    class Strategy < Cacher::Strategy
-      
-    end
-  end
-end

data/lib/picky/cacher/weights_generator.rb

@@ -1,15 +0,0 @@
-module Cacher
-
-  # Uses a logarithmic algorithm as default.
-  #
-  class WeightsGenerator < Generator
-
-    # Generate a weights index based on the given index.
-    #
-    def generate strategy = Weights::Logarithmic.new
-      strategy.generate_from self.index
-    end
-
-  end
-
-end

data/lib/picky/frontend_adapters/rack.rb

@@ -1,150 +0,0 @@
-require 'rack/mount'
-
-module FrontendAdapters
-
-  # TODO Rename to Routing again. Push everything back into appropriate Adapters.
-  #
-  class Rack # :nodoc:all
-  
-    @@defaults = {
-      query_key:    'query'.freeze,
-      offset_key:   'offset'.freeze,
-      content_type: 'application/octet-stream'.freeze # TODO Wrong.
-    }
-  
-    def initialize
-      @defaults = @@defaults.dup
-    end
-  
-    # 
-    #
-    def reset_routes
-      @routes = ::Rack::Mount::RouteSet.new
-    end
-    def routes
-      @routes || reset_routes
-    end
-    def finalize
-      routes.freeze
-    end
-  
-    # Routing simply delegates to the route set to handle a request.
-    #
-    def call env
-      routes.call env
-    end
-  
-    # API method.
-    #
-    def route options = {}
-      mappings, route_options = split options
-      mappings.each do |url, query|
-        route_one url, query, route_options
-      end
-    end
-    # Splits the route method options
-    # into real options and route options (/regexp/ => thing or 'some/path' => thing).
-    #
-    def split options
-      mappings      = {}
-      route_options = {}
-      options.each_pair do |key, value|
-        if Regexp === key or String === key
-          mappings[key] = value
-        else
-          route_options[key] = value
-        end
-      end
-      [mappings, route_options]
-    end
-    def route_one url, query, route_options = {}
-      raise RouteTargetNilError.new(url) unless query
-      routes.add_route Adapters::Rack.app_for(query, route_options), default_options(url, route_options)
-    end
-    class RouteTargetNilError < StandardError
-      def initialize url
-        @url = url
-      end
-      def to_s
-        "Routing for #{@url.inspect} was defined with a nil target object, i.e. #{@url.inspect} => nil."
-      end
-    end
-    #
-    #
-    def root status
-      answer %r{^/$}, STATUSES[status]
-    end
-    #
-    #
-    def default status
-      answer nil, STATUSES[status]
-    end
-  
-  
-  
-    # TODO Can Rack handle this for me?
-    #
-    # Note: Rack-mount already handles the 404.
-    #
-    STATUSES = {
-      200 => lambda { |_| [200, { 'Content-Type' => 'text/html', 'Content-Length' => '0' }, ['']] },
-      404 => lambda { |_| [404, { 'Content-Type' => 'text/html', 'Content-Length' => '0' }, ['']] }
-    }
-  
-    #
-    #
-    def default_options url, route_options = {}
-      url = normalized url
-    
-      options = { request_method: 'GET' }.merge route_options
-    
-      options[:path_info] = url if url
-    
-      options.delete :content_type
-      
-      query_params = options.delete :query
-      options[:query_string] = %r{#{generate_query_string(query_params)}} if query_params
-    
-      options
-    end
-    #
-    #
-    def generate_query_string query_params
-      raise "At least one query string condition is needed." if query_params.size.zero?
-      raise "Too many query param conditions (only 1 allowed): #{query_params}" if query_params.size > 1
-      k, v = query_params.first
-      "#{k}=#{v}"
-    end
-    
-    # Setup a route that answers using the given app.
-    #
-    def answer url = nil, app = nil
-      routes.add_route (app || STATUSES[200]), default_options(url)
-    end
-  
-    # Returns a regular expression for the url if it is given a String-like object.
-    #
-    def normalized url
-      url.respond_to?(:to_str) ? %r{#{url}} : url
-    end
-  
-    # Returns true if there are no routes defined.
-    #
-    def empty?
-      routes.length.zero?
-    end
-  
-    # TODO Beautify.
-    #
-    def to_s
-      routes.instance_variable_get(:@routes).map do |route|
-        path_info = route.conditions[:path_info]
-        anchored = ::Rack::Mount::Utils.regexp_anchored?(path_info)
-        anchored_ok = anchored ? "\u2713" : " "
-        "#{anchored_ok}  #{path_info.source}"
-      end.join "\n"
-    end
-  
-  end
-  
-end

data/lib/picky/index/bundle.rb

@@ -1,54 +0,0 @@
-module Index # :nodoc:all
-  # A Bundle is a number of indexes
-  # per [index, category] combination.
-  #
-  # At most, there are three indexes:
-  # * *core* index (always used)
-  # * *weights* index (always used)
-  # * *similarity* index (used with similarity)
-  # 
-  # In Picky, indexing is separated from the index
-  # handling itself through a parallel structure.
-  #
-  # Both use methods provided by this base class, but
-  # have very different goals:
-  #
-  # * *Indexing*::*Bundle* is just concerned with creating index files
-  #   and providing helper functions to e.g. check the indexes.
-  #
-  # * *Index*::*Bundle* is concerned with loading these index files into
-  #   memory and looking up search data as fast as possible.
-  #
-  class Bundle
-    
-    attr_reader   :identifier, :files
-    attr_accessor :index, :weights, :similarity, :configuration, :similarity_strategy
-    
-    delegate :clear,    :to => :index
-    delegate :[], :[]=, :to => :configuration
-    
-    def initialize name, configuration, similarity_strategy
-      @identifier = "#{configuration.identifier} (#{name})"
-      @files      = Files.new name, configuration
-      
-      @index         = {}
-      @weights       = {}
-      @similarity    = {}
-      @configuration = {} # A hash with config options.
-      
-      @similarity_strategy = similarity_strategy
-    end
-    
-    # Get a list of similar texts.
-    #
-    # Note: Does not return itself.
-    #
-    def similar text
-      code = similarity_strategy.encoded text
-      similar_codes = code && @similarity[code]
-      similar_codes.delete text if similar_codes
-      similar_codes || []
-    end
-    
-  end
-end