picky 2.7.0 → 3.0.0.pre1

data/lib/picky/category_indexing.rb

@@ -1,148 +1,146 @@
-#
-#
-class Category
+module Picky
 
-  attr_reader :indexing_exact,
-              :indexing_partial
-
-  # Prepares and caches this category.
   #
-  # This one should be used by users.
   #
-  def index
-    prepare
-    cache
-  end
+  class Category
 
-  # Indexes, creates the "prepared_..." file.
-  #
-  def prepare
-    with_data_snapshot do
-      indexer.index [self]
+    attr_reader :indexing_exact,
+                :indexing_partial
+
+    # Prepares and caches this category.
+    #
+    # This one should be used by users.
+    #
+    def index
+      prepare
+      cache
     end
-  end
 
-  # Take a data snapshot if the source offers it.
-  #
-  def with_data_snapshot
-    if source.respond_to? :with_snapshot
-      source.with_snapshot(@index) do
+    # Indexes, creates the "prepared_..." file.
+    #
+    def prepare
+      with_data_snapshot do
+        indexer.index [self]
+      end
+    end
+
+    # Take a data snapshot if the source offers it.
+    #
+    def with_data_snapshot
+      if source.respond_to? :with_snapshot
+        source.with_snapshot(@index) do
+          yield
+        end
+      else
         yield
       end
-    else
-      yield
     end
-  end
 
-  # Generates all caches for this category.
-  #
-  def cache
-    configure
-    generate_caches_from_source
-    generate_partial
-    generate_caches_from_memory
-    dump_caches
-    timed_exclaim %Q{"#{identifier}": Caching finished.}
-  end
-  # Generate the cache data.
-  #
-  def generate_caches_from_source
-    indexing_exact.generate_caches_from_source
-  end
-  def generate_partial
-    indexing_partial.generate_partial_from indexing_exact.inverted
-  end
-  def generate_caches_from_memory
-    indexing_partial.generate_caches_from_memory
-  end
-  def dump_caches
-    indexing_exact.dump
-    indexing_partial.dump
-  end
+    # Generates all caches for this category.
+    #
+    def cache
+      generate_caches_from_source
+      generate_partial
+      generate_caches_from_memory
+      dump_caches
+      timed_exclaim %Q{"#{identifier}": Caching finished.}
+    end
+    # Generate the cache data.
+    #
+    def generate_caches_from_source
+      indexing_exact.generate_caches_from_source
+    end
+    def generate_partial
+      indexing_partial.generate_partial_from indexing_exact.inverted
+    end
+    def generate_caches_from_memory
+      indexing_partial.generate_caches_from_memory
+    end
+    def dump_caches
+      indexing_exact.dump
+      indexing_partial.dump
+    end
 
-  # Return an appropriate source.
-  #
-  # If we have no explicit source, we'll check the index for one.
-  #
-  def source
-    @source || @index.source
-  end
+    # Return an appropriate source.
+    #
+    # If we have no explicit source, we'll check the index for one.
+    #
+    def source
+      (@source && extract_source) || @index.source
+    end
+    # Extract the actual source if it is wrapped in a time
+    # capsule, i.e. a block/lambda.
+    #
+    # TODO Extract into module.
+    #
+    def extract_source
+      @source = @source.respond_to?(:call) ? @source.call : @source
+    end
 
-  # Return the key format.
-  #
-  # If the source has no key format, then
-  # check for an explicit key format, and
-  # if none is defined, ask the index for
-  # one.
-  #
-  def key_format
-    source.respond_to?(:key_format) && source.key_format || @key_format || @index.key_format
-  end
+    # Return the key format.
+    #
+    # If the source has no key format, and
+    # none is defined on this category, ask
+    # the index for one.
+    #
+    def key_format
+      source.respond_to?(:key_format) && source.key_format || @key_format || @index.key_format
+    end
 
-  # Where the data is taken from.
-  #
-  def from
-    @from || name
-  end
+    # Where the data is taken from.
+    #
+    def from
+      @from || name
+    end
 
-  # The indexer is lazily generated and cached.
-  #
-  # TODO Really cache?
-  #
-  def indexer
-    @indexer ||= source.respond_to?(:each) ? Indexers::Parallel.new(self) : Indexers::Serial.new(self)
-  end
+    # The indexer is lazily generated and cached.
+    #
+    def indexer
+      @indexer ||= source.respond_to?(:each) ? Indexers::Parallel.new(self) : Indexers::Serial.new(self)
+    end
 
-  # Returns an appropriate tokenizer.
-  # If one isn't set on this category, will try the index,
-  # and finally the default index tokenizer.
-  #
-  def tokenizer
-    @tokenizer || @index.tokenizer
-  end
+    # Returns an appropriate tokenizer.
+    # If one isn't set on this category, will try the index,
+    # and finally the default index tokenizer.
+    #
+    def tokenizer
+      @tokenizer || @index.tokenizer
+    end
 
-  # We need to set what formatting method should be used.
-  # Uses the one defined in the indexer.
-  #
-  # TODO Make this more dynamic.
-  #
-  def configure
-    indexing_exact[:key_format] = self.key_format
-    indexing_partial[:key_format] = self.key_format
-  end
+    # Checks the caches for existence.
+    #
+    def check
+      timed_exclaim "Checking #{identifier}."
+      indexing_exact.raise_unless_cache_exists
+      indexing_partial.raise_unless_cache_exists
+    end
 
-  # Checks the caches for existence.
-  #
-  def check
-    timed_exclaim "Checking #{identifier}."
-    indexing_exact.raise_unless_cache_exists
-    indexing_partial.raise_unless_cache_exists
-  end
+    # Deletes the caches.
+    #
+    def clear
+      timed_exclaim "Deleting #{identifier}."
+      indexing_exact.delete
+      indexing_partial.delete
+    end
 
-  # Deletes the caches.
-  #
-  def clear
-    timed_exclaim "Deleting #{identifier}."
-    indexing_exact.delete
-    indexing_partial.delete
-  end
+    # Backup the caches.
+    # (Revert with restore_caches)
+    #
+    def backup
+      timed_exclaim "Backing up #{identifier}."
+      indexing_exact.backup
+      indexing_partial.backup
+    end
 
-  # Backup the caches.
-  # (Revert with restore_caches)
-  #
-  def backup
-    timed_exclaim "Backing up #{identifier}."
-    indexing_exact.backup
-    indexing_partial.backup
-  end
+    # Restore the caches.
+    # (Revert with backup_caches)
+    #
+    def restore
+      timed_exclaim "Restoring #{identifier}."
+      indexing_exact.restore
+      indexing_partial.restore
+    end
 
-  # Restore the caches.
-  # (Revert with backup_caches)
-  #
-  def restore
-    timed_exclaim "Restoring #{identifier}."
-    indexing_exact.restore
-    indexing_partial.restore
   end
 
 end

data/lib/picky/character_substituters/west_european.rb

@@ -1,37 +1,43 @@
 # encoding: utf-8
 #
-module CharacterSubstituters # :nodoc:all
-  # Substitutes Umlauts like
-  # ä, ö, ü => ae, oe, ue.
-  # (and more, see specs)
-  #
-  class WestEuropean
-
-    def initialize
-      @chars = ActiveSupport::Multibyte.proxy_class
-    end
+module Picky
 
-    def to_s
-      self.class.name
-    end
+  module CharacterSubstituters # :nodoc:all
+
+    # Substitutes Umlauts like
+    # ä, ö, ü => ae, oe, ue.
+    # (and more, see specs)
+    #
+    class WestEuropean
+
+      def initialize
+        @chars = ActiveSupport::Multibyte.proxy_class
+      end
 
-    def substitute text
-      trans = @chars.new(text).normalize(:kd)
+      def to_s
+        self.class.name
+      end
 
-      # substitute special cases
-      #
-      trans.gsub!('ß', 'ss')
+      def substitute text
+        trans = @chars.new(text).normalize(:kd)
 
-      # substitute umlauts (of A,O,U,a,o,u)
-      #
-      trans.gsub!(/([AOUaou])\314\210/u, '\1e')
+        # substitute special cases
+        #
+        trans.gsub!('ß', 'ss')
+
+        # substitute umlauts (of A,O,U,a,o,u)
+        #
+        trans.gsub!(/([AOUaou])\314\210/u, '\1e')
+
+        # get rid of ecutes, graves and …
+        #
+        trans.unpack('U*').select { |cp|
+          cp < 0x0300 || cp > 0x035F
+        }.pack('U*')
+      end
 
-      # get rid of ecutes, graves and …
-      #
-      trans.unpack('U*').select { |cp|
-        cp < 0x0300 || cp > 0x035F
-      }.pack('U*')
     end
 
   end
+
 end

data/lib/picky/cores.rb

@@ -1,104 +1,108 @@
-Infinity = 1.0/0
+module Picky
 
-# Handles processing over multiple cores.
-#
-class Cores # :nodoc:all
-  
-  # Pass it an ary or generator.
-  #
-  #   generator = (1..10).each
-  #   forked generator, :max => 5 do |element|
-  #     
-  #   end
-  #
-  # Options include:
-  #  * max: Maximum # of processors to use. Default is all it can get.
+  Infinity = 1.0/0
+
+  # Handles processing over multiple cores.
   #
-  def self.forked ary_or_generator, options = {}
-    return if ary_or_generator.empty?
-    raise "Block argument needed when running Cores.forked" unless block_given?
-    
-    ary_or_generator = ary_or_generator.sort_by { rand } if options[:randomly]
-    generator        = ary_or_generator.each
-    
-    # Get the maximum number of processors.
+  class Cores # :nodoc:all
+
+    # Pass it an ary or generator.
     #
-    max                  = max_processors options
-    currently_processing = 0
-    
-    # 
+    #   generator = (1..10).each
+    #   forked generator, :max => 5 do |element|
     #
-    loop do
-      # Ramp it up to num processors.
+    #   end
+    #
+    # Options include:
+    #  * max: Maximum # of processors to use. Default is all it can get.
+    #
+    def self.forked ary_or_generator, options = {}
+      return if ary_or_generator.empty?
+      raise "Block argument needed when running Cores.forked" unless block_given?
+
+      ary_or_generator = ary_or_generator.sort_by { rand } if options[:randomly]
+      generator        = ary_or_generator.each
+
+      # Get the maximum number of processors.
+      #
+      max                  = max_processors options
+      currently_processing = 0
+
       #
-      while currently_processing < max
-        
-        currently_processing = currently_processing + 1
-        
-        element = nil
+      #
+      loop do
+        # Ramp it up to num processors.
+        #
+        while currently_processing < max
+
+          currently_processing = currently_processing + 1
+
+          element = nil
+          begin
+            element = generator.next
+          rescue StopIteration => si
+            break
+          end
+          break unless element
+
+          Process.fork do
+            sleep 0.01*currently_processing
+            yield element
+          end
+
+        end
+
         begin
-          element = generator.next
-        rescue StopIteration => si
+          Process.wait 0 # Block and wait for any child to finish.
+        rescue Errno::ECHILD => e
           break
+        ensure
+          currently_processing = currently_processing - 1
         end
-        break unless element
-        
-        Process.fork do
-          sleep 0.01*currently_processing
-          yield element
-        end
-        
-      end
-      
-      begin
-        Process.wait 0 # Block and wait for any child to finish.
-      rescue Errno::ECHILD => e
-        break
-      ensure
-        currently_processing = currently_processing - 1
       end
     end
+
+    # Return the number of maximum usable processors.
+    #
+    def self.max_processors options = {}
+      options[:amount] || [number_of_cores, (options[:max] || Infinity)].min
+    end
+
+    # Gets the number of cores depending on OS.
+    #
+    def self.number_of_cores
+      extract_cores_for actual_platform
+    end
+    # Extracts the platform os from the platform.
+    #
+    # Note: Could also use 'rbconfig'.
+    #
+    def self.actual_platform
+      matched = platform.match(/-\b([a-z]*)/)
+      matched && matched[1]
+    end
+    # Returns a mapping
+    #   os_name => lambda_which_returns_a_number_of_cores
+    #
+    @@number_of_cores = {
+      'darwin' => lambda { `system_profiler SPHardwareDataType | grep -i 'Total Number Of Cores'`.gsub(/[^\d]/, '') },
+      'linux'  => lambda { `grep -ci ^processor /proc/cpuinfo` }
+    }
+    def self.os_to_core_mapping
+      @@number_of_cores
+    end
+    # Extracts the number of cores for the given os name.
+    #
+    # Note: Default is 1.
+    #
+    def self.extract_cores_for os
+      code_to_execute = os_to_core_mapping[os]
+      code_to_execute && code_to_execute.call.to_i || 1
+    end
+    def self.platform
+      RUBY_PLATFORM
+    end
+
   end
-  
-  # Return the number of maximum usable processors.
-  #
-  def self.max_processors options = {}
-    options[:amount] || [number_of_cores, (options[:max] || Infinity)].min
-  end
-  
-  # Gets the number of cores depending on OS.
-  #
-  def self.number_of_cores
-    extract_cores_for actual_platform
-  end
-  # Extracts the platform os from the platform.
-  #
-  # Note: Could also use 'rbconfig'.
-  #
-  def self.actual_platform
-    matched = platform.match(/-\b([a-z]*)/)
-    matched && matched[1]
-  end
-  # Returns a mapping
-  #   os_name => lambda_which_returns_a_number_of_cores
-  #
-  @@number_of_cores = {
-    'darwin' => lambda { `system_profiler SPHardwareDataType | grep -i 'Total Number Of Cores'`.gsub(/[^\d]/, '') },
-    'linux'  => lambda { `grep -ci ^processor /proc/cpuinfo` }
-  }
-  def self.os_to_core_mapping
-    @@number_of_cores
-  end
-  # Extracts the number of cores for the given os name.
-  #
-  # Note: Default is 1.
-  #
-  def self.extract_cores_for os
-    code_to_execute = os_to_core_mapping[os]
-    code_to_execute && code_to_execute.call.to_i || 1
-  end
-  def self.platform
-    RUBY_PLATFORM
-  end
-  
+
 end