tzinfo-data 1.2013.3

data/lib/tzinfo/data/tzdataparser.rb

@@ -0,0 +1,1270 @@
+#--
+# Copyright (c) 2005-2013 Philip Ross
+# 
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+# 
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+# THE SOFTWARE.
+#++
+
+require 'date'
+require 'fileutils'
+
+module TZInfo
+  module Data
+  
+    # Utility methods used by TZDataParser and associated classes.
+    module TZDataParserUtils #:nodoc:
+      
+      begin
+        Encoding
+        SUPPORTS_ENCODING = true
+      rescue NameError
+        SUPPORTS_ENCODING = false
+      end
+      
+      if SUPPORTS_ENCODING
+        def open_file(file_name, mode, opts, &block)
+          File.open(file_name, mode, opts, &block)
+        end
+      else
+        def open_file(file_name, mode, opts, &block)
+          File.open(file_name, mode, &block)
+        end
+      end
+      
+      private :open_file
+      
+      private
+      
+        # Parses a month specified in the tz data and converts it to a number
+        # between 1 and 12 representing January to December.
+        def parse_month(month)
+          lower = month.downcase
+          if lower =~ /^jan/
+            @month = 1
+          elsif lower =~ /^feb/
+            @month = 2
+          elsif lower =~ /^mar/
+            @month = 3
+          elsif lower =~ /^apr/
+            @month = 4
+          elsif lower =~ /^may/
+            @month = 5
+          elsif lower =~ /^jun/
+            @month = 6
+          elsif lower =~ /^jul/
+            @month = 7
+          elsif lower =~ /^aug/
+            @month = 8
+          elsif lower =~ /^sep/
+            @month = 9
+          elsif lower =~ /^oct/
+            @month = 10
+          elsif lower =~ /^nov/
+            @month = 11
+          elsif lower =~ /^dec/
+            @month = 12
+          else
+            raise "Invalid month: #{month}"
+          end
+        end
+        
+        # Parses an offset string [-]h:m:s (minutes and seconds are optional). Returns
+        # the offset in seconds.
+        def parse_offset(offset)
+          raise "Invalid time: #{offset}" if offset !~ /^(-)?(?:([0-9]+)(?::([0-9]+)(?::([0-9]+))?)?)?$/
+          
+          negative = !$1.nil?      
+          hour = $2.nil? ? 0 : $2.to_i
+          minute = $3.nil? ? 0 : $3.to_i
+          second = $4.nil? ? 0 : $4.to_i
+          
+          seconds = hour
+          seconds = seconds * 60
+          seconds = seconds + minute
+          seconds = seconds * 60
+          seconds = seconds + second
+          seconds = -seconds if negative
+          seconds
+        end
+        
+        # Encloses the string in single quotes and escapes any single quotes in
+        # the content.
+        def quote_str(str)
+          "'#{str.gsub('\'', '\\\\\'')}'"
+        end
+    end 
+  
+    # Parses Time Zone Data from the IANA Time Zone Database and transforms it 
+    # into a set of Ruby modules that can be used with TZInfo.
+    # 
+    # Normally, this class wouldn't be used. It is only run to update the 
+    # timezone data and index modules.
+    class TZDataParser
+      include TZDataParserUtils
+    
+      # Default earliest year that will be considered.
+      DEFAULT_MIN_YEAR = 1800
+      
+      # Default number of future years data to generate (not including the 
+      # current year).
+      DEFAULT_FUTURE_YEARS = 50
+      
+      # Earliest year that will be considered. Defaults to DEFAULT_MIN_YEAR.
+      attr_accessor :min_year
+      
+      # Latest year that will be considered. Defaults to the current year plus
+      # FUTURE_YEARS.
+      attr_accessor :max_year
+      
+      # Whether to generate zone definitions (set to false to stop zones being
+      # generated).    
+      attr_accessor :generate_zones
+      
+      # Whether to generate country definitions (set to false to stop countries
+      # being generated).
+      attr_accessor :generate_countries
+      
+      # Limit the set of zones to generate (set to an array containing zone
+      # identifiers).
+      attr_accessor :only_zones
+          
+      # Zones to exclude from generation when not using only_zones (set to an
+      # array containing zone identifiers).
+      attr_accessor :exclude_zones
+      
+      # Initializes a new TZDataParser. input_dir must contain the extracted
+      # tzdata tarball. output_dir is the location to output the modules
+      # (in definitions and indexes directories).
+      def initialize(input_dir, output_dir)
+        super()
+        @input_dir = input_dir
+        @output_dir = output_dir
+        @min_year = DEFAULT_MIN_YEAR
+        @max_year = Time.now.year + DEFAULT_FUTURE_YEARS
+        @rule_sets = {}
+        @zones = {}
+        @countries = {}
+        @no_rules = TZDataNoRules.new
+        @generate_zones = true
+        @generate_countries = true
+        @only_zones = []
+        @exclude_zones = []
+      end
+      
+      # Reads the tzdata source and generates the classes. Progress information
+      # is written to standard out.
+      def execute
+        files = Dir.entries(@input_dir).select do |file|
+          file =~ /\A[^\.]+\z/ &&            
+            !%w(README Makefile).include?(file) &&
+            File.file?(File.join(@input_dir, file))
+        end
+      
+        files.each {|file| load_rules(file) }
+        files.each {|file| load_zones(file) }
+        files.each {|file| load_links(file) }
+        
+        load_countries
+        
+        if @generate_zones
+          modules = []
+          
+          if @only_zones.nil? || @only_zones.empty?
+            @zones.each_value {|zone|
+              zone.write_module(@output_dir) unless @exclude_zones.include?(zone.name)
+            }
+          else
+            @only_zones.each {|id|
+              zone = @zones[id]
+              zone.write_module(@output_dir)            
+            }          
+          end
+          
+          write_timezones_index
+        end
+        
+        if @generate_countries        
+          write_countries_index
+        end
+      end
+          
+      private     
+        # Loads all the Rule definitions from the tz data and stores them in
+        # @rule_sets.
+        def load_rules(file)
+          puts 'load_rules: ' + file
+          
+          # Files are in ASCII, but may change to UTF-8 (a superset of ASCII)
+          # in the future.
+          open_file(File.join(@input_dir, file), 'r', :external_encoding => 'UTF-8', :internal_encoding => 'UTF-8') do |file|
+            file.each_line do |line|
+              line = line.gsub(/#.*$/, '')
+              line = line.gsub(/\s+$/, '')
+            
+              if line =~ /^Rule\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)/
+                
+                name = $1    
+                
+                if @rule_sets[name].nil? 
+                  @rule_sets[name] = TZDataRuleSet.new(name) 
+                end
+                
+                @rule_sets[name].add_rule(TZDataRule.new($2, $3, $4, $5, $6, $7, $8, $9))
+              end          
+            end
+          end
+        end
+        
+        # Gets a rules object for the given reference. Might be a named rule set,
+        # a fixed offset or an empty ruleset.
+        def get_rules(ref)
+          if ref == '-'
+            @no_rules
+          elsif ref =~ /^[0-9]+:[0-9]+$/
+            TZDataFixedOffsetRules.new(parse_offset(ref))
+          else
+            rule_set = @rule_sets[ref]
+            raise "Ruleset not found: #{ref}" if rule_set.nil?    
+            rule_set
+          end  
+        end
+        
+        # Loads in the Zone definitions from the tz data and stores them in @zones.
+        def load_zones(file)
+          puts 'load_zones: ' + file
+          
+          in_zone = nil
+          
+          # Files are in ASCII, but may change to UTF-8 (a superset of ASCII)
+          # in the future.
+          open_file(File.join(@input_dir, file), 'r', :external_encoding => 'UTF-8', :internal_encoding => 'UTF-8') do |file|
+            file.each_line do |line|
+              line = line.gsub(/#.*$/, '')
+              line = line.gsub(/\s+$/, '')
+            
+              if in_zone
+                if line =~ /^\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)(\s+([0-9]+(\s+.*)?))?$/              
+                  
+                  in_zone.add_observance(TZDataObservance.new($1, get_rules($2), $3, $5))
+                  
+                  in_zone = nil if $4.nil? 
+                end
+              else
+                if line =~ /^Zone\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)\s+([^\s]+)(\s+([0-9]+(\s+.*)?))?$/              
+                  name = $1
+                  
+                  if @zones[name].nil? 
+                    @zones[name] = TZDataZone.new(name, @min_year..@max_year)
+                  end                            
+                  
+                  @zones[name].add_observance(TZDataObservance.new($2, get_rules($3), $4, $6))
+                  
+                  in_zone = @zones[name] if !$5.nil?
+                end
+              end
+            end
+          end
+        end
+        
+        # Loads in the links and stores them in @zones.
+        def load_links(file)
+          puts 'load_links: ' + file
+          
+          # Files are in ASCII, but may change to UTF-8 (a superset of ASCII)
+          # in the future.
+          open_file(File.join(@input_dir, file), 'r', :external_encoding => 'UTF-8', :internal_encoding => 'UTF-8') do |file|
+            file.each_line do |line|
+              line = line.gsub(/#.*$/, '')
+              line = line.gsub(/\s+$/, '')
+            
+              if line =~ /^Link\s+([^\s]+)\s+([^\s]+)/            
+                name = $2           
+                link_to = @zones[$1]
+                raise "Link to zone not found (#{name}->#{link_to})" if link_to.nil?
+                raise "Zone already defined: #{name}" if !@zones[name].nil?
+                @zones[name] = TZDataLink.new(name, link_to)
+              end          
+            end
+          end
+        end
+        
+        # Loads countries from iso3166.tab and zone.tab and stores the result in
+        # @countries.
+        def load_countries
+          puts 'load_countries'
+          
+          # Files are in ASCII, but may change to UTF-8 (a superset of ASCII)
+          # in the future.
+          open_file(File.join(@input_dir, 'iso3166.tab'), 'r', :external_encoding => 'UTF-8', :internal_encoding => 'UTF-8') do |file|
+            file.each_line do |line|
+          
+              if line =~ /^([A-Z]{2})\t(.*)$/
+                code = $1
+                name = $2
+                @countries[code] = TZDataCountry.new(code, name)
+              end
+            end
+          end
+
+          # Files are in ASCII, but may change to UTF-8 (a superset of ASCII)
+          # in the future.
+          open_file(File.join(@input_dir, 'zone.tab'), 'r', :external_encoding => 'UTF-8', :internal_encoding => 'UTF-8') do |file|
+            file.each_line do |line|
+                    
+              line.chomp!          
+
+              if line =~ /^([A-Z]{2})\t([^\t]+)\t([^\t]+)(\t(.*))?$/
+                code = $1
+                location_str = $2
+                zone_name = $3
+                description = $5
+
+                country = @countries[code]
+                raise "Country not found: #{code}" if country.nil?
+                
+                location = TZDataLocation.new(location_str)
+                
+                zone = @zones[zone_name]
+                raise "Zone not found: #{zone_name}" if zone.nil?                        
+                
+                description = nil if description == ''
+                
+                country.add_zone(TZDataCountryTimezone.new(zone, description, location))
+              end
+            end
+          end
+        end
+        
+        # Writes a country index file.
+        def write_countries_index
+          dir = File.join(@output_dir, 'indexes')
+          FileUtils.mkdir_p(dir)
+          
+          open_file(File.join(dir, 'countries.rb'), 'w', :external_encoding => 'UTF-8', :universal_newline => true) do |file|
+            file.puts('# encoding: UTF-8')
+            file.puts('')
+            file.puts('# This file contains data derived from the IANA Time Zone Database')
+            file.puts('# (http://www.iana.org/time-zones).')
+            file.puts('')
+            
+            file.puts('module TZInfo')
+            file.puts('  module Data')
+            file.puts('    module Indexes')
+            file.puts('      module Countries')
+            file.puts('        include CountryIndexDefinition')
+            file.puts('')
+            
+            countries = @countries.values.sort {|c1,c2| c1.code <=> c2.code}  
+            countries.each {|country| country.write_index_record(file)}
+                      
+            file.puts('      end') # end module Countries                    
+            file.puts('    end') # end module Indexes
+            file.puts('  end') # end module Data
+            file.puts('end') # end module TZInfo
+          end                      
+        end
+        
+        # Writes a timezone index file.
+        def write_timezones_index
+          dir = File.join(@output_dir, 'indexes')
+          FileUtils.mkdir_p(dir)
+          
+          open_file(File.join(dir, 'timezones.rb'), 'w', :external_encoding => 'UTF-8', :universal_newline => true) do |file|
+            file.puts('# encoding: UTF-8')
+            file.puts('')
+            file.puts('# This file contains data derived from the IANA Time Zone Database')
+            file.puts('# (http://www.iana.org/time-zones).')
+            file.puts('')
+            
+            file.puts('module TZInfo')
+            file.puts('  module Data')
+            file.puts('    module Indexes')
+            file.puts('      module Timezones')
+            file.puts('        include TimezoneIndexDefinition')
+            file.puts('')
+            
+            zones = @zones.values.sort {|t1,t2| t1.name <=> t2.name}
+            zones.each {|zone| zone.write_index_record(file)}
+            
+            file.puts('      end') # end module Timezones
+            file.puts('    end') # end module Indexes
+            file.puts('  end') # end module Data
+            file.puts('end') # end module TZInfo
+          end
+          
+        end
+    end
+    
+    # Base class for all rule sets.
+    class TZDataRules #:nodoc:
+      # Name of the rule set, e.g. EU.
+      attr_reader :name
+      
+      def initialize(name)
+        @name = name
+      end
+      
+      def count
+        0
+      end
+    end
+    
+    # Empty rule set with a fixed daylight savings (std) offset.
+    class TZDataFixedOffsetRules < TZDataRules #:nodoc:
+      attr_reader :offset
+      
+      def initialize(offset)
+        super(offset.to_s)
+        @offset = offset
+      end        
+    end
+    
+    # An empty set of rules.
+    class TZDataNoRules < TZDataRules #:nodoc:
+      def initialize
+        super('-')
+      end
+    end
+    
+    # A rule set (as defined by Rule name in the tz data).
+    class TZDataRuleSet < TZDataRules #:nodoc:       
+      attr_reader :rules
+      
+      def initialize(name)
+        super
+        @rules = []
+      end
+      
+      # Adds a new rule to the set.
+      def add_rule(rule)
+        @rules << rule
+      end
+      
+      def count
+        @rules.length
+      end
+      
+      def each
+        @rules.each {|rule| yield rule}
+      end
+    end    
+    
+    # A rule in a RuleSet (a single Rule line in the tz data).
+    class TZDataRule #:nodoc:
+      include TZDataParserUtils
+    
+      attr_reader :from
+      attr_reader :to
+      attr_reader :type
+      attr_reader :in_month
+      attr_reader :on_day
+      attr_reader :at_time
+      attr_reader :save
+      attr_reader :letter
+      
+      def initialize(from, to, type, in_month, on_day, at_time, save, letter)
+        @from = parse_from(from)
+        @to = parse_to(to)
+        
+        # replace a to of :only with the from year
+        raise 'to cannot be only if from is minimum' if @to == :only && @from == :min
+        @to = @from if @to == :only      
+        
+        @type = parse_type(type)
+        @in_month = parse_month(in_month)
+        @on_day = TZDataDayOfMonth.new(on_day)
+        @at_time = TZDataTime.new(at_time)
+        @save = parse_offset(save)
+        @letter = parse_letter(letter)
+      end
+      
+      def activate(year)
+        # The following test ignores yearistype at present (currently unused in
+        # the data. parse_type currently excepts on encountering a year type.
+        if (@from == :min || @from <= year) && (@to == :max || @to >= year)                        
+          TZDataActivatedRule.new(self, year)
+        else
+          nil
+        end
+      end
+      
+      def at_utc_time(year, utc_offset, std_offset)
+        @at_time.to_utc(utc_offset, std_offset,
+          year, @in_month, @on_day.to_absolute(year, @in_month))
+      end
+      
+      private
+        def parse_from(from)
+          lower = from.downcase
+          if lower =~ /^min/
+            :min
+          elsif lower =~ /^[0-9]+$/
+            lower.to_i
+          else
+            raise "Invalid from: #{from}"
+          end
+        end
+        
+        def parse_to(to)
+          lower = to.downcase
+          if lower =~ /^max/
+            :max
+          elsif lower =~ /^o/
+            :only
+          elsif lower =~ /^[0-9]+$/
+            lower.to_i
+          else
+            raise "Invalid to: #{to}"
+          end
+        end
+        
+        def parse_type(type)
+          raise "Unsupported rule type: #{type}" if type != '-'
+          nil
+        end        
+        
+        def parse_letter(letter)
+          if letter == '-'
+            nil
+          else
+            letter
+          end
+        end
+    end
+    
+    # Base class for Zones and Links.
+    class TZDataDefinition #:nodoc:
+      include TZDataParserUtils
+      
+      attr_reader :name
+      attr_reader :name_elements
+      attr_reader :path_elements
+      
+      def initialize(name)
+        @name = name
+        
+        # + and - aren't allowed in class names
+        @name_elements = name.gsub(/-/, '__m__').gsub(/\+/, '__p__').split(/\//)
+        @path_elements = @name_elements.clone
+        @path_elements.pop
+      end        
+      
+      # Creates necessary directories, the file, writes the class header and footer
+      # and yields to a block to write the content.    
+      def create_file(output_dir) 
+        dir = File.join(output_dir, 'definitions', @path_elements.join(File::SEPARATOR))
+        FileUtils.mkdir_p(dir)
+        
+        open_file(File.join(output_dir, 'definitions', @name_elements.join(File::SEPARATOR)) + '.rb', 'w', :external_encoding => 'UTF-8', :universal_newline => true) do |file|
+          
+          def file.indent(by)
+            if @tz_indent
+              @tz_indent += by
+            else
+              @tz_indent = by
+            end
+          end
+          
+          def file.puts(s)
+            super("#{' ' * (@tz_indent || 0)}#{s}")
+          end
+          
+          file.puts('# encoding: UTF-8')
+          file.puts('')
+          file.puts('# This file contains data derived from the IANA Time Zone Database')
+          file.puts('# (http://www.iana.org/time-zones).')
+          file.puts('')
+          file.puts('module TZInfo')
+          file.indent(2)
+          file.puts('module Data')
+          file.indent(2)
+          file.puts('module Definitions')
+          file.indent(2)                
+          
+          @name_elements.each do |part| 
+            file.puts("module #{part}")
+            file.indent(2)
+          end
+          
+          file.puts('include TimezoneDefinition')
+          file.puts('')
+          
+          yield file
+                          
+          @name_elements.each do
+            file.indent(-2)  
+            file.puts('end')          
+          end
+          file.indent(-2)
+          file.puts('end') # end module Definitions
+          file.indent(-2)
+          file.puts('end') # end module Data
+          file.indent(-2)
+          file.puts('end') # end module TZInfo
+        end
+      end    
+    end
+    
+    # A tz data Link.
+    class TZDataLink < TZDataDefinition #:nodoc:
+      include TZDataParserUtils
+    
+      attr_reader :link_to
+      
+      def initialize(name, link_to)
+        super(name)
+        @link_to = link_to
+      end
+                  
+      # Writes a module for this link.
+      def write_module(output_dir)
+        puts "writing link #{name}"
+              
+        create_file(output_dir) {|file|
+          file.puts("linked_timezone #{quote_str(@name)}, #{quote_str(@link_to.name)}")        
+        }
+      end
+      
+      # Writes an index record for this link.    
+      def write_index_record(file)
+        file.puts("        linked_timezone #{quote_str(@name)}")
+      end
+    end
+    
+    # A tz data Zone. Each line from the tz data is loaded as a TZDataObservance.
+    class TZDataZone < TZDataDefinition #:nodoc:
+      include TZDataParserUtils
+      
+      attr_reader :observances
+      
+      def initialize(name, years)
+        super(name)
+        @years = years
+        @observances = []
+      end
+      
+      def add_observance(observance)      
+        @observances << observance
+      end                
+      
+      # Writes the module for the zone. Iterates all the periods and asks them
+      # to write all periods in the timezone.
+      def write_module(output_dir) 
+        puts "writing zone #{name}"
+        
+        create_file(output_dir) {|file|        
+          
+          file.puts("timezone #{quote_str(@name)} do |tz|")
+          file.indent(2)
+                
+          transitions = find_transitions
+          transitions.output_module(file)
+          
+          file.indent(-2)
+          file.puts('end')
+        }      
+      end
+      
+      # Writes an index record for this zone.    
+      def write_index_record(file)
+        file.puts("        timezone #{quote_str(@name)}")
+      end
+      
+      private
+        def find_transitions
+          transitions = TZDataTransitions.new(@years)
+          
+          # algorithm from zic.c outzone
+          
+          start_time = nil
+          until_time = nil
+          
+          
+          @observances.each_with_index {|observance, i|
+            std_offset = 0
+            use_start = i > 0
+            use_until = i < @observances.length - 1
+            utc_offset = observance.utc_offset
+            start_zone_id = nil
+            start_utc_offset = observance.utc_offset
+            start_std_offset = 0
+            
+            if observance.rule_set.count == 0
+              std_offset = observance.std_offset
+              start_zone_id = observance.format.expand(std_offset, nil)
+              
+              if use_start
+                transitions << TZDataTransition.new(start_time, utc_offset, std_offset, start_zone_id)
+                use_start = false
+              else
+                # zic algorithm only outputs this if std_offset is non-zero
+                # to get the initial LMT range, we output this regardless
+                transitions << TZDataTransition.new(nil, utc_offset, std_offset, start_zone_id)
+              end  
+            else
+              @years.each {|year|
+                if use_until && year > observance.valid_until.year
+                  break
+                end
+                
+                activated_rules = []
+                
+                observance.rule_set.each {|rule|
+                  activated_rule = rule.activate(year)
+                  activated_rules << activated_rule unless activated_rule.nil?
+                }
+                
+                while true 
+                  # turn until_time into UTC using the current utc_offset and std_offset
+                  until_time = observance.valid_until.to_utc(utc_offset, std_offset) if use_until
+                  
+                  earliest = nil
+
+                  activated_rules.each {|activated_rule| 
+                    # recalculate the time using the current std_offset
+                    activated_rule.calculate_time(utc_offset, std_offset) 
+                    earliest = activated_rule if earliest.nil? || activated_rule.at < earliest.at                   
+                  }
+                  
+                  break if earliest.nil?                
+                  activated_rules.delete(earliest)                               
+                  break if use_until && earliest.at >= until_time                
+                  std_offset = earliest.rule.save                   
+                  use_start = false if use_start && earliest.at == start_time
+                  
+                  if use_start
+                    if earliest.at < start_time
+                      start_utc_offset = observance.utc_offset
+                      start_std_offset = std_offset
+                      start_zone_id = observance.format.expand(earliest.rule.save, earliest.rule.letter)                    
+                      next
+                    end
+                    
+                    if start_zone_id.nil? && start_utc_offset + start_std_offset == observance.utc_offset + std_offset
+                      start_zone_id = observance.format.expand(earliest.rule.save, earliest.rule.letter)
+                    end
+                  end
+                  
+                  zone_id = observance.format.expand(earliest.rule.save, earliest.rule.letter)
+                  transitions << TZDataTransition.new(earliest.at, observance.utc_offset, earliest.rule.save, zone_id)
+                end              
+              }
+            end
+          
+            if use_start
+              start_zone_id = observance.format.expand(nil, nil) if start_zone_id.nil? && observance.format.fixed?            
+              raise 'Could not determine time zone abbreviation to use just after until time' if start_zone_id.nil?            
+              transitions << TZDataTransition.new(start_time, start_utc_offset, start_std_offset, start_zone_id)
+            end
+                      
+            start_time = observance.valid_until.to_utc(utc_offset, std_offset) if use_until 
+          }
+          
+          transitions
+        end
+    end
+    
+    # A observance within a zone (a line within the zone definition).
+    class TZDataObservance #:nodoc:
+      include TZDataParserUtils
+    
+      attr_reader :utc_offset
+      attr_reader :rule_set
+      attr_reader :format
+      attr_reader :valid_until
+      
+      def initialize(utc_offset, rule_set, format, valid_until)
+        @utc_offset = parse_offset(utc_offset)
+        @rule_set = rule_set      
+        @format = TZDataFormat.new(format)
+        @valid_until = valid_until.nil? ? nil : TZDataUntil.new(valid_until)      
+      end 
+      
+      def std_offset
+        if @rule_set.kind_of?(TZDataFixedOffsetRules)
+          @rule_set.offset
+        else
+          0
+        end
+      end
+    end
+    
+    # Collection of TZDataTransition instances used when building a zone class.
+    class TZDataTransitions #:nodoc:
+      include TZDataParserUtils
+      
+      def initialize(years)
+        @years = years
+        @transitions = []
+      end
+      
+      def << (transition)
+        @transitions << transition
+      end 
+      
+      def output_module(file)
+        optimize
+        
+        # Try and end on a transition to std if one happens in the last year.
+        if @transitions.length > 1 && 
+            @transitions.last.std_offset != 0 &&
+            @transitions[@transitions.length - 2].std_offset == 0 &&
+            @transitions[@transitions.length - 2].at_utc.year == @years.max
+          
+          transitions = @transitions[0..@transitions.length - 2]
+        else
+          transitions = @transitions
+        end
+        
+        process_offsets(file)
+        file.puts('')
+        
+        transitions.each do |t|        
+          t.write(file)
+        end            
+      end
+      
+      private
+        def optimize
+          @transitions.sort!
+          
+          # Optimization logic from zic.c writezone.
+          
+          from_i = 0
+          to_i = 0
+          
+          while from_i < @transitions.length
+            if to_i > 1 && 
+                !@transitions[from_i].at_utc.nil?  &&
+                !@transitions[to_i - 1].at_utc.nil? &&
+                @transitions[from_i].at_utc + Rational(@transitions[to_i - 1].total_offset, 86400) <=
+                @transitions[to_i - 1].at_utc + Rational(@transitions[to_i - 2].total_offset, 86400)
+                        
+              @transitions[to_i - 1] = @transitions[from_i].clone_with_at(@transitions[to_i - 1].at_utc)             
+              from_i += 1
+              next
+            end
+            
+            # Shuffle transitions up, eliminating any redundant transitions 
+            # along the way.
+            if to_i == 0 ||
+                @transitions[to_i - 1].utc_offset != @transitions[from_i].utc_offset ||
+                @transitions[to_i - 1].std_offset != @transitions[from_i].std_offset ||
+                @transitions[to_i - 1].zone_id != @transitions[from_i].zone_id
+                        
+              @transitions[to_i] = @transitions[from_i]
+              to_i += 1            
+            end
+            
+            from_i += 1
+          end                
+          
+          if to_i > 0
+            @transitions = @transitions[0..to_i - 1]
+          else
+            @transitions = []
+          end                
+        end
+        
+        def quote_zone_id(zone_id)     
+          if zone_id =~ %r{[\-+']}
+            ":#{quote_str(zone_id)}"          
+          else
+            ":#{zone_id}"
+          end
+        end  
+        
+        def process_offsets(file)
+          # A bit of a hack at the moment. The offset used to be output with
+          # each period (pair of transitions). They are now separated from the
+          # transition data. The code should probably be changed at some point to
+          # setup the offsets at an earlier stage.   
+          
+          # Assume that when this is called, the first transition is the Local
+          # Mean Time initial rule or a transition with no time that defines the
+          # offset for the entire zone.
+          
+          offsets = []
+          
+          # Find the first std offset. Timezones always start in std.
+          @transitions.each do |t|
+            if t.std_offset == 0
+              offset = {:utc_offset => t.utc_offset, 
+                :std_offset => t.std_offset,
+                :zone_id => t.zone_id,
+                :name => 'o0'}
+                
+              offsets << offset
+              break
+            end
+          end
+          
+          @transitions.each do |t|
+            offset = offsets.find do |o| 
+              o[:utc_offset] == t.utc_offset &&
+                o[:std_offset] == t.std_offset &&
+                o[:zone_id] == t.zone_id
+            end
+          
+            unless offset
+              offset = {:utc_offset => t.utc_offset, 
+                :std_offset => t.std_offset,
+                :zone_id => t.zone_id,
+                :name => "o#{offsets.length}"}
+                                         
+              offsets << offset
+            end
+            
+            t.offset_name = offset[:name]
+          end
+          
+          offsets.each do |offset|
+            file.puts("tz.offset :#{offset[:name]}, #{offset[:utc_offset]}, #{offset[:std_offset]}, #{quote_zone_id(offset[:zone_id])}")
+          end
+          
+        end
+    end
+    
+    # A transition that will be used to write the periods in a zone class.
+    class TZDataTransition #:nodoc:
+      include Comparable
+      
+      EPOCH = DateTime.new(1970, 1, 1)
+      
+      attr_reader :at_utc
+      attr_reader :utc_offset
+      attr_reader :std_offset
+      attr_reader :zone_id
+      attr_accessor :offset_name
+      
+      def initialize(at_utc, utc_offset, std_offset, zone_id)
+        @at_utc = at_utc
+        @utc_offset = utc_offset
+        @std_offset = std_offset
+        @zone_id = zone_id
+        @offset_name = nil
+      end
+      
+      def to_s
+        "At #{at_utc} UTC switch to UTC offset #{@utc_offset} with std offset #{@std_offset}, zone id #{@zone_id}"
+      end
+      
+      def <=>(transition)
+        if @at_utc == transition.at_utc
+          0
+        elsif @at_utc.nil?
+          -1
+        elsif transition.nil?
+          1
+        else       
+          @at_utc - transition.at_utc
+        end
+      end
+      
+      def total_offset
+        @utc_offset + @std_offset
+      end
+      
+      def clone_with_at(at_utc)
+        TZDataTransition.new(at_utc, @utc_offset, @std_offset, @zone_id)
+      end
+          
+      def write(file)        
+        if @at_utc
+          file.puts "tz.transition #{@at_utc.year}, #{@at_utc.mon}, :#{@offset_name}, #{timestamp_parameters(@at_utc)}" 
+        end        
+      end
+       
+      private
+        def timestamp_parameters(datetime)
+          timestamp = ((datetime - EPOCH) * 86400).to_i
+          
+          if timestamp < 0 || timestamp > 2147483647
+            # Output DateTime parameters as well as a timestamp for platforms
+            # where Time doesn't support negative or 64-bit values.
+            "#{timestamp}, #{datetime.ajd.numerator}, #{datetime.ajd.denominator}"
+          else
+            timestamp
+          end
+        end
+    end
+     
+    # An instance of a rule for a year.
+    class TZDataActivatedRule #:nodoc:
+      attr_reader :rule
+      attr_reader :year
+      attr_reader :at
+      
+      def initialize(rule, year)
+        @rule = rule
+        @year = year
+        @at = nil
+      end
+      
+      def calculate_time(utc_offset, std_offset)
+        @at = @rule.at_utc_time(@year, utc_offset, std_offset)
+      end
+    end
+    
+    # A tz data time definition - an hour, minute, second and reference. Reference
+    # is either :utc, :standard or :wall_clock.
+    class TZDataTime #:nodoc:
+      attr_reader :hour
+      attr_reader :minute
+      attr_reader :second
+      attr_reader :ref
+      
+      def initialize(spec)
+        raise "Invalid time: #{spec}" if spec !~ /^([0-9]+)(:([0-9]+)(:([0-9]+))?)?([wguzs])?$/
+        
+        @hour = $1.to_i
+        @minute = $3.nil? ? 0 : $3.to_i
+        @second = $5.nil? ? 0 : $5.to_i
+        
+        if $6 == 's'
+          @ref = :standard
+        elsif $6 == 'g' || $6 == 'u' || $6 == 'z'
+          @ref = :utc
+        else
+          @ref = :wall_clock
+        end
+      end
+      
+      # Converts the time to UTC given a utc_offset and std_offset.
+      def to_utc(utc_offset, std_offset, year, month, day)      
+        result = DateTime.new(year, month, day, @hour, @minute, @second)
+        offset = 0
+        offset = offset + utc_offset if @ref == :standard || @ref == :wall_clock            
+        offset = offset + std_offset if @ref == :wall_clock       
+        result - Rational(offset, 86400)            
+      end
+    end
+      
+    # A tz data day of the month reference. Can either be an absolute day,
+    # a last week day or a week day >= or <= than a specific day of month.
+    class TZDataDayOfMonth #:nodoc:
+      attr_reader :type
+      attr_reader :day_of_month
+      attr_reader :day_of_week
+      attr_reader :operator
+      
+      def initialize(spec)
+        raise "Invalid on: #{spec}" if spec !~ /^([0-9]+)|(last([A-z]+))|(([A-z]+)([<>]=)([0-9]+))$/
+        
+        if $1
+          @type = :absolute
+          @day_of_month = $1.to_i
+        elsif $3
+          @type = :last
+          @day_of_week = parse_day_of_week($3)
+        else
+          @type = :comparison
+          @day_of_week = parse_day_of_week($5)
+          @operator = parse_operator($6)
+          @day_of_month = $7.to_i
+        end
+      end
+      
+      # Returns the absolute day of month for the given year and month.
+      def to_absolute(year, month)
+        case @type
+          when :last
+            last_day_in_month = (Date.new(year, month, 1) >> 1) - 1          
+            offset = last_day_in_month.wday - @day_of_week
+            offset = offset + 7 if offset < 0
+            (last_day_in_month - offset).day
+          when :comparison
+            pivot = Date.new(year, month, @day_of_month)         
+            offset = @day_of_week - pivot.wday
+            offset = -offset if @operator == :less_equal
+            offset = offset + 7 if offset < 0
+            offset = -offset if @operator == :less_equal          
+            result = pivot + offset
+            if result.month != pivot.month
+              puts self.inspect
+              puts year
+              puts month
+            end
+            raise 'No suitable date found' if result.month != pivot.month
+            result.day          
+          else #absolute
+            @day_of_month
+        end
+      end
+      
+      private
+        def parse_day_of_week(day_of_week)
+          lower = day_of_week.downcase
+          if lower =~ /^mon/
+            1
+          elsif lower =~ /^tue/
+            2
+          elsif lower =~ /^wed/
+            3
+          elsif lower =~ /^thu/
+            4
+          elsif lower =~ /^fri/
+            5
+          elsif lower =~ /^sat/
+            6
+          elsif lower =~ /^sun/
+            0
+          else
+            raise "Invalid day of week: #{day_of_week}"
+          end
+        end
+        
+        def parse_operator(operator)
+          if operator == '>='
+            :greater_equal
+          elsif operator == '<='
+            :less_equal
+          else
+            raise "Invalid operator: #{operator}"
+          end
+        end
+    end   
+    
+    # A tz data Zone until reference.
+    class TZDataUntil #:nodoc:
+      include TZDataParserUtils
+    
+      attr_reader :year
+      attr_reader :month
+      attr_reader :day
+      attr_reader :time
+      
+      def initialize(spec)      
+        parts = spec.split(/\s+/)      
+        raise "Invalid until: #{spec}" if parts.length < 1
+        
+        @year = parts[0].to_i
+        @month = parts.length > 1 ? parse_month(parts[1]) : 1
+        @day = TZDataDayOfMonth.new(parts.length > 2 ? parts[2] : '1')
+        @time = TZDataTime.new(parts.length > 3 ? parts[3] : '00:00')
+      end
+      
+      # Converts the reference to a UTC DateTime.
+      def to_utc(utc_offset, std_offset)
+        @time.to_utc(utc_offset, std_offset, @year, @month, @day.to_absolute(@year, @month))            
+      end
+    end
+    
+    # A tz data Zone format string. Either alternate standard/daylight-savings,
+    # substitution (%s) format or a fixed string.
+    class TZDataFormat #:nodoc:        
+      def initialize(spec)
+        if spec =~ /([A-z]+)\/([A-z]+)/
+          @type = :alternate
+          @standard_abbrev = $1
+          @daylight_abbrev = $2
+        elsif spec =~ /%s/
+          @type = :subst
+          @abbrev = spec
+        else
+          @type = :fixed
+          @abbrev = spec
+        end
+      end
+      
+      # Expands given the current daylight savings offset and Rule string.
+      def expand(std_offset, rule_string)
+        if @type == :alternate
+          if std_offset == 0
+            @standard_abbrev
+          else
+            @daylight_abbrev
+          end
+        elsif @type == :subst
+          sprintf(@abbrev, rule_string)
+        else
+          @abbrev
+        end        
+      end
+      
+      # True if a string from the rule is required to expand this format.
+      def requires_rule_string?
+        @type == :subst
+      end
+      
+      # Is a fixed format string.
+      def fixed?
+        @type == :fixed
+      end
+    end
+    
+    # A location (latitude + longitude)
+    class TZDataLocation #:nodoc:
+      attr_reader :latitude
+      attr_reader :longitude
+      
+      # Constructs a new TZDataLocation from a string in ISO 6709 
+      # sign-degrees-minutes-seconds format, either +-DDMM+-DDDMM 
+      # or +-DDMMSS+-DDDMMSS, first latitude (+ is north), 
+      # then longitude (+ is east).
+      def initialize(coordinates)
+        if coordinates !~ /^([+\-])([0-9]{2})([0-9]{2})([0-9]{2})?([+\-])([0-9]{3})([0-9]{2})([0-9]{2})?$/
+          raise "Invalid coordinates: #{coordinates}"
+        end
+        
+        @latitude = Rational($2.to_i) + Rational($3.to_i, 60)
+        @latitude += Rational($4.to_i, 3600) unless $4.nil?
+        @latitude = -@latitude if $1 == '-'
+        
+        @longitude = Rational($6.to_i) + Rational($7.to_i, 60)
+        @longitude += Rational($8.to_i, 3600) unless $8.nil?
+        @longitude = -@longitude if $5 == '-'
+      end
+    end  
+    
+    TZDataCountryTimezone = Struct.new(:timezone, :description, :location)  #:nodoc:
+    
+    # An ISO 3166 country.
+    class TZDataCountry #:nodoc:
+      include TZDataParserUtils
+      
+      attr_reader :code
+      attr_reader :name
+      attr_reader :zones
+      
+      def initialize(code, name)
+        @code = code
+        @name = name
+        @zones = []
+      end
+      
+      # Adds a TZDataCountryTimezone
+      def add_zone(zone)      
+        @zones << zone      
+      end  
+      
+      def write_index_record(file)
+        s = "        country #{quote_str(@code)}, #{quote_str(@name)}"      
+        s << ' do |c|' if @zones.length > 0
+        
+        file.puts s
+        
+        @zones.each do |zone|
+          file.puts "          c.timezone #{quote_str(zone.timezone.name)}, #{zone.location.latitude.numerator}, #{zone.location.latitude.denominator}, #{zone.location.longitude.numerator}, #{zone.location.longitude.denominator}#{zone.description.nil? ? '' : ', ' + quote_str(zone.description)}" 
+        end
+        
+        file.puts '        end' if @zones.length > 0
+      end
+    end
+  end
+end