mojombo-chronic 0.3.0

data/lib/chronic/separator.rb

@@ -0,0 +1,91 @@
+module Chronic
+
+  class Separator < Tag #:nodoc:
+    def self.scan(tokens)
+      tokens.each_index do |i|
+        if t = self.scan_for_commas(tokens[i]) then tokens[i].tag(t); next end
+        if t = self.scan_for_slash_or_dash(tokens[i]) then tokens[i].tag(t); next end
+        if t = self.scan_for_at(tokens[i]) then tokens[i].tag(t); next end
+        if t = self.scan_for_in(tokens[i]) then tokens[i].tag(t); next end
+        if t = self.scan_for_on(tokens[i]) then tokens[i].tag(t); next end
+      end
+      tokens
+    end
+    
+    def self.scan_for_commas(token)
+      scanner = {/^,$/ => :comma}
+      scanner.keys.each do |scanner_item|
+        return SeparatorComma.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+    
+    def self.scan_for_slash_or_dash(token)
+      scanner = {/^-$/ => :dash,
+                 /^\/$/ => :slash}
+      scanner.keys.each do |scanner_item|
+        return SeparatorSlashOrDash.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+    
+    def self.scan_for_at(token)
+      scanner = {/^(at|@)$/ => :at}
+      scanner.keys.each do |scanner_item|
+        return SeparatorAt.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+    
+    def self.scan_for_in(token)
+      scanner = {/^in$/ => :in}
+      scanner.keys.each do |scanner_item|
+        return SeparatorIn.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+    
+    def self.scan_for_on(token)
+      scanner = {/^on$/ => :on}
+      scanner.keys.each do |scanner_item|
+        return SeparatorOn.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+    
+    def to_s
+      'separator'
+    end
+  end
+  
+  class SeparatorComma < Separator #:nodoc:
+    def to_s
+      super << '-comma'
+    end
+  end
+  
+  class SeparatorSlashOrDash < Separator #:nodoc:
+    def to_s
+      super << '-slashordash-' << @type.to_s
+    end
+  end
+  
+  class SeparatorAt < Separator #:nodoc:
+    def to_s
+      super << '-at'
+    end
+  end
+  
+  class SeparatorIn < Separator #:nodoc:
+    def to_s
+      super << '-in'
+    end
+  end
+  
+  class SeparatorOn < Separator #:nodoc:
+    def to_s
+      super << '-on'
+    end
+  end
+
+end

data/lib/chronic/time_zone.rb

@@ -0,0 +1,23 @@
+module Chronic
+  class TimeZone < Tag #:nodoc:
+    def self.scan(tokens)
+      tokens.each_index do |i|
+        if t = self.scan_for_all(tokens[i]) then tokens[i].tag(t); next end
+      end
+      tokens
+    end
+
+    def self.scan_for_all(token)
+      scanner = {/[PMCE][DS]T/i => :tz,
+                 /(tzminus)?\d{4}/ => :tz}
+      scanner.keys.each do |scanner_item|
+        return self.new(scanner[scanner_item]) if scanner_item =~ token.word
+      end
+      return nil
+    end
+
+    def to_s
+      'timezone'
+    end
+  end
+end

data/lib/numerizer/numerizer.rb

@@ -0,0 +1,97 @@
+require 'strscan'
+
+class Numerizer
+
+  DIRECT_NUMS = [
+                  ['eleven', '11'],
+                  ['twelve', '12'],
+                  ['thirteen', '13'],
+                  ['fourteen', '14'],
+                  ['fifteen', '15'],
+                  ['sixteen', '16'],
+                  ['seventeen', '17'],
+                  ['eighteen', '18'],
+                  ['nineteen', '19'],
+                  ['ninteen', '19'], # Common mis-spelling
+                  ['zero', '0'],
+                  ['one', '1'],
+                  ['two', '2'],
+                  ['three', '3'],
+                  ['four(\W|$)', '4\1'],  # The weird regex is so that it matches four but not fourty
+                  ['five', '5'],
+                  ['six(\W|$)', '6\1'],
+                  ['seven(\W|$)', '7\1'],
+                  ['eight(\W|$)', '8\1'],
+                  ['nine(\W|$)', '9\1'],
+                  ['ten', '10'],
+                  ['\ba[\b^$]', '1'] # doesn't make sense for an 'a' at the end to be a 1
+                ]
+
+  TEN_PREFIXES = [ ['twenty', 20],
+                    ['thirty', 30],
+                    ['fourty', 40],
+                    ['fifty', 50],
+                    ['sixty', 60],
+                    ['seventy', 70],
+                    ['eighty', 80],
+                    ['ninety', 90]
+                  ]
+
+  BIG_PREFIXES = [ ['hundred', 100],
+                    ['thousand', 1000],
+                    ['million', 1_000_000],
+                    ['billion', 1_000_000_000],
+                    ['trillion', 1_000_000_000_000],
+                  ]
+
+  def self.numerize(string)
+    string = string.dup
+  
+    # preprocess
+    string.gsub!(/ +|([^\d])-([^\d])/, '\1 \2') # will mutilate hyphenated-words but shouldn't matter for date extraction
+    string.gsub!(/a half/, 'haAlf') # take the 'a' out so it doesn't turn into a 1, save the half for the end
+    
+    # easy/direct replacements
+    
+    DIRECT_NUMS.each do |dn|
+      string.gsub!(/#{dn[0]}/i, '<num>' + dn[1])
+    end
+    
+    # ten, twenty, etc.
+    
+    TEN_PREFIXES.each do |tp|
+      string.gsub!(/(?:#{tp[0]}) *<num>(\d(?=[^\d]|$))*/i) { '<num>' + (tp[1] + $1.to_i).to_s }
+    end
+    
+    TEN_PREFIXES.each do |tp|
+      string.gsub!(/#{tp[0]}/i) { '<num>' + tp[1].to_s }
+    end
+    
+    # hundreds, thousands, millions, etc.
+    
+    BIG_PREFIXES.each do |bp|
+      string.gsub!(/(?:<num>)?(\d*) *#{bp[0]}/i) { '<num>' + (bp[1] * $1.to_i).to_s}
+      andition(string)
+    end
+    
+    # fractional addition
+    # I'm not combining this with the previous block as using float addition complicates the strings
+    # (with extraneous .0's and such )
+    string.gsub!(/(\d+)(?: | and |-)*haAlf/i) { ($1.to_f + 0.5).to_s }
+    
+    string.gsub(/<num>/, '')
+  end
+
+  private
+  
+  def self.andition(string)
+    sc = StringScanner.new(string)
+    while(sc.scan_until(/<num>(\d+)( | and )<num>(\d+)(?=[^\w]|$)/i))
+      if sc[2] =~ /and/ || sc[1].size > sc[3].size
+        string[(sc.pos - sc.matched_size)..(sc.pos-1)] = '<num>' + (sc[1].to_i + sc[3].to_i).to_s
+        sc.reset
+      end
+    end
+  end
+
+end

data/test/suite.rb

@@ -0,0 +1,9 @@
+require 'test/unit'
+
+tests = Dir["#{File.dirname(__FILE__)}/test_*.rb"]
+tests.delete_if { |o| o =~ /test_parsing/ }
+tests.each do |file|
+  require file
+end
+
+require File.dirname(__FILE__) + '/test_parsing.rb'

data/test/test_Chronic.rb

@@ -0,0 +1,50 @@
+require 'chronic'
+require 'test/unit'
+
+class TestChronic < Test::Unit::TestCase
+  
+  def setup
+    # Wed Aug 16 14:00:00 UTC 2006
+    @now = Time.local(2006, 8, 16, 14, 0, 0, 0)
+  end
+  
+  def test_post_normalize_am_pm_aliases
+    # affect wanted patterns
+    
+    tokens = [Chronic::Token.new("5:00"), Chronic::Token.new("morning")]
+    tokens[0].tag(Chronic::RepeaterTime.new("5:00"))
+    tokens[1].tag(Chronic::RepeaterDayPortion.new(:morning))
+    
+    assert_equal :morning, tokens[1].tags[0].type
+    
+    tokens = Chronic.dealias_and_disambiguate_times(tokens, {})
+    
+    assert_equal :am, tokens[1].tags[0].type
+    assert_equal 2, tokens.size
+    
+    # don't affect unwanted patterns
+    
+    tokens = [Chronic::Token.new("friday"), Chronic::Token.new("morning")]
+    tokens[0].tag(Chronic::RepeaterDayName.new(:friday))
+    tokens[1].tag(Chronic::RepeaterDayPortion.new(:morning))
+    
+    assert_equal :morning, tokens[1].tags[0].type
+    
+    tokens = Chronic.dealias_and_disambiguate_times(tokens, {})
+    
+    assert_equal :morning, tokens[1].tags[0].type
+    assert_equal 2, tokens.size
+  end
+  
+  def test_guess
+    span = Chronic::Span.new(Time.local(2006, 8, 16, 0), Time.local(2006, 8, 17, 0))
+    assert_equal Time.local(2006, 8, 16, 12), Chronic.guess(span)
+    
+    span = Chronic::Span.new(Time.local(2006, 8, 16, 0), Time.local(2006, 8, 17, 0, 0, 1))
+    assert_equal Time.local(2006, 8, 16, 12), Chronic.guess(span)
+    
+    span = Chronic::Span.new(Time.local(2006, 11), Time.local(2006, 12))
+    assert_equal Time.local(2006, 11, 16), Chronic.guess(span)
+  end
+  
+end

data/test/test_Handler.rb

@@ -0,0 +1,110 @@
+require 'chronic'
+require 'test/unit'
+
+class TestHandler < Test::Unit::TestCase
+  
+  def setup
+    # Wed Aug 16 14:00:00 UTC 2006
+    @now = Time.local(2006, 8, 16, 14, 0, 0, 0)
+  end
+
+  def test_handler_class_1
+    handler = Chronic::Handler.new([:repeater], :handler)
+    
+    tokens = [Chronic::Token.new('friday')]
+    tokens[0].tag(Chronic::RepeaterDayName.new(:friday))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('afternoon')
+    tokens[1].tag(Chronic::RepeaterDayPortion.new(:afternoon))
+    
+    assert !handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_handler_class_2
+    handler = Chronic::Handler.new([:repeater, :repeater?], :handler)
+    
+    tokens = [Chronic::Token.new('friday')]
+    tokens[0].tag(Chronic::RepeaterDayName.new(:friday))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('afternoon')
+    tokens[1].tag(Chronic::RepeaterDayPortion.new(:afternoon))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('afternoon')
+    tokens[2].tag(Chronic::RepeaterDayPortion.new(:afternoon))
+    
+    assert !handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_handler_class_3
+    handler = Chronic::Handler.new([:repeater, 'time?'], :handler)
+    
+    tokens = [Chronic::Token.new('friday')]
+    tokens[0].tag(Chronic::RepeaterDayName.new(:friday))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('afternoon')
+    tokens[1].tag(Chronic::RepeaterDayPortion.new(:afternoon))
+    
+    assert !handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_handler_class_4
+    handler = Chronic::Handler.new([:repeater_month_name, :scalar_day, 'time?'], :handler)
+    
+    tokens = [Chronic::Token.new('may')]
+    tokens[0].tag(Chronic::RepeaterMonthName.new(:may))
+    
+    assert !handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('27')
+    tokens[1].tag(Chronic::ScalarDay.new(27))
+    
+    assert handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_handler_class_5
+    handler = Chronic::Handler.new([:repeater, 'time?'], :handler)
+    
+    tokens = [Chronic::Token.new('friday')]
+    tokens[0].tag(Chronic::RepeaterDayName.new(:friday))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('5:00')
+    tokens[1].tag(Chronic::RepeaterTime.new('5:00'))
+    
+    assert handler.match(tokens, Chronic.definitions)
+    
+    tokens << Chronic::Token.new('pm')
+    tokens[2].tag(Chronic::RepeaterDayPortion.new(:pm))
+    
+    assert handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_handler_class_6
+    handler = Chronic::Handler.new([:scalar, :repeater, :pointer], :handler)
+    
+    tokens = [Chronic::Token.new('3'),
+              Chronic::Token.new('years'),
+              Chronic::Token.new('past')]
+              
+    tokens[0].tag(Chronic::Scalar.new(3))
+    tokens[1].tag(Chronic::RepeaterYear.new(:year))
+    tokens[2].tag(Chronic::Pointer.new(:past))
+    
+    assert handler.match(tokens, Chronic.definitions)
+  end
+  
+  def test_constantize
+    handler = Chronic::Handler.new([], :handler)
+    assert_equal Chronic::RepeaterTime, handler.constantize(:repeater_time)
+  end
+  
+end

data/test/test_Numerizer.rb

@@ -0,0 +1,52 @@
+require 'test/unit'
+require 'chronic'
+
+class ParseNumbersTest < Test::Unit::TestCase
+
+  def test_straight_parsing
+    strings = { 1 => 'one',
+               5 => 'five',
+               10 => 'ten',
+               11 => 'eleven',
+               12 => 'twelve',
+               13 => 'thirteen',
+               14 => 'fourteen',
+               15 => 'fifteen',
+               16 => 'sixteen',
+               17 => 'seventeen',
+               18 => 'eighteen',
+               19 => 'nineteen',
+               20 => 'twenty',
+               27 => 'twenty seven',
+               31 => 'thirty-one',
+               59 => 'fifty nine',
+               100 => 'a hundred',
+               100 => 'one hundred',
+               150 => 'one hundred and fifty',
+               # 150 => 'one fifty',
+               200 => 'two-hundred',
+               500 => '5 hundred',
+               999 => 'nine hundred and ninety nine',
+               1_000 => 'one thousand',
+               1_200 => 'twelve hundred',
+               1_200 => 'one thousand two hundred',
+               17_000 => 'seventeen thousand',
+               21_473 => 'twentyone-thousand-four-hundred-and-seventy-three',
+               74_002 => 'seventy four thousand and two',
+               99_999 => 'ninety nine thousand nine hundred ninety nine',
+               100_000 => '100 thousand',
+               250_000 => 'two hundred fifty thousand',
+               1_000_000 => 'one million',
+               1_250_007 => 'one million two hundred fifty thousand and seven',
+               1_000_000_000 => 'one billion',
+               1_000_000_001 => 'one billion and one' }
+               
+    strings.keys.sort.each do |key|
+      assert_equal key, Numerizer.numerize(strings[key]).to_i
+    end
+  end
+  
+  def test_edges
+    assert_equal "27 Oct 2006 7:30am", Numerizer.numerize("27 Oct 2006 7:30am")
+  end
+end

data/test/test_RepeaterDayName.rb

@@ -0,0 +1,52 @@
+require 'chronic'
+require 'test/unit'
+
+class TestRepeaterDayName < Test::Unit::TestCase
+  
+  def setup
+    @now = Time.local(2006, 8, 16, 14, 0, 0, 0)
+  end
+  
+  def test_match
+    token = Chronic::Token.new('saturday')
+    repeater = Chronic::Repeater.scan_for_day_names(token)
+    assert_equal Chronic::RepeaterDayName, repeater.class
+    assert_equal :saturday, repeater.type
+    
+    token = Chronic::Token.new('sunday')
+    repeater = Chronic::Repeater.scan_for_day_names(token)
+    assert_equal Chronic::RepeaterDayName, repeater.class
+    assert_equal :sunday, repeater.type
+  end
+
+  def test_next_future
+    mondays = Chronic::RepeaterDayName.new(:monday)
+    mondays.start = @now
+    
+    span = mondays.next(:future)
+    
+    assert_equal Time.local(2006, 8, 21), span.begin
+    assert_equal Time.local(2006, 8, 22), span.end 
+
+    span = mondays.next(:future)
+    
+    assert_equal Time.local(2006, 8, 28), span.begin
+    assert_equal Time.local(2006, 8, 29), span.end
+  end
+  
+  def test_next_past
+    mondays = Chronic::RepeaterDayName.new(:monday)
+    mondays.start = @now
+    
+    span = mondays.next(:past)
+    
+    assert_equal Time.local(2006, 8, 14), span.begin
+    assert_equal Time.local(2006, 8, 15), span.end 
+
+    span = mondays.next(:past)
+    
+    assert_equal Time.local(2006, 8, 7), span.begin
+    assert_equal Time.local(2006, 8, 8), span.end
+  end
+  
+end