u 0.5.0 → 1.0.0

data/build/lib/u/build/data/compositionexclusions.rb

@@ -0,0 +1,14 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::CompositionExclusions
+  def initialize(path)
+    @entries = {}
+    U::Build::Data::File.each(path, 1) do |point|
+      @entries[point] = true
+    end
+  end
+
+  def include?(code)
+    @entries.include? code
+  end
+end

data/build/lib/u/build/data/derivedeastasianwidth.rb

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::DerivedEastAsianWidth
+  include Enumerable
+
+  def initialize(path)
+    @path = path
+  end
+
+  def each
+    U::Build::Data::File.each(@path, 2) do |point, width|
+      yield point, width
+    end
+  end
+end

data/build/lib/u/build/data/file.rb

@@ -0,0 +1,88 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::File
+  class << self
+    def open(path, fields)
+      file = new(path, fields)
+      begin
+        yield file
+      ensure
+        file.close
+      end
+    end
+
+    def each(path, fields)
+      open(path, fields) do |file|
+        file.each do |point, *fs|
+          yield point, *fs
+        end
+      end
+    end
+
+    def each_with_unlisted(path, fields)
+      open(path, fields) do |file|
+        file.each_with_unlisted do |point, previous, unlisted, *fs|
+          yield point, previous, unlisted, *fs
+        end
+      end
+    end
+  end
+
+  def initialize(path, fields)
+    @path, @fields = path, fields
+    @file = File.open(@path, 'rb')
+  end
+
+  def close
+    @file.close
+    self
+  end
+
+  def each
+    @file.each_line.with_index do |line, index|
+      next if comment_or_empty? line
+      fields = line.sub(/\s*#.*\Z/, '').strip.split(/\s*;\s*/, -1)
+      raise RuntimeError,
+        '%s:%d: wrong number of fields: %d instead of %d' %
+          [@path, index + 1, fields.size, @fields] unless
+            @fields === fields.size
+      code = fields.shift
+      if code =~ /\A([0-9A-Fa-f]{4,6})\.\.([0-9A-Fa-f]{4,6})\Z/
+        first, last = $1.hex, $2.hex
+      else
+        first = last = code.hex
+      end
+      begin
+        first.upto(last) do |point|
+          yield point, *fields
+        end
+      rescue => e
+        raise e, '%s:%d: %s' % [@path, index + 1, e]
+      end
+    end
+    self
+  end
+
+  def each_with_unlisted
+    previous = -1
+    each do |point, *fields|
+      (previous + 1).upto(point - 1) do |i|
+        yield i, previous, true, *fields
+      end
+      yield point, point - 1, false, *fields
+      previous = point
+    end
+    (previous + 1).upto(0x10ffff) do |i|
+      yield i, previous, true
+    end
+    self
+  end
+
+private
+
+  Code = 0
+
+  def comment_or_empty?(line)
+    line =~ /\A(?:#|\s*\Z)/
+  end
+end

data/build/lib/u/build/data/linebreak.rb

@@ -0,0 +1,14 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::LineBreak
+  def initialize(data, path)
+    @entries = []
+    U::Build::Data::File.each_with_unlisted(path, 2) do |point, previous, unlisted, property|
+      @entries[point] = unlisted ? (data[point].type == 'Cn' ? 'XX' : 'AL') : property
+    end
+  end
+
+  def [](point)
+    @entries[point]
+  end
+end

data/build/lib/u/build/data/proplist.rb

@@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::PropList
+  include Enumerable
+
+  def initialize(path)
+    @entries = Hash.new{ |h, k| h[k] = [] }
+    U::Build::Data::File.each(path, 2) do |point, property|
+      @entries[point] << property
+    end
+  end
+
+  def each
+    @entries.each do |point, properties|
+      yield point, properties
+    end
+  end
+end

data/build/lib/u/build/data/scripts.rb

@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::Scripts
+  include Enumerable
+
+  def initialize(path)
+    @entries = {}
+    U::Build::Data::File.each(path, 2) do |point, script|
+      @entries[point] = script
+    end
+  end
+
+  def [](point)
+    @entries.fetch(point, 'Unknown')
+  end
+
+  def each
+    @entries.each do |point, script|
+      yield point, script
+    end
+  end
+end

data/build/lib/u/build/data/specialcasing.rb

@@ -0,0 +1,106 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::SpecialCasing
+  include Enumerable
+
+  def initialize(data, path)
+    @entries = {}
+    File.new(data, path).each do |point, entry|
+      @entries[point] = entry
+    end
+  end
+
+  def include?(point)
+    @entries.include? point
+  end
+
+  def [](point)
+    @entries[point]
+  end
+
+  def each
+    @entries.values.sort.each do |entry|
+      yield entry
+    end
+  end
+
+  # TODO: Move to its own file
+  class Entry
+    include Comparable
+
+    def initialize(points, offset)
+      @points, @offset = points, offset
+    end
+
+    def <=>(other)
+      self.class == other.class ?
+        offset <=> other.offset :
+        nil
+    end
+
+    def to_s
+      @points.to_s
+    end
+
+    def to_escaped_s
+      @points.to_escaped_s
+    end
+
+    attr_reader :offset
+  end
+
+private
+
+  class File
+    def initialize(data, path)
+      @data, @path = data, path
+      @offset = 0
+    end
+
+    def each
+      U::Build::Data::File.each(@path, 5..6) do |point, lower, title, upper, condition, _comment|
+        raise RuntimeError,
+          'special case for code point that does not have a type: %04X' %
+            point unless
+              @data[point].type
+        # We deal with conditional special cases in code.
+        next unless condition.empty?
+        yield point, case @data[point].type
+        when 'Lu'
+          verify upper, point, 'Lu', 'Upper'
+          entry(@data[point].value, lower, title)
+        when 'Lt'
+          verify title, point, 'Lt', 'Title'
+          entry(nil, lower, upper)
+        when 'Ll'
+          verify lower, point, 'Ll', 'Lower'
+          entry(@data[point].value, upper, title)
+        else
+          raise RuntimeError,
+            'special case entry of type %s for non-alphabetic code point: %04X' %
+              [@data[point].type, point]
+        end
+      end
+    end
+
+  private
+
+    def verify(field, point, type, ccase)
+      raise RuntimeError,
+        'special case entry has type %s, but UCD_%s(%04X) != %04X: %04X' %
+          [type, ccase, point, point, point] unless
+            field.hex == point
+    end
+
+    def entry(single, field1, field2)
+      Entry.new(U::Build::Data::Unicode::Points.new([
+        single ? single.to_s(16) : nil,
+        field1,
+        "\0",
+        field2
+      ].compact.join(' ')), @offset).tap{ |e|
+        @offset += e.to_s.bytesize + 1
+      }
+    end
+  end
+end

data/build/lib/u/build/data/unicode.rb

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::Unicode
+  include Enumerable
+
+  require 'u/build/data/unicode/points'
+  require 'u/build/data/unicode/entry'
+
+  Name, Category, CombiningClasses, _, Decomposition, DecimalValue, _, _, _, _, _, Upper, Lower, _ = (0..13).to_a
+
+  def initialize(path)
+    @entries = []
+    U::Build::Data::File.each_with_unlisted(path, 15) do |point, previous, unlisted, *fields|
+      if point >= 0xe0000 and previous < 0xe0000
+        @pages_before_e0000 = (previous >> 8) + 1
+        @last_char_part1_i = @pages_before_e0000 * 256 - 1
+        @last_char_part1_x = '0x%04x' % @last_char_part1_i
+        @last_char_part1_X = '%04X' % @last_char_part1_i
+      end
+      @entries << Entry.new(point,
+                            (unlisted and fields[Name] !~ /Last>\Z/) ? Unlisted : fields)
+    end
+    @last = 0x10ffff
+  end
+
+  attr_reader :pages_before_e0000, :last_char_part1_i, :last_char_part1_x, :last_char_part1_X, :last
+
+  def [](index)
+    @entries[index]
+  end
+
+  def each
+    @entries.each do |entry|
+      yield entry
+    end
+  end
+
+private
+
+  Unlisted = ['', 'Cn', '0', '', '', '', '', '', '', '', '', '', '', ''].freeze
+end

data/build/lib/u/build/data/unicode/entry.rb

@@ -0,0 +1,27 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::Unicode::Entry
+  require 'u/build/data/unicode/entry/decomposition'
+
+  def initialize(code, fields)
+    # TODO: Why not processes Nl?
+    case @type = fields[U::Build::Data::Unicode::Category]
+    when 'Nd'
+      @value = fields[U::Build::Data::Unicode::DecimalValue].to_i
+    when 'Ll'
+      @value = fields[U::Build::Data::Unicode::Upper].hex
+    when 'Lu'
+      @value = fields[U::Build::Data::Unicode::Lower].hex
+    when 'Lt'
+      @value = nil
+      @title_to_lower = fields[U::Build::Data::Unicode::Lower].hex
+      @title_to_upper = fields[U::Build::Data::Unicode::Upper].hex
+    else
+      @value = nil
+    end
+    @cclass = fields[U::Build::Data::Unicode::CombiningClasses].to_i
+    @decomposition = Decomposition.new(fields[U::Build::Data::Unicode::Decomposition])
+  end
+
+  attr_reader :code, :type, :value, :title_to_lower, :title_to_upper, :cclass, :decomposition
+end

data/build/lib/u/build/data/unicode/entry/decomposition.rb

@@ -0,0 +1,29 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::Unicode::Entry::Decomposition
+  include Enumerable
+
+  def initialize(decomposition)
+    @canonical = decomposition !~ /\A<.*>\s*(.*)/
+    @points = U::Build::Data::Unicode::Points.new(@canonical ? decomposition : $1)
+  end
+
+  def canonical?
+    @canonical
+  end
+
+  def empty?
+    @points.empty?
+  end
+
+  def length
+    @points.count
+  end
+
+  def each
+    @points.each do |point|
+      yield point
+    end
+    self
+  end
+end

data/build/lib/u/build/data/unicode/points.rb

@@ -0,0 +1,32 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Data::Unicode::Points
+  include Enumerable
+
+  def initialize(string)
+    @points = string.split(/\s+/).map{ |s| s.hex }
+  end
+
+  def empty?
+    @points.empty?
+  end
+
+  def length
+    @points.length
+  end
+
+  def each
+    @points.each do |point|
+      yield point
+    end
+    self
+  end
+
+  def to_s
+    @points.pack('U*')
+  end
+
+  def to_escaped_s
+    to_s.unpack('H*').first.gsub(/../, '\\x\0')
+  end
+end

data/build/lib/u/build/header.rb

@@ -0,0 +1,11 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Header
+  require 'u/build/header/table'
+  require 'u/build/header/tables'
+
+  def initialize(io = $stdout)
+    io.puts "/* Automatically generated file */\n\n"
+    yield
+  end
+end

data/build/lib/u/build/header/table.rb

@@ -0,0 +1,19 @@
+# -*- coding: utf-8 -*-
+
+class U::Build::Header::Table
+  require 'u/build/header/table/row'
+
+  def initialize(declaration)
+    @declaration = declaration
+    @rows = []
+  end
+
+  def <<(row)
+    @rows << row
+    self
+  end
+
+  def to_s
+    "\n\n%s = {\n%s\n};" % [@declaration, @rows.join(",\n")]
+  end
+end