ffi-icu 0.0.5 → 0.0.6

data/lib/ffi-icu.rb

@@ -10,6 +10,8 @@ module ICU
       :osx
     when /linux/
       :linux
+    when /mswin|msys|mingw|cygwin|bccwin|wince|emc/
+      :windows
     else
       os
     end
@@ -20,6 +22,11 @@ module ICU
   end
 end
 
+unless ICU.ruby19?
+  require 'jcode'
+  $KCODE = 'u'
+end
+
 require "ffi-icu/core_ext/string"
 require "ffi-icu/lib"
 require "ffi-icu/uchar"
@@ -29,7 +36,3 @@ require "ffi-icu/transliteration"
 require "ffi-icu/normalization"
 require "ffi-icu/break_iterator"
 
-unless ICU.ruby19?
-  require 'jcode'
-  $KCODE = 'u'
-end

data/lib/ffi-icu/break_iterator.rb

@@ -18,11 +18,13 @@ module ICU
 
     def text=(str)
       Lib.check_error { |err|
-        Lib.ubrk_setText @iterator, UCharPointer.from_string(str), str.length, err
+        Lib.ubrk_setText @iterator, UCharPointer.from_string(str), str.jlength, err
       }
     end
 
     def each(&blk)
+      return to_enum(:each) unless block_given?
+
       int = first
 
       while int != UBRK_DONE

data/lib/ffi-icu/chardet.rb

@@ -2,18 +2,15 @@ module ICU
   module CharDet
 
     def self.detect(string)
-      detector = Detector.new
-      res = detector.detect string
-      detector.close
-
-      res
+      Detector.new.detect string
     end
 
     class Detector
       Match = Struct.new(:name, :confidence, :language)
 
       def initialize
-        @detector = Lib.check_error { |ptr| Lib.ucsdet_open(ptr) }
+        ptr = Lib.check_error { |err| Lib.ucsdet_open err }
+        @detector = FFI::AutoPointer.new(ptr, Lib.method(:ucsdet_close))
       end
 
       def input_filter_enabled?
@@ -30,10 +27,6 @@ module ICU
         end
       end
 
-      def close
-        Lib.ucsdet_close @detector
-      end
-
       def detect(str)
         set_text(str)
 

data/lib/ffi-icu/collation.rb

@@ -2,11 +2,7 @@ module ICU
   module Collation
 
     def self.collate(locale, arr)
-      collator = Collator.new(locale)
-      res = collator.collate(arr)
-      collator.close
-
-      res
+      Collator.new(locale).collate(arr)
     end
 
     def self.keywords
@@ -34,7 +30,8 @@ module ICU
       ULOC_VALID_LOCALE = 1
 
       def initialize(locale)
-        @c = Lib.check_error { |error| Lib.ucol_open(locale, error) }
+        ptr = Lib.check_error { |error| Lib.ucol_open(locale, error) }
+        @c = FFI::AutoPointer.new(ptr, Lib.method(:ucol_close))
       end
 
       def locale
@@ -44,19 +41,19 @@ module ICU
       def compare(a, b)
         Lib.ucol_strcoll(
           @c,
-          UCharPointer.from_string(a), a.length,
-          UCharPointer.from_string(b), b.length
+          UCharPointer.from_string(a), a.jlength,
+          UCharPointer.from_string(b), b.jlength
         )
       end
 
       def greater?(a, b)
-        Lib.ucol_greater(@c, UCharPointer.from_string(a), a.length,
-                             UCharPointer.from_string(b), b.length)
+        Lib.ucol_greater(@c, UCharPointer.from_string(a), a.jlength,
+                             UCharPointer.from_string(b), b.jlength)
       end
 
       def greater_or_equal?(a, b)
-        Lib.ucol_greaterOrEqual(@c, UCharPointer.from_string(a), a.length,
-                                    UCharPointer.from_string(b), b.length)
+        Lib.ucol_greaterOrEqual(@c, UCharPointer.from_string(a), a.jlength,
+                                    UCharPointer.from_string(b), b.jlength)
       end
 
       def equal?(*args)
@@ -68,16 +65,16 @@ module ICU
 
         a, b = args
 
-        Lib.ucol_equal(@c, UCharPointer.from_string(a), a.length,
-                           UCharPointer.from_string(b), b.length)
+        Lib.ucol_equal(@c, UCharPointer.from_string(a), a.jlength,
+                           UCharPointer.from_string(b), b.jlength)
       end
 
-      def collate(array)
-        array.sort { |a,b| compare a, b }
-      end
+      def collate(sortable)
+        unless sortable.respond_to?(:sort)
+          raise ArgumentError, "argument must respond to :sort with arity of 2"
+        end
 
-      def close
-        Lib.ucol_close(@c)
+        sortable.sort { |a, b| compare a, b }
       end
     end # Collator
 

data/lib/ffi-icu/core_ext/string.rb

@@ -2,4 +2,8 @@ class String
   unless method_defined?(:bytesize)
     alias_method :bytesize, :length
   end
+
+  unless method_defined?(:jlength)
+    alias_method :jlength, :length
+  end
 end

data/lib/ffi-icu/lib.rb

@@ -9,10 +9,11 @@ module ICU
     extend FFI::Library
 
     VERSIONS = {
-      "42" => "_4_2",
-      "44" => "_44",
+      "48" => "_48",
+      "46" => "_46",
       "45" => "_45",
-      "46" => "_46"
+      "44" => "_44",
+      "42" => "_4_2",
     }
 
     # FIXME: this is incredibly ugly, figure out some better way
@@ -33,27 +34,34 @@ module ICU
         end
       end
 
+      libs = nil
+      versions = VERSIONS.keys
+
       # ok, try to find it
       case ICU.platform
       when :osx
         ffi_lib "icucore"
       when :linux
-        versions = VERSIONS.keys
         libs = ffi_lib versions.map { |v| "libicui18n.so.#{v}" },
                        versions.map { |v| "libicutu.so.#{v}"   }
 
-        VERSIONS.find do |so_version, func_version|
-          if libs.first.name =~ /#{so_version}$/
-            suffix = func_version
-          end
-        end
+      when :windows
+        libs = ffi_lib versions.map { |v| "icuin#{v}.dll" }
       else
         raise LoadError
       end
 
+      if libs
+        lib_name = libs.first.name
+        version  = VERSIONS.find { |object, func| lib_name =~ /#{object}(\.dll)?$/ }
+
+        version or raise "unable to find suffix in #{lib_name}"
+        suffix = version.last
+      end
+
       suffix
     rescue LoadError => ex
-      raise LoadError, "no idea how to load ICU on #{ICU.platform}, patches appreciated! (#{ex.message})"
+      raise LoadError, "no idea how to load ICU on #{ICU.platform.inspect}, patches appreciated! (#{ex.message})"
     end
 
     def self.check_error
@@ -98,8 +106,8 @@ module ICU
 
     attach_function :u_errorName,     "u_errorName#{suffix}",     [:int],      :string
     attach_function :uenum_count,     "uenum_count#{suffix}",     [:pointer,   :pointer], :int
-    attach_function :uenum_close,     "uenum_close#{suffix}",      [:pointer], :void
-    attach_function :uenum_next,      "uenum_next#{suffix}",       [:pointer,  :pointer,  :pointer], :string
+    attach_function :uenum_close,     "uenum_close#{suffix}",     [:pointer], :void
+    attach_function :uenum_next,      "uenum_next#{suffix}",      [:pointer,  :pointer,  :pointer], :string
     attach_function :u_charsToUChars, "u_charsToUChars#{suffix}", [:string,    :pointer,  :int32_t], :void
     attach_function :u_UCharsToChars, "u_UCharsToChars#{suffix}", [:pointer,   :string,   :int32_t], :void
 
@@ -126,11 +134,11 @@ module ICU
     # http://icu-project.org/apiref/icu4c/ucol_8h.html
     #
 
-    attach_function :ucol_open,             "ucol_open#{suffix}",              [:string,   :pointer], :pointer
-    attach_function :ucol_close,            "ucol_close#{suffix}",             [:pointer], :void
-    attach_function :ucol_strcoll,          "ucol_strcoll#{suffix}",           [:pointer,  :pointer,  :int32_t,  :pointer, :int32_t], :int
-    attach_function :ucol_getKeywords,      "ucol_getKeywords#{suffix}",       [:pointer], :pointer
-    attach_function :ucol_getKeywordValues, "ucol_getKeywordValues#{suffix}",  [:string,   :pointer], :pointer
+    attach_function :ucol_open,             "ucol_open#{suffix}",             [:string,    :pointer], :pointer
+    attach_function :ucol_close,            "ucol_close#{suffix}",            [:pointer],  :void
+    attach_function :ucol_strcoll,          "ucol_strcoll#{suffix}",          [:pointer,   :pointer,  :int32_t,  :pointer, :int32_t], :int
+    attach_function :ucol_getKeywords,      "ucol_getKeywords#{suffix}",      [:pointer],  :pointer
+    attach_function :ucol_getKeywordValues, "ucol_getKeywordValues#{suffix}", [:string,    :pointer], :pointer
     attach_function :ucol_getAvailable,     "ucol_getAvailable#{suffix}",     [:int32_t],  :string
     attach_function :ucol_countAvailable,   "ucol_countAvailable#{suffix}",   [],          :int32_t
     attach_function :ucol_getLocale,        "ucol_getLocale#{suffix}",        [:pointer,   :int,      :pointer], :string
@@ -189,6 +197,8 @@ module ICU
     #
     # Text Boundary Analysis
     #
+    # http://icu-project.org/apiref/icu4c/ubrk_8h.html
+    #
 
     enum :iterator_type, [ :character, :word, :line, :sentence, :title]
     enum :word_break, [ :none,         0,

data/lib/ffi-icu/normalization.rb

@@ -2,7 +2,7 @@ module ICU
   module Normalization
 
     def self.normalize(input, mode = :default)
-      input_length  = input.unpack("U*").size
+      input_length  = input.jlength
       needed_length = out_length = options = 0
       in_ptr        = UCharPointer.from_string(input)
       out_ptr       = UCharPointer.new(out_length)

data/lib/ffi-icu/transliteration.rb

@@ -4,10 +4,7 @@ module ICU
     class << self
       def transliterate(translit_id, str, rules = nil)
         t = Transliterator.new translit_id, rules
-        res = t.transliterate str
-        t.close
-
-        res
+        t.transliterate str
       end
       alias_method :translit, :transliterate
 
@@ -29,7 +26,7 @@ module ICU
         rules_length = 0
 
         if rules
-          rules_length = rules.length + 1
+          rules_length = rules.jlength + 1
           rules = UCharPointer.from_string(rules)
         end
 
@@ -37,7 +34,8 @@ module ICU
         begin
           Lib.check_error do |status|
             # couldn't get utrans_openU to work properly, so using deprecated utrans_open for now
-            @tr = Lib.utrans_openU(UCharPointer.from_string(id), id.length, direction, rules, rules_length, @parse_error, status)
+            ptr = Lib.utrans_openU(UCharPointer.from_string(id), id.jlength, direction, rules, rules_length, @parse_error, status)
+            @tr = FFI::AutoPointer.new(ptr, Lib.method(:utrans_close))
           end
         rescue ICU::Error => ex
           raise ex, "#{ex.message} (#{parse_error})"
@@ -80,10 +78,6 @@ module ICU
         buf.string text_length.get_int32(0)
       end
 
-      def close
-        Lib.utrans_close @tr
-      end
-
     end # Transliterator
   end # Translit
 end # ICU

data/lib/ffi-icu/version.rb

@@ -1,3 +1,3 @@
 module ICU
-  VERSION = "0.0.5"
+  VERSION = "0.0.6"
 end

data/spec/break_iterator_spec.rb

@@ -18,6 +18,12 @@ module ICU
       iterator.to_a.should == [0, 5, 6, 11, 12, 17, 18, 21, 22, 26, 27, 28, 39, 40, 51, 52, 56, 57, 58, 61, 62, 64, 65, 72, 73, 79, 80, 90, 91, 93, 94, 100, 101, 103, 104, 110, 111, 116, 117, 123, 124]
     end
 
+    it "finds all word boundaries in a Thai string" do
+      iterator = BreakIterator.new :word, "th_TH"
+      iterator.text = "การทดลอง"
+      iterator.to_a.should == [0, 3, 8]
+    end
+
     it "finds all sentence boundaries in an English string" do
       iterator = BreakIterator.new :sentence, "en_US"
       iterator.text = "This is a sentence. This is another sentence, with a comma in it."
@@ -45,5 +51,12 @@ module ICU
       iterator.should_not be_boundary(10)
     end
 
+    it "returns an Enumerator if no block was given" do
+      iterator = BreakIterator.new :word, "nb"
+      expected = ICU.ruby19? ? Enumerator : Enumerable::Enumerator
+
+      iterator.each.should be_kind_of(expected)
+    end
+
   end # BreakIterator
 end # ICU

data/spec/chardet_spec.rb

@@ -5,7 +5,6 @@ require 'spec_helper'
 describe ICU::CharDet::Detector do
 
   before { @d = ICU::CharDet::Detector.new }
-  after { @d.close }
 
   it "should recognize UTF-8" do
     m = @d.detect("æåø")

data/spec/collation_spec.rb

@@ -4,15 +4,24 @@ require 'spec_helper'
 
 module ICU
   module Collation
+    describe "Collation" do
+      it "should collate an array of strings" do
+        Collation.collate("nb", %w[æ å ø]).should == %w[æ ø å]
+      end
+    end
+
     describe Collator do
 
       before { @c = Collator.new("nb") }
-      after { @c.close }
 
       it "should collate an array of strings" do
         @c.collate(%w[å ø æ]).should == %w[æ ø å]
       end
 
+      it "raises an error if argument does not respond to :sort" do
+        lambda { @c.collate(1) }.should raise_error(ArgumentError)
+      end
+
       it "should return available locales" do
         locales = ICU::Collation.available_locales
         locales.should be_kind_of(Array)
@@ -22,7 +31,6 @@ module ICU
 
       it "should return the locale of the collator" do
         l = @c.locale
-        l.should be_kind_of(String)
         l.should == "nb"
       end
 

data/spec/transliteration_spec.rb

@@ -9,8 +9,6 @@ module ICU
       @t = Transliteration::Transliterator.new(*args)
     end
 
-    after { @t.close if @t }
-
     [
      { :id => "Any-Hex",     :input => "abcde",  :output => "\\u0061\\u0062\\u0063\\u0064\\u0065"  },
      { :id => "Lower",       :input => "ABC",    :output => "abc"                                  },

metadata

@@ -2,7 +2,7 @@
 name: ffi-icu
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.0.5
+  version: 0.0.6
 platform: ruby
 authors: 
 - Jari Bakken