Fix int32_t, simpler spec for transliteration

lib/ffi-icu/chardet.rb

@@ -44,7 +44,7 @@ module ICU
       def detect_all(str)
         set_text(str)
 
-        matches_found_ptr = FFI::MemoryPointer.new :int32
+        matches_found_ptr = FFI::MemoryPointer.new :int32_t
         array_ptr = Lib.check_error do |status|
           Lib.ucsdet_detectAll(@detector, matches_found_ptr, status)
         end

lib/ffi-icu/lib.rb

@@ -81,8 +81,8 @@ module ICU
     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 :u_charsToUChars, "u_charsToUChars#{suffix}", [:string, :pointer, :int32], :void
-    attach_function :u_UCharsToChars, "u_UCharsToChars#{suffix}", [:pointer, :string, :int32], :void
+    attach_function :u_charsToUChars, "u_charsToUChars#{suffix}", [:string, :pointer, :int32_t], :void
+    attach_function :u_UCharsToChars, "u_UCharsToChars#{suffix}", [:pointer, :string, :int32_t], :void
 
     # CharDet
     #
@@ -91,12 +91,12 @@ module ICU
 
     attach_function :ucsdet_open, "ucsdet_open#{suffix}",  [:pointer], :pointer
     attach_function :ucsdet_close, "ucsdet_close#{suffix}",  [:pointer], :void
-    attach_function :ucsdet_setText, "ucsdet_setText#{suffix}",  [:pointer, :string, :int32, :pointer], :void
-    attach_function :ucsdet_setDeclaredEncoding, "ucsdet_setDeclaredEncoding#{suffix}",  [:pointer, :string, :int32, :pointer], :void
+    attach_function :ucsdet_setText, "ucsdet_setText#{suffix}",  [:pointer, :string, :int32_t, :pointer], :void
+    attach_function :ucsdet_setDeclaredEncoding, "ucsdet_setDeclaredEncoding#{suffix}",  [:pointer, :string, :int32_t, :pointer], :void
     attach_function :ucsdet_detect, "ucsdet_detect#{suffix}",  [:pointer, :pointer], :pointer
     attach_function :ucsdet_detectAll, "ucsdet_detectAll#{suffix}",  [:pointer, :pointer, :pointer], :pointer
     attach_function :ucsdet_getName, "ucsdet_getName#{suffix}",  [:pointer, :pointer], :string
-    attach_function :ucsdet_getConfidence, "ucsdet_getConfidence#{suffix}",  [:pointer, :pointer], :int32
+    attach_function :ucsdet_getConfidence, "ucsdet_getConfidence#{suffix}",  [:pointer, :pointer], :int32_t
     attach_function :ucsdet_getLanguage, "ucsdet_getLanguage#{suffix}",  [:pointer, :pointer], :string
     attach_function :ucsdet_getAllDetectableCharsets, "ucsdet_getAllDetectableCharsets#{suffix}",  [:pointer, :pointer], :pointer
     attach_function :ucsdet_isInputFilterEnabled, "ucsdet_isInputFilterEnabled#{suffix}",  [:pointer], :bool
@@ -109,15 +109,15 @@ module ICU
 
     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, :pointer, :int32], :int
+    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], :string
-    attach_function :ucol_countAvailable, "ucol_countAvailable#{suffix}", [], :int32
+    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
-    attach_function :ucol_greater, "ucol_greater#{suffix}", [:pointer, :pointer, :int32, :pointer, :int32], :bool
-    attach_function :ucol_greaterOrEqual, "ucol_greaterOrEqual#{suffix}", [:pointer, :pointer, :int32, :pointer, :int32], :bool
-    attach_function :ucol_equal, "ucol_equal#{suffix}", [:pointer, :pointer, :int32, :pointer, :int32], :bool
+    attach_function :ucol_greater, "ucol_greater#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
+    attach_function :ucol_greaterOrEqual, "ucol_greaterOrEqual#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
+    attach_function :ucol_equal, "ucol_equal#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
 
     # Transliteration
     #
@@ -125,8 +125,8 @@ module ICU
     #
 
     class UParseError < FFI::Struct
-      layout :line,         :int32,
-             :offset,       :int32,
+      layout :line,         :int32_t,
+             :offset,       :int32_t,
              :pre_context,  :pointer,
              :post_context, :pointer
 
@@ -134,19 +134,19 @@ module ICU
     end
 
     class UTransPosition < FFI::Struct
-      layout :context_start, :int32,
-             :context_limit, :int32,
-             :start,         :int32,
-             :end,           :int32
+      layout :context_start, :int32_t,
+             :context_limit, :int32_t,
+             :start,         :int32_t,
+             :end,           :int32_t
 
     end
 
     enum :trans_direction, [:forward, :reverse]
 
     attach_function :utrans_openIDs, "utrans_openIDs#{suffix}", [:pointer], :pointer
-    attach_function :utrans_openU, "utrans_openU#{suffix}", [:pointer, :int32, :trans_direction, :pointer, :int32, :pointer, :pointer], :pointer
-    attach_function :utrans_open, "utrans_open#{suffix}", [:string, :trans_direction, :pointer, :int32, :pointer, :pointer], :pointer
-    attach_function :utrans_transUChars, "utrans_transUChars#{suffix}", [:pointer, :pointer, :pointer, :int32, :int32, :pointer, :pointer], :void
+    attach_function :utrans_openU, "utrans_openU#{suffix}", [:pointer, :int32_t, :trans_direction, :pointer, :int32_t, :pointer, :pointer], :pointer
+    attach_function :utrans_open, "utrans_open#{suffix}", [:string, :trans_direction, :pointer, :int32_t, :pointer, :pointer], :pointer
+    attach_function :utrans_transUChars, "utrans_transUChars#{suffix}", [:pointer, :pointer, :pointer, :int32_t, :int32_t, :pointer, :pointer], :void
 
     # Normalization
     #
@@ -162,6 +162,6 @@ module ICU
                                 :fcd,     6
                               ]
 
-    attach_function :unorm_normalize, "unorm_normalize#{suffix}", [:pointer, :int32, :normalization_mode, :int32, :pointer, :int32, :pointer], :int32
+    attach_function :unorm_normalize, "unorm_normalize#{suffix}", [:pointer, :int32_t, :normalization_mode, :int32_t, :pointer, :int32_t, :pointer], :int32_t
   end # Lib
 end # ICU

lib/ffi-icu/transliteration.rb

@@ -26,33 +26,29 @@ module ICU
         @parse_error = Lib::UParseError.new
         Lib.check_error do |status|
           # couldn't get utrans_openU to work properly, so using deprecated utrans_open for now
-          @tr = Lib.utrans_open(id, direction, nil, 0, nil, status)
+          @tr = Lib.utrans_open(id, direction, nil, -1, @parse_error, status)
         end
       end
 
-      def transliterate(string)
-        capacity = string.length + 1
+      def transliterate(from)
+        capacity = from.bytesize + 1
+        limit    = FFI::MemoryPointer.new :int32_t
 
-        text_length = FFI::MemoryPointer.new :int32
-        text_length.put_int32(0, string.length)
+        text_length = FFI::MemoryPointer.new :int32_t
+        text_length.put_int32(0, from.length)
 
-        limit = FFI::MemoryPointer.new :int32
-
-        uchar_ptr = UCharPointer.from_string(string)
-        # pos = Lib::UTransPosition.new
-        # pos[:context_start] = pos[:start] = 0
-        # pos[:context_limit] = pos[:end]   = length
+        buf = UCharPointer.from_string(from)
 
         Lib.check_error do |error|
-          Lib.utrans_transUChars(@tr, uchar_ptr, text_length, capacity, 0, limit, error)
+          Lib.utrans_transUChars(@tr, buf, text_length, capacity, 0, limit, error)
         end
 
         # TODO: clean up UCharPointer
-        uchar_ptr.string
+        buf.string
       end
 
       def close
-
+        Lib.utrans_close @tr
       end
     end # Transliterator
 

lib/ffi-icu/uchar.rb

@@ -13,8 +13,10 @@ module ICU
       ptr
     end
 
-    def string
-      wstring = get_array_of_uint16(0, size / FFI.type_size(UCHAR_TYPE))
+    def string(length = nil)
+      length ||= size / FFI.type_size(UCHAR_TYPE)
+
+      wstring = get_array_of_uint16(0, length)
       wstring.pack("U*")
     end
 

spec/transliteration_spec.rb

@@ -5,11 +5,17 @@ module ICU
 
     describe Transliterator do
 
-      before { @t = Transliterator.new("Greek-Latin", :reverse) }
-      after { @t.close }
+      def transliterator(*args)
+        @t = Transliterator.new(*args)
+      end
+      after { @t.close if @t }
 
-      it "should transliterate a string" do
-        @t.transliterate("Hello World").should == "Χελλο Ωορλδ"
+      it "should transliterate Greek-Latin" do
+        transliterator("Greek-Latin").transliterate("Hello World").should == "Χελλο Ωορλδ"
+      end
+
+      it "should transliterate Lower" do
+        transliterator("Lower").transliterate("ABC").should == "abc"
       end
 
     end