Merge pull request #62 from michaelnmmeyer/master

Reduce the size of the binary.

CMakeLists.txt

@@ -7,9 +7,9 @@ disallow_intree_builds()
 project (utf8proc C)
 
 # Be sure to also update these in Makefile!
-set(SO_MAJOR 1)
-set(SO_MINOR 3)
-set(SO_PATCH 1)
+set(SO_MAJOR 2)
+set(SO_MINOR 0)
+set(SO_PATCH 0)
 
 add_definitions (
   -DUTF8PROC_EXPORTS

MANIFEST

@@ -2,6 +2,6 @@ include/
 include/utf8proc.h
 lib/
 lib/libutf8proc.a
-lib/libutf8proc.so -> libutf8proc.so.1.3.1
-lib/libutf8proc.so.1 -> libutf8proc.so.1.3.1
-lib/libutf8proc.so.1.3.1
+lib/libutf8proc.so -> libutf8proc.so.2.0.0
+lib/libutf8proc.so.2 -> libutf8proc.so.2.0.0
+lib/libutf8proc.so.2.0.0

Makefile

@@ -19,9 +19,9 @@ UCFLAGS = $(CFLAGS) $(PICFLAG) $(C99FLAG) $(WCFLAGS) -DUTF8PROC_EXPORTS
 # not API compatibility: MAJOR should be incremented whenever *binary*
 # compatibility is broken, even if the API is backward-compatible
 # Be sure to also update these in MANIFEST and CMakeLists.txt!
-MAJOR=1
-MINOR=3
-PATCH=1
+MAJOR=2
+MINOR=0
+PATCH=0
 
 OS := $(shell uname)
 ifeq ($(OS),Darwin) # MacOS X

data/data_generator.rb

@@ -116,13 +116,14 @@ def str2c(string, prefix)
   return "UTF8PROC_#{prefix}_#{string.upcase}"
 end
 def ary2c(array)
-  return "NULL" if array.nil?
+  return "UINT16_MAX" if array.nil?
   unless $int_array_indicies[array]
     $int_array_indicies[array] = $int_array.length
     array.each { |entry| $int_array << entry }
     $int_array << -1
   end
-  return "utf8proc_sequences + #{$int_array_indicies[array]}"
+  raise "Array index out of bound" if $int_array_indicies[array] >= 65535
+  return "#{$int_array_indicies[array]}"
 end
 
 class UnicodeChar
@@ -305,7 +306,7 @@ end
 $stdout << "};\n\n"
 
 $stdout << "const utf8proc_property_t utf8proc_properties[] = {\n"
-$stdout << "  {0, 0, 0, 0, NULL, NULL, -1, -1, -1, -1, -1, false,false,false,false, UTF8PROC_BOUNDCLASS_OTHER, 0},\n"
+$stdout << "  {0, 0, 0, 0, UINT16_MAX, UINT16_MAX, -1, -1, -1, -1, -1, false,false,false,false, UTF8PROC_BOUNDCLASS_OTHER, 0},\n"
 properties.each { |line|
   $stdout << line
 }

utf8proc.c

@@ -357,10 +357,10 @@ UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_decompose_char(utf8proc_int32_t uc,
       category == UTF8PROC_CATEGORY_ME) return 0;
   }
   if (options & UTF8PROC_CASEFOLD) {
-    if (property->casefold_mapping) {
+    if (property->casefold_mapping != UINT16_MAX) {
       const utf8proc_int32_t *casefold_entry;
       utf8proc_ssize_t written = 0;
-      for (casefold_entry = property->casefold_mapping;
+      for (casefold_entry = &utf8proc_sequences[property->casefold_mapping];
           *casefold_entry >= 0; casefold_entry++) {
         written += utf8proc_decompose_char(*casefold_entry, dst+written,
           (bufsize > written) ? (bufsize - written) : 0, options,
@@ -371,11 +371,11 @@ UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_decompose_char(utf8proc_int32_t uc,
     }
   }
   if (options & (UTF8PROC_COMPOSE|UTF8PROC_DECOMPOSE)) {
-    if (property->decomp_mapping &&
+    if (property->decomp_mapping != UINT16_MAX &&
         (!property->decomp_type || (options & UTF8PROC_COMPAT))) {
       const utf8proc_int32_t *decomp_entry;
       utf8proc_ssize_t written = 0;
-      for (decomp_entry = property->decomp_mapping;
+      for (decomp_entry = &utf8proc_sequences[property->decomp_mapping];
           *decomp_entry >= 0; decomp_entry++) {
         written += utf8proc_decompose_char(*decomp_entry, dst+written,
           (bufsize > written) ? (bufsize - written) : 0, options,

utf8proc.h

@@ -132,6 +132,10 @@ extern "C" {
 #define SSIZE_MAX ((size_t)SIZE_MAX/2)
 #endif
 
+#ifndef UINT16_MAX
+#  define UINT16_MAX ~(utf8proc_uint16_t)0
+#endif
+
 /**
  * Option flags used by several functions in the library.
  */
@@ -238,8 +242,8 @@ typedef struct utf8proc_property_struct {
    * @see utf8proc_decomp_type_t.
    */
   utf8proc_propval_t decomp_type;
-  const utf8proc_int32_t *decomp_mapping;
-  const utf8proc_int32_t *casefold_mapping;
+  utf8proc_uint16_t decomp_mapping;
+  utf8proc_uint16_t casefold_mapping;
   utf8proc_int32_t uppercase_mapping;
   utf8proc_int32_t lowercase_mapping;
   utf8proc_int32_t titlecase_mapping;