Impelment Unicode normalization Form D in libarchive itself in order …

…not to rely on CoreServices framework on Mac OS.

It passes all NormalizationTest.txt released The Unicode Consortium(http://www.unicode.org).

CMakeLists.txt

@@ -1369,11 +1369,6 @@ IF(MSVC)
   ADD_DEFINITIONS(-D_CRT_SECURE_NO_DEPRECATE)
 ENDIF(MSVC)
 
-# We need CoreServices on Mac OS.
-IF(APPLE)
-  LIST(APPEND ADDITIONAL_LIBS "-framework CoreServices")
-ENDIF(APPLE)
-
 IF(ENABLE_TEST)
   ADD_CUSTOM_TARGET(run_all_tests)
 ENDIF(ENABLE_TEST)

build/utils/gen_archive_string_composition_h.sh

@@ -1,10 +1,13 @@
 #!/bin/sh
 #
-# This needs http://unicode.org/Public/UNIDATA/UnicodeData.txt
+# This needs http://unicode.org/Public/6.0.0/ucd/UnicodeData.txt
 #
 inputfile="$1"	# Expect UnicodeData.txt
 outfile=archive_string_composition.h
 pickout=/tmp/mk_unicode_composition_tbl$$.awk
+pickout2=/tmp/mk_unicode_composition_tbl2$$.awk
+#nfdtmp=/tmp/mk_unicode_decomposition_tmp$$.txt
+nfdtmp="nfdtmpx"
 #################################################################################
 #
 # Append the file header of "archive_string_composition.h"
@@ -14,7 +17,7 @@ append_copyright()
 {
 cat > ${outfile} <<CR_END
 /*-
- * Copyright (c) 2011 libarchive Project
+ * Copyright (c) 2011-2012 libarchive Project
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -44,7 +47,7 @@ cat > ${outfile} <<CR_END
 /*
  * ATTENTION!
  *  This file is generated by build/utils/gen_archive_string_composition_h.sh
- *  from http://unicode.org/Public/UNIDATA/UnicodeData.txt
+ *  from http://unicode.org/Public/6.0.0/ucd/UnicodeData.txt
  *
  *  See also http://unicode.org/report/tr15/
  */
@@ -76,6 +79,7 @@ BEGIN {
   min = "";
   max = "";
   cmd="sort | awk -F ' ' '{printf \"\\\\t{ 0x%s , 0x%s , 0x%s },\\\\n\",\$1,\$2,\$3}'"
+  nfdtbl="${nfdtmp}"
   print "static const struct unicode_composition_table u_composition_table[] = {"
 }
 END {
@@ -178,7 +182,6 @@ END {
   }
   print "};"
   print ""
-  print "#endif /* ARCHIVE_STRING_COMPOSITION_H_INCLUDED */"
 }
 #
 #
@@ -241,7 +244,7 @@ function hextoi(hex)
 #}
 #
 # Exclusion code points specified by  
-# http://unicode.org/Public/UNIDATA/CompositionExclusions.txt
+# http://unicode.org/Public/6.0.0/ucd/CompositionExclusions.txt
 ##
 # 1. Script Specifices
 ##
@@ -404,6 +407,35 @@ function hextoi(hex)
         print "0"cp[1], "0"cp[2], "0"\$1 | cmd
     else
         print cp[1], cp[2], \$1 | cmd
+    # NFC ==> NFD table.
+    if (length(\$1) == 4)
+        print "0"\$1, "0"cp[1], "0"cp[2] >>nfdtbl
+    else
+        print \$1, cp[1], cp[2] >>nfdtbl
+}
+AWK_END
+#################################################################################
+# awk script
+#
+#################################################################################
+cat > ${pickout2} <<AWK_END
+#
+BEGIN {
+  FS = " "
+  print "struct unicode_decomposition_table {"
+  print "\tuint32_t nfc;"
+  print "\tuint32_t cp1;"
+  print "\tuint32_t cp2;"
+  print "};"
+  print ""
+  print "static const struct unicode_decomposition_table u_decomposition_table[] = {"
+}
+END {
+  print "};"
+  print ""
+}
+{
+printf "\t{ 0x%s , 0x%s , 0x%s },\n", \$1, \$2, \$3;
 }
 AWK_END
 #################################################################################
@@ -413,6 +445,11 @@ AWK_END
 #################################################################################
 append_copyright
 awk -f ${pickout} ${inputfile} >> ${outfile}
+awk -f ${pickout2} ${nfdtmp} >> ${outfile}
+echo "#endif /* ARCHIVE_STRING_COMPOSITION_H_INCLUDED */" >> ${outfile}
+echo "" >> ${outfile}
 #
 # Remove awk the script.
 rm ${pickout}
+rm ${pickout2}
+rm ${nfdtmp}

configure.ac

@@ -197,13 +197,6 @@ case $host in
   ;;
 esac
 
-# We need CoreServices on Mac OS.
-case $host in
-  *darwin* )
-  LIBS="${LIBS} -framework CoreServices"
-  ;;
-esac
-
 # Checks for header files.
 AC_HEADER_DIRENT
 AC_HEADER_SYS_WAIT