Add m-esr52 at 52.6.0

5 files changed, 2071 insertions, 0 deletions

diff --git a/intl/icu/source/tools/gennorm2/Makefile.in b/intl/icu/source/tools/gennorm2/Makefile.in
new file mode 100644
index 000000000..228076346
--- /dev/null
+++ b/intl/icu/source/tools/gennorm2/Makefile.in
@@ -0,0 +1,81 @@
+## Makefile.in for ICU - tools/gennorm2
+## Copyright (C) 2016 and later: Unicode, Inc. and others.
+## License & terms of use: http://www.unicode.org/copyright.html
+## Copyright (c) 2009-2011, International Business Machines Corporation and
+## others. All Rights Reserved.
+## Steven R. Loomis/Markus W. Scherer
+
+## Source directory information
+srcdir = @srcdir@
+top_srcdir = @top_srcdir@
+
+top_builddir = ../..
+
+include $(top_builddir)/icudefs.mk
+
+## Build directory information
+subdir = tools/gennorm2
+
+TARGET_STUB_NAME = gennorm2
+
+## Extra files to remove for 'make clean'
+CLEANFILES = *~ $(DEPS)
+
+## Target information
+TARGET = $(BINDIR)/$(TARGET_STUB_NAME)$(EXEEXT)
+
+CPPFLAGS += -I$(top_srcdir)/common -I$(srcdir)/../toolutil
+LIBS = $(LIBICUTOOLUTIL) $(LIBICUI18N) $(LIBICUUC) $(DEFAULT_LIBS) $(LIB_M)
+
+OBJECTS = gennorm2.o n2builder.o
+
+DEPS = $(OBJECTS:.o=.d)
+
+## List of phony targets
+.PHONY : all all-local install install-local clean clean-local	\
+distclean distclean-local dist dist-local check check-local install-man
+
+## Clear suffix list
+.SUFFIXES :
+
+## List of standard targets
+all: all-local
+install: install-local
+clean: clean-local
+distclean : distclean-local
+dist: dist-local
+check: all check-local
+
+all-local: $(TARGET)
+
+install-local: all-local
+	$(MKINSTALLDIRS) $(DESTDIR)$(sbindir)
+	$(INSTALL) $(TARGET) $(DESTDIR)$(sbindir)
+
+dist-local:
+
+clean-local:
+	test -z "$(CLEANFILES)" || $(RMV) $(CLEANFILES)
+	$(RMV) $(TARGET) $(OBJECTS)
+
+distclean-local: clean-local
+	$(RMV) Makefile
+
+check-local: all-local
+
+Makefile: $(srcdir)/Makefile.in  $(top_builddir)/config.status
+	cd $(top_builddir) \
+	 && CONFIG_FILES=$(subdir)/$@ CONFIG_HEADERS= $(SHELL) ./config.status
+
+$(TARGET) : $(OBJECTS)
+	$(LINK.cc) $(OUTOPT)$@ $^ $(LIBS)
+	$(POST_BUILD_STEP)
+
+
+ifeq (,$(MAKECMDGOALS))
+-include $(DEPS)
+else
+ifneq ($(patsubst %clean,,$(MAKECMDGOALS)),)
+-include $(DEPS)
+endif
+endif
diff --git a/intl/icu/source/tools/gennorm2/gennorm2.cpp b/intl/icu/source/tools/gennorm2/gennorm2.cpp
new file mode 100644
index 000000000..9c85e0bc9
--- /dev/null
+++ b/intl/icu/source/tools/gennorm2/gennorm2.cpp
@@ -0,0 +1,286 @@
+// Copyright (C) 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2009-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  gennorm2.cpp
+*   encoding:   US-ASCII
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2009nov25
+*   created by: Markus W. Scherer
+*
+*   This program reads text files that define Unicode normalization,
+*   parses them, and builds a binary data file.
+*/
+
+#include "unicode/utypes.h"
+#include "n2builder.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "unicode/errorcode.h"
+#include "unicode/localpointer.h"
+#include "unicode/putil.h"
+#include "unicode/uchar.h"
+#include "unicode/unistr.h"
+#include "charstr.h"
+#include "normalizer2impl.h"
+#include "toolutil.h"
+#include "uoptions.h"
+#include "uparse.h"
+
+#if UCONFIG_NO_NORMALIZATION
+#include "unewdata.h"
+#endif
+
+U_NAMESPACE_BEGIN
+
+UBool beVerbose=FALSE, haveCopyright=TRUE;
+
+U_DEFINE_LOCAL_OPEN_POINTER(LocalStdioFilePointer, FILE, fclose);
+
+#if !UCONFIG_NO_NORMALIZATION
+void parseFile(FILE *f, Normalizer2DataBuilder &builder);
+#endif
+
+/* -------------------------------------------------------------------------- */
+
+enum {
+    HELP_H,
+    HELP_QUESTION_MARK,
+    VERBOSE,
+    COPYRIGHT,
+    SOURCEDIR,
+    OUTPUT_FILENAME,
+    UNICODE_VERSION,
+    WRITE_C_SOURCE,
+    OPT_FAST
+};
+
+static UOption options[]={
+    UOPTION_HELP_H,
+    UOPTION_HELP_QUESTION_MARK,
+    UOPTION_VERBOSE,
+    UOPTION_COPYRIGHT,
+    UOPTION_SOURCEDIR,
+    UOPTION_DEF("output", 'o', UOPT_REQUIRES_ARG),
+    UOPTION_DEF("unicode", 'u', UOPT_REQUIRES_ARG),
+    UOPTION_DEF("csource", '\1', UOPT_NO_ARG),
+    UOPTION_DEF("fast", '\1', UOPT_NO_ARG)
+};
+
+extern "C" int
+main(int argc, char* argv[]) {
+    U_MAIN_INIT_ARGS(argc, argv);
+
+    /* preset then read command line options */
+    options[SOURCEDIR].value="";
+    argc=u_parseArgs(argc, argv, sizeof(options)/sizeof(options[HELP_H]), options);
+
+    /* error handling, printing usage message */
+    if(argc<0) {
+        fprintf(stderr,
+            "error in command line argument \"%s\"\n",
+            argv[-argc]);
+    }
+    if(!options[OUTPUT_FILENAME].doesOccur) {
+        argc=-1;
+    }
+    if( argc<2 ||
+        options[HELP_H].doesOccur || options[HELP_QUESTION_MARK].doesOccur
+    ) {
+        /*
+         * Broken into chunks because the C89 standard says the minimum
+         * required supported string length is 509 bytes.
+         */
+        fprintf(stderr,
+            "Usage: %s [-options] infiles+ -o outputfilename\n"
+            "\n"
+            "Reads the infiles with normalization data and\n"
+            "creates a binary or C source file (outputfilename) with the data.\n"
+            "\n",
+            argv[0]);
+        fprintf(stderr,
+            "Options:\n"
+            "\t-h or -? or --help  this usage text\n"
+            "\t-v or --verbose     verbose output\n"
+            "\t-c or --copyright   include a copyright notice\n"
+            "\t-u or --unicode     Unicode version, followed by the version like 5.2.0\n");
+        fprintf(stderr,
+            "\t-s or --sourcedir   source directory, followed by the path\n"
+            "\t-o or --output      output filename\n"
+            "\t      --csource     writes a C source file with initializers\n");
+        fprintf(stderr,
+            "\t      --fast        optimize the data for fast normalization,\n"
+            "\t                    which might increase its size  (Writes fully decomposed\n"
+            "\t                    regular mappings instead of delta mappings.\n"
+            "\t                    You should measure the runtime speed to make sure that\n"
+            "\t                    this is a good trade-off.)\n");
+        return argc<0 ? U_ILLEGAL_ARGUMENT_ERROR : U_ZERO_ERROR;
+    }
+
+    beVerbose=options[VERBOSE].doesOccur;
+    haveCopyright=options[COPYRIGHT].doesOccur;
+
+    IcuToolErrorCode errorCode("gennorm2/main()");
+
+#if UCONFIG_NO_NORMALIZATION
+
+    fprintf(stderr,
+        "gennorm2 writes a dummy binary data file "
+        "because UCONFIG_NO_NORMALIZATION is set, \n"
+        "see icu/source/common/unicode/uconfig.h\n");
+    udata_createDummy(NULL, NULL, options[OUTPUT_FILENAME].value, errorCode);
+    // Should not return an error since this is the expected behaviour if UCONFIG_NO_NORMALIZATION is on.
+    // return U_UNSUPPORTED_ERROR;
+    return 0;
+
+#else
+
+    LocalPointer<Normalizer2DataBuilder> builder(new Normalizer2DataBuilder(errorCode), errorCode);
+    errorCode.assertSuccess();
+
+    if(options[UNICODE_VERSION].doesOccur) {
+        builder->setUnicodeVersion(options[UNICODE_VERSION].value);
+    }
+
+    if(options[OPT_FAST].doesOccur) {
+        builder->setOptimization(Normalizer2DataBuilder::OPTIMIZE_FAST);
+    }
+
+    // prepare the filename beginning with the source dir
+    CharString filename(options[SOURCEDIR].value, errorCode);
+    int32_t pathLength=filename.length();
+    if( pathLength>0 &&
+        filename[pathLength-1]!=U_FILE_SEP_CHAR &&
+        filename[pathLength-1]!=U_FILE_ALT_SEP_CHAR
+    ) {
+        filename.append(U_FILE_SEP_CHAR, errorCode);
+        pathLength=filename.length();
+    }
+
+    for(int i=1; i<argc; ++i) {
+        printf("gennorm2: processing %s\n", argv[i]);
+        filename.append(argv[i], errorCode);
+        LocalStdioFilePointer f(fopen(filename.data(), "r"));
+        if(f==NULL) {
+            fprintf(stderr, "gennorm2 error: unable to open %s\n", filename.data());
+            exit(U_FILE_ACCESS_ERROR);
+        }
+        builder->setOverrideHandling(Normalizer2DataBuilder::OVERRIDE_PREVIOUS);
+        parseFile(f.getAlias(), *builder);
+        filename.truncate(pathLength);
+    }
+
+    if(options[WRITE_C_SOURCE].doesOccur) {
+        builder->writeCSourceFile(options[OUTPUT_FILENAME].value);
+    } else {
+        builder->writeBinaryFile(options[OUTPUT_FILENAME].value);
+    }
+
+    return errorCode.get();
+
+#endif
+}
+
+#if !UCONFIG_NO_NORMALIZATION
+
+void parseFile(FILE *f, Normalizer2DataBuilder &builder) {
+    IcuToolErrorCode errorCode("gennorm2/parseFile()");
+    char line[300];
+    uint32_t startCP, endCP;
+    while(NULL!=fgets(line, (int)sizeof(line), f)) {
+        char *comment=(char *)strchr(line, '#');
+        if(comment!=NULL) {
+            *comment=0;
+        }
+        u_rtrim(line);
+        if(line[0]==0) {
+            continue;  // skip empty and comment-only lines
+        }
+        if(line[0]=='*') {
+            const char *s=u_skipWhitespace(line+1);
+            if(0==strncmp(s, "Unicode", 7)) {
+                s=u_skipWhitespace(s+7);
+                builder.setUnicodeVersion(s);
+            }
+            continue;  // reserved syntax
+        }
+        const char *delimiter;
+        int32_t rangeLength=
+            u_parseCodePointRangeAnyTerminator(line, &startCP, &endCP, &delimiter, errorCode);
+        if(errorCode.isFailure()) {
+            fprintf(stderr, "gennorm2 error: parsing code point range from %s\n", line);
+            exit(errorCode.reset());
+        }
+        delimiter=u_skipWhitespace(delimiter);
+        if(*delimiter==':') {
+            const char *s=u_skipWhitespace(delimiter+1);
+            char *end;
+            unsigned long value=strtoul(s, &end, 10);
+            if(end<=s || *u_skipWhitespace(end)!=0 || value>=0xff) {
+                fprintf(stderr, "gennorm2 error: parsing ccc from %s\n", line);
+                exit(U_PARSE_ERROR);
+            }
+            for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
+                builder.setCC(c, (uint8_t)value);
+            }
+            continue;
+        }
+        if(*delimiter=='-') {
+            if(*u_skipWhitespace(delimiter+1)!=0) {
+                fprintf(stderr, "gennorm2 error: parsing remove-mapping %s\n", line);
+                exit(U_PARSE_ERROR);
+            }
+            for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
+                builder.removeMapping(c);
+            }
+            continue;
+        }
+        if(*delimiter=='=' || *delimiter=='>') {
+            UChar uchars[Normalizer2Impl::MAPPING_LENGTH_MASK];
+            int32_t length=u_parseString(delimiter+1, uchars, UPRV_LENGTHOF(uchars), NULL, errorCode);
+            if(errorCode.isFailure()) {
+                fprintf(stderr, "gennorm2 error: parsing mapping string from %s\n", line);
+                exit(errorCode.reset());
+            }
+            UnicodeString mapping(FALSE, uchars, length);
+            if(*delimiter=='=') {
+                if(rangeLength!=1) {
+                    fprintf(stderr,
+                            "gennorm2 error: round-trip mapping for more than 1 code point on %s\n",
+                            line);
+                    exit(U_PARSE_ERROR);
+                }
+                builder.setRoundTripMapping((UChar32)startCP, mapping);
+            } else {
+                for(UChar32 c=(UChar32)startCP; c<=(UChar32)endCP; ++c) {
+                    builder.setOneWayMapping(c, mapping);
+                }
+            }
+            continue;
+        }
+        fprintf(stderr, "gennorm2 error: unrecognized data line %s\n", line);
+        exit(U_PARSE_ERROR);
+    }
+}
+
+#endif // !UCONFIG_NO_NORMALIZATION
+
+U_NAMESPACE_END
+
+/*
+ * Hey, Emacs, please set the following:
+ *
+ * Local Variables:
+ * indent-tabs-mode: nil
+ * End:
+ *
+ */
diff --git a/intl/icu/source/tools/gennorm2/gennorm2.vcxproj b/intl/icu/source/tools/gennorm2/gennorm2.vcxproj
new file mode 100644
index 000000000..99625ebde
--- /dev/null
+++ b/intl/icu/source/tools/gennorm2/gennorm2.vcxproj
@@ -0,0 +1,270 @@
+<?xml version="1.0" encoding="utf-8"?>

+<Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">

+  <ItemGroup Label="ProjectConfigurations">

+    <ProjectConfiguration Include="Debug|Win32">

+      <Configuration>Debug</Configuration>

+      <Platform>Win32</Platform>

+    </ProjectConfiguration>

+    <ProjectConfiguration Include="Debug|x64">

+      <Configuration>Debug</Configuration>

+      <Platform>x64</Platform>

+    </ProjectConfiguration>

+    <ProjectConfiguration Include="Release|Win32">

+      <Configuration>Release</Configuration>

+      <Platform>Win32</Platform>

+    </ProjectConfiguration>

+    <ProjectConfiguration Include="Release|x64">

+      <Configuration>Release</Configuration>

+      <Platform>x64</Platform>

+    </ProjectConfiguration>

+  </ItemGroup>

+  <PropertyGroup Label="Globals">

+    <ProjectGuid>{C7891A65-80AB-4245-912E-5F1E17B0E6C4}</ProjectGuid>

+    <RootNamespace>gennorm2</RootNamespace>

+    <Keyword>Win32Proj</Keyword>

+  </PropertyGroup>

+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />

+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">

+    <ConfigurationType>Application</ConfigurationType>

+    <CharacterSet>Unicode</CharacterSet>

+    <PlatformToolset>v140</PlatformToolset>

+  </PropertyGroup>

+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">

+    <ConfigurationType>Application</ConfigurationType>

+    <CharacterSet>Unicode</CharacterSet>

+    <WholeProgramOptimization>true</WholeProgramOptimization>

+    <PlatformToolset>v140</PlatformToolset>

+  </PropertyGroup>

+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">

+    <ConfigurationType>Application</ConfigurationType>

+    <CharacterSet>Unicode</CharacterSet>

+    <PlatformToolset>v140</PlatformToolset>

+  </PropertyGroup>

+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">

+    <ConfigurationType>Application</ConfigurationType>

+    <CharacterSet>Unicode</CharacterSet>

+    <WholeProgramOptimization>true</WholeProgramOptimization>

+    <PlatformToolset>v140</PlatformToolset>

+  </PropertyGroup>

+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />

+  <ImportGroup Label="ExtensionSettings">

+  </ImportGroup>

+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="PropertySheets">

+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />

+  </ImportGroup>

+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="PropertySheets">

+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />

+  </ImportGroup>

+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="PropertySheets">

+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />

+  </ImportGroup>

+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="PropertySheets">

+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />

+  </ImportGroup>

+  <PropertyGroup Label="UserMacros" />

+  <PropertyGroup>

+    <_ProjectFileVersion>10.0.30319.1</_ProjectFileVersion>

+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">.\x86\Release\</OutDir>

+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">.\x86\Release\</IntDir>

+    <LinkIncremental Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">false</LinkIncremental>

+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">.\x86\Debug\</OutDir>

+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">.\x86\Debug\</IntDir>

+    <LinkIncremental Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">true</LinkIncremental>

+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Release|x64'">.\x64\Release\</OutDir>

+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Release|x64'">.\x64\Release\</IntDir>

+    <LinkIncremental Condition="'$(Configuration)|$(Platform)'=='Release|x64'">false</LinkIncremental>

+    <OutDir Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">.\x64\Debug\</OutDir>

+    <IntDir Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">.\x64\Debug\</IntDir>

+    <LinkIncremental Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">true</LinkIncremental>

+  </PropertyGroup>

+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">

+    <CustomBuildStep>

+      <Command>copy "$(TargetPath)" ..\..\..\bin
+</Command>

+      <Outputs>..\..\..\bin\$(TargetFileName);%(Outputs)</Outputs>

+    </CustomBuildStep>

+    <ClCompile>

+      <AdditionalIncludeDirectories>..\..\common;..\toolutil;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>

+      <PreprocessorDefinitions>WIN32;NDEBUG;_CRT_SECURE_NO_DEPRECATE;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Optimization>MaxSpeed</Optimization>

+      <IntrinsicFunctions>true</IntrinsicFunctions>

+      <StringPooling>true</StringPooling>

+      <RuntimeLibrary>MultiThreadedDLL</RuntimeLibrary>

+      <FunctionLevelLinking>true</FunctionLevelLinking>

+      <DisableLanguageExtensions>false</DisableLanguageExtensions>

+      <TreatWChar_tAsBuiltInType>true</TreatWChar_tAsBuiltInType>

+      <PrecompiledHeaderOutputFile>.\x86\Release\gennorm2.pch</PrecompiledHeaderOutputFile>

+      <AssemblerListingLocation>.\x86\Release\</AssemblerListingLocation>

+      <ObjectFileName>.\x86\Release\</ObjectFileName>

+      <ProgramDataBaseFileName>.\x86\Release\</ProgramDataBaseFileName>

+      <WarningLevel>Level3</WarningLevel>

+      <DebugInformationFormat>ProgramDatabase</DebugInformationFormat>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <CompileAs>Default</CompileAs>

+    </ClCompile>

+    <ResourceCompile>

+      <PreprocessorDefinitions>NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Culture>0x0409</Culture>

+    </ResourceCompile>

+    <Link>

+      <OutputFile>.\x86\Release\gennorm2.exe</OutputFile>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <ProgramDatabaseFile>.\x86\Release\gennorm2.pdb</ProgramDatabaseFile>

+      <GenerateDebugInformation>true</GenerateDebugInformation>

+      <SubSystem>Console</SubSystem>

+      <OptimizeReferences>true</OptimizeReferences>

+      <EnableCOMDATFolding>true</EnableCOMDATFolding>

+      <TargetMachine>MachineX86</TargetMachine>

+      <RandomizedBaseAddress>false</RandomizedBaseAddress>

+      <DataExecutionPrevention>false</DataExecutionPrevention>

+    </Link>

+  </ItemDefinitionGroup>

+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">

+    <CustomBuildStep>

+      <Command>copy "$(TargetPath)" ..\..\..\bin
+</Command>

+      <Outputs>..\..\..\bin\$(TargetFileName);%(Outputs)</Outputs>

+    </CustomBuildStep>

+    <ClCompile>

+      <Optimization>Disabled</Optimization>

+      <AdditionalIncludeDirectories>..\..\common;..\toolutil;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>

+      <PreprocessorDefinitions>WIN32;_DEBUG;_CRT_SECURE_NO_DEPRECATE;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <IntrinsicFunctions>true</IntrinsicFunctions>

+      <MinimalRebuild>true</MinimalRebuild>

+      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>

+      <StringPooling>true</StringPooling>

+      <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>

+      <BufferSecurityCheck>true</BufferSecurityCheck>

+      <DisableLanguageExtensions>false</DisableLanguageExtensions>

+      <TreatWChar_tAsBuiltInType>true</TreatWChar_tAsBuiltInType>

+      <PrecompiledHeaderOutputFile>.\x86\Debug\gennorm2.pch</PrecompiledHeaderOutputFile>

+      <AssemblerListingLocation>.\x86\Debug\</AssemblerListingLocation>

+      <ObjectFileName>.\x86\Debug\</ObjectFileName>

+      <ProgramDataBaseFileName>.\x86\Debug\</ProgramDataBaseFileName>

+      <BrowseInformation>true</BrowseInformation>

+      <WarningLevel>Level3</WarningLevel>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <DebugInformationFormat>EditAndContinue</DebugInformationFormat>

+      <CompileAs>Default</CompileAs>

+    </ClCompile>

+    <ResourceCompile>

+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Culture>0x0409</Culture>

+    </ResourceCompile>

+    <Link>

+      <OutputFile>.\x86\Debug\gennorm2.exe</OutputFile>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <GenerateDebugInformation>true</GenerateDebugInformation>

+      <ProgramDatabaseFile>.\x86\Debug\gennorm2.pdb</ProgramDatabaseFile>

+      <SubSystem>Console</SubSystem>

+      <TargetMachine>MachineX86</TargetMachine>

+      <RandomizedBaseAddress>false</RandomizedBaseAddress>

+      <DataExecutionPrevention>false</DataExecutionPrevention>

+    </Link>

+  </ItemDefinitionGroup>

+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">

+    <CustomBuildStep>

+      <Command>copy "$(TargetPath)" ..\..\..\bin64
+</Command>

+      <Outputs>..\..\..\bin64\$(TargetFileName);%(Outputs)</Outputs>

+    </CustomBuildStep>

+    <ClCompile>

+      <AdditionalIncludeDirectories>..\..\common;..\toolutil;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>

+      <PreprocessorDefinitions>WIN64;WIN32;NDEBUG;_CRT_SECURE_NO_DEPRECATE;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Optimization>MaxSpeed</Optimization>

+      <IntrinsicFunctions>true</IntrinsicFunctions>

+      <StringPooling>true</StringPooling>

+      <RuntimeLibrary>MultiThreadedDLL</RuntimeLibrary>

+      <FunctionLevelLinking>true</FunctionLevelLinking>

+      <DisableLanguageExtensions>false</DisableLanguageExtensions>

+      <TreatWChar_tAsBuiltInType>true</TreatWChar_tAsBuiltInType>

+      <PrecompiledHeaderOutputFile>.\x64\Release\gennorm2.pch</PrecompiledHeaderOutputFile>

+      <AssemblerListingLocation>.\x64\Release\</AssemblerListingLocation>

+      <ObjectFileName>.\x64\Release\</ObjectFileName>

+      <ProgramDataBaseFileName>.\x64\Release\</ProgramDataBaseFileName>

+      <WarningLevel>Level3</WarningLevel>

+      <DebugInformationFormat>ProgramDatabase</DebugInformationFormat>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <CompileAs>Default</CompileAs>

+    </ClCompile>

+    <ResourceCompile>

+      <PreprocessorDefinitions>NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Culture>0x0409</Culture>

+    </ResourceCompile>

+    <Link>

+      <OutputFile>.\x64\Release\gennorm2.exe</OutputFile>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <ProgramDatabaseFile>.\x64\Release\gennorm2.pdb</ProgramDatabaseFile>

+      <GenerateDebugInformation>true</GenerateDebugInformation>

+      <SubSystem>Console</SubSystem>

+      <OptimizeReferences>true</OptimizeReferences>

+      <EnableCOMDATFolding>true</EnableCOMDATFolding>

+      <TargetMachine>MachineX64</TargetMachine>

+      <RandomizedBaseAddress>false</RandomizedBaseAddress>

+      <DataExecutionPrevention>false</DataExecutionPrevention>

+    </Link>

+  </ItemDefinitionGroup>

+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">

+    <CustomBuildStep>

+      <Command>copy "$(TargetPath)" ..\..\..\bin64
+</Command>

+      <Outputs>..\..\..\bin64\$(TargetFileName);%(Outputs)</Outputs>

+    </CustomBuildStep>

+    <ClCompile>

+      <Optimization>Disabled</Optimization>

+      <AdditionalIncludeDirectories>..\..\common;..\toolutil;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>

+      <PreprocessorDefinitions>WIN64;WIN32;_DEBUG;_CRT_SECURE_NO_DEPRECATE;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <IntrinsicFunctions>true</IntrinsicFunctions>

+      <MinimalRebuild>true</MinimalRebuild>

+      <BasicRuntimeChecks>EnableFastChecks</BasicRuntimeChecks>

+      <StringPooling>true</StringPooling>

+      <RuntimeLibrary>MultiThreadedDebugDLL</RuntimeLibrary>

+      <BufferSecurityCheck>true</BufferSecurityCheck>

+      <DisableLanguageExtensions>false</DisableLanguageExtensions>

+      <TreatWChar_tAsBuiltInType>true</TreatWChar_tAsBuiltInType>

+      <PrecompiledHeaderOutputFile>.\x64\Debug\gennorm2.pch</PrecompiledHeaderOutputFile>

+      <AssemblerListingLocation>.\x64\Debug\</AssemblerListingLocation>

+      <ObjectFileName>.\x64\Debug\</ObjectFileName>

+      <ProgramDataBaseFileName>.\x64\Debug\</ProgramDataBaseFileName>

+      <BrowseInformation>true</BrowseInformation>

+      <WarningLevel>Level3</WarningLevel>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <DebugInformationFormat>EditAndContinue</DebugInformationFormat>

+      <CompileAs>Default</CompileAs>

+    </ClCompile>

+    <ResourceCompile>

+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>

+      <Culture>0x0409</Culture>

+    </ResourceCompile>

+    <Link>

+      <OutputFile>.\x64\Debug\gennorm2.exe</OutputFile>

+      <SuppressStartupBanner>true</SuppressStartupBanner>

+      <GenerateDebugInformation>true</GenerateDebugInformation>

+      <ProgramDatabaseFile>.\x64\Debug\gennorm2.pdb</ProgramDatabaseFile>

+      <SubSystem>Console</SubSystem>

+      <TargetMachine>MachineX64</TargetMachine>

+      <RandomizedBaseAddress>false</RandomizedBaseAddress>

+      <DataExecutionPrevention>false</DataExecutionPrevention>

+    </Link>

+  </ItemDefinitionGroup>

+  <ItemGroup>

+    <ClCompile Include="gennorm2.cpp" />

+    <ClCompile Include="n2builder.cpp" />

+  </ItemGroup>

+  <ItemGroup>

+    <ClInclude Include="n2builder.h" />

+  </ItemGroup>

+  <ItemGroup>

+    <ProjectReference Include="..\..\common\common.vcxproj">

+      <Project>{73c0a65b-d1f2-4de1-b3a6-15dad2c23f3d}</Project>

+      <ReferenceOutputAssembly>false</ReferenceOutputAssembly>

+    </ProjectReference>

+    <ProjectReference Include="..\toolutil\toolutil.vcxproj">

+      <Project>{6b231032-3cb5-4eed-9210-810d666a23a0}</Project>

+      <ReferenceOutputAssembly>false</ReferenceOutputAssembly>

+    </ProjectReference>

+  </ItemGroup>

+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />

+  <ImportGroup Label="ExtensionTargets">

+  </ImportGroup>

+</Project>
\ No newline at end of file
diff --git a/intl/icu/source/tools/gennorm2/n2builder.cpp b/intl/icu/source/tools/gennorm2/n2builder.cpp
new file mode 100644
index 000000000..54ad860f3
--- /dev/null
+++ b/intl/icu/source/tools/gennorm2/n2builder.cpp
@@ -0,0 +1,1307 @@
+// Copyright (C) 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2009-2016, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  n2builder.cpp
+*   encoding:   US-ASCII
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2009nov25
+*   created by: Markus W. Scherer
+*
+* Builds Normalizer2 data and writes a binary .nrm file.
+* For the file format see source/common/normalizer2impl.h.
+*/
+
+#include "unicode/utypes.h"
+#include "n2builder.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#if U_HAVE_STD_STRING
+#include <vector>
+#endif
+#include "unicode/errorcode.h"
+#include "unicode/localpointer.h"
+#include "unicode/putil.h"
+#include "unicode/udata.h"
+#include "unicode/uniset.h"
+#include "unicode/unistr.h"
+#include "unicode/ustring.h"
+#include "charstr.h"
+#include "hash.h"
+#include "normalizer2impl.h"
+#include "toolutil.h"
+#include "unewdata.h"
+#include "utrie2.h"
+#include "uvectr32.h"
+#include "writesrc.h"
+
+#if !UCONFIG_NO_NORMALIZATION
+
+/* UDataInfo cf. udata.h */
+static UDataInfo dataInfo={
+    sizeof(UDataInfo),
+    0,
+
+    U_IS_BIG_ENDIAN,
+    U_CHARSET_FAMILY,
+    U_SIZEOF_UCHAR,
+    0,
+
+    { 0x4e, 0x72, 0x6d, 0x32 }, /* dataFormat="Nrm2" */
+    { 2, 0, 0, 0 },             /* formatVersion */
+    { 5, 2, 0, 0 }              /* dataVersion (Unicode version) */
+};
+
+U_NAMESPACE_BEGIN
+
+class HangulIterator {
+public:
+    struct Range {
+        UChar32 start, limit;
+        uint16_t norm16;
+    };
+
+    HangulIterator() : rangeIndex(0) {}
+    const Range *nextRange() {
+        if(rangeIndex<UPRV_LENGTHOF(ranges)) {
+            return ranges+rangeIndex++;
+        } else {
+            return NULL;
+        }
+    }
+    void reset() { rangeIndex=0; }
+private:
+    static const Range ranges[4];
+    int32_t rangeIndex;
+};
+
+const HangulIterator::Range HangulIterator::ranges[4]={
+    { Hangul::JAMO_L_BASE, Hangul::JAMO_L_BASE+Hangul::JAMO_L_COUNT, 1 },
+    { Hangul::JAMO_V_BASE, Hangul::JAMO_V_BASE+Hangul::JAMO_V_COUNT, Normalizer2Impl::JAMO_VT },
+    // JAMO_T_BASE+1: not U+11A7
+    { Hangul::JAMO_T_BASE+1, Hangul::JAMO_T_BASE+Hangul::JAMO_T_COUNT, Normalizer2Impl::JAMO_VT },
+    { Hangul::HANGUL_BASE, Hangul::HANGUL_BASE+Hangul::HANGUL_COUNT, 0 },  // will become minYesNo
+};
+
+struct CompositionPair {
+    CompositionPair(UChar32 t, UChar32 c) : trail(t), composite(c) {}
+    UChar32 trail, composite;
+};
+
+struct Norm {
+    enum MappingType { NONE, REMOVED, ROUND_TRIP, ONE_WAY };
+
+    UBool hasMapping() const { return mappingType>REMOVED; }
+
+    // Requires hasMapping() and well-formed mapping.
+    void setMappingCP() {
+        UChar32 c;
+        if(!mapping->isEmpty() && mapping->length()==U16_LENGTH(c=mapping->char32At(0))) {
+            mappingCP=c;
+        } else {
+            mappingCP=U_SENTINEL;
+        }
+    }
+
+    const CompositionPair *getCompositionPairs(int32_t &length) const {
+        if(compositions==NULL) {
+            length=0;
+            return NULL;
+        } else {
+            length=compositions->size()/2;
+            return reinterpret_cast<const CompositionPair *>(compositions->getBuffer());
+        }
+    }
+
+    UnicodeString *mapping;
+    UnicodeString *rawMapping;  // non-NULL if the mapping is further decomposed
+    UChar32 mappingCP;  // >=0 if mapping to 1 code point
+    int32_t mappingPhase;
+    MappingType mappingType;
+
+    UVector32 *compositions;  // (trail, composite) pairs
+    uint8_t cc;
+    UBool combinesBack;
+    UBool hasNoCompBoundaryAfter;
+
+    enum OffsetType {
+        OFFSET_NONE,
+        // Composition for back-combining character. Allowed, but not normally used.
+        OFFSET_MAYBE_YES,
+        // Composition for a starter that does not have a decomposition mapping.
+        OFFSET_YES_YES,
+        // Round-trip mapping & composition for a starter.
+        OFFSET_YES_NO_MAPPING_AND_COMPOSITION,
+        // Round-trip mapping for a starter that itself does not combine-forward.
+        OFFSET_YES_NO_MAPPING_ONLY,
+        // One-way mapping.
+        OFFSET_NO_NO,
+        // Delta for an algorithmic one-way mapping.
+        OFFSET_DELTA
+    };
+    enum { OFFSET_SHIFT=4, OFFSET_MASK=(1<<OFFSET_SHIFT)-1 };
+    int32_t offset;
+};
+
+class Normalizer2DBEnumerator {
+public:
+    Normalizer2DBEnumerator(Normalizer2DataBuilder &b) : builder(b) {}
+    virtual ~Normalizer2DBEnumerator() {}
+    virtual UBool rangeHandler(UChar32 start, UChar32 end, uint32_t value) = 0;
+    Normalizer2DBEnumerator *ptr() { return this; }
+protected:
+    Normalizer2DataBuilder &builder;
+};
+
+U_CDECL_BEGIN
+
+static UBool U_CALLCONV
+enumRangeHandler(const void *context, UChar32 start, UChar32 end, uint32_t value) {
+    return ((Normalizer2DBEnumerator *)context)->rangeHandler(start, end, value);
+}
+
+U_CDECL_END
+
+Normalizer2DataBuilder::Normalizer2DataBuilder(UErrorCode &errorCode) :
+        phase(0), overrideHandling(OVERRIDE_PREVIOUS), optimization(OPTIMIZE_NORMAL),
+        norm16TrieLength(0) {
+    memset(unicodeVersion, 0, sizeof(unicodeVersion));
+    normTrie=utrie2_open(0, 0, &errorCode);
+    normMem=utm_open("gennorm2 normalization structs", 10000, 0x110100, sizeof(Norm));
+    norms=allocNorm();  // unused Norm struct at index 0
+    memset(indexes, 0, sizeof(indexes));
+    memset(smallFCD, 0, sizeof(smallFCD));
+}
+
+Normalizer2DataBuilder::~Normalizer2DataBuilder() {
+    utrie2_close(normTrie);
+    int32_t normsLength=utm_countItems(normMem);
+    for(int32_t i=1; i<normsLength; ++i) {
+        delete norms[i].mapping;
+        delete norms[i].rawMapping;
+        delete norms[i].compositions;
+    }
+    utm_close(normMem);
+    utrie2_close(norm16Trie);
+}
+
+void
+Normalizer2DataBuilder::setUnicodeVersion(const char *v) {
+    UVersionInfo nullVersion={ 0, 0, 0, 0 };
+    UVersionInfo version;
+    u_versionFromString(version, v);
+    if( 0!=memcmp(version, unicodeVersion, U_MAX_VERSION_LENGTH) &&
+        0!=memcmp(nullVersion, unicodeVersion, U_MAX_VERSION_LENGTH)
+    ) {
+        char buffer[U_MAX_VERSION_STRING_LENGTH];
+        u_versionToString(unicodeVersion, buffer);
+        fprintf(stderr, "gennorm2 error: multiple inconsistent Unicode version numbers %s vs. %s\n",
+                buffer, v);
+        exit(U_ILLEGAL_ARGUMENT_ERROR);
+    }
+    memcpy(unicodeVersion, version, U_MAX_VERSION_LENGTH);
+}
+
+Norm *Normalizer2DataBuilder::allocNorm() {
+    Norm *p=(Norm *)utm_alloc(normMem);
+    norms=(Norm *)utm_getStart(normMem);  // in case it got reallocated
+    return p;
+}
+
+/* get an existing Norm unit */
+Norm *Normalizer2DataBuilder::getNorm(UChar32 c) {
+    uint32_t i=utrie2_get32(normTrie, c);
+    if(i==0) {
+        return NULL;
+    }
+    return norms+i;
+}
+
+const Norm &Normalizer2DataBuilder::getNormRef(UChar32 c) const {
+    return norms[utrie2_get32(normTrie, c)];
+}
+
+/*
+ * get or create a Norm unit;
+ * get or create the intermediate trie entries for it as well
+ */
+Norm *Normalizer2DataBuilder::createNorm(UChar32 c) {
+    uint32_t i=utrie2_get32(normTrie, c);
+    if(i!=0) {
+        return norms+i;
+    } else {
+        /* allocate Norm */
+        Norm *p=allocNorm();
+        IcuToolErrorCode errorCode("gennorm2/createNorm()");
+        utrie2_set32(normTrie, c, (uint32_t)(p-norms), errorCode);
+        return p;
+    }
+}
+
+Norm *Normalizer2DataBuilder::checkNormForMapping(Norm *p, UChar32 c) {
+    if(p!=NULL) {
+        if(p->mappingType!=Norm::NONE) {
+            if( overrideHandling==OVERRIDE_NONE ||
+                (overrideHandling==OVERRIDE_PREVIOUS && p->mappingPhase==phase)
+            ) {
+                fprintf(stderr,
+                        "error in gennorm2 phase %d: "
+                        "not permitted to override mapping for U+%04lX from phase %d\n",
+                        (int)phase, (long)c, (int)p->mappingPhase);
+                exit(U_INVALID_FORMAT_ERROR);
+            }
+            delete p->mapping;
+            p->mapping=NULL;
+        }
+        p->mappingPhase=phase;
+    }
+    return p;
+}
+
+void Normalizer2DataBuilder::setOverrideHandling(OverrideHandling oh) {
+    overrideHandling=oh;
+    ++phase;
+}
+
+void Normalizer2DataBuilder::setCC(UChar32 c, uint8_t cc) {
+    createNorm(c)->cc=cc;
+}
+
+uint8_t Normalizer2DataBuilder::getCC(UChar32 c) const {
+    return getNormRef(c).cc;
+}
+
+static UBool isWellFormed(const UnicodeString &s) {
+    UErrorCode errorCode=U_ZERO_ERROR;
+    u_strToUTF8(NULL, 0, NULL, s.getBuffer(), s.length(), &errorCode);
+    return U_SUCCESS(errorCode) || errorCode==U_BUFFER_OVERFLOW_ERROR;
+}
+
+void Normalizer2DataBuilder::setOneWayMapping(UChar32 c, const UnicodeString &m) {
+    if(!isWellFormed(m)) {
+        fprintf(stderr,
+                "error in gennorm2 phase %d: "
+                "illegal one-way mapping from U+%04lX to malformed string\n",
+                (int)phase, (long)c);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    Norm *p=checkNormForMapping(createNorm(c), c);
+    p->mapping=new UnicodeString(m);
+    p->mappingType=Norm::ONE_WAY;
+    p->setMappingCP();
+}
+
+void Normalizer2DataBuilder::setRoundTripMapping(UChar32 c, const UnicodeString &m) {
+    if(U_IS_SURROGATE(c)) {
+        fprintf(stderr,
+                "error in gennorm2 phase %d: "
+                "illegal round-trip mapping from surrogate code point U+%04lX\n",
+                (int)phase, (long)c);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    if(!isWellFormed(m)) {
+        fprintf(stderr,
+                "error in gennorm2 phase %d: "
+                "illegal round-trip mapping from U+%04lX to malformed string\n",
+                (int)phase, (long)c);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    int32_t numCP=u_countChar32(m.getBuffer(), m.length());
+    if(numCP!=2) {
+        fprintf(stderr,
+                "error in gennorm2 phase %d: "
+                "illegal round-trip mapping from U+%04lX to %d!=2 code points\n",
+                (int)phase, (long)c, (int)numCP);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    Norm *p=checkNormForMapping(createNorm(c), c);
+    p->mapping=new UnicodeString(m);
+    p->mappingType=Norm::ROUND_TRIP;
+    p->mappingCP=U_SENTINEL;
+}
+
+void Normalizer2DataBuilder::removeMapping(UChar32 c) {
+    Norm *p=checkNormForMapping(getNorm(c), c);
+    if(p!=NULL) {
+        p->mappingType=Norm::REMOVED;
+    }
+}
+
+class CompositionBuilder : public Normalizer2DBEnumerator {
+public:
+    CompositionBuilder(Normalizer2DataBuilder &b) : Normalizer2DBEnumerator(b) {}
+    virtual UBool rangeHandler(UChar32 start, UChar32 end, uint32_t value) {
+        builder.addComposition(start, end, value);
+        return TRUE;
+    }
+};
+
+void
+Normalizer2DataBuilder::addComposition(UChar32 start, UChar32 end, uint32_t value) {
+    if(norms[value].mappingType==Norm::ROUND_TRIP) {
+        if(start!=end) {
+            fprintf(stderr,
+                    "gennorm2 error: same round-trip mapping for "
+                    "more than 1 code point U+%04lX..U+%04lX\n",
+                    (long)start, (long)end);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        if(norms[value].cc!=0) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "U+%04lX has a round-trip mapping and ccc!=0, "
+                    "not possible in Unicode normalization\n",
+                    (long)start);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        // setRoundTripMapping() ensured that there are exactly two code points.
+        const UnicodeString &m=*norms[value].mapping;
+        UChar32 lead=m.char32At(0);
+        UChar32 trail=m.char32At(m.length()-1);
+        if(getCC(lead)!=0) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "U+%04lX's round-trip mapping's starter U+%04lX has ccc!=0, "
+                    "not possible in Unicode normalization\n",
+                    (long)start, (long)lead);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        // Flag for trailing character.
+        createNorm(trail)->combinesBack=TRUE;
+        // Insert (trail, composite) pair into compositions list for the lead character.
+        IcuToolErrorCode errorCode("gennorm2/addComposition()");
+        Norm *leadNorm=createNorm(lead);
+        UVector32 *compositions=leadNorm->compositions;
+        int32_t i;
+        if(compositions==NULL) {
+            compositions=leadNorm->compositions=new UVector32(errorCode);
+            i=0;  // "insert" the first pair at index 0
+        } else {
+            // Insertion sort, and check for duplicate trail characters.
+            int32_t length;
+            const CompositionPair *pairs=leadNorm->getCompositionPairs(length);
+            for(i=0; i<length; ++i) {
+                if(trail==pairs[i].trail) {
+                    fprintf(stderr,
+                            "gennorm2 error: same round-trip mapping for "
+                            "more than 1 code point (e.g., U+%04lX) to U+%04lX + U+%04lX\n",
+                            (long)start, (long)lead, (long)trail);
+                    exit(U_INVALID_FORMAT_ERROR);
+                }
+                if(trail<pairs[i].trail) {
+                    break;
+                }
+            }
+        }
+        compositions->insertElementAt(trail, 2*i, errorCode);
+        compositions->insertElementAt(start, 2*i+1, errorCode);
+    }
+}
+
+UBool Normalizer2DataBuilder::combinesWithCCBetween(const Norm &norm,
+                                                    uint8_t lowCC, uint8_t highCC) const {
+    if((highCC-lowCC)>=2) {
+        int32_t length;
+        const CompositionPair *pairs=norm.getCompositionPairs(length);
+        for(int32_t i=0; i<length; ++i) {
+            uint8_t trailCC=getCC(pairs[i].trail);
+            if(lowCC<trailCC && trailCC<highCC) {
+                return TRUE;
+            }
+        }
+    }
+    return FALSE;
+}
+
+UChar32 Normalizer2DataBuilder::combine(const Norm &norm, UChar32 trail) const {
+    int32_t length;
+    const CompositionPair *pairs=norm.getCompositionPairs(length);
+    for(int32_t i=0; i<length; ++i) {
+        if(trail==pairs[i].trail) {
+            return pairs[i].composite;
+        }
+        if(trail<pairs[i].trail) {
+            break;
+        }
+    }
+    return U_SENTINEL;
+}
+
+class Decomposer : public Normalizer2DBEnumerator {
+public:
+    Decomposer(Normalizer2DataBuilder &b) : Normalizer2DBEnumerator(b), didDecompose(FALSE) {}
+    virtual UBool rangeHandler(UChar32 start, UChar32 end, uint32_t value) {
+        didDecompose|=builder.decompose(start, end, value);
+        return TRUE;
+    }
+    UBool didDecompose;
+};
+
+UBool
+Normalizer2DataBuilder::decompose(UChar32 start, UChar32 end, uint32_t value) {
+    if(norms[value].hasMapping()) {
+        Norm &norm=norms[value];
+        const UnicodeString &m=*norm.mapping;
+        UnicodeString *decomposed=NULL;
+        const UChar *s=m.getBuffer();
+        int32_t length=m.length();
+        int32_t prev, i=0;
+        UChar32 c;
+        while(i<length) {
+            prev=i;
+            U16_NEXT(s, i, length, c);
+            if(start<=c && c<=end) {
+                fprintf(stderr,
+                        "gennorm2 error: U+%04lX maps to itself directly or indirectly\n",
+                        (long)c);
+                exit(U_INVALID_FORMAT_ERROR);
+            }
+            const Norm &cNorm=getNormRef(c);
+            if(cNorm.hasMapping()) {
+                if(norm.mappingType==Norm::ROUND_TRIP) {
+                    if(prev==0) {
+                        if(cNorm.mappingType!=Norm::ROUND_TRIP) {
+                            fprintf(stderr,
+                                    "gennorm2 error: "
+                                    "U+%04lX's round-trip mapping's starter "
+                                    "U+%04lX one-way-decomposes, "
+                                    "not possible in Unicode normalization\n",
+                                    (long)start, (long)c);
+                            exit(U_INVALID_FORMAT_ERROR);
+                        }
+                        uint8_t myTrailCC=getCC(m.char32At(i));
+                        UChar32 cTrailChar=cNorm.mapping->char32At(cNorm.mapping->length()-1);
+                        uint8_t cTrailCC=getCC(cTrailChar);
+                        if(cTrailCC>myTrailCC) {
+                            fprintf(stderr,
+                                    "gennorm2 error: "
+                                    "U+%04lX's round-trip mapping's starter "
+                                    "U+%04lX decomposes and the "
+                                    "inner/earlier tccc=%hu > outer/following tccc=%hu, "
+                                    "not possible in Unicode normalization\n",
+                                    (long)start, (long)c,
+                                    (short)cTrailCC, (short)myTrailCC);
+                            exit(U_INVALID_FORMAT_ERROR);
+                        }
+                    } else {
+                        fprintf(stderr,
+                                "gennorm2 error: "
+                                "U+%04lX's round-trip mapping's non-starter "
+                                "U+%04lX decomposes, "
+                                "not possible in Unicode normalization\n",
+                                (long)start, (long)c);
+                        exit(U_INVALID_FORMAT_ERROR);
+                    }
+                }
+                if(decomposed==NULL) {
+                    decomposed=new UnicodeString(m, 0, prev);
+                }
+                decomposed->append(*cNorm.mapping);
+            } else if(Hangul::isHangul(c)) {
+                UChar buffer[3];
+                int32_t hangulLength=Hangul::decompose(c, buffer);
+                if(norm.mappingType==Norm::ROUND_TRIP && prev!=0) {
+                    fprintf(stderr,
+                            "gennorm2 error: "
+                            "U+%04lX's round-trip mapping's non-starter "
+                            "U+%04lX decomposes, "
+                            "not possible in Unicode normalization\n",
+                            (long)start, (long)c);
+                    exit(U_INVALID_FORMAT_ERROR);
+                }
+                if(decomposed==NULL) {
+                    decomposed=new UnicodeString(m, 0, prev);
+                }
+                decomposed->append(buffer, hangulLength);
+            } else if(decomposed!=NULL) {
+                decomposed->append(m, prev, i-prev);
+            }
+        }
+        if(decomposed!=NULL) {
+            if(norm.rawMapping==NULL) {
+                // Remember the original mapping when decomposing recursively.
+                norm.rawMapping=norm.mapping;
+            } else {
+                delete norm.mapping;
+            }
+            norm.mapping=decomposed;
+            // Not  norm.setMappingCP();  because the original mapping
+            // is most likely to be encodable as a delta.
+            return TRUE;
+        }
+    }
+    return FALSE;
+}
+
+class BuilderReorderingBuffer {
+public:
+    BuilderReorderingBuffer() : fLength(0), fLastStarterIndex(-1), fDidReorder(FALSE) {}
+    void reset() {
+        fLength=0;
+        fLastStarterIndex=-1;
+        fDidReorder=FALSE;
+    }
+    int32_t length() const { return fLength; }
+    UBool isEmpty() const { return fLength==0; }
+    int32_t lastStarterIndex() const { return fLastStarterIndex; }
+    UChar32 charAt(int32_t i) const { return fArray[i]>>8; }
+    uint8_t ccAt(int32_t i) const { return (uint8_t)fArray[i]; }
+    UBool didReorder() const { return fDidReorder; }
+    void append(UChar32 c, uint8_t cc) {
+        if(cc==0 || fLength==0 || ccAt(fLength-1)<=cc) {
+            if(cc==0) {
+                fLastStarterIndex=fLength;
+            }
+            fArray[fLength++]=(c<<8)|cc;
+            return;
+        }
+        // Let this character bubble back to its canonical order.
+        int32_t i=fLength-1;
+        while(i>fLastStarterIndex && ccAt(i)>cc) {
+            --i;
+        }
+        ++i;  // after the last starter or prevCC<=cc
+        // Move this and the following characters forward one to make space.
+        for(int32_t j=fLength; i<j; --j) {
+            fArray[j]=fArray[j-1];
+        }
+        fArray[i]=(c<<8)|cc;
+        ++fLength;
+        fDidReorder=TRUE;
+    }
+    void toString(UnicodeString &dest) {
+        dest.remove();
+        for(int32_t i=0; i<fLength; ++i) {
+            dest.append(charAt(i));
+        }
+    }
+    void setComposite(UChar32 composite, int32_t combMarkIndex) {
+        fArray[fLastStarterIndex]=composite<<8;
+        // Remove the combining mark that contributed to the composite.
+        --fLength;
+        while(combMarkIndex<fLength) {
+            fArray[combMarkIndex]=fArray[combMarkIndex+1];
+            ++combMarkIndex;
+        }
+    }
+private:
+    int32_t fArray[Normalizer2Impl::MAPPING_LENGTH_MASK];
+    int32_t fLength;
+    int32_t fLastStarterIndex;
+    UBool fDidReorder;
+};
+
+void
+Normalizer2DataBuilder::reorder(Norm *p, BuilderReorderingBuffer &buffer) {
+    UnicodeString &m=*p->mapping;
+    int32_t length=m.length();
+    if(length>Normalizer2Impl::MAPPING_LENGTH_MASK) {
+        return;  // writeMapping() will complain about it and print the code point.
+    }
+    const UChar *s=m.getBuffer();
+    int32_t i=0;
+    UChar32 c;
+    while(i<length) {
+        U16_NEXT(s, i, length, c);
+        buffer.append(c, getCC(c));
+    }
+    if(buffer.didReorder()) {
+        buffer.toString(m);
+    }
+}
+
+/*
+ * Computes the flag for the last code branch in Normalizer2Impl::hasCompBoundaryAfter().
+ * A starter character with a mapping does not have a composition boundary after it
+ * if the character itself combines-forward (which is tested by the caller of this function),
+ * or it is deleted (mapped to the empty string),
+ * or its mapping contains no starter,
+ * or the last starter combines-forward.
+ */
+UBool Normalizer2DataBuilder::hasNoCompBoundaryAfter(BuilderReorderingBuffer &buffer) {
+    if(buffer.isEmpty()) {
+        return TRUE;  // maps-to-empty-string is no boundary of any kind
+    }
+    int32_t lastStarterIndex=buffer.lastStarterIndex();
+    if(lastStarterIndex<0) {
+        return TRUE;  // no starter
+    }
+    UChar32 starter=buffer.charAt(lastStarterIndex);
+    if( Hangul::isJamoL(starter) ||
+        (Hangul::isJamoV(starter) &&
+         0<lastStarterIndex && Hangul::isJamoL(buffer.charAt(lastStarterIndex-1)))
+    ) {
+        // A Jamo leading consonant or an LV pair combines-forward if it is at the end,
+        // otherwise it is blocked.
+        return lastStarterIndex==buffer.length()-1;
+    }
+    // Note: There can be no Hangul syllable in the fully decomposed mapping.
+    const Norm *starterNorm=&getNormRef(starter);
+    if(starterNorm->compositions==NULL) {
+        return FALSE;  // the last starter does not combine forward
+    }
+    // Compose as far as possible, and see if further compositions are possible.
+    uint8_t prevCC=0;
+    for(int32_t combMarkIndex=lastStarterIndex+1; combMarkIndex<buffer.length();) {
+        uint8_t cc=buffer.ccAt(combMarkIndex);  // !=0 because after last starter
+        if(combinesWithCCBetween(*starterNorm, prevCC, cc)) {
+            return TRUE;
+        }
+        if( prevCC<cc &&
+            (starter=combine(*starterNorm, buffer.charAt(combMarkIndex)))>=0
+        ) {
+            buffer.setComposite(starter, combMarkIndex);
+            starterNorm=&getNormRef(starter);
+            if(starterNorm->compositions==NULL) {
+                return FALSE;  // the composite does not combine further
+            }
+        } else {
+            prevCC=cc;
+            ++combMarkIndex;
+        }
+    }
+    // TRUE if the final, forward-combining starter is at the end.
+    return prevCC==0;
+}
+
+// Requires p->hasMapping().
+// Returns the offset of the "first unit" from the beginning of the extraData for c.
+// That is the same as the length of the optional data for the raw mapping and the ccc/lccc word.
+int32_t Normalizer2DataBuilder::writeMapping(UChar32 c, const Norm *p, UnicodeString &dataString) {
+    UnicodeString &m=*p->mapping;
+    int32_t length=m.length();
+    if(length>Normalizer2Impl::MAPPING_LENGTH_MASK) {
+        fprintf(stderr,
+                "gennorm2 error: "
+                "mapping for U+%04lX longer than maximum of %d\n",
+                (long)c, Normalizer2Impl::MAPPING_LENGTH_MASK);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    int32_t leadCC, trailCC;
+    if(length==0) {
+        leadCC=trailCC=0;
+    } else {
+        leadCC=getCC(m.char32At(0));
+        trailCC=getCC(m.char32At(length-1));
+    }
+    if(c<Normalizer2Impl::MIN_CCC_LCCC_CP && (p->cc!=0 || leadCC!=0)) {
+        fprintf(stderr,
+                "gennorm2 error: "
+                "U+%04lX below U+0300 has ccc!=0 or lccc!=0, not supported by ICU\n",
+                (long)c);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    // Write small-FCD data.
+    if((leadCC|trailCC)!=0) {
+        UChar32 lead= c<=0xffff ? c : U16_LEAD(c);
+        smallFCD[lead>>8]|=(uint8_t)1<<((lead>>5)&7);
+    }
+    // Write the mapping & raw mapping extraData.
+    int32_t firstUnit=length|(trailCC<<8);
+    int32_t preMappingLength=0;
+    if(p->rawMapping!=NULL) {
+        UnicodeString &rm=*p->rawMapping;
+        int32_t rmLength=rm.length();
+        if(rmLength>Normalizer2Impl::MAPPING_LENGTH_MASK) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "raw mapping for U+%04lX longer than maximum of %d\n",
+                    (long)c, Normalizer2Impl::MAPPING_LENGTH_MASK);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        UChar rm0=rm.charAt(0);
+        if( rmLength==length-1 &&
+            // 99: overlong substring lengths get pinned to remainder lengths anyway
+            0==rm.compare(1, 99, m, 2, 99) &&
+            rm0>Normalizer2Impl::MAPPING_LENGTH_MASK
+        ) {
+            // Compression:
+            // rawMapping=rm0+mapping.substring(2) -> store only rm0
+            //
+            // The raw mapping is the same as the final mapping after replacing
+            // the final mapping's first two code units with the raw mapping's first one.
+            // In this case, we store only that first unit, rm0.
+            // This helps with a few hundred mappings.
+            dataString.append(rm0);
+            preMappingLength=1;
+        } else {
+            // Store the raw mapping with its length.
+            dataString.append(rm);
+            dataString.append((UChar)rmLength);
+            preMappingLength=rmLength+1;
+        }
+        firstUnit|=Normalizer2Impl::MAPPING_HAS_RAW_MAPPING;
+    }
+    int32_t cccLccc=p->cc|(leadCC<<8);
+    if(cccLccc!=0) {
+        dataString.append((UChar)cccLccc);
+        ++preMappingLength;
+        firstUnit|=Normalizer2Impl::MAPPING_HAS_CCC_LCCC_WORD;
+    }
+    if(p->hasNoCompBoundaryAfter) {
+        firstUnit|=Normalizer2Impl::MAPPING_NO_COMP_BOUNDARY_AFTER;
+    }
+    dataString.append((UChar)firstUnit);
+    dataString.append(m);
+    return preMappingLength;
+}
+
+// Requires p->compositions!=NULL.
+void Normalizer2DataBuilder::writeCompositions(UChar32 c, const Norm *p, UnicodeString &dataString) {
+    if(p->cc!=0) {
+        fprintf(stderr,
+                "gennorm2 error: "
+                "U+%04lX combines-forward and has ccc!=0, not possible in Unicode normalization\n",
+                (long)c);
+        exit(U_INVALID_FORMAT_ERROR);
+    }
+    int32_t length;
+    const CompositionPair *pairs=p->getCompositionPairs(length);
+    for(int32_t i=0; i<length; ++i) {
+        const CompositionPair &pair=pairs[i];
+        // 22 bits for the composite character and whether it combines forward.
+        UChar32 compositeAndFwd=pair.composite<<1;
+        if(getNormRef(pair.composite).compositions!=NULL) {
+            compositeAndFwd|=1;  // The composite character also combines-forward.
+        }
+        // Encode most pairs in two units and some in three.
+        int32_t firstUnit, secondUnit, thirdUnit;
+        if(pair.trail<Normalizer2Impl::COMP_1_TRAIL_LIMIT) {
+            if(compositeAndFwd<=0xffff) {
+                firstUnit=pair.trail<<1;
+                secondUnit=compositeAndFwd;
+                thirdUnit=-1;
+            } else {
+                firstUnit=(pair.trail<<1)|Normalizer2Impl::COMP_1_TRIPLE;
+                secondUnit=compositeAndFwd>>16;
+                thirdUnit=compositeAndFwd;
+            }
+        } else {
+            firstUnit=(Normalizer2Impl::COMP_1_TRAIL_LIMIT+
+                       (pair.trail>>Normalizer2Impl::COMP_1_TRAIL_SHIFT))|
+                      Normalizer2Impl::COMP_1_TRIPLE;
+            secondUnit=(pair.trail<<Normalizer2Impl::COMP_2_TRAIL_SHIFT)|
+                       (compositeAndFwd>>16);
+            thirdUnit=compositeAndFwd;
+        }
+        // Set the high bit of the first unit if this is the last composition pair.
+        if(i==(length-1)) {
+            firstUnit|=Normalizer2Impl::COMP_1_LAST_TUPLE;
+        }
+        dataString.append((UChar)firstUnit).append((UChar)secondUnit);
+        if(thirdUnit>=0) {
+            dataString.append((UChar)thirdUnit);
+        }
+    }
+}
+
+class ExtraDataWriter : public Normalizer2DBEnumerator {
+public:
+    ExtraDataWriter(Normalizer2DataBuilder &b) :
+        Normalizer2DBEnumerator(b),
+        yesYesCompositions(1000, (UChar32)0xffff, 2),  // 0=inert, 1=Jamo L, 2=start of compositions
+        yesNoMappingsAndCompositions(1000, (UChar32)0, 1) {}  // 0=Hangul, 1=start of normal data
+    virtual UBool rangeHandler(UChar32 start, UChar32 end, uint32_t value) {
+        if(value!=0) {
+            if(start!=end) {
+                fprintf(stderr,
+                        "gennorm2 error: unexpected shared data for "
+                        "multiple code points U+%04lX..U+%04lX\n",
+                        (long)start, (long)end);
+                exit(U_INTERNAL_PROGRAM_ERROR);
+            }
+            builder.writeExtraData(start, value, *this);
+        }
+        return TRUE;
+    }
+    UnicodeString maybeYesCompositions;
+    UnicodeString yesYesCompositions;
+    UnicodeString yesNoMappingsAndCompositions;
+    UnicodeString yesNoMappingsOnly;
+    UnicodeString noNoMappings;
+    Hashtable previousNoNoMappings;  // If constructed in runtime code, pass in UErrorCode.
+};
+
+void Normalizer2DataBuilder::writeExtraData(UChar32 c, uint32_t value, ExtraDataWriter &writer) {
+    Norm *p=norms+value;
+    if(!p->hasMapping()) {
+        // Write small-FCD data.
+        // There is similar code in writeMapping() for characters that do have a mapping.
+        if(c<Normalizer2Impl::MIN_CCC_LCCC_CP && p->cc!=0) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "U+%04lX below U+0300 has ccc!=0, not supported by ICU\n",
+                    (long)c);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        if(p->cc!=0) {
+            UChar32 lead= c<=0xffff ? c : U16_LEAD(c);
+            smallFCD[lead>>8]|=(uint8_t)1<<((lead>>5)&7);
+        }
+    }
+    if(p->combinesBack) {
+        if(p->hasMapping()) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "U+%04lX combines-back and decomposes, not possible in Unicode normalization\n",
+                    (long)c);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        if(p->compositions!=NULL) {
+            p->offset=
+                (writer.maybeYesCompositions.length()<<Norm::OFFSET_SHIFT)|
+                Norm::OFFSET_MAYBE_YES;
+            writeCompositions(c, p, writer.maybeYesCompositions);
+        }
+    } else if(!p->hasMapping()) {
+        if(p->compositions!=NULL) {
+            p->offset=
+                (writer.yesYesCompositions.length()<<Norm::OFFSET_SHIFT)|
+                Norm::OFFSET_YES_YES;
+            writeCompositions(c, p, writer.yesYesCompositions);
+        }
+    } else if(p->mappingType==Norm::ROUND_TRIP) {
+        if(p->compositions!=NULL) {
+            int32_t offset=writer.yesNoMappingsAndCompositions.length()+
+                           writeMapping(c, p, writer.yesNoMappingsAndCompositions);
+            p->offset=(offset<<Norm::OFFSET_SHIFT)|Norm::OFFSET_YES_NO_MAPPING_AND_COMPOSITION;
+            writeCompositions(c, p, writer.yesNoMappingsAndCompositions);
+        } else {
+            int32_t offset=writer.yesNoMappingsOnly.length()+
+                           writeMapping(c, p, writer.yesNoMappingsOnly);
+            p->offset=(offset<<Norm::OFFSET_SHIFT)|Norm::OFFSET_YES_NO_MAPPING_ONLY;
+        }
+    } else /* one-way */ {
+        if(p->compositions!=NULL) {
+            fprintf(stderr,
+                    "gennorm2 error: "
+                    "U+%04lX combines-forward and has a one-way mapping, "
+                    "not possible in Unicode normalization\n",
+                    (long)c);
+            exit(U_INVALID_FORMAT_ERROR);
+        }
+        if(p->cc==0 && optimization!=OPTIMIZE_FAST) {
+            // Try a compact, algorithmic encoding.
+            // Only for ccc=0, because we can't store additional information
+            // and we do not recursively follow an algorithmic encoding for access to the ccc.
+            //
+            // Also, if hasNoCompBoundaryAfter is set, we can only use the algorithmic encoding
+            // if the mappingCP decomposes further, to ensure that there is a place to store it.
+            // We want to see that the final mapping does not have exactly 1 code point,
+            // or else we would have to recursively ensure that the final mapping is stored
+            // in normal extraData.
+            if(p->mappingCP>=0 && (!p->hasNoCompBoundaryAfter || 1!=p->mapping->countChar32())) {
+                int32_t delta=p->mappingCP-c;
+                if(-Normalizer2Impl::MAX_DELTA<=delta && delta<=Normalizer2Impl::MAX_DELTA) {
+                    p->offset=(delta<<Norm::OFFSET_SHIFT)|Norm::OFFSET_DELTA;
+                }
+            }
+        }
+        if(p->offset==0) {
+            int32_t oldNoNoLength=writer.noNoMappings.length();
+            int32_t offset=oldNoNoLength+writeMapping(c, p, writer.noNoMappings);
+            UnicodeString newMapping=writer.noNoMappings.tempSubString(oldNoNoLength);
+            int32_t previousOffset=writer.previousNoNoMappings.geti(newMapping);
+            if(previousOffset!=0) {
+                // Duplicate, remove the new units and point to the old ones.
+                writer.noNoMappings.truncate(oldNoNoLength);
+                p->offset=((previousOffset-1)<<Norm::OFFSET_SHIFT)|Norm::OFFSET_NO_NO;
+            } else {
+                // Enter this new mapping into the hashtable, avoiding value 0 which is "not found".
+                IcuToolErrorCode errorCode("gennorm2/writeExtraData()/Hashtable.puti()");
+                writer.previousNoNoMappings.puti(newMapping, offset+1, errorCode);
+                p->offset=(offset<<Norm::OFFSET_SHIFT)|Norm::OFFSET_NO_NO;
+            }
+        }
+    }
+}
+
+class Norm16Writer : public Normalizer2DBEnumerator {
+public:
+    Norm16Writer(Normalizer2DataBuilder &b) : Normalizer2DBEnumerator(b) {}
+    virtual UBool rangeHandler(UChar32 start, UChar32 end, uint32_t value) {
+        builder.writeNorm16(start, end, value);
+        return TRUE;
+    }
+};
+
+void Normalizer2DataBuilder::writeNorm16(UChar32 start, UChar32 end, uint32_t value) {
+    if(value!=0) {
+        const Norm *p=norms+value;
+        int32_t offset=p->offset>>Norm::OFFSET_SHIFT;
+        int32_t norm16=0;
+        UBool isDecompNo=FALSE;
+        UBool isCompNoMaybe=FALSE;
+        switch(p->offset&Norm::OFFSET_MASK) {
+        case Norm::OFFSET_NONE:
+            // No mapping, no compositions list.
+            if(p->combinesBack) {
+                norm16=Normalizer2Impl::MIN_NORMAL_MAYBE_YES+p->cc;
+                isDecompNo=(UBool)(p->cc!=0);
+                isCompNoMaybe=TRUE;
+            } else if(p->cc!=0) {
+                norm16=Normalizer2Impl::MIN_YES_YES_WITH_CC-1+p->cc;
+                isDecompNo=isCompNoMaybe=TRUE;
+            }
+            break;
+        case Norm::OFFSET_MAYBE_YES:
+            norm16=indexes[Normalizer2Impl::IX_MIN_MAYBE_YES]+offset;
+            isCompNoMaybe=TRUE;
+            break;
+        case Norm::OFFSET_YES_YES:
+            norm16=offset;
+            break;
+        case Norm::OFFSET_YES_NO_MAPPING_AND_COMPOSITION:
+            norm16=indexes[Normalizer2Impl::IX_MIN_YES_NO]+offset;
+            isDecompNo=TRUE;
+            break;
+        case Norm::OFFSET_YES_NO_MAPPING_ONLY:
+            norm16=indexes[Normalizer2Impl::IX_MIN_YES_NO_MAPPINGS_ONLY]+offset;
+            isDecompNo=TRUE;
+            break;
+        case Norm::OFFSET_NO_NO:
+            norm16=indexes[Normalizer2Impl::IX_MIN_NO_NO]+offset;
+            isDecompNo=isCompNoMaybe=TRUE;
+            break;
+        case Norm::OFFSET_DELTA:
+            norm16=getCenterNoNoDelta()+offset;
+            isDecompNo=isCompNoMaybe=TRUE;
+            break;
+        default:  // Should not occur.
+            exit(U_INTERNAL_PROGRAM_ERROR);
+        }
+        IcuToolErrorCode errorCode("gennorm2/writeNorm16()");
+        utrie2_setRange32(norm16Trie, start, end, (uint32_t)norm16, TRUE, errorCode);
+        if(isDecompNo && start<indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]) {
+            indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]=start;
+        }
+        if(isCompNoMaybe && start<indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]) {
+            indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]=start;
+        }
+    }
+}
+
+void Normalizer2DataBuilder::setHangulData() {
+    HangulIterator hi;
+    const HangulIterator::Range *range;
+    // Check that none of the Hangul/Jamo code points have data.
+    while((range=hi.nextRange())!=NULL) {
+        for(UChar32 c=range->start; c<range->limit; ++c) {
+            if(utrie2_get32(norm16Trie, c)!=0) {
+                fprintf(stderr,
+                        "gennorm2 error: "
+                        "illegal mapping/composition/ccc data for Hangul or Jamo U+%04lX\n",
+                        (long)c);
+                exit(U_INVALID_FORMAT_ERROR);
+            }
+        }
+    }
+    // Set data for algorithmic runtime handling.
+    IcuToolErrorCode errorCode("gennorm2/setHangulData()");
+    hi.reset();
+    while((range=hi.nextRange())!=NULL) {
+        uint16_t norm16=range->norm16;
+        if(norm16==0) {
+            norm16=(uint16_t)indexes[Normalizer2Impl::IX_MIN_YES_NO];  // Hangul LV/LVT encoded as minYesNo
+            if(range->start<indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]) {
+                indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]=range->start;
+            }
+        } else {
+            if(range->start<indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]) {  // Jamo V/T are maybeYes
+                indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]=range->start;
+            }
+        }
+        utrie2_setRange32(norm16Trie, range->start, range->limit-1, norm16, TRUE, errorCode);
+        errorCode.assertSuccess();
+    }
+}
+
+U_CDECL_BEGIN
+
+static UBool U_CALLCONV
+enumRangeMaxValue(const void *context, UChar32 /*start*/, UChar32 /*end*/, uint32_t value) {
+    uint32_t *pMaxValue=(uint32_t *)context;
+    if(value>*pMaxValue) {
+        *pMaxValue=value;
+    }
+    return TRUE;
+}
+
+U_CDECL_END
+
+void Normalizer2DataBuilder::processData() {
+    IcuToolErrorCode errorCode("gennorm2/processData()");
+    norm16Trie=utrie2_open(0, 0, errorCode);
+    errorCode.assertSuccess();
+
+    utrie2_enum(normTrie, NULL, enumRangeHandler, CompositionBuilder(*this).ptr());
+
+    Decomposer decomposer(*this);
+    do {
+        decomposer.didDecompose=FALSE;
+        utrie2_enum(normTrie, NULL, enumRangeHandler, &decomposer);
+    } while(decomposer.didDecompose);
+
+    BuilderReorderingBuffer buffer;
+    int32_t normsLength=utm_countItems(normMem);
+    for(int32_t i=1; i<normsLength; ++i) {
+        // Set the hasNoCompBoundaryAfter flag for use by the last code branch
+        // in Normalizer2Impl::hasCompBoundaryAfter().
+        // For details see the comments on hasNoCompBoundaryAfter(buffer).
+        const Norm &norm=norms[i];
+        if(norm.hasMapping()) {
+            if(norm.compositions!=NULL) {
+                norms[i].hasNoCompBoundaryAfter=TRUE;
+            } else {
+                buffer.reset();
+                reorder(norms+i, buffer);
+                norms[i].hasNoCompBoundaryAfter=hasNoCompBoundaryAfter(buffer);
+            }
+        }
+    }
+
+    indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]=0x110000;
+    indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]=0x110000;
+
+    ExtraDataWriter extraDataWriter(*this);
+    utrie2_enum(normTrie, NULL, enumRangeHandler, &extraDataWriter);
+
+    extraData=extraDataWriter.maybeYesCompositions;
+    extraData.append(extraDataWriter.yesYesCompositions).
+              append(extraDataWriter.yesNoMappingsAndCompositions).
+              append(extraDataWriter.yesNoMappingsOnly).
+              append(extraDataWriter.noNoMappings);
+    // Pad to even length for 4-byte alignment of following data.
+    if(extraData.length()&1) {
+        extraData.append((UChar)0);
+    }
+
+    indexes[Normalizer2Impl::IX_MIN_YES_NO]=
+        extraDataWriter.yesYesCompositions.length();
+    indexes[Normalizer2Impl::IX_MIN_YES_NO_MAPPINGS_ONLY]=
+        indexes[Normalizer2Impl::IX_MIN_YES_NO]+
+        extraDataWriter.yesNoMappingsAndCompositions.length();
+    indexes[Normalizer2Impl::IX_MIN_NO_NO]=
+        indexes[Normalizer2Impl::IX_MIN_YES_NO_MAPPINGS_ONLY]+
+        extraDataWriter.yesNoMappingsOnly.length();
+    indexes[Normalizer2Impl::IX_LIMIT_NO_NO]=
+        indexes[Normalizer2Impl::IX_MIN_NO_NO]+
+        extraDataWriter.noNoMappings.length();
+    indexes[Normalizer2Impl::IX_MIN_MAYBE_YES]=
+        Normalizer2Impl::MIN_NORMAL_MAYBE_YES-
+        extraDataWriter.maybeYesCompositions.length();
+
+    int32_t minNoNoDelta=getCenterNoNoDelta()-Normalizer2Impl::MAX_DELTA;
+    if(indexes[Normalizer2Impl::IX_LIMIT_NO_NO]>minNoNoDelta) {
+        fprintf(stderr,
+                "gennorm2 error: "
+                "data structure overflow, too much mapping composition data\n");
+        exit(U_BUFFER_OVERFLOW_ERROR);
+    }
+
+    utrie2_enum(normTrie, NULL, enumRangeHandler, Norm16Writer(*this).ptr());
+
+    setHangulData();
+
+    // Look for the "worst" norm16 value of any supplementary code point
+    // corresponding to a lead surrogate, and set it as that surrogate's value.
+    // Enables quick check inner loops to look at only code units.
+    //
+    // We could be more sophisticated:
+    // We could collect a bit set for whether there are values in the different
+    // norm16 ranges (yesNo, maybeYes, yesYesWithCC etc.)
+    // and select the best value that only breaks the composition and/or decomposition
+    // inner loops if necessary.
+    // However, that seems like overkill for an optimization for supplementary characters.
+    for(UChar lead=0xd800; lead<0xdc00; ++lead) {
+        uint32_t maxValue=utrie2_get32(norm16Trie, lead);
+        utrie2_enumForLeadSurrogate(norm16Trie, lead, NULL, enumRangeMaxValue, &maxValue);
+        if( maxValue>=(uint32_t)indexes[Normalizer2Impl::IX_LIMIT_NO_NO] &&
+            maxValue>(uint32_t)indexes[Normalizer2Impl::IX_MIN_NO_NO]
+        ) {
+            // Set noNo ("worst" value) if it got into "less-bad" maybeYes or ccc!=0.
+            // Otherwise it might end up at something like JAMO_VT which stays in
+            // the inner decomposition quick check loop.
+            maxValue=(uint32_t)indexes[Normalizer2Impl::IX_LIMIT_NO_NO]-1;
+        }
+        utrie2_set32ForLeadSurrogateCodeUnit(norm16Trie, lead, maxValue, errorCode);
+    }
+
+    // Adjust supplementary minimum code points to break quick check loops at their lead surrogates.
+    // For an empty data file, minCP=0x110000 turns into 0xdc00 (first trail surrogate)
+    // which is harmless.
+    // As a result, the minimum code points are always BMP code points.
+    int32_t minCP=indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP];
+    if(minCP>=0x10000) {
+        indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]=U16_LEAD(minCP);
+    }
+    minCP=indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP];
+    if(minCP>=0x10000) {
+        indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]=U16_LEAD(minCP);
+    }
+
+    utrie2_freeze(norm16Trie, UTRIE2_16_VALUE_BITS, errorCode);
+    norm16TrieLength=utrie2_serialize(norm16Trie, NULL, 0, errorCode);
+    if(errorCode.get()!=U_BUFFER_OVERFLOW_ERROR) {
+        fprintf(stderr, "gennorm2 error: unable to freeze/serialize the normalization trie - %s\n",
+                errorCode.errorName());
+        exit(errorCode.reset());
+    }
+    errorCode.reset();
+
+    int32_t offset=(int32_t)sizeof(indexes);
+    indexes[Normalizer2Impl::IX_NORM_TRIE_OFFSET]=offset;
+    offset+=norm16TrieLength;
+    indexes[Normalizer2Impl::IX_EXTRA_DATA_OFFSET]=offset;
+    offset+=extraData.length()*2;
+    indexes[Normalizer2Impl::IX_SMALL_FCD_OFFSET]=offset;
+    offset+=sizeof(smallFCD);
+    int32_t totalSize=offset;
+    for(int32_t i=Normalizer2Impl::IX_RESERVED3_OFFSET; i<=Normalizer2Impl::IX_TOTAL_SIZE; ++i) {
+        indexes[i]=totalSize;
+    }
+
+    if(beVerbose) {
+        printf("size of normalization trie:         %5ld bytes\n", (long)norm16TrieLength);
+        printf("size of 16-bit extra data:          %5ld uint16_t\n", (long)extraData.length());
+        printf("size of small-FCD data:             %5ld bytes\n", (long)sizeof(smallFCD));
+        printf("size of binary data file contents:  %5ld bytes\n", (long)totalSize);
+        printf("minDecompNoCodePoint:              U+%04lX\n", (long)indexes[Normalizer2Impl::IX_MIN_DECOMP_NO_CP]);
+        printf("minCompNoMaybeCodePoint:           U+%04lX\n", (long)indexes[Normalizer2Impl::IX_MIN_COMP_NO_MAYBE_CP]);
+        printf("minYesNo:                          0x%04x\n", (int)indexes[Normalizer2Impl::IX_MIN_YES_NO]);
+        printf("minYesNoMappingsOnly:              0x%04x\n", (int)indexes[Normalizer2Impl::IX_MIN_YES_NO_MAPPINGS_ONLY]);
+        printf("minNoNo:                           0x%04x\n", (int)indexes[Normalizer2Impl::IX_MIN_NO_NO]);
+        printf("limitNoNo:                         0x%04x\n", (int)indexes[Normalizer2Impl::IX_LIMIT_NO_NO]);
+        printf("minMaybeYes:                       0x%04x\n", (int)indexes[Normalizer2Impl::IX_MIN_MAYBE_YES]);
+    }
+
+    UVersionInfo nullVersion={ 0, 0, 0, 0 };
+    if(0==memcmp(nullVersion, unicodeVersion, 4)) {
+        u_versionFromString(unicodeVersion, U_UNICODE_VERSION);
+    }
+    memcpy(dataInfo.dataVersion, unicodeVersion, 4);
+}
+
+void Normalizer2DataBuilder::writeBinaryFile(const char *filename) {
+    processData();
+
+    IcuToolErrorCode errorCode("gennorm2/writeBinaryFile()");
+    LocalArray<uint8_t> norm16TrieBytes(new uint8_t[norm16TrieLength]);
+    utrie2_serialize(norm16Trie, norm16TrieBytes.getAlias(), norm16TrieLength, errorCode);
+    errorCode.assertSuccess();
+
+    UNewDataMemory *pData=
+        udata_create(NULL, NULL, filename, &dataInfo,
+                     haveCopyright ? U_COPYRIGHT_STRING : NULL, errorCode);
+    if(errorCode.isFailure()) {
+        fprintf(stderr, "gennorm2 error: unable to create the output file %s - %s\n",
+                filename, errorCode.errorName());
+        exit(errorCode.reset());
+    }
+    udata_writeBlock(pData, indexes, sizeof(indexes));
+    udata_writeBlock(pData, norm16TrieBytes.getAlias(), norm16TrieLength);
+    udata_writeUString(pData, extraData.getBuffer(), extraData.length());
+    udata_writeBlock(pData, smallFCD, sizeof(smallFCD));
+    int32_t writtenSize=udata_finish(pData, errorCode);
+    if(errorCode.isFailure()) {
+        fprintf(stderr, "gennorm2: error %s writing the output file\n", errorCode.errorName());
+        exit(errorCode.reset());
+    }
+    int32_t totalSize=indexes[Normalizer2Impl::IX_TOTAL_SIZE];
+    if(writtenSize!=totalSize) {
+        fprintf(stderr, "gennorm2 error: written size %ld != calculated size %ld\n",
+            (long)writtenSize, (long)totalSize);
+        exit(U_INTERNAL_PROGRAM_ERROR);
+    }
+}
+
+void
+Normalizer2DataBuilder::writeCSourceFile(const char *filename) {
+    processData();
+
+    IcuToolErrorCode errorCode("gennorm2/writeCSourceFile()");
+    const char *basename=findBasename(filename);
+    CharString path(filename, (int32_t)(basename-filename), errorCode);
+    CharString dataName(basename, errorCode);
+    const char *extension=strrchr(basename, '.');
+    if(extension!=NULL) {
+        dataName.truncate((int32_t)(extension-basename));
+    }
+    errorCode.assertSuccess();
+
+    LocalArray<uint8_t> norm16TrieBytes(new uint8_t[norm16TrieLength]);
+    utrie2_serialize(norm16Trie, norm16TrieBytes.getAlias(), norm16TrieLength, errorCode);
+    errorCode.assertSuccess();
+
+    FILE *f=usrc_create(path.data(), basename, "icu/source/tools/gennorm2/n2builder.cpp");
+    if(f==NULL) {
+        fprintf(stderr, "gennorm2/writeCSourceFile() error: unable to create the output file %s\n",
+                filename);
+        exit(U_FILE_ACCESS_ERROR);
+        return;
+    }
+    fputs("#ifdef INCLUDED_FROM_NORMALIZER2_CPP\n\n", f);
+    char line[100];
+    sprintf(line, "static const UVersionInfo %s_formatVersion={", dataName.data());
+    usrc_writeArray(f, line, dataInfo.formatVersion, 8, 4, "};\n");
+    sprintf(line, "static const UVersionInfo %s_dataVersion={", dataName.data());
+    usrc_writeArray(f, line, dataInfo.dataVersion, 8, 4, "};\n\n");
+    sprintf(line, "static const int32_t %s_indexes[Normalizer2Impl::IX_COUNT]={\n",
+            dataName.data());
+    usrc_writeArray(f,
+        line,
+        indexes, 32, Normalizer2Impl::IX_COUNT,
+        "\n};\n\n");
+    sprintf(line, "static const uint16_t %s_trieIndex[%%ld]={\n", dataName.data());
+    usrc_writeUTrie2Arrays(f,
+        line, NULL,
+        norm16Trie,
+        "\n};\n\n");
+    sprintf(line, "static const uint16_t %s_extraData[%%ld]={\n", dataName.data());
+    usrc_writeArray(f,
+        line,
+        extraData.getBuffer(), 16, extraData.length(),
+        "\n};\n\n");
+    sprintf(line, "static const uint8_t %s_smallFCD[%%ld]={\n", dataName.data());
+    usrc_writeArray(f,
+        line,
+        smallFCD, 8, sizeof(smallFCD),
+        "\n};\n\n");
+    /*fputs(  // TODO
+        "static const UCaseProps %s_singleton={\n"
+        "  NULL,\n"
+        "  %s_indexes,\n"
+        "  %s_extraData,\n"
+        "  %s_smallFCD,\n",
+        f);*/
+    sprintf(line, "static const UTrie2 %s_trie={\n", dataName.data());
+    char line2[100];
+    sprintf(line2, "%s_trieIndex", dataName.data());
+    usrc_writeUTrie2Struct(f,
+        line,
+        norm16Trie, line2, NULL,
+        "};\n");
+    fputs("\n#endif  // INCLUDED_FROM_NORMALIZER2_CPP\n", f);
+    fclose(f);
+}
+
+U_NAMESPACE_END
+
+#endif /* #if !UCONFIG_NO_NORMALIZATION */
+
+/*
+ * Hey, Emacs, please set the following:
+ *
+ * Local Variables:
+ * indent-tabs-mode: nil
+ * End:
+ */
diff --git a/intl/icu/source/tools/gennorm2/n2builder.h b/intl/icu/source/tools/gennorm2/n2builder.h
new file mode 100644
index 000000000..6f8b8afff
--- /dev/null
+++ b/intl/icu/source/tools/gennorm2/n2builder.h
@@ -0,0 +1,127 @@
+// Copyright (C) 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
+/*
+*******************************************************************************
+*
+*   Copyright (C) 2009-2014, International Business Machines
+*   Corporation and others.  All Rights Reserved.
+*
+*******************************************************************************
+*   file name:  n2builder.h
+*   encoding:   US-ASCII
+*   tab size:   8 (not used)
+*   indentation:4
+*
+*   created on: 2009nov25
+*   created by: Markus W. Scherer
+*/
+
+#ifndef __N2BUILDER_H__
+#define __N2BUILDER_H__
+
+#include "unicode/utypes.h"
+
+#if !UCONFIG_NO_NORMALIZATION
+
+#include "unicode/errorcode.h"
+#include "unicode/unistr.h"
+#include "normalizer2impl.h"  // for IX_COUNT
+#include "toolutil.h"
+#include "utrie2.h"
+
+U_NAMESPACE_BEGIN
+
+extern UBool beVerbose, haveCopyright;
+
+struct Norm;
+
+class BuilderReorderingBuffer;
+class ExtraDataWriter;
+
+class Normalizer2DataBuilder {
+public:
+    Normalizer2DataBuilder(UErrorCode &errorCode);
+    ~Normalizer2DataBuilder();
+
+    enum OverrideHandling {
+        OVERRIDE_NONE,
+        OVERRIDE_ANY,
+        OVERRIDE_PREVIOUS
+    };
+
+    void setOverrideHandling(OverrideHandling oh);
+
+    enum Optimization {
+        OPTIMIZE_NORMAL,
+        OPTIMIZE_FAST
+    };
+
+    void setOptimization(Optimization opt) { optimization=opt; }
+
+    void setCC(UChar32 c, uint8_t cc);
+    void setOneWayMapping(UChar32 c, const UnicodeString &m);
+    void setRoundTripMapping(UChar32 c, const UnicodeString &m);
+    void removeMapping(UChar32 c);
+
+    void setUnicodeVersion(const char *v);
+
+    void writeBinaryFile(const char *filename);
+    void writeCSourceFile(const char *filename);
+
+private:
+    friend class CompositionBuilder;
+    friend class Decomposer;
+    friend class ExtraDataWriter;
+    friend class Norm16Writer;
+
+    // No copy constructor nor assignment operator.
+    Normalizer2DataBuilder(const Normalizer2DataBuilder &other);
+    Normalizer2DataBuilder &operator=(const Normalizer2DataBuilder &other);
+
+    Norm *allocNorm();
+    Norm *getNorm(UChar32 c);
+    Norm *createNorm(UChar32 c);
+    Norm *checkNormForMapping(Norm *p, UChar32 c);  // check for permitted overrides
+
+    const Norm &getNormRef(UChar32 c) const;
+    uint8_t getCC(UChar32 c) const;
+    UBool combinesWithCCBetween(const Norm &norm, uint8_t lowCC, uint8_t highCC) const;
+    UChar32 combine(const Norm &norm, UChar32 trail) const;
+
+    void addComposition(UChar32 start, UChar32 end, uint32_t value);
+    UBool decompose(UChar32 start, UChar32 end, uint32_t value);
+    void reorder(Norm *p, BuilderReorderingBuffer &buffer);
+    UBool hasNoCompBoundaryAfter(BuilderReorderingBuffer &buffer);
+    void setHangulData();
+    int32_t writeMapping(UChar32 c, const Norm *p, UnicodeString &dataString);
+    void writeCompositions(UChar32 c, const Norm *p, UnicodeString &dataString);
+    void writeExtraData(UChar32 c, uint32_t value, ExtraDataWriter &writer);
+    int32_t getCenterNoNoDelta() {
+        return indexes[Normalizer2Impl::IX_MIN_MAYBE_YES]-Normalizer2Impl::MAX_DELTA-1;
+    }
+    void writeNorm16(UChar32 start, UChar32 end, uint32_t value);
+    void processData();
+
+    UTrie2 *normTrie;
+    UToolMemory *normMem;
+    Norm *norms;
+
+    int32_t phase;
+    OverrideHandling overrideHandling;
+
+    Optimization optimization;
+
+    int32_t indexes[Normalizer2Impl::IX_COUNT];
+    UTrie2 *norm16Trie;
+    int32_t norm16TrieLength;
+    UnicodeString extraData;
+    uint8_t smallFCD[0x100];
+
+    UVersionInfo unicodeVersion;
+};
+
+U_NAMESPACE_END
+
+#endif // #if !UCONFIG_NO_NORMALIZATION
+
+#endif  // __N2BUILDER_H__