ICU-11802 add 4 Emoji properties from emoji-data.txt 2.0

X-SVN-Rev: 38182

diff --git a/tools/unicode/c/genprops/corepropsbuilder.cpp b/tools/unicode/c/genprops/corepropsbuilder.cpp
index 531595c3e744d88c7a7bdebcfbdafa548eaa1df2..26b254ad19faebd4a9b3d3c1628da01129b87d76 100644 (file)
--- a/tools/unicode/c/genprops/corepropsbuilder.cpp
+++ b/tools/unicode/c/genprops/corepropsbuilder.cpp
@@ -1,7 +1,7 @@
 /*
 *******************************************************************************
 *
-*   Copyright (C) 1999-2015, International Business Machines
+*   Copyright (C) 1999-2016, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
@@ -45,7 +45,7 @@ the udata API for loading ICU data. Especially, a UDataInfo structure
 precedes the actual data. It contains platform properties values and the
 file format version.
 
-The following is a description of format version 7.1 .
+The following is a description of format version 7.2 .
 
 Data contents:
 
@@ -244,6 +244,12 @@ Unicode 6.2 adds sexagesimal (base-60) numeric values:
 
 The encoding of numeric values was extended to handle such values.
 
+--- Changes in format version 7.2 ---
+
+ICU 57 adds 4 Emoji properties to vector word 2.
+http://bugs.icu-project.org/trac/ticket/11802
+http://www.unicode.org/reports/tr51/#Emoji_Properties
+
 ----------------------------------------------------------------------------- */
 
 U_NAMESPACE_USE
@@ -259,8 +265,8 @@ static UDataInfo dataInfo={
     0,
 
     { 0x55, 0x50, 0x72, 0x6f },                 /* dataFormat="UPro" */
-    { 7, 1, 0, 0 },                             /* formatVersion */
-    { 6, 2, 0, 0 }                              /* dataVersion */
+    { 7, 2, 0, 0 },                             /* formatVersion */
+    { 8, 0, 0, 0 }                              /* dataVersion */
 };
 
 class CorePropsBuilder : public PropsBuilder {
@@ -528,6 +534,11 @@ propToBinaries[]={
     { UCHAR_ID_START,                       1, UPROPS_ID_START },
     { UCHAR_ID_CONTINUE,                    1, UPROPS_ID_CONTINUE },
     { UCHAR_GRAPHEME_BASE,                  1, UPROPS_GRAPHEME_BASE },
+
+    { UCHAR_EMOJI,                          2, UPROPS_2_EMOJI },
+    { UCHAR_EMOJI_PRESENTATION,             2, UPROPS_2_EMOJI_PRESENTATION },
+    { UCHAR_EMOJI_MODIFIER,                 2, UPROPS_2_EMOJI_MODIFIER },
+    { UCHAR_EMOJI_MODIFIER_BASE,            2, UPROPS_2_EMOJI_MODIFIER_BASE },
 };
 
 struct PropToEnum {

diff --git a/tools/unicode/c/genprops/pnames_data.h b/tools/unicode/c/genprops/pnames_data.h
index e89e782b0ebf3e02209aabc307aba2df4a4580e1..fca71f9e383abc30e214458e06e689927bfc592e 100644 (file)
--- a/tools/unicode/c/genprops/pnames_data.h
+++ b/tools/unicode/c/genprops/pnames_data.h
@@ -1,5 +1,5 @@
 /**
- * Copyright (C) 2002-2015, International Business Machines Corporation and
+ * Copyright (C) 2002-2016, International Business Machines Corporation and
  * others. All Rights Reserved.
  *
  * machine-generated by: icu/tools/unicode/py/preparseucd.py
@@ -997,7 +997,7 @@ static const Value VALUES_gcm[38] = {
     Value((int32_t)U_GC_ZS_MASK, "Zs Space_Separator"),
 };
 
-static const Property PROPERTIES[96] = {
+static const Property PROPERTIES[100] = {
     Property(UCHAR_ALPHABETIC, "Alpha Alphabetic"),
     Property(UCHAR_ASCII_HEX_DIGIT, "AHex ASCII_Hex_Digit"),
     Property(UCHAR_BIDI_CONTROL, "Bidi_C Bidi_Control"),
@@ -1055,6 +1055,10 @@ static const Property PROPERTIES[96] = {
     Property(UCHAR_CHANGES_WHEN_CASEFOLDED, "CWCF Changes_When_Casefolded"),
     Property(UCHAR_CHANGES_WHEN_CASEMAPPED, "CWCM Changes_When_Casemapped"),
     Property(UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED, "CWKCF Changes_When_NFKC_Casefolded"),
+    Property(UCHAR_EMOJI, "Emoji Emoji"),
+    Property(UCHAR_EMOJI_PRESENTATION, "Emoji_Presentation Emoji_Presentation"),
+    Property(UCHAR_EMOJI_MODIFIER, "Emoji_Modifier Emoji_Modifier"),
+    Property(UCHAR_EMOJI_MODIFIER_BASE, "Emoji_Modifier_Base Emoji_Modifier_Base"),
     Property(UCHAR_BIDI_CLASS, "bc Bidi_Class", VALUES_bc, 23),
     Property(UCHAR_BLOCK, "blk Block", VALUES_blk, 263),
     Property(UCHAR_CANONICAL_COMBINING_CLASS, "ccc Canonical_Combining_Class", VALUES_ccc, 57),

diff --git a/tools/unicode/py/preparseucd.py b/tools/unicode/py/preparseucd.py
index e68d5bd1f5b0dabc032556697e5e4fd9ddfc40a2..00c63ba923955a47f7fac813acf4150f75909994 100755 (executable)
--- a/tools/unicode/py/preparseucd.py
+++ b/tools/unicode/py/preparseucd.py
@@ -1,6 +1,6 @@
 #!/usr/bin/python
 # -*- coding: utf-8 -*-
-# Copyright (c) 2009-2015 International Business Machines
+# Copyright (c) 2009-2016 International Business Machines
 # Corporation and others. All Rights Reserved.
 #
 #   file name:  preparseucd.py
@@ -516,6 +516,11 @@ def ParsePropertyAliases(in_file):
   AddBinaryProperty("nfcinert", "NFC_Inert")
   AddBinaryProperty("nfkcinert", "NFKC_Inert")
   AddBinaryProperty("segstart", "Segment_Starter")
+  # http://www.unicode.org/reports/tr51/#Emoji_Properties
+  AddBinaryProperty("Emoji", "Emoji")
+  AddBinaryProperty("Emoji_Presentation", "Emoji_Presentation")
+  AddBinaryProperty("Emoji_Modifier", "Emoji_Modifier")
+  AddBinaryProperty("Emoji_Modifier_Base", "Emoji_Modifier_Base")
   # C/POSIX character classes that do not have Unicode property [value] aliases.
   # See uchar.h.
   AddPOSIXBinaryProperty("alnum")
@@ -1535,6 +1540,7 @@ _files = {
   "DerivedNormalizationProps.txt": (CopyAndStrip, ParseNamedProperties),
   "DerivedNumericValues.txt": (DontCopy, ParseDerivedNumericValues),
   "EastAsianWidth.txt": (DontCopy, ParseEastAsianWidth),
+  "emoji-data.txt": (DontCopy, ParseNamedProperties),
   "GraphemeBreakProperty.txt": (DontCopy, ParseGraphemeBreakProperty),
   "GraphemeBreakTest.txt": (PrependBOM, "testdata"),
   "IndicPositionalCategory.txt": (DontCopy, ParseIndicPositionalCategory),