From 84fa0ef97db9dd72693f10a1fa72c766f2b0838f Mon Sep 17 00:00:00 2001
From: Hiroki Sato <hrs@FreeBSD.org>
Date: Sat, 15 Aug 2020 07:19:37 +0000
Subject: [PATCH] - Update to Unicode CLDR v35 (Unicode 12.0).

- Update tools/tools/locale to add make targets to automatically
  generate locale source files.  With this change, just typing
  "make obj && make -j4" will rebuild them.  Check README for more details.

- Fix issues in ja_JP ctypedef and range specification support
  in utf8-rollup.pl.

- Add a temporary patch for UnicodeData.txt to fix code ranges of
  CJK Ideograph Extension A and Extension B.

- tools/cldr2def.pl:
    Use eucJP for ja_JP ctypedef because eucJP is not compatible with UTF-8.

- tools/convert_map.pl:
    Add a verbose error message.

- tools/utf8-rollup.pl:
    Normalize entries to use Unicode, not UTF-8.

Reviewed by:		bapt
Differential Revision:	https://reviews.freebsd.org/D25503
---
 tools/tools/locale/Makefile                   | 193 ++++--
 tools/tools/locale/README                     |  80 ++-
 tools/tools/locale/etc/final-maps/map.UTF-8   | 554 ++++++++++++++++++
 .../tools/locale/patch/patch-UnicodeData.txt  |  29 +
 tools/tools/locale/tools/cldr2def.pl          |   5 +
 tools/tools/locale/tools/convert_map.pl       |   5 +-
 tools/tools/locale/tools/finalize             |  38 +-
 tools/tools/locale/tools/utf8-rollup.pl       |  31 +-
 8 files changed, 821 insertions(+), 114 deletions(-)
 create mode 100644 tools/tools/locale/patch/patch-UnicodeData.txt

diff --git a/tools/tools/locale/Makefile b/tools/tools/locale/Makefile
index 508034d19695..aad2c2160bb5 100644
--- a/tools/tools/locale/Makefile
+++ b/tools/tools/locale/Makefile
@@ -7,19 +7,30 @@
 #
 # Modified by John Marino to suit DragonFly needs
 #
-
-.OBJDIR:	.
-
-.if !defined(UNIDIR)
-.error UNIDIR is not set
+.if ${.CURDIR} == ${.OBJDIR}
+.error Do make obj first.
 .endif
-PASSON=		UNIDIR="${UNIDIR}"
 
-ETCDIR=		${.CURDIR}/etc
+LOCALESRCDIR?=	${DESTDIR}/usr/src/share
+TMPDIR?=	/tmp
+
+BASEDIR=	${.CURDIR}
+ETCDIR=		${BASEDIR}/etc
+TOOLSDIR=	${BASEDIR}/tools
+PATCHDIR=	${BASEDIR}/patch
+UNIDIR=		${.OBJDIR:tA}/unicode
+
+PKGS=	openjdk8 \
+	apache-ant \
+	p5-XML-Parser \
+	p5-Tie-IxHash \
+	p5-Text-Iconv
+tools-test:
+	pkg info -e ${PKGS}
+	@echo tools ok.
 
 KNOWN=		monetdef numericdef msgdef colldef ctypedef # timedef
 TYPES?=		${KNOWN}
-LOCALE_DESTDIR?=	/tmp/generated-locales/
 
 COLLATION_SPECIAL?= \
 	cs_CZ ISO8859-2 \
@@ -44,65 +55,79 @@ COLLATION_SPECIAL?= \
 .for area enc in ${COLLATION_SPECIAL}
 COLLATIONS_SPECIAL_ENV+=	${area}.${enc}
 .endfor
-PASSON+=	COLLATIONS_SPECIAL="${COLLATIONS_SPECIAL_ENV}"
+SETENV=	env -i \
+	PATH="${PATH}" \
+	TMPDIR="${TMPDIR}" \
+	COLLATIONS_SPECIAL="${COLLATIONS_SPECIAL_ENV}" \
+	UNIDIR="${UNIDIR}" \
+	BASEDIR="${BASEDIR}" \
+	TOOLSDIR="${TOOLSDIR}" \
+	ETCDIR="${ETCDIR}"
 
-all:
-.for t in ${TYPES}
-.  if ${KNOWN:M${t}}
-	test -d ${t} || mkdir ${t}
-	make build-${t}
-.  endif
-.endfor
+all: posix build afterbuild
+.ORDER: posix build afterbuild
+
+afterbuild: build
 	@echo ""
 	@find . -name *failed
 
 .for t in ${TYPES}
+.  if ${KNOWN:M${t}}
+build: build-${t}
+.ORDER: build-${t} afterbuild
+.  endif
+.endfor
+
+diff:
+.for t in ${TYPES}
+.  if ${KNOWN:M${t}}
+diff: diff-${t}
+diff-${t}:
+	-/usr/bin/diff -ruN -x Makefile -x Makefile.depend \
+	    ${LOCALESRCDIR}/${t} ${t}
+.  endif
+.endfor
+
+install:
+.for t in ${TYPES}
+.  if ${KNOWN:M${t}}
 install: install-${t}
 install-${t}:
-.  if ${KNOWN:M${t}}
-	rm -rf ${.CURDIR}/${t}.draft
-	rm -f ${.CURDIR}/../../../share/${t}/Makefile
-	rm -f ${.CURDIR}/../../../share/${t}/*.src
-	mv ${.CURDIR}/${t}/* ${.CURDIR}/../../../share/${t}/
+	cd ${LOCALESRCDIR}/${t} && \
+	    rm -f Makefile *.src && \
+	    install -c ${t}/* ${LOCALESRCDIR}/${t}
 .  endif
 .endfor
 
 post-install:
 .for t in ${TYPES}
 .  if ${KNOWN:M${t}}
-	(cd ${.CURDIR}/../../../share/${t} && \
-		make && make install && make clean)
+	cd ${LOCALSRCDIR}/${t} && \
+	    make && make install && make clean
 .  endif
 .endfor
 
 .for t in ${TYPES}
-gen-${t}:
-	mkdir -p ${t} ${t}.draft
-	perl -I tools tools/cldr2def.pl \
-		--unidir=$$(realpath ${UNIDIR}) \
-		--etc=$$(realpath ${ETCDIR}) \
+CLEANDIRS+=	${t} ${t}.draft
+${t}:
+	mkdir -p ${t} ${t}.draft && \
+	perl -I ${TOOLSDIR} ${TOOLSDIR}/cldr2def.pl \
+		--unidir=${UNIDIR:tA} \
+		--etc=${ETCDIR:tA} \
 		--type=${t}
 
-build-${t}: gen-${t}
-	env ${PASSON} tools/finalize ${t}
+build-${t}: ${t}
+	${SETENV} OUTBASEDIR="${.OBJDIR}/${t}" ${TOOLSDIR}/finalize ${t}
 .endfor
 
-gen-ctypedef:	ctype-rollup
-static-colldef: gen-colldef
+static-colldef: colldef
 build-colldef:	static-colldef
 
 static-colldef:
 .for area enc in ${COLLATION_SPECIAL}
-	awk -f tools/extract-colldef.awk ${UNIDIR}/posix/${area}.${enc}.src > \
-	    colldef.draft/${area}.${enc}.src
-.endfor
-
-ctype-rollup:
-	perl -I tools tools/utf8-rollup.pl --unidir=$$(realpath ${UNIDIR})
-
-clean:
-.for t in ${TYPES}
-	rm -rf ${t} ${t}.draft
+colldef.draft/${area}.${enc}.src: posix/${area}.${enc}.src
+	awk -f ${TOOLSDIR}/extract-colldef.awk \
+	    ${.ALLSRC} > ${.TARGET} || (rm -f ${.TARGET} && false)
 .endfor
 
 BASE_LOCALES_OF_INTEREST?= \
@@ -145,31 +170,71 @@ ENCODINGS=	Big5 \
 		US-ASCII \
 		UTF-8
 
-POSIX:	posixsrc posixcol posixcm
-.if !exists(${UNIDIR}/tools/java/cldr.jar)
-.error check README about building cldr.jar
+# CLDR files
+CLDRFILES_CORE= https://unicode.org/Public/cldr/35/core.zip
+CLDRFILES_KEY=	https://unicode.org/Public/cldr/35/keyboards.zip
+CLDRFILES_TOOLS=https://unicode.org/Public/cldr/35/tools.zip
+CLDRFILES_UCD=	http://www.unicode.org/Public/zipped/latest/UCD.zip
+
+# fetch and extract targets
+${UNIDIR}:
+	mkdir -p ${UNIDIR}
+.for N in CORE KEY TOOLS UCD
+${CLDRFILES_${N}:T}:
+	fetch ${CLDRFILES_${N}}
+fetch: ${CLDRFILES_${N}:T}
+extract-${CLDRFILES_${N}:T}:: ${CLDRFILES_${N}:T} ${UNIDIR}
+	cd ${UNIDIR} && unzip -o ../${CLDRFILES_${N}:T}
+extract: extract-${CLDRFILES_${N}:T}
+.endfor
+patch::
+.if exists(${PATCHDIR})
+	cd ${UNIDIR} && cat ${PATCHDIR}/patch-* | patch
 .endif
+
+.if !exists(${UNIDIR}/tools/java/cldr.jar)
+.ORDER: extract patch
+build-tools: extract patch tools-test ${UNIDIR}
+	cd ${UNIDIR}/tools/java && ${SETENV} ant all jar
+.else
+build-tools:
+	@echo cldr.jar is ready.
+.endif
+
+JAVA_CLDR= java -DCLDR_DIR=${UNIDIR:Q} -jar ${UNIDIR}/tools/java/cldr.jar
+
+posix: posixcm post-posixcm posixsrc posixcol
+.ORDER: posixcm post-posixcm posixsrc posixcol
+${UNIDIR}/posix:
+	ln -s -f ../posix ${.TARGET}
+clean-posix:
+	rm -rf posix ${UNIDIR}/posix
+post-posixcm: ${UNIDIR}/posix
+	perl -I ${TOOLSDIR} ${TOOLSDIR}/utf8-rollup.pl \
+	    --unidir=${UNIDIR}
+.for enc in ${ENCODINGS}
+posixcm: build-tools posix/${enc}.cm
+.ORDER: build-tools posix/${enc}.cm
+posix/${enc}.cm:
+	mkdir -p posix && \
+	    ${JAVA_CLDR} org.unicode.cldr.posix.GenerateCharmap \
+		-d posix -c ${enc}
+.endfor
 .for area in ${BASE_LOCALES_OF_INTEREST}
-posixsrc: ${UNIDIR}/posix/${area}.UTF-8.src
-${UNIDIR}/posix/${area}.UTF-8.src:
-	java -DCLDR_DIR=${UNIDIR:Q} -jar ${UNIDIR}/tools/java/cldr.jar \
-		org.unicode.cldr.posix.GeneratePOSIX \
-		-d ${UNIDIR}/posix -m ${area} -c UTF-8
+posixsrc: build-tools posix/${area}.UTF-8.src
+.ORDER: build-tools posix/${area}.UTF-8.src
+posix/${area}.UTF-8.src: 
+	mkdir -p posix && \
+	    ${JAVA_CLDR} org.unicode.cldr.posix.GeneratePOSIX \
+		-d posix -m ${area} -c UTF-8
 .endfor
 .for area encoding in ${COLLATION_SPECIAL}
-posixcol: ${UNIDIR}/posix/${area}.${encoding}.src
-${UNIDIR}/posix/${area}.${encoding}.src:
-	java -DCLDR_DIR=${UNIDIR:Q} -jar ${UNIDIR}/tools/java/cldr.jar \
-		org.unicode.cldr.posix.GeneratePOSIX \
-		-d ${UNIDIR}/posix -m ${area} -c ${encoding}
-.endfor
-.for enc in ${ENCODINGS}
-posixcm: ${UNIDIR}/posix/${enc}.cm
-${UNIDIR}/posix/${enc}.cm:
-	java -DCLDR_DIR=${UNIDIR:Q} -jar ${UNIDIR}/tools/java/cldr.jar \
-		org.unicode.cldr.posix.GenerateCharmap \
-		-d ${UNIDIR}/posix -c ${enc}
+posixcol: build-tools posix/${area}.${encoding}.src
+.ORDER: build-tools posix/${area}.${encoding}.src
+posix/${area}.${encoding}.src:
+	mkdir -p posix && \
+	    ${JAVA_CLDR} org.unicode.cldr.posix.GeneratePOSIX \
+		-d posix -m ${area} -c ${encoding}
 .endfor
 
-clean-POSIX:
-	rm -f ${UNIDIR}/posix/*
+.include <bsd.obj.mk>
diff --git a/tools/tools/locale/README b/tools/tools/locale/README
index 60357d4d8dfe..8dc8c2a25fc1 100644
--- a/tools/tools/locale/README
+++ b/tools/tools/locale/README
@@ -1,32 +1,58 @@
 # $FreeBSD$
 
-To generate the locales:
+Files in this directory are used to generate locale source files
+from files in CLDR (Unicode Common Locale Data Repository). 
 
-Tools needed:
-	java (openjdk >= 8)
-	perl
-	converters/p5-Text-Iconv
-	devel/apache-ant
-	devel/p5-Tie-IxHash
-	textproc/p5-XML-Parser
+To generate the files, do the following:
 
-1. Fetch CLDR data from: http://unicode.org/Public/cldr/.  You need all of the
-core.zip, keyboards.zip, and tools.zip.
-2. Fetch unidata (UCD.zip) from http://www.unicode.org/Public/zipped/latest.
-3. Extract:
-	mkdir -p ~/unicode
-	cd ~/unicode
-	unzip ~/core.zip
-	unzip ~/keyboards.zip
-	unzip ~/tools.zip
-	unzip ~/UCD.zip
-4. Export variable:
-	UNIDIR=~/unicode; export UNIDIR
-5. Build the CLDR tools:
-	cd $UNIDIR/tools/java
-	ant jar
-6. Build POSIX data files from CLDR data:
-	make POSIX
-7. Build and install new locale data:
-	make
+	cd /usr/src/tools/tools/locale
+	make obj	(mandatory)
+	make -j16	(-jN recommended)
+	make diff	(check if the changes are reasonable)
 	make install
+
+"make" downloads the necessary files, build them, and install the
+results into /usr/src/share/* as source files for locales.
+
+More details are as follows:
+
+Variables:
+	LOCALESRCDIR
+		Destination path for the generated locale files.
+		Default: $DESTDIR/usr/src/share.
+	TMPDIR
+		Temporary directory.
+		Default: /tmp
+
+Targets:
+	make obj
+		Create a temporary directory for building.
+
+	make clean
+		Clean up the obj directories.
+
+	make cleandir
+		Remove the obj directories completely.
+
+	make tools-test
+		Check if necessary tools are installed or not.
+		If something is missing, install them.
+
+	make fetch
+		Download necessary files from CLDR.
+
+	make build-tools
+		Build a tool to generate locale source files.
+
+	make posix
+		Build POSIX locale source files.
+
+	make build
+		Build locale files.
+
+	make diff
+		Run diff(1) the build results against $LOCALESRCDIR.
+
+	make install
+		Install the build results into $LOCALESRCDIR.
+[EOF]
diff --git a/tools/tools/locale/etc/final-maps/map.UTF-8 b/tools/tools/locale/etc/final-maps/map.UTF-8
index 031b8545982d..ad0dfd2c7de6 100644
--- a/tools/tools/locale/etc/final-maps/map.UTF-8
+++ b/tools/tools/locale/etc/final-maps/map.UTF-8
@@ -2969,6 +2969,7 @@ CHARMAP
 <TELUGU_DIGIT_SEVEN> \xE0\xB1\xAD
 <TELUGU_DIGIT_EIGHT> \xE0\xB1\xAE
 <TELUGU_DIGIT_NINE> \xE0\xB1\xAF
+<TELUGU_SIGN_SIDDHAM> \xE0\xB1\xB7
 <TELUGU_FRACTION_DIGIT_ZERO_FOR_ODD_POWERS_OF_FOUR> \xE0\xB1\xB8
 <TELUGU_FRACTION_DIGIT_ONE_FOR_ODD_POWERS_OF_FOUR> \xE0\xB1\xB9
 <TELUGU_FRACTION_DIGIT_TWO_FOR_ODD_POWERS_OF_FOUR> \xE0\xB1\xBA
@@ -3363,14 +3364,24 @@ CHARMAP
 <LAO_LETTER_KO> \xE0\xBA\x81
 <LAO_LETTER_KHO_SUNG> \xE0\xBA\x82
 <LAO_LETTER_KHO_TAM> \xE0\xBA\x84
+<LAO_LETTER_PALI_GHA> \xE0\xBA\x86
 <LAO_LETTER_NGO> \xE0\xBA\x87
 <LAO_LETTER_CO> \xE0\xBA\x88
+<LAO_LETTER_PALI_CHA> \xE0\xBA\x89
 <LAO_LETTER_SO_TAM> \xE0\xBA\x8A
+<LAO_LETTER_PALI_JHA> \xE0\xBA\x8C
 <LAO_LETTER_NYO> \xE0\xBA\x8D
+<LAO_LETTER_PALI_NYA> \xE0\xBA\x8E
+<LAO_LETTER_PALI_TTA> \xE0\xBA\x8F
+<LAO_LETTER_PALI_TTHA> \xE0\xBA\x90
+<LAO_LETTER_PALI_DDA> \xE0\xBA\x91
+<LAO_LETTER_PALI_DDHA> \xE0\xBA\x92
+<LAO_LETTER_PALI_NNA> \xE0\xBA\x93
 <LAO_LETTER_DO> \xE0\xBA\x94
 <LAO_LETTER_TO> \xE0\xBA\x95
 <LAO_LETTER_THO_SUNG> \xE0\xBA\x96
 <LAO_LETTER_THO_TAM> \xE0\xBA\x97
+<LAO_LETTER_PALI_DHA> \xE0\xBA\x98
 <LAO_LETTER_NO> \xE0\xBA\x99
 <LAO_LETTER_BO> \xE0\xBA\x9A
 <LAO_LETTER_PO> \xE0\xBA\x9B
@@ -3378,13 +3389,17 @@ CHARMAP
 <LAO_LETTER_FO_TAM> \xE0\xBA\x9D
 <LAO_LETTER_PHO_TAM> \xE0\xBA\x9E
 <LAO_LETTER_FO_SUNG> \xE0\xBA\x9F
+<LAO_LETTER_PALI_BHA> \xE0\xBA\xA0
 <LAO_LETTER_MO> \xE0\xBA\xA1
 <LAO_LETTER_YO> \xE0\xBA\xA2
 <LAO_LETTER_LO_LING> \xE0\xBA\xA3
 <LAO_LETTER_LO_LOOT> \xE0\xBA\xA5
 <LAO_LETTER_WO> \xE0\xBA\xA7
+<LAO_LETTER_SANSKRIT_SHA> \xE0\xBA\xA8
+<LAO_LETTER_SANSKRIT_SSA> \xE0\xBA\xA9
 <LAO_LETTER_SO_SUNG> \xE0\xBA\xAA
 <LAO_LETTER_HO_SUNG> \xE0\xBA\xAB
+<LAO_LETTER_PALI_LLA> \xE0\xBA\xAC
 <LAO_LETTER_O> \xE0\xBA\xAD
 <LAO_LETTER_HO_TAM> \xE0\xBA\xAE
 <LAO_ELLIPSIS> \xE0\xBA\xAF
@@ -3398,6 +3413,7 @@ CHARMAP
 <LAO_VOWEL_SIGN_YY> \xE0\xBA\xB7
 <LAO_VOWEL_SIGN_U> \xE0\xBA\xB8
 <LAO_VOWEL_SIGN_UU> \xE0\xBA\xB9
+<LAO_SIGN_PALI_VIRAMA> \xE0\xBA\xBA
 <LAO_VOWEL_SIGN_MAI_KON> \xE0\xBA\xBB
 <LAO_SEMIVOWEL_SIGN_LO> \xE0\xBA\xBC
 <LAO_SEMIVOWEL_SIGN_NYO> \xE0\xBA\xBD
@@ -6656,6 +6672,7 @@ CHARMAP
 <VEDIC_SIGN_ATIKRAMA> \xE1\xB3\xB7
 <VEDIC_TONE_RING_ABOVE> \xE1\xB3\xB8
 <VEDIC_TONE_DOUBLE_RING_ABOVE> \xE1\xB3\xB9
+<VEDIC_SIGN_DOUBLE_ANUSVARA_ANTARGOMUKHA> \xE1\xB3\xBA
 <LATIN_LETTER_SMALL_CAPITAL_A> \xE1\xB4\x80
 <LATIN_LETTER_SMALL_CAPITAL_AE> \xE1\xB4\x81
 <LATIN_SMALL_LETTER_TURNED_AE> \xE1\xB4\x82
@@ -10325,6 +10342,7 @@ CHARMAP
 <BLACK_MEDIUM_DOWN-POINTING_TRIANGLE_CENTRED> \xE2\xAF\x86
 <BLACK_MEDIUM_LEFT-POINTING_TRIANGLE_CENTRED> \xE2\xAF\x87
 <BLACK_MEDIUM_RIGHT-POINTING_TRIANGLE_CENTRED> \xE2\xAF\x88
+<NEPTUNE_FORM_TWO> \xE2\xAF\x89
 <TOP_HALF_BLACK_CIRCLE> \xE2\xAF\x8A
 <BOTTOM_HALF_BLACK_CIRCLE> \xE2\xAF\x8B
 <LIGHT_FOUR_POINTED_BLACK_CUSP> \xE2\xAF\x8C
@@ -10378,6 +10396,7 @@ CHARMAP
 <DOUBLED_SYMBOL> \xE2\xAF\xBC
 <PASSED_SYMBOL> \xE2\xAF\xBD
 <REVERSED_RIGHT_ANGLE> \xE2\xAF\xBE
+<HELLSCHREIBER_PAUSE_SYMBOL> \xE2\xAF\xBF
 <GLAGOLITIC_CAPITAL_LETTER_AZU> \xE2\xB0\x80
 <GLAGOLITIC_CAPITAL_LETTER_BUKY> \xE2\xB0\x81
 <GLAGOLITIC_CAPITAL_LETTER_VEDE> \xE2\xB0\x82
@@ -10916,6 +10935,7 @@ CHARMAP
 <MEDIEVAL_COMMA> \xE2\xB9\x8C
 <PARAGRAPHUS_MARK> \xE2\xB9\x8D
 <PUNCTUS_ELEVATUS_MARK> \xE2\xB9\x8E
+<CORNISH_VERSE_DIVIDER> \xE2\xB9\x8F
 <CJK_RADICAL_REPEAT> \xE2\xBA\x80
 <CJK_RADICAL_CLIFF> \xE2\xBA\x81
 <CJK_RADICAL_SECOND_ONE> \xE2\xBA\x82
@@ -41812,6 +41832,17 @@ CHARMAP
 <LATIN_SMALL_LETTER_OMEGA> \xEA\x9E\xB7
 <LATIN_CAPITAL_LETTER_U_WITH_STROKE> \xEA\x9E\xB8
 <LATIN_SMALL_LETTER_U_WITH_STROKE> \xEA\x9E\xB9
+<LATIN_CAPITAL_LETTER_GLOTTAL_A> \xEA\x9E\xBA
+<LATIN_SMALL_LETTER_GLOTTAL_A> \xEA\x9E\xBB
+<LATIN_CAPITAL_LETTER_GLOTTAL_I> \xEA\x9E\xBC
+<LATIN_SMALL_LETTER_GLOTTAL_I> \xEA\x9E\xBD
+<LATIN_CAPITAL_LETTER_GLOTTAL_U> \xEA\x9E\xBE
+<LATIN_SMALL_LETTER_GLOTTAL_U> \xEA\x9E\xBF
+<LATIN_CAPITAL_LETTER_ANGLICANA_W> \xEA\x9F\x82
+<LATIN_SMALL_LETTER_ANGLICANA_W> \xEA\x9F\x83
+<LATIN_CAPITAL_LETTER_C_WITH_PALATAL_HOOK> \xEA\x9F\x84
+<LATIN_CAPITAL_LETTER_S_WITH_HOOK> \xEA\x9F\x85
+<LATIN_CAPITAL_LETTER_Z_WITH_PALATAL_HOOK> \xEA\x9F\x86
 <LATIN_EPIGRAPHIC_LETTER_SIDEWAYS_I> \xEA\x9F\xB7
 <MODIFIER_LETTER_CAPITAL_H_WITH_STROKE> \xEA\x9F\xB8
 <MODIFIER_LETTER_SMALL_LIGATURE_OE> \xEA\x9F\xB9
@@ -42577,6 +42608,8 @@ CHARMAP
 <LATIN_SMALL_LETTER_UO> \xEA\xAD\xA3
 <LATIN_SMALL_LETTER_INVERTED_ALPHA> \xEA\xAD\xA4
 <GREEK_LETTER_SMALL_CAPITAL_OMEGA> \xEA\xAD\xA5
+<LATIN_SMALL_LETTER_DZ_DIGRAPH_WITH_RETROFLEX_HOOK> \xEA\xAD\xA6
+<LATIN_SMALL_LETTER_TS_DIGRAPH_WITH_RETROFLEX_HOOK> \xEA\xAD\xA7
 <CHEROKEE_SMALL_LETTER_A> \xEA\xAD\xB0
 <CHEROKEE_SMALL_LETTER_E> \xEA\xAD\xB1
 <CHEROKEE_SMALL_LETTER_I> \xEA\xAD\xB2
@@ -64381,6 +64414,29 @@ CHARMAP
 <SOGDIAN_PUNCTUATION_CIRCLE_WITH_DOT> \xF0\x90\xBD\x97
 <SOGDIAN_PUNCTUATION_TWO_CIRCLES_WITH_DOTS> \xF0\x90\xBD\x98
 <SOGDIAN_PUNCTUATION_HALF_CIRCLE_WITH_DOT> \xF0\x90\xBD\x99
+<ELYMAIC_LETTER_ALEPH> \xF0\x90\xBF\xA0
+<ELYMAIC_LETTER_BETH> \xF0\x90\xBF\xA1
+<ELYMAIC_LETTER_GIMEL> \xF0\x90\xBF\xA2
+<ELYMAIC_LETTER_DALETH> \xF0\x90\xBF\xA3
+<ELYMAIC_LETTER_HE> \xF0\x90\xBF\xA4
+<ELYMAIC_LETTER_WAW> \xF0\x90\xBF\xA5
+<ELYMAIC_LETTER_ZAYIN> \xF0\x90\xBF\xA6
+<ELYMAIC_LETTER_HETH> \xF0\x90\xBF\xA7
+<ELYMAIC_LETTER_TETH> \xF0\x90\xBF\xA8
+<ELYMAIC_LETTER_YODH> \xF0\x90\xBF\xA9
+<ELYMAIC_LETTER_KAPH> \xF0\x90\xBF\xAA
+<ELYMAIC_LETTER_LAMEDH> \xF0\x90\xBF\xAB
+<ELYMAIC_LETTER_MEM> \xF0\x90\xBF\xAC
+<ELYMAIC_LETTER_NUN> \xF0\x90\xBF\xAD
+<ELYMAIC_LETTER_SAMEKH> \xF0\x90\xBF\xAE
+<ELYMAIC_LETTER_AYIN> \xF0\x90\xBF\xAF
+<ELYMAIC_LETTER_PE> \xF0\x90\xBF\xB0
+<ELYMAIC_LETTER_SADHE> \xF0\x90\xBF\xB1
+<ELYMAIC_LETTER_QOPH> \xF0\x90\xBF\xB2
+<ELYMAIC_LETTER_RESH> \xF0\x90\xBF\xB3
+<ELYMAIC_LETTER_SHIN> \xF0\x90\xBF\xB4
+<ELYMAIC_LETTER_TAW> \xF0\x90\xBF\xB5
+<ELYMAIC_LIGATURE_ZAYIN-YODH> \xF0\x90\xBF\xB6
 <BRAHMI_SIGN_CANDRABINDU> \xF0\x91\x80\x80
 <BRAHMI_SIGN_ANUSVARA> \xF0\x91\x80\x81
 <BRAHMI_SIGN_VISARGA> \xF0\x91\x80\x82
@@ -65163,6 +65219,7 @@ CHARMAP
 <NEWA_PLACEHOLDER_MARK> \xF0\x91\x91\x9B
 <NEWA_INSERTION_SIGN> \xF0\x91\x91\x9D
 <NEWA_SANDHI_MARK> \xF0\x91\x91\x9E
+<NEWA_LETTER_VEDIC_ANUSVARA> \xF0\x91\x91\x9F
 <TIRHUTA_ANJI> \xF0\x91\x92\x80
 <TIRHUTA_LETTER_A> \xF0\x91\x92\x81
 <TIRHUTA_LETTER_AA> \xF0\x91\x92\x82
@@ -65485,6 +65542,7 @@ CHARMAP
 <TAKRI_VOWEL_SIGN_AU> \xF0\x91\x9A\xB5
 <TAKRI_SIGN_VIRAMA> \xF0\x91\x9A\xB6
 <TAKRI_SIGN_NUKTA> \xF0\x91\x9A\xB7
+<TAKRI_LETTER_ARCHAIC_KHA> \xF0\x91\x9A\xB8
 <TAKRI_DIGIT_ZERO> \xF0\x91\x9B\x80
 <TAKRI_DIGIT_ONE> \xF0\x91\x9B\x81
 <TAKRI_DIGIT_TWO> \xF0\x91\x9B\x82
@@ -65697,6 +65755,71 @@ CHARMAP
 <WARANG_CITI_NUMBER_EIGHTY> \xF0\x91\xA3\xB1
 <WARANG_CITI_NUMBER_NINETY> \xF0\x91\xA3\xB2
 <WARANG_CITI_OM> \xF0\x91\xA3\xBF
+<NANDINAGARI_LETTER_A> \xF0\x91\xA6\xA0
+<NANDINAGARI_LETTER_AA> \xF0\x91\xA6\xA1
+<NANDINAGARI_LETTER_I> \xF0\x91\xA6\xA2
+<NANDINAGARI_LETTER_II> \xF0\x91\xA6\xA3
+<NANDINAGARI_LETTER_U> \xF0\x91\xA6\xA4
+<NANDINAGARI_LETTER_UU> \xF0\x91\xA6\xA5
+<NANDINAGARI_LETTER_VOCALIC_R> \xF0\x91\xA6\xA6
+<NANDINAGARI_LETTER_VOCALIC_RR> \xF0\x91\xA6\xA7
+<NANDINAGARI_LETTER_E> \xF0\x91\xA6\xAA
+<NANDINAGARI_LETTER_AI> \xF0\x91\xA6\xAB
+<NANDINAGARI_LETTER_O> \xF0\x91\xA6\xAC
+<NANDINAGARI_LETTER_AU> \xF0\x91\xA6\xAD
+<NANDINAGARI_LETTER_KA> \xF0\x91\xA6\xAE
+<NANDINAGARI_LETTER_KHA> \xF0\x91\xA6\xAF
+<NANDINAGARI_LETTER_GA> \xF0\x91\xA6\xB0
+<NANDINAGARI_LETTER_GHA> \xF0\x91\xA6\xB1
+<NANDINAGARI_LETTER_NGA> \xF0\x91\xA6\xB2
+<NANDINAGARI_LETTER_CA> \xF0\x91\xA6\xB3
+<NANDINAGARI_LETTER_CHA> \xF0\x91\xA6\xB4
+<NANDINAGARI_LETTER_JA> \xF0\x91\xA6\xB5
+<NANDINAGARI_LETTER_JHA> \xF0\x91\xA6\xB6
+<NANDINAGARI_LETTER_NYA> \xF0\x91\xA6\xB7
+<NANDINAGARI_LETTER_TTA> \xF0\x91\xA6\xB8
+<NANDINAGARI_LETTER_TTHA> \xF0\x91\xA6\xB9
+<NANDINAGARI_LETTER_DDA> \xF0\x91\xA6\xBA
+<NANDINAGARI_LETTER_DDHA> \xF0\x91\xA6\xBB
+<NANDINAGARI_LETTER_NNA> \xF0\x91\xA6\xBC
+<NANDINAGARI_LETTER_TA> \xF0\x91\xA6\xBD
+<NANDINAGARI_LETTER_THA> \xF0\x91\xA6\xBE
+<NANDINAGARI_LETTER_DA> \xF0\x91\xA6\xBF
+<NANDINAGARI_LETTER_DHA> \xF0\x91\xA7\x80
+<NANDINAGARI_LETTER_NA> \xF0\x91\xA7\x81
+<NANDINAGARI_LETTER_PA> \xF0\x91\xA7\x82
+<NANDINAGARI_LETTER_PHA> \xF0\x91\xA7\x83
+<NANDINAGARI_LETTER_BA> \xF0\x91\xA7\x84
+<NANDINAGARI_LETTER_BHA> \xF0\x91\xA7\x85
+<NANDINAGARI_LETTER_MA> \xF0\x91\xA7\x86
+<NANDINAGARI_LETTER_YA> \xF0\x91\xA7\x87
+<NANDINAGARI_LETTER_RA> \xF0\x91\xA7\x88
+<NANDINAGARI_LETTER_LA> \xF0\x91\xA7\x89
+<NANDINAGARI_LETTER_VA> \xF0\x91\xA7\x8A
+<NANDINAGARI_LETTER_SHA> \xF0\x91\xA7\x8B
+<NANDINAGARI_LETTER_SSA> \xF0\x91\xA7\x8C
+<NANDINAGARI_LETTER_SA> \xF0\x91\xA7\x8D
+<NANDINAGARI_LETTER_HA> \xF0\x91\xA7\x8E
+<NANDINAGARI_LETTER_LLA> \xF0\x91\xA7\x8F
+<NANDINAGARI_LETTER_RRA> \xF0\x91\xA7\x90
+<NANDINAGARI_VOWEL_SIGN_AA> \xF0\x91\xA7\x91
+<NANDINAGARI_VOWEL_SIGN_I> \xF0\x91\xA7\x92
+<NANDINAGARI_VOWEL_SIGN_II> \xF0\x91\xA7\x93
+<NANDINAGARI_VOWEL_SIGN_U> \xF0\x91\xA7\x94
+<NANDINAGARI_VOWEL_SIGN_UU> \xF0\x91\xA7\x95
+<NANDINAGARI_VOWEL_SIGN_VOCALIC_R> \xF0\x91\xA7\x96
+<NANDINAGARI_VOWEL_SIGN_VOCALIC_RR> \xF0\x91\xA7\x97
+<NANDINAGARI_VOWEL_SIGN_E> \xF0\x91\xA7\x9A
+<NANDINAGARI_VOWEL_SIGN_AI> \xF0\x91\xA7\x9B
+<NANDINAGARI_VOWEL_SIGN_O> \xF0\x91\xA7\x9C
+<NANDINAGARI_VOWEL_SIGN_AU> \xF0\x91\xA7\x9D
+<NANDINAGARI_SIGN_ANUSVARA> \xF0\x91\xA7\x9E
+<NANDINAGARI_SIGN_VISARGA> \xF0\x91\xA7\x9F
+<NANDINAGARI_SIGN_VIRAMA> \xF0\x91\xA7\xA0
+<NANDINAGARI_SIGN_AVAGRAHA> \xF0\x91\xA7\xA1
+<NANDINAGARI_SIGN_SIDDHAM> \xF0\x91\xA7\xA2
+<NANDINAGARI_HEADSTROKE> \xF0\x91\xA7\xA3
+<NANDINAGARI_VOWEL_SIGN_PRISHTHAMATRA_E> \xF0\x91\xA7\xA4
 <ZANABAZAR_SQUARE_LETTER_A> \xF0\x91\xA8\x80
 <ZANABAZAR_SQUARE_VOWEL_SIGN_I> \xF0\x91\xA8\x81
 <ZANABAZAR_SQUARE_VOWEL_SIGN_UE> \xF0\x91\xA8\x82
@@ -65821,6 +65944,8 @@ CHARMAP
 <SOYOMBO_LETTER_SA> \xF0\x91\xAA\x81
 <SOYOMBO_LETTER_HA> \xF0\x91\xAA\x82
 <SOYOMBO_LETTER_KSSA> \xF0\x91\xAA\x83
+<SOYOMBO_SIGN_JIHVAMULIYA> \xF0\x91\xAA\x84
+<SOYOMBO_SIGN_UPADHMANIYA> \xF0\x91\xAA\x85
 <SOYOMBO_CLUSTER-INITIAL_LETTER_RA> \xF0\x91\xAA\x86
 <SOYOMBO_CLUSTER-INITIAL_LETTER_LA> \xF0\x91\xAA\x87
 <SOYOMBO_CLUSTER-INITIAL_LETTER_SHA> \xF0\x91\xAA\x88
@@ -66235,6 +66360,57 @@ CHARMAP
 <MAKASAR_VOWEL_SIGN_O> \xF0\x91\xBB\xB6
 <MAKASAR_PASSIMBANG> \xF0\x91\xBB\xB7
 <MAKASAR_END_OF_SECTION> \xF0\x91\xBB\xB8
+<TAMIL_FRACTION_ONE_THREE-HUNDRED-AND-TWENTIETH> \xF0\x91\xBF\x80
+<TAMIL_FRACTION_ONE_ONE-HUNDRED-AND-SIXTIETH> \xF0\x91\xBF\x81
+<TAMIL_FRACTION_ONE_EIGHTIETH> \xF0\x91\xBF\x82
+<TAMIL_FRACTION_ONE_SIXTY-FOURTH> \xF0\x91\xBF\x83
+<TAMIL_FRACTION_ONE_FORTIETH> \xF0\x91\xBF\x84
+<TAMIL_FRACTION_ONE_THIRTY-SECOND> \xF0\x91\xBF\x85
+<TAMIL_FRACTION_THREE_EIGHTIETHS> \xF0\x91\xBF\x86
+<TAMIL_FRACTION_THREE_SIXTY-FOURTHS> \xF0\x91\xBF\x87
+<TAMIL_FRACTION_ONE_TWENTIETH> \xF0\x91\xBF\x88
+<TAMIL_FRACTION_ONE_SIXTEENTH-1> \xF0\x91\xBF\x89
+<TAMIL_FRACTION_ONE_SIXTEENTH-2> \xF0\x91\xBF\x8A
+<TAMIL_FRACTION_ONE_TENTH> \xF0\x91\xBF\x8B
+<TAMIL_FRACTION_ONE_EIGHTH> \xF0\x91\xBF\x8C
+<TAMIL_FRACTION_THREE_TWENTIETHS> \xF0\x91\xBF\x8D
+<TAMIL_FRACTION_THREE_SIXTEENTHS> \xF0\x91\xBF\x8E
+<TAMIL_FRACTION_ONE_FIFTH> \xF0\x91\xBF\x8F
+<TAMIL_FRACTION_ONE_QUARTER> \xF0\x91\xBF\x90
+<TAMIL_FRACTION_ONE_HALF-1> \xF0\x91\xBF\x91
+<TAMIL_FRACTION_ONE_HALF-2> \xF0\x91\xBF\x92
+<TAMIL_FRACTION_THREE_QUARTERS> \xF0\x91\xBF\x93
+<TAMIL_FRACTION_DOWNSCALING_FACTOR_KIIZH> \xF0\x91\xBF\x94
+<TAMIL_SIGN_NEL> \xF0\x91\xBF\x95
+<TAMIL_SIGN_CEVITU> \xF0\x91\xBF\x96
+<TAMIL_SIGN_AAZHAAKKU> \xF0\x91\xBF\x97
+<TAMIL_SIGN_UZHAKKU> \xF0\x91\xBF\x98
+<TAMIL_SIGN_MUUVUZHAKKU> \xF0\x91\xBF\x99
+<TAMIL_SIGN_KURUNI> \xF0\x91\xBF\x9A
+<TAMIL_SIGN_PATHAKKU> \xF0\x91\xBF\x9B
+<TAMIL_SIGN_MUKKURUNI> \xF0\x91\xBF\x9C
+<TAMIL_SIGN_KAACU> \xF0\x91\xBF\x9D
+<TAMIL_SIGN_PANAM> \xF0\x91\xBF\x9E
+<TAMIL_SIGN_PON> \xF0\x91\xBF\x9F
+<TAMIL_SIGN_VARAAKAN> \xF0\x91\xBF\xA0
+<TAMIL_SIGN_PAARAM> \xF0\x91\xBF\xA1
+<TAMIL_SIGN_KUZHI> \xF0\x91\xBF\xA2
+<TAMIL_SIGN_VELI> \xF0\x91\xBF\xA3
+<TAMIL_WET_CULTIVATION_SIGN> \xF0\x91\xBF\xA4
+<TAMIL_DRY_CULTIVATION_SIGN> \xF0\x91\xBF\xA5
+<TAMIL_LAND_SIGN> \xF0\x91\xBF\xA6
+<TAMIL_SALT_PAN_SIGN> \xF0\x91\xBF\xA7
+<TAMIL_TRADITIONAL_CREDIT_SIGN> \xF0\x91\xBF\xA8
+<TAMIL_TRADITIONAL_NUMBER_SIGN> \xF0\x91\xBF\xA9
+<TAMIL_CURRENT_SIGN> \xF0\x91\xBF\xAA
+<TAMIL_AND_ODD_SIGN> \xF0\x91\xBF\xAB
+<TAMIL_SPENT_SIGN> \xF0\x91\xBF\xAC
+<TAMIL_TOTAL_SIGN> \xF0\x91\xBF\xAD
+<TAMIL_IN_POSSESSION_SIGN> \xF0\x91\xBF\xAE
+<TAMIL_STARTING_FROM_SIGN> \xF0\x91\xBF\xAF
+<TAMIL_SIGN_MUTHALIYA> \xF0\x91\xBF\xB0
+<TAMIL_SIGN_VAKAIYARAA> \xF0\x91\xBF\xB1
+<TAMIL_PUNCTUATION_END_OF_TEXT> \xF0\x91\xBF\xBF
 <CUNEIFORM_SIGN_A> \xF0\x92\x80\x80
 <CUNEIFORM_SIGN_A_TIMES_A> \xF0\x92\x80\x81
 <CUNEIFORM_SIGN_A_TIMES_BAD> \xF0\x92\x80\x82
@@ -68540,6 +68716,15 @@ CHARMAP
 <EGYPTIAN_HIEROGLYPH_AA030> \xF0\x93\x90\xAC
 <EGYPTIAN_HIEROGLYPH_AA031> \xF0\x93\x90\xAD
 <EGYPTIAN_HIEROGLYPH_AA032> \xF0\x93\x90\xAE
+<EGYPTIAN_HIEROGLYPH_VERTICAL_JOINER> \xF0\x93\x90\xB0
+<EGYPTIAN_HIEROGLYPH_HORIZONTAL_JOINER> \xF0\x93\x90\xB1
+<EGYPTIAN_HIEROGLYPH_INSERT_AT_TOP_START> \xF0\x93\x90\xB2
+<EGYPTIAN_HIEROGLYPH_INSERT_AT_BOTTOM_START> \xF0\x93\x90\xB3
+<EGYPTIAN_HIEROGLYPH_INSERT_AT_TOP_END> \xF0\x93\x90\xB4
+<EGYPTIAN_HIEROGLYPH_INSERT_AT_BOTTOM_END> \xF0\x93\x90\xB5
+<EGYPTIAN_HIEROGLYPH_OVERLAY_MIDDLE> \xF0\x93\x90\xB6
+<EGYPTIAN_HIEROGLYPH_BEGIN_SEGMENT> \xF0\x93\x90\xB7
+<EGYPTIAN_HIEROGLYPH_END_SEGMENT> \xF0\x93\x90\xB8
 <ANATOLIAN_HIEROGLYPH_A001> \xF0\x94\x90\x80
 <ANATOLIAN_HIEROGLYPH_A002> \xF0\x94\x90\x81
 <ANATOLIAN_HIEROGLYPH_A003> \xF0\x94\x90\x82
@@ -70058,6 +70243,13 @@ CHARMAP
 <MIAO_LETTER_WA> \xF0\x96\xBD\x82
 <MIAO_LETTER_AH> \xF0\x96\xBD\x83
 <MIAO_LETTER_HHA> \xF0\x96\xBD\x84
+<MIAO_LETTER_BRI> \xF0\x96\xBD\x85
+<MIAO_LETTER_SYI> \xF0\x96\xBD\x86
+<MIAO_LETTER_DZYI> \xF0\x96\xBD\x87
+<MIAO_LETTER_TE> \xF0\x96\xBD\x88
+<MIAO_LETTER_TSE> \xF0\x96\xBD\x89
+<MIAO_LETTER_RTE> \xF0\x96\xBD\x8A
+<MIAO_SIGN_CONSONANT_MODIFIER_BAR> \xF0\x96\xBD\x8F
 <MIAO_LETTER_NASALIZATION> \xF0\x96\xBD\x90
 <MIAO_SIGN_ASPIRATION> \xF0\x96\xBD\x91
 <MIAO_SIGN_REFORMED_VOICING> \xF0\x96\xBD\x92
@@ -70105,6 +70297,15 @@ CHARMAP
 <MIAO_VOWEL_SIGN_OU> \xF0\x96\xBD\xBC
 <MIAO_VOWEL_SIGN_N> \xF0\x96\xBD\xBD
 <MIAO_VOWEL_SIGN_NG> \xF0\x96\xBD\xBE
+<MIAO_VOWEL_SIGN_UOG> \xF0\x96\xBD\xBF
+<MIAO_VOWEL_SIGN_YUI> \xF0\x96\xBE\x80
+<MIAO_VOWEL_SIGN_OG> \xF0\x96\xBE\x81
+<MIAO_VOWEL_SIGN_OER> \xF0\x96\xBE\x82
+<MIAO_VOWEL_SIGN_VW> \xF0\x96\xBE\x83
+<MIAO_VOWEL_SIGN_IG> \xF0\x96\xBE\x84
+<MIAO_VOWEL_SIGN_EA> \xF0\x96\xBE\x85
+<MIAO_VOWEL_SIGN_IONG> \xF0\x96\xBE\x86
+<MIAO_VOWEL_SIGN_UI> \xF0\x96\xBE\x87
 <MIAO_TONE_RIGHT> \xF0\x96\xBE\x8F
 <MIAO_TONE_TOP_RIGHT> \xF0\x96\xBE\x90
 <MIAO_TONE_ABOVE> \xF0\x96\xBE\x91
@@ -70124,6 +70325,8 @@ CHARMAP
 <MIAO_LETTER_REFORMED_TONE-8> \xF0\x96\xBE\x9F
 <TANGUT_ITERATION_MARK> \xF0\x96\xBF\xA0
 <NUSHU_ITERATION_MARK> \xF0\x96\xBF\xA1
+<OLD_CHINESE_HOOK_MARK> \xF0\x96\xBF\xA2
+<OLD_CHINESE_ITERATION_MARK> \xF0\x96\xBF\xA3
 <TANGUT_IDEOGRAPH-17000> \xF0\x97\x80\x80
 <TANGUT_IDEOGRAPH-17001> \xF0\x97\x80\x81
 <TANGUT_IDEOGRAPH-17002> \xF0\x97\x80\x82
@@ -76254,6 +76457,12 @@ CHARMAP
 <TANGUT_IDEOGRAPH-187EF> \xF0\x98\x9F\xAF
 <TANGUT_IDEOGRAPH-187F0> \xF0\x98\x9F\xB0
 <TANGUT_IDEOGRAPH-187F1> \xF0\x98\x9F\xB1
+<TANGUT_IDEOGRAPH-187F2> \xF0\x98\x9F\xB2
+<TANGUT_IDEOGRAPH-187F3> \xF0\x98\x9F\xB3
+<TANGUT_IDEOGRAPH-187F4> \xF0\x98\x9F\xB4
+<TANGUT_IDEOGRAPH-187F5> \xF0\x98\x9F\xB5
+<TANGUT_IDEOGRAPH-187F6> \xF0\x98\x9F\xB6
+<TANGUT_IDEOGRAPH-187F7> \xF0\x98\x9F\xB7
 <TANGUT_COMPONENT-001> \xF0\x98\xA0\x80
 <TANGUT_COMPONENT-002> \xF0\x98\xA0\x81
 <TANGUT_COMPONENT-003> \xF0\x98\xA0\x82
@@ -77296,6 +77505,13 @@ CHARMAP
 <HENTAIGANA_LETTER_WO-7> \xF0\x9B\x84\x9C
 <HENTAIGANA_LETTER_N-MU-MO-1> \xF0\x9B\x84\x9D
 <HENTAIGANA_LETTER_N-MU-MO-2> \xF0\x9B\x84\x9E
+<HIRAGANA_LETTER_SMALL_WI> \xF0\x9B\x85\x90
+<HIRAGANA_LETTER_SMALL_WE> \xF0\x9B\x85\x91
+<HIRAGANA_LETTER_SMALL_WO> \xF0\x9B\x85\x92
+<KATAKANA_LETTER_SMALL_WI> \xF0\x9B\x85\xA4
+<KATAKANA_LETTER_SMALL_WE> \xF0\x9B\x85\xA5
+<KATAKANA_LETTER_SMALL_WO> \xF0\x9B\x85\xA6
+<KATAKANA_LETTER_SMALL_N> \xF0\x9B\x85\xA7
 <NUSHU_CHARACTER-1B170> \xF0\x9B\x85\xB0
 <NUSHU_CHARACTER-1B171> \xF0\x9B\x85\xB1
 <NUSHU_CHARACTER-1B172> \xF0\x9B\x85\xB2
@@ -80224,6 +80440,136 @@ CHARMAP
 <COMBINING_GLAGOLITIC_LETTER_BIG_YUS> \xF0\x9E\x80\xA8
 <COMBINING_GLAGOLITIC_LETTER_IOTATED_BIG_YUS> \xF0\x9E\x80\xA9
 <COMBINING_GLAGOLITIC_LETTER_FITA> \xF0\x9E\x80\xAA
+<NYIAKENG_PUACHUE_HMONG_LETTER_MA> \xF0\x9E\x84\x80
+<NYIAKENG_PUACHUE_HMONG_LETTER_TSA> \xF0\x9E\x84\x81
+<NYIAKENG_PUACHUE_HMONG_LETTER_NTA> \xF0\x9E\x84\x82
+<NYIAKENG_PUACHUE_HMONG_LETTER_TA> \xF0\x9E\x84\x83
+<NYIAKENG_PUACHUE_HMONG_LETTER_HA> \xF0\x9E\x84\x84
+<NYIAKENG_PUACHUE_HMONG_LETTER_NA> \xF0\x9E\x84\x85
+<NYIAKENG_PUACHUE_HMONG_LETTER_XA> \xF0\x9E\x84\x86
+<NYIAKENG_PUACHUE_HMONG_LETTER_NKA> \xF0\x9E\x84\x87
+<NYIAKENG_PUACHUE_HMONG_LETTER_CA> \xF0\x9E\x84\x88
+<NYIAKENG_PUACHUE_HMONG_LETTER_LA> \xF0\x9E\x84\x89
+<NYIAKENG_PUACHUE_HMONG_LETTER_SA> \xF0\x9E\x84\x8A
+<NYIAKENG_PUACHUE_HMONG_LETTER_ZA> \xF0\x9E\x84\x8B
+<NYIAKENG_PUACHUE_HMONG_LETTER_NCA> \xF0\x9E\x84\x8C
+<NYIAKENG_PUACHUE_HMONG_LETTER_NTSA> \xF0\x9E\x84\x8D
+<NYIAKENG_PUACHUE_HMONG_LETTER_KA> \xF0\x9E\x84\x8E
+<NYIAKENG_PUACHUE_HMONG_LETTER_DA> \xF0\x9E\x84\x8F
+<NYIAKENG_PUACHUE_HMONG_LETTER_NYA> \xF0\x9E\x84\x90
+<NYIAKENG_PUACHUE_HMONG_LETTER_NRA> \xF0\x9E\x84\x91
+<NYIAKENG_PUACHUE_HMONG_LETTER_VA> \xF0\x9E\x84\x92
+<NYIAKENG_PUACHUE_HMONG_LETTER_NTXA> \xF0\x9E\x84\x93
+<NYIAKENG_PUACHUE_HMONG_LETTER_TXA> \xF0\x9E\x84\x94
+<NYIAKENG_PUACHUE_HMONG_LETTER_FA> \xF0\x9E\x84\x95
+<NYIAKENG_PUACHUE_HMONG_LETTER_RA> \xF0\x9E\x84\x96
+<NYIAKENG_PUACHUE_HMONG_LETTER_QA> \xF0\x9E\x84\x97
+<NYIAKENG_PUACHUE_HMONG_LETTER_YA> \xF0\x9E\x84\x98
+<NYIAKENG_PUACHUE_HMONG_LETTER_NQA> \xF0\x9E\x84\x99
+<NYIAKENG_PUACHUE_HMONG_LETTER_PA> \xF0\x9E\x84\x9A
+<NYIAKENG_PUACHUE_HMONG_LETTER_XYA> \xF0\x9E\x84\x9B
+<NYIAKENG_PUACHUE_HMONG_LETTER_NPA> \xF0\x9E\x84\x9C
+<NYIAKENG_PUACHUE_HMONG_LETTER_DLA> \xF0\x9E\x84\x9D
+<NYIAKENG_PUACHUE_HMONG_LETTER_NPLA> \xF0\x9E\x84\x9E
+<NYIAKENG_PUACHUE_HMONG_LETTER_HAH> \xF0\x9E\x84\x9F
+<NYIAKENG_PUACHUE_HMONG_LETTER_MLA> \xF0\x9E\x84\xA0
+<NYIAKENG_PUACHUE_HMONG_LETTER_PLA> \xF0\x9E\x84\xA1
+<NYIAKENG_PUACHUE_HMONG_LETTER_GA> \xF0\x9E\x84\xA2
+<NYIAKENG_PUACHUE_HMONG_LETTER_RRA> \xF0\x9E\x84\xA3
+<NYIAKENG_PUACHUE_HMONG_LETTER_A> \xF0\x9E\x84\xA4
+<NYIAKENG_PUACHUE_HMONG_LETTER_AA> \xF0\x9E\x84\xA5
+<NYIAKENG_PUACHUE_HMONG_LETTER_I> \xF0\x9E\x84\xA6
+<NYIAKENG_PUACHUE_HMONG_LETTER_U> \xF0\x9E\x84\xA7
+<NYIAKENG_PUACHUE_HMONG_LETTER_O> \xF0\x9E\x84\xA8
+<NYIAKENG_PUACHUE_HMONG_LETTER_OO> \xF0\x9E\x84\xA9
+<NYIAKENG_PUACHUE_HMONG_LETTER_E> \xF0\x9E\x84\xAA
+<NYIAKENG_PUACHUE_HMONG_LETTER_EE> \xF0\x9E\x84\xAB
+<NYIAKENG_PUACHUE_HMONG_LETTER_W> \xF0\x9E\x84\xAC
+<NYIAKENG_PUACHUE_HMONG_TONE-B> \xF0\x9E\x84\xB0
+<NYIAKENG_PUACHUE_HMONG_TONE-M> \xF0\x9E\x84\xB1
+<NYIAKENG_PUACHUE_HMONG_TONE-J> \xF0\x9E\x84\xB2
+<NYIAKENG_PUACHUE_HMONG_TONE-V> \xF0\x9E\x84\xB3
+<NYIAKENG_PUACHUE_HMONG_TONE-S> \xF0\x9E\x84\xB4
+<NYIAKENG_PUACHUE_HMONG_TONE-G> \xF0\x9E\x84\xB5
+<NYIAKENG_PUACHUE_HMONG_TONE-D> \xF0\x9E\x84\xB6
+<NYIAKENG_PUACHUE_HMONG_SIGN_FOR_PERSON> \xF0\x9E\x84\xB7
+<NYIAKENG_PUACHUE_HMONG_SIGN_FOR_THING> \xF0\x9E\x84\xB8
+<NYIAKENG_PUACHUE_HMONG_SIGN_FOR_LOCATION> \xF0\x9E\x84\xB9
+<NYIAKENG_PUACHUE_HMONG_SIGN_FOR_ANIMAL> \xF0\x9E\x84\xBA
+<NYIAKENG_PUACHUE_HMONG_SIGN_FOR_INVERTEBRATE> \xF0\x9E\x84\xBB
+<NYIAKENG_PUACHUE_HMONG_SIGN_XW_XW> \xF0\x9E\x84\xBC
+<NYIAKENG_PUACHUE_HMONG_SYLLABLE_LENGTHENER> \xF0\x9E\x84\xBD
+<NYIAKENG_PUACHUE_HMONG_DIGIT_ZERO> \xF0\x9E\x85\x80
+<NYIAKENG_PUACHUE_HMONG_DIGIT_ONE> \xF0\x9E\x85\x81
+<NYIAKENG_PUACHUE_HMONG_DIGIT_TWO> \xF0\x9E\x85\x82
+<NYIAKENG_PUACHUE_HMONG_DIGIT_THREE> \xF0\x9E\x85\x83
+<NYIAKENG_PUACHUE_HMONG_DIGIT_FOUR> \xF0\x9E\x85\x84
+<NYIAKENG_PUACHUE_HMONG_DIGIT_FIVE> \xF0\x9E\x85\x85
+<NYIAKENG_PUACHUE_HMONG_DIGIT_SIX> \xF0\x9E\x85\x86
+<NYIAKENG_PUACHUE_HMONG_DIGIT_SEVEN> \xF0\x9E\x85\x87
+<NYIAKENG_PUACHUE_HMONG_DIGIT_EIGHT> \xF0\x9E\x85\x88
+<NYIAKENG_PUACHUE_HMONG_DIGIT_NINE> \xF0\x9E\x85\x89
+<NYIAKENG_PUACHUE_HMONG_LOGOGRAM_NYAJ> \xF0\x9E\x85\x8E
+<NYIAKENG_PUACHUE_HMONG_CIRCLED_CA> \xF0\x9E\x85\x8F
+<WANCHO_LETTER_AA> \xF0\x9E\x8B\x80
+<WANCHO_LETTER_A> \xF0\x9E\x8B\x81
+<WANCHO_LETTER_BA> \xF0\x9E\x8B\x82
+<WANCHO_LETTER_CA> \xF0\x9E\x8B\x83
+<WANCHO_LETTER_DA> \xF0\x9E\x8B\x84
+<WANCHO_LETTER_GA> \xF0\x9E\x8B\x85
+<WANCHO_LETTER_YA> \xF0\x9E\x8B\x86
+<WANCHO_LETTER_PHA> \xF0\x9E\x8B\x87
+<WANCHO_LETTER_LA> \xF0\x9E\x8B\x88
+<WANCHO_LETTER_NA> \xF0\x9E\x8B\x89
+<WANCHO_LETTER_PA> \xF0\x9E\x8B\x8A
+<WANCHO_LETTER_TA> \xF0\x9E\x8B\x8B
+<WANCHO_LETTER_THA> \xF0\x9E\x8B\x8C
+<WANCHO_LETTER_FA> \xF0\x9E\x8B\x8D
+<WANCHO_LETTER_SA> \xF0\x9E\x8B\x8E
+<WANCHO_LETTER_SHA> \xF0\x9E\x8B\x8F
+<WANCHO_LETTER_JA> \xF0\x9E\x8B\x90
+<WANCHO_LETTER_ZA> \xF0\x9E\x8B\x91
+<WANCHO_LETTER_WA> \xF0\x9E\x8B\x92
+<WANCHO_LETTER_VA> \xF0\x9E\x8B\x93
+<WANCHO_LETTER_KA> \xF0\x9E\x8B\x94
+<WANCHO_LETTER_O> \xF0\x9E\x8B\x95
+<WANCHO_LETTER_AU> \xF0\x9E\x8B\x96
+<WANCHO_LETTER_RA> \xF0\x9E\x8B\x97
+<WANCHO_LETTER_MA> \xF0\x9E\x8B\x98
+<WANCHO_LETTER_KHA> \xF0\x9E\x8B\x99
+<WANCHO_LETTER_HA> \xF0\x9E\x8B\x9A
+<WANCHO_LETTER_E> \xF0\x9E\x8B\x9B
+<WANCHO_LETTER_I> \xF0\x9E\x8B\x9C
+<WANCHO_LETTER_NGA> \xF0\x9E\x8B\x9D
+<WANCHO_LETTER_U> \xF0\x9E\x8B\x9E
+<WANCHO_LETTER_LLHA> \xF0\x9E\x8B\x9F
+<WANCHO_LETTER_TSA> \xF0\x9E\x8B\xA0
+<WANCHO_LETTER_TRA> \xF0\x9E\x8B\xA1
+<WANCHO_LETTER_ONG> \xF0\x9E\x8B\xA2
+<WANCHO_LETTER_AANG> \xF0\x9E\x8B\xA3
+<WANCHO_LETTER_ANG> \xF0\x9E\x8B\xA4
+<WANCHO_LETTER_ING> \xF0\x9E\x8B\xA5
+<WANCHO_LETTER_ON> \xF0\x9E\x8B\xA6
+<WANCHO_LETTER_EN> \xF0\x9E\x8B\xA7
+<WANCHO_LETTER_AAN> \xF0\x9E\x8B\xA8
+<WANCHO_LETTER_NYA> \xF0\x9E\x8B\xA9
+<WANCHO_LETTER_UEN> \xF0\x9E\x8B\xAA
+<WANCHO_LETTER_YIH> \xF0\x9E\x8B\xAB
+<WANCHO_TONE_TUP> \xF0\x9E\x8B\xAC
+<WANCHO_TONE_TUPNI> \xF0\x9E\x8B\xAD
+<WANCHO_TONE_KOI> \xF0\x9E\x8B\xAE
+<WANCHO_TONE_KOINI> \xF0\x9E\x8B\xAF
+<WANCHO_DIGIT_ZERO> \xF0\x9E\x8B\xB0
+<WANCHO_DIGIT_ONE> \xF0\x9E\x8B\xB1
+<WANCHO_DIGIT_TWO> \xF0\x9E\x8B\xB2
+<WANCHO_DIGIT_THREE> \xF0\x9E\x8B\xB3
+<WANCHO_DIGIT_FOUR> \xF0\x9E\x8B\xB4
+<WANCHO_DIGIT_FIVE> \xF0\x9E\x8B\xB5
+<WANCHO_DIGIT_SIX> \xF0\x9E\x8B\xB6
+<WANCHO_DIGIT_SEVEN> \xF0\x9E\x8B\xB7
+<WANCHO_DIGIT_EIGHT> \xF0\x9E\x8B\xB8
+<WANCHO_DIGIT_NINE> \xF0\x9E\x8B\xB9
+<WANCHO_NGUN_SIGN> \xF0\x9E\x8B\xBF
 <MENDE_KIKAKUI_SYLLABLE_M001_KI> \xF0\x9E\xA0\x80
 <MENDE_KIKAKUI_SYLLABLE_M002_KA> \xF0\x9E\xA0\x81
 <MENDE_KIKAKUI_SYLLABLE_M003_KU> \xF0\x9E\xA0\x82
@@ -80512,6 +80858,7 @@ CHARMAP
 <ADLAM_CONSONANT_MODIFIER> \xF0\x9E\xA5\x88
 <ADLAM_GEMINATE_CONSONANT_MODIFIER> \xF0\x9E\xA5\x89
 <ADLAM_NUKTA> \xF0\x9E\xA5\x8A
+<ADLAM_NASALIZATION_MARK> \xF0\x9E\xA5\x8B
 <ADLAM_DIGIT_ZERO> \xF0\x9E\xA5\x90
 <ADLAM_DIGIT_ONE> \xF0\x9E\xA5\x91
 <ADLAM_DIGIT_TWO> \xF0\x9E\xA5\x92
@@ -80592,6 +80939,67 @@ CHARMAP
 <INDIC_SIYAQ_NUMBER_ALTERNATE_TWO> \xF0\x9E\xB2\xB2
 <INDIC_SIYAQ_NUMBER_ALTERNATE_TEN_THOUSAND> \xF0\x9E\xB2\xB3
 <INDIC_SIYAQ_ALTERNATE_LAKH_MARK> \xF0\x9E\xB2\xB4
+<OTTOMAN_SIYAQ_NUMBER_ONE> \xF0\x9E\xB4\x81
+<OTTOMAN_SIYAQ_NUMBER_TWO> \xF0\x9E\xB4\x82
+<OTTOMAN_SIYAQ_NUMBER_THREE> \xF0\x9E\xB4\x83
+<OTTOMAN_SIYAQ_NUMBER_FOUR> \xF0\x9E\xB4\x84
+<OTTOMAN_SIYAQ_NUMBER_FIVE> \xF0\x9E\xB4\x85
+<OTTOMAN_SIYAQ_NUMBER_SIX> \xF0\x9E\xB4\x86
+<OTTOMAN_SIYAQ_NUMBER_SEVEN> \xF0\x9E\xB4\x87
+<OTTOMAN_SIYAQ_NUMBER_EIGHT> \xF0\x9E\xB4\x88
+<OTTOMAN_SIYAQ_NUMBER_NINE> \xF0\x9E\xB4\x89
+<OTTOMAN_SIYAQ_NUMBER_TEN> \xF0\x9E\xB4\x8A
+<OTTOMAN_SIYAQ_NUMBER_TWENTY> \xF0\x9E\xB4\x8B
+<OTTOMAN_SIYAQ_NUMBER_THIRTY> \xF0\x9E\xB4\x8C
+<OTTOMAN_SIYAQ_NUMBER_FORTY> \xF0\x9E\xB4\x8D
+<OTTOMAN_SIYAQ_NUMBER_FIFTY> \xF0\x9E\xB4\x8E
+<OTTOMAN_SIYAQ_NUMBER_SIXTY> \xF0\x9E\xB4\x8F
+<OTTOMAN_SIYAQ_NUMBER_SEVENTY> \xF0\x9E\xB4\x90
+<OTTOMAN_SIYAQ_NUMBER_EIGHTY> \xF0\x9E\xB4\x91
+<OTTOMAN_SIYAQ_NUMBER_NINETY> \xF0\x9E\xB4\x92
+<OTTOMAN_SIYAQ_NUMBER_ONE_HUNDRED> \xF0\x9E\xB4\x93
+<OTTOMAN_SIYAQ_NUMBER_TWO_HUNDRED> \xF0\x9E\xB4\x94
+<OTTOMAN_SIYAQ_NUMBER_THREE_HUNDRED> \xF0\x9E\xB4\x95
+<OTTOMAN_SIYAQ_NUMBER_FOUR_HUNDRED> \xF0\x9E\xB4\x96
+<OTTOMAN_SIYAQ_NUMBER_FIVE_HUNDRED> \xF0\x9E\xB4\x97
+<OTTOMAN_SIYAQ_NUMBER_SIX_HUNDRED> \xF0\x9E\xB4\x98
+<OTTOMAN_SIYAQ_NUMBER_SEVEN_HUNDRED> \xF0\x9E\xB4\x99
+<OTTOMAN_SIYAQ_NUMBER_EIGHT_HUNDRED> \xF0\x9E\xB4\x9A
+<OTTOMAN_SIYAQ_NUMBER_NINE_HUNDRED> \xF0\x9E\xB4\x9B
+<OTTOMAN_SIYAQ_NUMBER_ONE_THOUSAND> \xF0\x9E\xB4\x9C
+<OTTOMAN_SIYAQ_NUMBER_TWO_THOUSAND> \xF0\x9E\xB4\x9D
+<OTTOMAN_SIYAQ_NUMBER_THREE_THOUSAND> \xF0\x9E\xB4\x9E
+<OTTOMAN_SIYAQ_NUMBER_FOUR_THOUSAND> \xF0\x9E\xB4\x9F
+<OTTOMAN_SIYAQ_NUMBER_FIVE_THOUSAND> \xF0\x9E\xB4\xA0
+<OTTOMAN_SIYAQ_NUMBER_SIX_THOUSAND> \xF0\x9E\xB4\xA1
+<OTTOMAN_SIYAQ_NUMBER_SEVEN_THOUSAND> \xF0\x9E\xB4\xA2
+<OTTOMAN_SIYAQ_NUMBER_EIGHT_THOUSAND> \xF0\x9E\xB4\xA3
+<OTTOMAN_SIYAQ_NUMBER_NINE_THOUSAND> \xF0\x9E\xB4\xA4
+<OTTOMAN_SIYAQ_NUMBER_TEN_THOUSAND> \xF0\x9E\xB4\xA5
+<OTTOMAN_SIYAQ_NUMBER_TWENTY_THOUSAND> \xF0\x9E\xB4\xA6
+<OTTOMAN_SIYAQ_NUMBER_THIRTY_THOUSAND> \xF0\x9E\xB4\xA7
+<OTTOMAN_SIYAQ_NUMBER_FORTY_THOUSAND> \xF0\x9E\xB4\xA8
+<OTTOMAN_SIYAQ_NUMBER_FIFTY_THOUSAND> \xF0\x9E\xB4\xA9
+<OTTOMAN_SIYAQ_NUMBER_SIXTY_THOUSAND> \xF0\x9E\xB4\xAA
+<OTTOMAN_SIYAQ_NUMBER_SEVENTY_THOUSAND> \xF0\x9E\xB4\xAB
+<OTTOMAN_SIYAQ_NUMBER_EIGHTY_THOUSAND> \xF0\x9E\xB4\xAC
+<OTTOMAN_SIYAQ_NUMBER_NINETY_THOUSAND> \xF0\x9E\xB4\xAD
+<OTTOMAN_SIYAQ_MARRATAN> \xF0\x9E\xB4\xAE
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_TWO> \xF0\x9E\xB4\xAF
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_THREE> \xF0\x9E\xB4\xB0
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_FOUR> \xF0\x9E\xB4\xB1
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_FIVE> \xF0\x9E\xB4\xB2
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_SIX> \xF0\x9E\xB4\xB3
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_SEVEN> \xF0\x9E\xB4\xB4
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_EIGHT> \xF0\x9E\xB4\xB5
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_NINE> \xF0\x9E\xB4\xB6
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_TEN> \xF0\x9E\xB4\xB7
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_FOUR_HUNDRED> \xF0\x9E\xB4\xB8
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_SIX_HUNDRED> \xF0\x9E\xB4\xB9
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_TWO_THOUSAND> \xF0\x9E\xB4\xBA
+<OTTOMAN_SIYAQ_ALTERNATE_NUMBER_TEN_THOUSAND> \xF0\x9E\xB4\xBB
+<OTTOMAN_SIYAQ_FRACTION_ONE_HALF> \xF0\x9E\xB4\xBC
+<OTTOMAN_SIYAQ_FRACTION_ONE_SIXTH> \xF0\x9E\xB4\xBD
 <ARABIC_MATHEMATICAL_ALEF> \xF0\x9E\xB8\x80
 <ARABIC_MATHEMATICAL_BEH> \xF0\x9E\xB8\x81
 <ARABIC_MATHEMATICAL_JEEM> \xF0\x9E\xB8\x82
@@ -81066,6 +81474,7 @@ CHARMAP
 <NEGATIVE_CIRCLED_LATIN_CAPITAL_LETTER_Z> \xF0\x9F\x85\xA9
 <RAISED_MC_SIGN> \xF0\x9F\x85\xAA
 <RAISED_MD_SIGN> \xF0\x9F\x85\xAB
+<RAISED_MR_SIGN> \xF0\x9F\x85\xAC
 <NEGATIVE_SQUARED_LATIN_CAPITAL_LETTER_A> \xF0\x9F\x85\xB0
 <NEGATIVE_SQUARED_LATIN_CAPITAL_LETTER_B> \xF0\x9F\x85\xB1
 <NEGATIVE_SQUARED_LATIN_CAPITAL_LETTER_C> \xF0\x9F\x85\xB2
@@ -82198,6 +82607,7 @@ CHARMAP
 <SHOPPING_TROLLEY> \xF0\x9F\x9B\x92
 <STUPA> \xF0\x9F\x9B\x93
 <PAGODA> \xF0\x9F\x9B\x94
+<HINDU_TEMPLE> \xF0\x9F\x9B\x95
 <HAMMER_AND_WRENCH> \xF0\x9F\x9B\xA0
 <SHIELD> \xF0\x9F\x9B\xA1
 <OIL_DRUM> \xF0\x9F\x9B\xA2
@@ -82221,6 +82631,7 @@ CHARMAP
 <SLED> \xF0\x9F\x9B\xB7
 <FLYING_SAUCER> \xF0\x9F\x9B\xB8
 <SKATEBOARD> \xF0\x9F\x9B\xB9
+<AUTO_RICKSHAW> \xF0\x9F\x9B\xBA
 <ALCHEMICAL_SYMBOL_FOR_QUINTESSENCE> \xF0\x9F\x9C\x80
 <ALCHEMICAL_SYMBOL_FOR_AIR> \xF0\x9F\x9C\x81
 <ALCHEMICAL_SYMBOL_FOR_FIRE> \xF0\x9F\x9C\x82
@@ -82426,6 +82837,18 @@ CHARMAP
 <NEGATIVE_CIRCLED_TRIANGLE> \xF0\x9F\x9F\x96
 <CIRCLED_SQUARE> \xF0\x9F\x9F\x97
 <NEGATIVE_CIRCLED_SQUARE> \xF0\x9F\x9F\x98
+<LARGE_ORANGE_CIRCLE> \xF0\x9F\x9F\xA0
+<LARGE_YELLOW_CIRCLE> \xF0\x9F\x9F\xA1
+<LARGE_GREEN_CIRCLE> \xF0\x9F\x9F\xA2
+<LARGE_PURPLE_CIRCLE> \xF0\x9F\x9F\xA3
+<LARGE_BROWN_CIRCLE> \xF0\x9F\x9F\xA4
+<LARGE_RED_SQUARE> \xF0\x9F\x9F\xA5
+<LARGE_BLUE_SQUARE> \xF0\x9F\x9F\xA6
+<LARGE_ORANGE_SQUARE> \xF0\x9F\x9F\xA7
+<LARGE_YELLOW_SQUARE> \xF0\x9F\x9F\xA8
+<LARGE_GREEN_SQUARE> \xF0\x9F\x9F\xA9
+<LARGE_PURPLE_SQUARE> \xF0\x9F\x9F\xAA
+<LARGE_BROWN_SQUARE> \xF0\x9F\x9F\xAB
 <LEFTWARDS_ARROW_WITH_SMALL_TRIANGLE_ARROWHEAD> \xF0\x9F\xA0\x80
 <UPWARDS_ARROW_WITH_SMALL_TRIANGLE_ARROWHEAD> \xF0\x9F\xA0\x81
 <RIGHTWARDS_ARROW_WITH_SMALL_TRIANGLE_ARROWHEAD> \xF0\x9F\xA0\x82
@@ -82586,6 +83009,9 @@ CHARMAP
 <DOWNWARD_FACING_NOTCHED_HOOK> \xF0\x9F\xA4\x89
 <DOWNWARD_FACING_HOOK_WITH_DOT> \xF0\x9F\xA4\x8A
 <DOWNWARD_FACING_NOTCHED_HOOK_WITH_DOT> \xF0\x9F\xA4\x8B
+<WHITE_HEART> \xF0\x9F\xA4\x8D
+<BROWN_HEART> \xF0\x9F\xA4\x8E
+<PINCHING_HAND> \xF0\x9F\xA4\x8F
 <ZIPPER-MOUTH_FACE> \xF0\x9F\xA4\x90
 <MONEY-MOUTH_FACE> \xF0\x9F\xA4\x91
 <FACE_WITH_THERMOMETER> \xF0\x9F\xA4\x92
@@ -82633,6 +83059,7 @@ CHARMAP
 <WRESTLERS> \xF0\x9F\xA4\xBC
 <WATER_POLO> \xF0\x9F\xA4\xBD
 <HANDBALL> \xF0\x9F\xA4\xBE
+<DIVING_MASK> \xF0\x9F\xA4\xBF
 <WILTED_FLOWER> \xF0\x9F\xA5\x80
 <DRUM_WITH_DRUMSTICKS> \xF0\x9F\xA5\x81
 <CLINKING_GLASSES> \xF0\x9F\xA5\x82
@@ -82682,11 +83109,13 @@ CHARMAP
 <MOON_CAKE> \xF0\x9F\xA5\xAE
 <BAGEL> \xF0\x9F\xA5\xAF
 <SMILING_FACE_WITH_SMILING_EYES_AND_THREE_HEARTS> \xF0\x9F\xA5\xB0
+<YAWNING_FACE> \xF0\x9F\xA5\xB1
 <FACE_WITH_PARTY_HORN_AND_PARTY_HAT> \xF0\x9F\xA5\xB3
 <FACE_WITH_UNEVEN_EYES_AND_WAVY_MOUTH> \xF0\x9F\xA5\xB4
 <OVERHEATED_FACE> \xF0\x9F\xA5\xB5
 <FREEZING_FACE> \xF0\x9F\xA5\xB6
 <FACE_WITH_PLEADING_EYES> \xF0\x9F\xA5\xBA
+<SARI> \xF0\x9F\xA5\xBB
 <LAB_COAT> \xF0\x9F\xA5\xBC
 <GOGGLES> \xF0\x9F\xA5\xBD
 <HIKING_BOOT> \xF0\x9F\xA5\xBE
@@ -82726,6 +83155,14 @@ CHARMAP
 <MICROBE> \xF0\x9F\xA6\xA0
 <BADGER> \xF0\x9F\xA6\xA1
 <SWAN> \xF0\x9F\xA6\xA2
+<SLOTH> \xF0\x9F\xA6\xA5
+<OTTER> \xF0\x9F\xA6\xA6
+<ORANGUTAN> \xF0\x9F\xA6\xA7
+<SKUNK> \xF0\x9F\xA6\xA8
+<FLAMINGO> \xF0\x9F\xA6\xA9
+<OYSTER> \xF0\x9F\xA6\xAA
+<GUIDE_DOG> \xF0\x9F\xA6\xAE
+<PROBING_CANE> \xF0\x9F\xA6\xAF
 <EMOJI_COMPONENT_RED_HAIR> \xF0\x9F\xA6\xB0
 <EMOJI_COMPONENT_CURLY_HAIR> \xF0\x9F\xA6\xB1
 <EMOJI_COMPONENT_BALD> \xF0\x9F\xA6\xB2
@@ -82736,9 +83173,26 @@ CHARMAP
 <TOOTH> \xF0\x9F\xA6\xB7
 <SUPERHERO> \xF0\x9F\xA6\xB8
 <SUPERVILLAIN> \xF0\x9F\xA6\xB9
+<SAFETY_VEST> \xF0\x9F\xA6\xBA
+<EAR_WITH_HEARING_AID> \xF0\x9F\xA6\xBB
+<MOTORIZED_WHEELCHAIR> \xF0\x9F\xA6\xBC
+<MANUAL_WHEELCHAIR> \xF0\x9F\xA6\xBD
+<MECHANICAL_ARM> \xF0\x9F\xA6\xBE
+<MECHANICAL_LEG> \xF0\x9F\xA6\xBF
 <CHEESE_WEDGE> \xF0\x9F\xA7\x80
 <CUPCAKE> \xF0\x9F\xA7\x81
 <SALT_SHAKER> \xF0\x9F\xA7\x82
+<BEVERAGE_BOX> \xF0\x9F\xA7\x83
+<GARLIC> \xF0\x9F\xA7\x84
+<ONION> \xF0\x9F\xA7\x85
+<FALAFEL> \xF0\x9F\xA7\x86
+<WAFFLE> \xF0\x9F\xA7\x87
+<BUTTER> \xF0\x9F\xA7\x88
+<MATE_DRINK> \xF0\x9F\xA7\x89
+<ICE_CUBE> \xF0\x9F\xA7\x8A
+<STANDING_PERSON> \xF0\x9F\xA7\x8D
+<KNEELING_PERSON> \xF0\x9F\xA7\x8E
+<DEAF_PERSON> \xF0\x9F\xA7\x8F
 <FACE_WITH_MONOCLE> \xF0\x9F\xA7\x90
 <ADULT> \xF0\x9F\xA7\x91
 <CHILD> \xF0\x9F\xA7\x92
@@ -82787,6 +83241,90 @@ CHARMAP
 <SPONGE> \xF0\x9F\xA7\xBD
 <RECEIPT> \xF0\x9F\xA7\xBE
 <NAZAR_AMULET> \xF0\x9F\xA7\xBF
+<NEUTRAL_CHESS_KING> \xF0\x9F\xA8\x80
+<NEUTRAL_CHESS_QUEEN> \xF0\x9F\xA8\x81
+<NEUTRAL_CHESS_ROOK> \xF0\x9F\xA8\x82
+<NEUTRAL_CHESS_BISHOP> \xF0\x9F\xA8\x83
+<NEUTRAL_CHESS_KNIGHT> \xF0\x9F\xA8\x84
+<NEUTRAL_CHESS_PAWN> \xF0\x9F\xA8\x85
+<WHITE_CHESS_KNIGHT_ROTATED_FORTY-FIVE_DEGREES> \xF0\x9F\xA8\x86
+<BLACK_CHESS_KNIGHT_ROTATED_FORTY-FIVE_DEGREES> \xF0\x9F\xA8\x87
+<NEUTRAL_CHESS_KNIGHT_ROTATED_FORTY-FIVE_DEGREES> \xF0\x9F\xA8\x88
+<WHITE_CHESS_KING_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x89
+<WHITE_CHESS_QUEEN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8A
+<WHITE_CHESS_ROOK_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8B
+<WHITE_CHESS_BISHOP_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8C
+<WHITE_CHESS_KNIGHT_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8D
+<WHITE_CHESS_PAWN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8E
+<BLACK_CHESS_KING_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x8F
+<BLACK_CHESS_QUEEN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x90
+<BLACK_CHESS_ROOK_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x91
+<BLACK_CHESS_BISHOP_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x92
+<BLACK_CHESS_KNIGHT_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x93
+<BLACK_CHESS_PAWN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x94
+<NEUTRAL_CHESS_KING_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x95
+<NEUTRAL_CHESS_QUEEN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x96
+<NEUTRAL_CHESS_ROOK_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x97
+<NEUTRAL_CHESS_BISHOP_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x98
+<NEUTRAL_CHESS_KNIGHT_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x99
+<NEUTRAL_CHESS_PAWN_ROTATED_NINETY_DEGREES> \xF0\x9F\xA8\x9A
+<WHITE_CHESS_KNIGHT_ROTATED_ONE_HUNDRED_THIRTY-FIVE_DEGREES> \xF0\x9F\xA8\x9B
+<BLACK_CHESS_KNIGHT_ROTATED_ONE_HUNDRED_THIRTY-FIVE_DEGREES> \xF0\x9F\xA8\x9C
+<NEUTRAL_CHESS_KNIGHT_ROTATED_ONE_HUNDRED_THIRTY-FIVE_DEGREES> \xF0\x9F\xA8\x9D
+<WHITE_CHESS_TURNED_KING> \xF0\x9F\xA8\x9E
+<WHITE_CHESS_TURNED_QUEEN> \xF0\x9F\xA8\x9F
+<WHITE_CHESS_TURNED_ROOK> \xF0\x9F\xA8\xA0
+<WHITE_CHESS_TURNED_BISHOP> \xF0\x9F\xA8\xA1
+<WHITE_CHESS_TURNED_KNIGHT> \xF0\x9F\xA8\xA2
+<WHITE_CHESS_TURNED_PAWN> \xF0\x9F\xA8\xA3
+<BLACK_CHESS_TURNED_KING> \xF0\x9F\xA8\xA4
+<BLACK_CHESS_TURNED_QUEEN> \xF0\x9F\xA8\xA5
+<BLACK_CHESS_TURNED_ROOK> \xF0\x9F\xA8\xA6
+<BLACK_CHESS_TURNED_BISHOP> \xF0\x9F\xA8\xA7
+<BLACK_CHESS_TURNED_KNIGHT> \xF0\x9F\xA8\xA8
+<BLACK_CHESS_TURNED_PAWN> \xF0\x9F\xA8\xA9
+<NEUTRAL_CHESS_TURNED_KING> \xF0\x9F\xA8\xAA
+<NEUTRAL_CHESS_TURNED_QUEEN> \xF0\x9F\xA8\xAB
+<NEUTRAL_CHESS_TURNED_ROOK> \xF0\x9F\xA8\xAC
+<NEUTRAL_CHESS_TURNED_BISHOP> \xF0\x9F\xA8\xAD
+<NEUTRAL_CHESS_TURNED_KNIGHT> \xF0\x9F\xA8\xAE
+<NEUTRAL_CHESS_TURNED_PAWN> \xF0\x9F\xA8\xAF
+<WHITE_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_TWENTY-FIVE_DEGREES> \xF0\x9F\xA8\xB0
+<BLACK_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_TWENTY-FIVE_DEGREES> \xF0\x9F\xA8\xB1
+<NEUTRAL_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_TWENTY-FIVE_DEGREES> \xF0\x9F\xA8\xB2
+<WHITE_CHESS_KING_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB3
+<WHITE_CHESS_QUEEN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB4
+<WHITE_CHESS_ROOK_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB5
+<WHITE_CHESS_BISHOP_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB6
+<WHITE_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB7
+<WHITE_CHESS_PAWN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB8
+<BLACK_CHESS_KING_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xB9
+<BLACK_CHESS_QUEEN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBA
+<BLACK_CHESS_ROOK_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBB
+<BLACK_CHESS_BISHOP_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBC
+<BLACK_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBD
+<BLACK_CHESS_PAWN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBE
+<NEUTRAL_CHESS_KING_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA8\xBF
+<NEUTRAL_CHESS_QUEEN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA9\x80
+<NEUTRAL_CHESS_ROOK_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA9\x81
+<NEUTRAL_CHESS_BISHOP_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA9\x82
+<NEUTRAL_CHESS_KNIGHT_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA9\x83
+<NEUTRAL_CHESS_PAWN_ROTATED_TWO_HUNDRED_SEVENTY_DEGREES> \xF0\x9F\xA9\x84
+<WHITE_CHESS_KNIGHT_ROTATED_THREE_HUNDRED_FIFTEEN_DEGREES> \xF0\x9F\xA9\x85
+<BLACK_CHESS_KNIGHT_ROTATED_THREE_HUNDRED_FIFTEEN_DEGREES> \xF0\x9F\xA9\x86
+<NEUTRAL_CHESS_KNIGHT_ROTATED_THREE_HUNDRED_FIFTEEN_DEGREES> \xF0\x9F\xA9\x87
+<WHITE_CHESS_EQUIHOPPER> \xF0\x9F\xA9\x88
+<BLACK_CHESS_EQUIHOPPER> \xF0\x9F\xA9\x89
+<NEUTRAL_CHESS_EQUIHOPPER> \xF0\x9F\xA9\x8A
+<WHITE_CHESS_EQUIHOPPER_ROTATED_NINETY_DEGREES> \xF0\x9F\xA9\x8B
+<BLACK_CHESS_EQUIHOPPER_ROTATED_NINETY_DEGREES> \xF0\x9F\xA9\x8C
+<NEUTRAL_CHESS_EQUIHOPPER_ROTATED_NINETY_DEGREES> \xF0\x9F\xA9\x8D
+<WHITE_CHESS_KNIGHT-QUEEN> \xF0\x9F\xA9\x8E
+<WHITE_CHESS_KNIGHT-ROOK> \xF0\x9F\xA9\x8F
+<WHITE_CHESS_KNIGHT-BISHOP> \xF0\x9F\xA9\x90
+<BLACK_CHESS_KNIGHT-QUEEN> \xF0\x9F\xA9\x91
+<BLACK_CHESS_KNIGHT-ROOK> \xF0\x9F\xA9\x92
+<BLACK_CHESS_KNIGHT-BISHOP> \xF0\x9F\xA9\x93
 <XIANGQI_RED_GENERAL> \xF0\x9F\xA9\xA0
 <XIANGQI_RED_MANDARIN> \xF0\x9F\xA9\xA1
 <XIANGQI_RED_ELEPHANT> \xF0\x9F\xA9\xA2
@@ -82801,6 +83339,22 @@ CHARMAP
 <XIANGQI_BLACK_CHARIOT> \xF0\x9F\xA9\xAB
 <XIANGQI_BLACK_CANNON> \xF0\x9F\xA9\xAC
 <XIANGQI_BLACK_SOLDIER> \xF0\x9F\xA9\xAD
+<BALLET_SHOES> \xF0\x9F\xA9\xB0
+<ONE-PIECE_SWIMSUIT> \xF0\x9F\xA9\xB1
+<BRIEFS> \xF0\x9F\xA9\xB2
+<SHORTS> \xF0\x9F\xA9\xB3
+<DROP_OF_BLOOD> \xF0\x9F\xA9\xB8
+<ADHESIVE_BANDAGE> \xF0\x9F\xA9\xB9
+<STETHOSCOPE> \xF0\x9F\xA9\xBA
+<YO-YO> \xF0\x9F\xAA\x80
+<KITE> \xF0\x9F\xAA\x81
+<PARACHUTE> \xF0\x9F\xAA\x82
+<RINGED_PLANET> \xF0\x9F\xAA\x90
+<CHAIR> \xF0\x9F\xAA\x91
+<RAZOR> \xF0\x9F\xAA\x92
+<AXE> \xF0\x9F\xAA\x93
+<DIYA_LAMP> \xF0\x9F\xAA\x94
+<BANJO> \xF0\x9F\xAA\x95
 <CJK_UNIFIED_IDEOGRAPH-20000> \xF0\xA0\x80\x80
 <CJK_UNIFIED_IDEOGRAPH-20001> \xF0\xA0\x80\x81
 <CJK_UNIFIED_IDEOGRAPH-20002> \xF0\xA0\x80\x82
diff --git a/tools/tools/locale/patch/patch-UnicodeData.txt b/tools/tools/locale/patch/patch-UnicodeData.txt
new file mode 100644
index 000000000000..fe65ebacd16a
--- /dev/null
+++ b/tools/tools/locale/patch/patch-UnicodeData.txt
@@ -0,0 +1,29 @@
+--- UnicodeData.txt.orig	2020-06-29 14:05:49.483379000 +0900
++++ UnicodeData.txt	2020-06-29 14:12:09.808622000 +0900
+@@ -12138,7 +12138,7 @@
+ 33FE;IDEOGRAPHIC TELEGRAPH SYMBOL FOR DAY THIRTY-ONE;So;0;L;<compat> 0033 0031 65E5;;;;N;;;;;
+ 33FF;SQUARE GAL;So;0;ON;<square> 0067 0061 006C;;;;N;;;;;
+ 3400;<CJK Ideograph Extension A, First>;Lo;0;L;;;;;N;;;;;
+-4DBF;<CJK Ideograph Extension A, Last>;Lo;0;L;;;;;N;;;;;
++4DB5;<CJK Ideograph Extension A, Last>;Lo;0;L;;;;;N;;;;;
+ 4DC0;HEXAGRAM FOR THE CREATIVE HEAVEN;So;0;ON;;;;;N;;;;;
+ 4DC1;HEXAGRAM FOR THE RECEPTIVE EARTH;So;0;ON;;;;;N;;;;;
+ 4DC2;HEXAGRAM FOR DIFFICULTY AT THE BEGINNING;So;0;ON;;;;;N;;;;;
+@@ -12204,7 +12204,7 @@
+ 4DFE;HEXAGRAM FOR AFTER COMPLETION;So;0;ON;;;;;N;;;;;
+ 4DFF;HEXAGRAM FOR BEFORE COMPLETION;So;0;ON;;;;;N;;;;;
+ 4E00;<CJK Ideograph, First>;Lo;0;L;;;;;N;;;;;
+-9FFC;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;;
++9FEF;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;;
+ A000;YI SYLLABLE IT;Lo;0;L;;;;;N;;;;;
+ A001;YI SYLLABLE IX;Lo;0;L;;;;;N;;;;;
+ A002;YI SYLLABLE I;Lo;0;L;;;;;N;;;;;
+@@ -32901,7 +32901,7 @@
+ 1FBF8;SEGMENTED DIGIT EIGHT;Nd;0;EN;<font> 0038;8;8;8;N;;;;;
+ 1FBF9;SEGMENTED DIGIT NINE;Nd;0;EN;<font> 0039;9;9;9;N;;;;;
+ 20000;<CJK Ideograph Extension B, First>;Lo;0;L;;;;;N;;;;;
+-2A6DD;<CJK Ideograph Extension B, Last>;Lo;0;L;;;;;N;;;;;
++2A6D6;<CJK Ideograph Extension B, Last>;Lo;0;L;;;;;N;;;;;
+ 2A700;<CJK Ideograph Extension C, First>;Lo;0;L;;;;;N;;;;;
+ 2B734;<CJK Ideograph Extension C, Last>;Lo;0;L;;;;;N;;;;;
+ 2B740;<CJK Ideograph Extension D, First>;Lo;0;L;;;;;N;;;;;
diff --git a/tools/tools/locale/tools/cldr2def.pl b/tools/tools/locale/tools/cldr2def.pl
index 3a6f8ac79d18..5f756cc3895a 100755
--- a/tools/tools/locale/tools/cldr2def.pl
+++ b/tools/tools/locale/tools/cldr2def.pl
@@ -460,6 +460,11 @@ sub transform_ctypes {
 		foreach my $enc (sort keys(%{$languages{$l}{$f}{data}{$c}})) {
 			next if ($enc eq $DEFENCODING);
 			$filename = "$UNIDIR/posix/$file.$DEFENCODING.src";
+			if ($file eq 'ja_JP') {
+				# Override $filename for ja_JP because
+				# its CTYPE is not compatible with UTF-8.
+				$filename = "$UNIDIR/posix/$file.eucJP.src";
+			}
 			if (! -f $filename) {
 				print STDERR "Cannot open $filename\n";
 				next;
diff --git a/tools/tools/locale/tools/convert_map.pl b/tools/tools/locale/tools/convert_map.pl
index 88222531d064..8b54ff33381b 100755
--- a/tools/tools/locale/tools/convert_map.pl
+++ b/tools/tools/locale/tools/convert_map.pl
@@ -87,7 +87,7 @@ sub load_utf8_cm
 {
     my $file = shift;
 
-    open(UTF8, "$file") || die "open";
+    open(UTF8, "$file") || die "$!: open: $file";
 
     while (<UTF8>) {
 	next if (/^#/);
@@ -158,7 +158,8 @@ $mf = shift(@ARGV);
 $codeset = shift(@ARGV);
 my $max_mb;
 
-load_utf8_cm("etc/final-maps/map.UTF-8");
+my $etcdir = (exists $ENV{'ETCDIR'}) ? $ENV{'ETCDIR'} : "etc";
+load_utf8_cm("${etcdir}/final-maps/map.UTF-8");
 load_map($mf);
 
 
diff --git a/tools/tools/locale/tools/finalize b/tools/tools/locale/tools/finalize
index f4dfd7d0892f..88dfcad0cb24 100755
--- a/tools/tools/locale/tools/finalize
+++ b/tools/tools/locale/tools/finalize
@@ -47,15 +47,21 @@ usage ()
   $1 = "numericdef" -o $1 = "timedef" -o $1 = "ctypedef" ] || usage
 
 self=$(realpath $0)
-base=$(dirname ${self})
-old=${base}/../${1}.draft
-new=${base}/../${1}
-TEMP=/tmp/${1}.locales
-TEMP2=/tmp/${1}.hashes
-TEMP3=/tmp/${1}.symlinks
-TEMP4=/tmp/${1}.mapped
-FULLMAP=/tmp/utf8-map
-FULLEXTRACT=/tmp/extracted-names
+base=${BASEDIR:-$(dirname ${self})}
+: ${ETCDIR:=${base}/../etc}
+: ${TOOLSDIR:=${base}}
+: ${OUTBASEDIR:=${base}/../${1}}
+: ${OLD_DIR:=${OUTBASEDIR}.draft}
+: ${NEW_DIR:=${OUTBASEDIR}}
+old=${OLD_DIR}
+new=${NEW_DIR}
+: ${TMPDIR:=/tmp}
+TEMP=${TMPDIR}/${1}.locales
+TEMP2=${TMPDIR}/${1}.hashes
+TEMP3=${TMPDIR}/${1}.symlinks
+TEMP4=${TMPDIR}/${1}.mapped
+FULLMAP=${TMPDIR}/utf8-map
+FULLEXTRACT=${TMPDIR}/extracted-names
 AWKCMD="/## PLACEHOLDER/ { \
 	  while ( getline line < \"${TEMP}\" ) {print line} } \
 	/## SYMPAIRS/ { \
@@ -65,6 +71,7 @@ AWKCMD="/## PLACEHOLDER/ { \
 	!/## / { print \$0 }"
 
 # Rename the sources with 3 components name into the POSIX version of the name using @modifier
+mkdir -p $old $new
 cd $old
 pwd
 for i in *_*_*.*.src; do
@@ -142,13 +149,13 @@ then
 	rm -f ${TEMP2}
 	/usr/bin/sed -E -e 's/[ ]+/ /g' \
 		${UNIDIR}/posix/UTF-8.cm \
-		> ${base}/../etc/final-maps/map.UTF-8
+		> ${ETCDIR}/final-maps/map.UTF-8
 	/usr/bin/sed -E -e 's/[ ]+/ /g' \
 		${UNIDIR}/posix/eucCN.cm \
-		> ${base}/../etc/final-maps/map.eucCN
+		> ${ETCDIR}/final-maps/map.eucCN
 	/usr/bin/sed -E -e 's/[ ]+/ /g' \
 		${UNIDIR}/posix/eucCN.cm \
-		> ${base}/../etc/final-maps/map.GB2312
+		> ${ETCDIR}/final-maps/map.GB2312
 
 	# GB18030 and Big5 are pre-generated from CLDR data
 	CHARMAPS="ARMSCII-8 CP1131 CP1251 \
@@ -160,10 +167,11 @@ then
 	for map in ${CHARMAPS}
 	do
 		encoding=${map}
-		/usr/local/bin/perl ${base}/convert_map.pl \
-			${base}/../etc/charmaps/${map}.TXT ${encoding} \
+		env ETCDIR="${ETCDIR}" \
+		/usr/local/bin/perl ${TOOLSDIR}/convert_map.pl \
+			${ETCDIR}/charmaps/${map}.TXT ${encoding} \
 			| /usr/bin/sed -E -e 's/	+/ /g' \
-		> ${base}/../etc/final-maps/map.${map}
+		> ${ETCDIR}/final-maps/map.${map}
 		echo map ${map} converted.
 	done
 
diff --git a/tools/tools/locale/tools/utf8-rollup.pl b/tools/tools/locale/tools/utf8-rollup.pl
index da93d2f4398a..b275828d52c9 100755
--- a/tools/tools/locale/tools/utf8-rollup.pl
+++ b/tools/tools/locale/tools/utf8-rollup.pl
@@ -30,6 +30,7 @@
 
 use strict;
 use Getopt::Long;
+use Encode qw(encode decode);
 
 if ($#ARGV != 0) {
 	print "Usage: $0 --unidir=<unidir>\n";
@@ -52,6 +53,23 @@ generate_footer ();
 
 ############################
 
+sub utf8to32 {
+	my @kl = split /\\x/, $_[0];
+
+	shift @kl if ($kl[0] eq '');
+	my $k = pack('H2' x scalar @kl, @kl);
+	my $ux = encode('UTF-32BE', decode('UTF-8', $k));
+	my $u = uc(unpack('H*', $ux));
+	# Remove BOM
+	$u =~ s/^0000FEFF//;
+	# Remove heading bytes of 0
+	while ($u =~ m/^0/ and length($u) > 4) {
+		$u =~ s/^0//;
+	}
+
+	return $u;
+}
+
 sub get_utf8map {
 	my $file = shift;
 
@@ -75,9 +93,10 @@ sub get_utf8map {
 		last if ($l eq "END CHARMAP");
 
 		$l =~ /^(<[^\s]+>)\s+(.*)/;
-		my $k = $2;
+		my $k = utf8to32($2);	# UTF-8 char code
 		my $v = $1;
-		$k =~ s/\\x//g;		# UTF-8 char code
+
+#		print STDERR "register: $k - $v\n";
 		$utf8map{$k} = $v;
 	}
 }
@@ -143,7 +162,7 @@ sub parse_unidata {
 
 	foreach my $l (@lines) {
 		my @d = split(/;/, $l, -1);
-		my $mb = wctomb($d[0]);
+		my $mb = $d[0];
 		my $cat;
 
 		# XXX There are code points present in UnicodeData.txt
@@ -180,9 +199,9 @@ sub parse_unidata {
 
 		# Check if there's upper/lower mapping
 		if ($d[12] ne "") {
-			$data{'toupper'}{$mb} = wctomb($d[12]);
+			$data{'toupper'}{$mb} = $d[12];
 		} elsif ($d[13] ne "") {
-			$data{'tolower'}{$mb} = wctomb($d[13]);
+			$data{'tolower'}{$mb} = $d[13];
 		}
 	}
 
@@ -193,7 +212,7 @@ sub parse_unidata {
 	foreach my $cat (sort keys (%data)) {
 		print FOUT "$cat\t";
 		$first = 1;
-	foreach my $mb (sort keys (%{$data{$cat}})) {
+	foreach my $mb (sort {hex($a) <=> hex($b)} keys (%{$data{$cat}})) {
 		if ($first == 1) {
 			$first = 0;
 		} elsif ($inrange == 1) {