Initial checkin for UcdXML

unicode-org · Jun 6, 2024 · 2f29705 · 2f29705
1 parent 184d7e5
commit 2f29705
Show file tree

Hide file tree

Showing 5 changed files with 49 additions and 8 deletions.
diff --git a/unicodetools/src/main/java/org/unicode/props/PropertyParsingInfo.java b/unicodetools/src/main/java/org/unicode/props/PropertyParsingInfo.java
@@ -516,7 +516,8 @@ static void parseSourceFile(
         } else {
             indexUnicodeProperties.getFileNames().add(fullFilename);
             UcdLineParser parser = new UcdLineParser(FileUtilities.in("", fullFilename));
-            if (fileName.startsWith("Unihan") || fileName.startsWith("k")) {
+            if (fileName.startsWith("Unihan") || fileName.startsWith("k")
+                    || fileName.startsWith("NushuSources") || fileName.startsWith("TangutSources")) {
                 parser.withTabs(true);
             }
             PropertyParsingInfo propInfo;

diff --git a/unicodetools/src/main/java/org/unicode/props/UcdProperty.java b/unicodetools/src/main/java/org/unicode/props/UcdProperty.java
@@ -84,12 +84,16 @@ public enum UcdProperty {
     Emoji_SB(PropertyType.Miscellaneous, "ESB"),
     ISO_Comment(PropertyType.Miscellaneous, "isc"),
     Jamo_Short_Name(PropertyType.Miscellaneous, "JSN"),
+    NC_Corrected(PropertyType.Miscellaneous, "ncCorrected"),
+    NC_Original(PropertyType.Miscellaneous, "ncOriginal"),
+    NC_Version(PropertyType.Miscellaneous, "ncVersion"),
     Name(PropertyType.Miscellaneous, "na"),
     Name_Alias(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "Name_Alias"),
     Named_Sequences(PropertyType.Miscellaneous, "NS"),
     Named_Sequences_Prov(PropertyType.Miscellaneous, "NSP"),
     Standardized_Variant(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "SV"),
     Unicode_1_Name(PropertyType.Miscellaneous, "na1"),
+    emoji_variation_sequence(PropertyType.Miscellaneous, "EVS"),
     kAlternateTotalStrokes(PropertyType.Miscellaneous, "cjkAlternateTotalStrokes"),
     kBigFive(PropertyType.Miscellaneous, "cjkBigFive"),
     kCCCII(PropertyType.Miscellaneous, "cjkCCCII"),
@@ -182,13 +186,15 @@ public enum UcdProperty {
     kRSKanWa(PropertyType.Miscellaneous, "cjkRSKanWa"),
     kRSKangXi(PropertyType.Miscellaneous, "cjkRSKangXi"),
     kRSKorean(PropertyType.Miscellaneous, "cjkRSKorean"),
+    kRSTUnicode(PropertyType.Miscellaneous, "cjkRSTUnicode"),
     kRSUnicode(
             PropertyType.Miscellaneous,
             null,
             ValueCardinality.Ordered,
             "cjkRSUnicode",
             "Unicode_Radical_Stroke",
             "URS"),
+    kReading(PropertyType.Miscellaneous, "cjkReading"),
     kSBGY(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "cjkSBGY"),
     kSMSZD2003Index(PropertyType.Miscellaneous, "cjkSMSZD2003Index"),
     kSMSZD2003Readings(PropertyType.Miscellaneous, "cjkSMSZD2003Readings"),
@@ -200,9 +206,11 @@ public enum UcdProperty {
             ValueCardinality.Unordered,
             "cjkSpecializedSemanticVariant"),
     kSpoofingVariant(PropertyType.Miscellaneous, "cjkSpoofingVariant"),
+    kSrc_NushuDuben(PropertyType.Miscellaneous, "cjkSrc_NushuDuben"),
     kStrange(PropertyType.Miscellaneous, "cjkStrange"),
     kTGH(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "cjkTGH"),
     kTGHZ2013(PropertyType.Miscellaneous, "cjkTGHZ2013"),
+    kTGT_MergedSrc(PropertyType.Miscellaneous, "cjkTGT_MergedSrc"),
     kTaiwanTelegraph(PropertyType.Miscellaneous, "cjkTaiwanTelegraph"),
     kTang(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "cjkTang"),
     kTotalStrokes(PropertyType.Miscellaneous, null, ValueCardinality.Ordered, "cjkTotalStrokes"),
@@ -212,6 +220,7 @@ public enum UcdProperty {
     kXHC1983(PropertyType.Miscellaneous, null, ValueCardinality.Unordered, "cjkXHC1983"),
     kXerox(PropertyType.Miscellaneous, "cjkXerox"),
     kZVariant(PropertyType.Miscellaneous, "cjkZVariant"),
+    kZhuang(PropertyType.Miscellaneous, "cjkZhuang"),
     kZhuangNumeric(PropertyType.Miscellaneous, "cjkZhuangNumeric"),
 
     // Catalog

diff --git a/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyAliases.txt b/unicodetools/src/main/resources/org/unicode/props/ExtraPropertyAliases.txt
@@ -62,6 +62,7 @@ CJKR ; CJK_Radical
 EDCM ; Emoji_DCM
 EKDDI ; Emoji_KDDI
 ESB ; Emoji_SB
+EVS ; emoji_variation_sequence
 NS ; Named_Sequences
 NSP ; Named_Sequences_Prov
 SV ; Standardized_Variant
@@ -145,6 +146,13 @@ cjkJoyoKanji ; kJoyoKanji
 cjkKoreanEducationHanja ; kKoreanEducationHanja
 cjkKoreanName ; kKoreanName
 cjkTGH ; kTGH
+cjkRSTUnicode ; kRSTUnicode
+cjkReading ; kReading
+cjkSrc_NushuDuben ; kSrc_NushuDuben
+cjkTGT_MergedSrc ; kTGT_MergedSrc
+ncCorrected ; NC_Corrected
+ncOriginal ; NC_Original
+ncVersion ; NC_Version
 # 13.0
 cjkSpoofingVariant       ; kSpoofingVariant
 cjkTGHZ2013              ; kTGHZ2013
@@ -162,3 +170,4 @@ cjkVietnameseNumeric     ; kVietnameseNumeric
 cjkZhuangNumeric         ; kZhuangNumeric
 # 16.0
 cjkFanqie                ; kFanqie
+cjkZhuang                ; kZhuang
diff --git a/unicodetools/src/main/resources/org/unicode/props/IndexPropertyRegex.txt b/unicodetools/src/main/resources/org/unicode/props/IndexPropertyRegex.txt
@@ -142,7 +142,7 @@ kHanYu ;                      MULTI_VALUED ;                 [1-8][0-9]{4}\.[0-3
 kIRGHanyuDaZidian ;           SINGLE_VALUED ;                [1-8][0-9]{4}\.[0-3][0-9][01]
 kCNS1992 ;                    SINGLE_VALUED ;                [1-9]-[0-9A-F]{4}
 kTotalStrokes ;               ORDERED ;                      [1-9][0-9]{0,2}
-kRSUnicode ;                  ORDERED ;                 [1-9][0-9]{0,2}\'?\.[0-9]{1,2}
+kRSUnicode ;                  ORDERED ;                      [1-9][0-9]{0,2}\'?\.[0-9]{1,2}
 kRSJapanese ;                 EXTENSIBLE ;                   [1-9][0-9]{0,2}\.[0-9]{1,2}
 kRSKanWa ;                    EXTENSIBLE ;                   [1-9][0-9]{0,2}\.[0-9]{1,2}
 kRSKangXi ;                   EXTENSIBLE ;                   [1-9][0-9]{0,2}\.[0-9]{1,2}
@@ -170,16 +170,25 @@ kHanyuPinlu ;                 MULTI_VALUED ;                 [a-z\x{308}]+[1-5]\
 kCantonese ;                  MULTI_VALUED ;                 [a-z]{1,6}[1-6]
 kTang ;                       MULTI_VALUED ;                 \*?[A-Za-z()\x{E6}\x{251}\x{259}\x{25B}\x{300}\x{30C}]+
 
-kJinmeiyoKanji ; 		      MULTI_VALUED ;                 (20[0-9]{2})(:U\+2?[0-9A-F]{4})?
-kJoyoKanji ; 		          MULTI_VALUED ;                 (20[0-9]{2})|(U\+2?[0-9A-F]{4})
+kJinmeiyoKanji ;              MULTI_VALUED ;                 (20[0-9]{2})(:U\+2?[0-9A-F]{4})?
+kJoyoKanji ;                  MULTI_VALUED ;                 (20[0-9]{2})|(U\+2?[0-9A-F]{4})
 kKoreanEducationHanja ;       MULTI_VALUED ;                 20[0-9]{2}
-kKoreanName ; 		          MULTI_VALUED ;                 (20[0-9]{2})(:U\+2?[0-9A-F]{4})*
-kTGH ; 		                  MULTI_VALUED ;      			 20[0-9]{2}:[1-9][0-9]{0,3}
+kKoreanName ;                 MULTI_VALUED ;                 (20[0-9]{2})(:U\+2?[0-9A-F]{4})*
+kTGH ;                        MULTI_VALUED ;                 20[0-9]{2}:[1-9][0-9]{0,3}
 
 
-kIRG_UKSource ;                SINGLE_VALUED ;                V[0-4]-[0-9A-F]{4}
+kIRG_UKSource ;               SINGLE_VALUED ;                V[0-4]-[0-9A-F]{4}
 kIRG_SSource ;                SINGLE_VALUED ;                V[0-4]-[0-9A-F]{4}
 
+kSrc_NushuDuben ;             SINGLE_VALUED ;                [0-9]+\.[0-9]+
+kReading ;                    SINGLE_VALUED ;                [a-z]{1,6}[1-6]+
+kRSTUnicode ;                 SINGLE_VALUED ;                [0-9]+\.[0-9]+
+kTGT_MergedSrc ;              SINGLE_VALUED ;                L2008-[0-9A-F]{4,5}(-[0-9]{4,5})?
+
+NC_Original ;                 SINGLE_VALUED ;                [0-9A-F]{4,5}
+NC_Corrected ;                SINGLE_VALUED ;                [0-9A-F]{4,5}
+NC_Version ;                  SINGLE_VALUED ;                [0-9]\.[0-9]\.[0-9]
+
 
 # =============================
 # Catalog/Enum/Binary Properties
@@ -204,5 +213,5 @@ Confusable_MA			;	SINGLE_VALUED	;				$codePoints
 #Emoji					;	SINGLE_VALUED	;				<enum>
 #Emoji_Presentation		;	SINGLE_VALUED	;				<enum>
 #Emoji_Modifier			;	SINGLE_VALUED	;				<enum>
-#Emoji_Modifier_Base		;	SINGLE_VALUED	;				<enum>
+#Emoji_Modifier_Base 	;	SINGLE_VALUED	;				<enum>
 
diff --git a/unicodetools/src/main/resources/org/unicode/props/IndexUnicodeProperties.txt b/unicodetools/src/main/resources/org/unicode/props/IndexUnicodeProperties.txt
@@ -36,13 +36,16 @@ FileType ; Unihan_OtherMappings ; PropertyValue
 FileType ; Unihan_RadicalStrokeCounts ; PropertyValue
 FileType ; Unihan_Readings ; PropertyValue
 FileType ; Unihan_Variants ; PropertyValue
+FileType ; NushuSources ; PropertyValue
+FileType ; TangutSources ; PropertyValue
 
 # NameAliases File Type
 # Contains a multivalued property, where successive values are not in the same line, but are divided out on successive lines with the same code point
 
 FileType ; NameAliases ; NameAliases
 FileType ; NameAliasesProv ; NameAliases
 FileType ; StandardizedVariants ; StandardizedVariants
+FileType ; emoji-variation-sequences ; StandardizedVariants
 
 # CJKRadicals File Type
 
@@ -309,6 +312,15 @@ Unihan_Variants ; kSpoofingVariant
 Unihan_Variants ; kTraditionalVariant
 Unihan_Variants ; kZVariant
 
+NushuSources ; kSrc_NushuDuben
+NushuSources ; kReading
+TangutSources ; kRSTUnicode
+TangutSources ; kTGT_MergedSrc
+
+NormalizationCorrections ; NC_Original
+NormalizationCorrections ; NC_Corrected
+NormalizationCorrections ; NC_Version
+
 # Extras
 
 ScriptExtensions ; Script_Extensions
@@ -319,6 +331,7 @@ EmojiSources ; Emoji_SB ; 3
 NamedSequences ; Named_Sequences
 NamedSequencesProv ; Named_Sequences_Prov
 StandardizedVariants ; Standardized_Variant
+emoji-variation-sequences ; emoji-variation-sequence
 DoNotEmit ; Do_Not_Emit_Preferred ; 1
 DoNotEmit ; Do_Not_Emit_Type ; 2