Add tests for character-properties to help refactoring

Author: Deraen

Diff for: clj/.gitignore

@@ -10,4 +10,5 @@ pom.xml.asc
 .lein-failures
 .lein-plugins
 .lein-repl-history
+.nrepl-port
 /tmp

Diff for: clj/test/vim_clojure_static/generate_test.clj

@@ -0,0 +1,249 @@
+(ns vim-clojure-static.generate-test
+  (:require [clojure.test :as t :refer [deftest is]]
+            [vim-clojure-static.generate :as sut]))
+
+(deftest character-properties-test
+  (is (= #{"Lower" "Space" "XDigit" "Alnum" "Cntrl" "Graph" "Alpha" "Print"
+           "Blank" "Digit" "Upper" "Punct" "ASCII"}
+         (:posix sut/character-properties)))
+
+  (is (= #{"javaSpaceChar" "javaUnicodeIdentifierPart" "javaLetterOrDigit"
+           "javaTitleCase" "javaLowerCase" "javaDefined" "javaAlphabetic"
+           "javaIdentifierIgnorable" "javaJavaIdentifierStart"
+           "javaIdeographic" "javaWhitespace" "javaMirrored"
+           "javaUnicodeIdentifierStart" "javaISOControl" "javaUpperCase"
+           "javaDigit" "javaLetter" "javaJavaIdentifierPart"}
+         (:java sut/character-properties)))
+
+  (is (= #{"IDEOGRAPHIC" "HEX_DIGIT" "ALPHABETIC" "NONCHARACTERCODEPOINT"
+           "GRAPH" "PUNCTUATION" "WORD" "LETTER" "TITLECASE" "JOIN_CONTROL"
+           "CONTROL" "HEXDIGIT" "LOWERCASE" "NONCHARACTER_CODE_POINT"
+           "JOINCONTROL" "BLANK" "WHITESPACE" "ALNUM" "DIGIT" "WHITE_SPACE"
+           "ASSIGNED" "UPPERCASE" "PRINT"}
+         (:binary sut/character-properties)))
+
+  (is (= #{"LC" "Lo" "So" "Lm" "Sc" "Lt" "Co" "Me" "Zp" "L" "Sm" "Ps" "Cf" "M"
+           "LD" "S" "Ll" "Z" "Pi" "Cc" "C" "Pe" "Sk" "Pf" "Nd" "Nl" "P" "Cn"
+           "Zs" "Zl" "Mn" "Mc" "Lu" "N" "Cs" "Pc" "Pd" "Po" "No"}
+         (:category sut/character-properties)))
+
+  ;; Consider removing following once character-properties is updated
+  (is (= #{"COPT" "BAMU" "HAN" "LIMB" "TIBT" "MALAYALAM" "UNKNOWN" "ZYYY"
+           "MYANMAR" "CYPRIOT" "ZINH" "KANNADA" "TELU" "GUJARATI" "GREEK"
+           "AVESTAN" "LINEAR_B" "ARABIC" "UGAR" "GOTH" "DEVA" "CARI" "OGAM"
+           "OGHAM" "LANA" "SYRC" "HANO" "TAGB" "HANG" "NEW_TAI_LUE" "KHMER"
+           "MEROITIC_CURSIVE" "CANADIAN_ABORIGINAL" "JAVA" "LIMBU" "SUND" "GUJR"
+           "LYDIAN" "PLRD" "KANA" "OL_CHIKI" "ETHIOPIC" "HANGUL" "RUNR" "CHAM"
+           "CPRT" "CHAKMA" "LEPCHA" "TAI_THAM" "LATIN" "KALI" "GURMUKHI"
+           "SYRIAC" "PHOENICIAN" "KHAR" "XSUX" "TAI_LE" "ITAL" "ORKH" "BENGALI"
+           "PHAGS_PA" "OLD_SOUTH_ARABIAN" "MERO" "SAMR" "GEORGIAN" "TAMIL"
+           "BATK" "PHAG" "TAML" "OLD_TURKIC" "LATN" "CUNEIFORM" "MLYM" "LAOO"
+           "THAI" "BUHD" "PHNX" "BRAHMI" "SYLO" "ARMN" "BUHID" "HEBREW" "BAMUM"
+           "SAUR" "TAVT" "AVST" "ETHI" "REJANG" "LYCI" "CYRILLIC" "BRAI" "BRAH"
+           "TALU" "HANI" "SHRD" "VAII" "SINHALA" "MONG" "TFNG" "GREK" "ARMENIAN"
+           "TIFINAGH" "LISU" "HANUNOO" "THAANA" "NKOO" "CAKM" "TALE" "CANS"
+           "YIII" "TAGBANWA" "BALINESE" "BOPO" "ARAB" "MERC" "TELUGU" "TAGALOG"
+           "GLAG" "COPTIC" "CYRL" "SUNDANESE" "LINB" "ZZZZ" "RUNIC" "MANDAIC"
+           "INSCRIPTIONAL_PAHLAVI" "ORIYA" "GURU" "SORA" "DESERET" "VAI" "MAND"
+           "IMPERIAL_ARAMAIC" "GEOR" "KNDA" "THAA" "DEVANAGARI" "PHLI" "TIBETAN"
+           "BUGINESE" "KHAROSHTHI" "HIRA" "COMMON" "SINH" "KTHI" "ARMI" "MIAO"
+           "SYLOTI_NAGRI" "LYCIAN" "INSCRIPTIONAL_PARTHIAN" "UGARITIC"
+           "SORA_SOMPENG" "BALI" "SHARADA" "BENG" "MYMR" "GOTHIC" "HEBR"
+           "BRAILLE" "YI" "SHAW" "EGYP" "CHEROKEE" "HIRAGANA" "PRTI" "OSMA"
+           "EGYPTIAN_HIEROGLYPHS" "SARB" "MEETEI_MAYEK" "LEPC" "LYDI" "TAI_VIET"
+           "KAITHI" "INHERITED" "OLD_PERSIAN" "OSMANYA" "BUGI" "TGLG" "RJNG"
+           "NKO" "BATAK" "ORYA" "MTEI" "SHAVIAN" "SAURASHTRA" "OLD_ITALIC"
+           "DSRT" "SAMARITAN" "CHER" "TAKR" "BOPOMOFO" "JAVANESE" "GLAGOLITIC"
+           "LAO" "KATAKANA" "MEROITIC_HIEROGLYPHS" "XPEO" "OLCK" "KAYAH_LI"
+           "TAKRI" "KHMR" "MONGOLIAN" "CARIAN"}
+         (:script sut/character-properties)))
+
+  (is (= #{"ETHIOPICEXTENDED" "ARABICPRESENTATIONFORMS-A" "SUNDANESESUPPLEMENT"
+           "CYRILLIC SUPPLEMENTARY" "CONTROLPICTURES" "BAMUMSUPPLEMENT"
+           "IMPERIALARAMAIC" "LATIN-1SUPPLEMENT"
+           "CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A" "DOMINO TILES"
+           "SUPPLEMENTARYPRIVATEUSEAREA-B" "LATIN EXTENDED ADDITIONAL"
+           "VARIATION_SELECTORS_SUPPLEMENT" "CONTROL PICTURES"
+           "MEROITIC HIEROGLYPHS" "CUNEIFORM NUMBERS AND PUNCTUATION"
+           "HIGH_SURROGATES" "HANGUL_JAMO" "SUPERSCRIPTS_AND_SUBSCRIPTS"
+           "SYLOTI NAGRI" "MALAYALAM" "CYRILLIC_SUPPLEMENTARY" "ARROWS"
+           "ALCHEMICALSYMBOLS" "SUPPLEMENTALARROWS-A" "MYANMAR"
+           "LATINEXTENDED-B" "BASICLATIN" "PHAISTOSDISC"
+           "ANCIENTGREEKMUSICALNOTATION" "KANNADA" "GUJARATI" "OLDPERSIAN"
+           "LATINEXTENDEDADDITIONAL" "GREEK" "AVESTAN" "HIGH SURROGATES"
+           "ARABIC" "IDEOGRAPHIC_DESCRIPTION_CHARACTERS"
+           "CJK SYMBOLS AND PUNCTUATION" "BOX_DRAWING"
+           "ARABICMATHEMATICALALPHABETICSYMBOLS"
+           "PHONETIC EXTENSIONS SUPPLEMENT" "VEDIC_EXTENSIONS" "MEETEI MAYEK"
+           "OPTICAL CHARACTER RECOGNITION" "GENERAL PUNCTUATION"
+           "VERTICAL_FORMS" "LATINEXTENDED-C" "COMBINING HALF MARKS"
+           "CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT" "TAIVIET" "MEROITICCURSIVE"
+           "ENCLOSED_ALPHANUMERICS" "SUPPLEMENTAL_ARROWS_A" "KHMER_SYMBOLS"
+           "LINEAR_B_SYLLABARY" "DOMINO_TILES" "GREEK EXTENDED"
+           "BOPOMOFO_EXTENDED" "OGHAM" "IDEOGRAPHIC DESCRIPTION CHARACTERS"
+           "ARABIC SUPPLEMENT" "LATIN EXTENDED-B" "ETHIOPIC EXTENDED-A"
+           "ARABIC EXTENDED-A" "SUPPLEMENTALARROWS-B" "PHONETICEXTENSIONS"
+           "VARIATION_SELECTORS" "MISCELLANEOUS SYMBOLS" "LINEARBSYLLABARY"
+           "CYPRIOT SYLLABARY" "COMBININGDIACRITICALMARKSSUPPLEMENT"
+           "ENCLOSED CJK LETTERS AND MONTHS" "ALCHEMICAL_SYMBOLS"
+           "COMBINING DIACRITICAL MARKS SUPPLEMENT" "ENCLOSEDALPHANUMERICS"
+           "MISCELLANEOUS MATHEMATICAL SYMBOLS-A" "NEW_TAI_LUE" "KHMER"
+           "CJKUNIFIEDIDEOGRAPHS" "CUNEIFORM_NUMBERS_AND_PUNCTUATION"
+           "YIJING HEXAGRAM SYMBOLS" "KANASUPPLEMENT" "CJK_RADICALS_SUPPLEMENT"
+           "AEGEAN NUMBERS" "LATIN_EXTENDED_ADDITIONAL" "MEROITIC_CURSIVE"
+           "DEVANAGARIEXTENDED" "SORA SOMPENG" "LIMBU"
+           "TRANSPORT AND MAP SYMBOLS" "SUPPLEMENTALMATHEMATICALOPERATORS"
+           "LYDIAN" "BASIC_LATIN" "ARABIC_PRESENTATION_FORMS_B"
+           "YIJING_HEXAGRAM_SYMBOLS"
+           "UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED"
+           "MISCELLANEOUS SYMBOLS AND ARROWS" "CYRILLIC EXTENDED-A"
+           "COMMON_INDIC_NUMBER_FORMS" "MUSICAL_SYMBOLS" "SUNDANESE_SUPPLEMENT"
+           "UNIFIED CANADIAN ABORIGINAL SYLLABICS"
+           "UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS" "PHONETIC_EXTENSIONS"
+           "ALPHABETICPRESENTATIONFORMS" "OL_CHIKI" "ETHIOPIC"
+           "GEOMETRIC_SHAPES" "MAHJONG TILES" "ALCHEMICAL SYMBOLS" "CJK STROKES"
+           "CYRILLIC_EXTENDED_B" "CHAM" "CJKUNIFIEDIDEOGRAPHSEXTENSIONB"
+           "MEETEI MAYEK EXTENSIONS" "YI_SYLLABLES" "CHAKMA" "LEPCHA"
+           "ETHIOPIC_SUPPLEMENT" "TAI_THAM" "MISCELLANEOUS_SYMBOLS_AND_ARROWS"
+           "GEOMETRICSHAPES" "MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A"
+           "HALFWIDTHANDFULLWIDTHFORMS" "CJK_UNIFIED_IDEOGRAPHS" "PHAISTOS_DISC"
+           "MATHEMATICAL_ALPHANUMERIC_SYMBOLS" "MYANMAREXTENDED-A" "TAI THAM"
+           "CJK COMPATIBILITY IDEOGRAPHS" "IPA EXTENSIONS" "TAI VIET" "GURMUKHI"
+           "RUMI_NUMERAL_SYMBOLS" "MISCELLANEOUS_SYMBOLS" "CJK_STROKES" "SYRIAC"
+           "COMBINING MARKS FOR SYMBOLS" "VARIATION SELECTORS SUPPLEMENT"
+           "PHOENICIAN" "OLD PERSIAN" "ETHIOPIC_EXTENDED_A" "AEGEANNUMBERS"
+           "TAI_LE" "MUSICAL SYMBOLS" "LATIN_EXTENDED_B" "AEGEAN_NUMBERS"
+           "GENERALPUNCTUATION" "LINEARBIDEOGRAMS" "COMBINING DIACRITICAL MARKS"
+           "HANGUL COMPATIBILITY JAMO" "COMBININGMARKSFORSYMBOLS"
+           "CYPRIOTSYLLABARY" "OPTICALCHARACTERRECOGNITION" "BENGALI"
+           "CYRILLIC EXTENDED-B" "PHAGS_PA" "SUPPLEMENTALPUNCTUATION"
+           "LATINEXTENDED-A" "OLD_SOUTH_ARABIAN" "KANGXI_RADICALS"
+           "OPTICAL_CHARACTER_RECOGNITION" "GENERAL_PUNCTUATION" "GEORGIAN"
+           "PRIVATE_USE_AREA" "ENCLOSED_IDEOGRAPHIC_SUPPLEMENT"
+           "SURROGATES_AREA" "YI SYLLABLES" "GEORGIAN SUPPLEMENT" "TAMIL"
+           "BRAILLEPATTERNS" "LATIN_EXTENDED_A" "VARIATIONSELECTORSSUPPLEMENT"
+           "PHAISTOS DISC" "SPACING MODIFIER LETTERS" "OLDTURKIC"
+           "SPACINGMODIFIERLETTERS" "OLD_TURKIC" "OLDSOUTHARABIAN"
+           "UNIFIED CANADIAN ABORIGINAL SYLLABICS EXTENDED" "CUNEIFORM"
+           "CJK RADICALS SUPPLEMENT" "SUPERSCRIPTS AND SUBSCRIPTS"
+           "SUPPLEMENTAL ARROWS-A" "THAI" "COUNTINGRODNUMERALS"
+           "ARABICSUPPLEMENT" "MISCELLANEOUS SYMBOLS AND PICTOGRAPHS"
+           "PRIVATE USE AREA" "HANGULJAMOEXTENDED-A" "ANCIENTGREEKNUMBERS"
+           "TAIXUANJINGSYMBOLS" "ARABIC_EXTENDED_A" "PHAGS-PA"
+           "BOPOMOFO EXTENDED" "ENCLOSED_CJK_LETTERS_AND_MONTHS" "BRAHMI"
+           "SUPPLEMENTARY_PRIVATE_USE_AREA_A" "KAYAHLI" "MODIFIERTONELETTERS"
+           "MEETEIMAYEK" "MATHEMATICALOPERATORS" "CYPRIOT_SYLLABARY" "BUHID"
+           "MEROITIC CURSIVE" "MUSICALSYMBOLS" "ANCIENT_GREEK_MUSICAL_NOTATION"
+           "OLDITALIC" "MYANMAR_EXTENDED_A" "ENCLOSED IDEOGRAPHIC SUPPLEMENT"
+           "HEBREW" "BYZANTINE MUSICAL SYMBOLS" "BAMUM" "OL CHIKI"
+           "ARABIC_MATHEMATICAL_ALPHABETIC_SYMBOLS"
+           "HIGH PRIVATE USE SURROGATES" "CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D"
+           "HANGULJAMO" "DINGBATS" "MISCELLANEOUSSYMBOLS" "EMOTICONS"
+           "ARABIC_SUPPLEMENT" "CJKRADICALSSUPPLEMENT" "GREEK AND COPTIC"
+           "MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS" "REJANG"
+           "SMALL FORM VARIANTS" "ARABIC_PRESENTATION_FORMS_A"
+           "CJK COMPATIBILITY IDEOGRAPHS SUPPLEMENT" "CYRILLIC"
+           "VARIATION SELECTORS" "SUPPLEMENTARYPRIVATEUSEAREA-A"
+           "MATHEMATICAL ALPHANUMERIC SYMBOLS" "OLD TURKIC"
+           "SUNDANESE SUPPLEMENT" "TAILE" "BAMUM SUPPLEMENT"
+           "MEETEIMAYEKEXTENSIONS" "NUMBERFORMS" "SINHALA" "VEDICEXTENSIONS"
+           "PHONETIC EXTENSIONS" "SUPPLEMENTARY PRIVATE USE AREA-A"
+           "ENCLOSED_ALPHANUMERIC_SUPPLEMENT" "SUPPLEMENTAL ARROWS-B"
+           "BOXDRAWING" "NEWTAILUE" "VEDIC EXTENSIONS" "IPAEXTENSIONS"
+           "GEOMETRIC SHAPES" "ARABIC PRESENTATION FORMS-A" "KANGXIRADICALS"
+           "MISCELLANEOUSSYMBOLSANDPICTOGRAPHS" "MODIFIER_TONE_LETTERS"
+           "CJKUNIFIEDIDEOGRAPHSEXTENSIONC" "MEROITICHIEROGLYPHS"
+           "INSCRIPTIONALPAHLAVI" "ARMENIAN" "MEETEI_MAYEK_EXTENSIONS"
+           "YIJINGHEXAGRAMSYMBOLS" "SORASOMPENG" "KANGXI RADICALS"
+           "LATIN_EXTENDED_D" "CYRILLICEXTENDED-B"
+           "CJKUNIFIEDIDEOGRAPHSEXTENSIONA" "COMBINING_HALF_MARKS"
+           "HIGHPRIVATEUSESURROGATES" "ARABICEXTENDED-A"
+           "COMMON INDIC NUMBER FORMS" "TIFINAGH" "MISCELLANEOUSTECHNICAL"
+           "IPA_EXTENSIONS" "LISU" "CJK COMPATIBILITY"
+           "COMBINING_DIACRITICAL_MARKS" "CJK_COMPATIBILITY_IDEOGRAPHS"
+           "HANUNOO" "THAANA" "COMBININGDIACRITICALMARKS"
+           "ENCLOSEDCJKLETTERSANDMONTHS" "CURRENCY_SYMBOLS" "KHMER SYMBOLS"
+           "PRIVATEUSEAREA" "HANGUL JAMO EXTENDED-A" "YISYLLABLES"
+           "CYRILLICEXTENDED-A" "IMPERIAL ARAMAIC" "LOW_SURROGATES"
+           "KANA SUPPLEMENT" "LINEAR B IDEOGRAMS" "LINEAR B SYLLABARY"
+           "TAGBANWA" "COMMONINDICNUMBERFORMS" "HANGUL_COMPATIBILITY_JAMO"
+           "ALPHABETIC PRESENTATION FORMS" "BALINESE" "LATIN_1_SUPPLEMENT"
+           "CJKCOMPATIBILITY" "TELUGU" "TAGALOG" "MAHJONG_TILES"
+           "COMBINING DIACRITICAL MARKS FOR SYMBOLS"
+           "ENCLOSEDALPHANUMERICSUPPLEMENT" "LATIN-1 SUPPLEMENT"
+           "COMBINING_MARKS_FOR_SYMBOLS" "KANA_SUPPLEMENT" "COPTIC"
+           "INSCRIPTIONAL PAHLAVI" "DEVANAGARI_EXTENDED" "COMBININGHALFMARKS"
+           "SUNDANESE" "GEORGIANSUPPLEMENT" "EGYPTIAN HIEROGLYPHS"
+           "LATIN EXTENDED-A" "NUMBER_FORMS" "HANGUL_JAMO_EXTENDED_A"
+           "YI RADICALS" "HANGUL_SYLLABLES" "LATINEXTENDED-D"
+           "SUPPLEMENTARY_PRIVATE_USE_AREA_B" "COUNTING ROD NUMERALS"
+           "ETHIOPICEXTENDED-A" "TAI_XUAN_JING_SYMBOLS" "RUNIC" "MANDAIC"
+           "CJK_COMPATIBILITY_FORMS" "INSCRIPTIONAL_PAHLAVI"
+           "SUPPLEMENTARY PRIVATE USE AREA-B" "PHONETICEXTENSIONSSUPPLEMENT"
+           "INSCRIPTIONAL PARTHIAN" "ANCIENT_SYMBOLS"
+           "ARABICPRESENTATIONFORMS-B" "ORIYA" "BLOCKELEMENTS"
+           "HALFWIDTH AND FULLWIDTH FORMS" "YIRADICALS" "OLD ITALIC"
+           "LOW SURROGATES" "SUPERSCRIPTSANDSUBSCRIPTS" "ENCLOSED ALPHANUMERICS"
+           "DESERET" "PLAYINGCARDS" "MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B" "VAI"
+           "BLOCK_ELEMENTS" "LETTERLIKESYMBOLS" "CJK UNIFIED IDEOGRAPHS"
+           "TAI XUAN JING SYMBOLS" "ENCLOSED ALPHANUMERIC SUPPLEMENT"
+           "GREEKEXTENDED" "SUPPLEMENTAL_PUNCTUATION" "IMPERIAL_ARAMAIC"
+           "KATAKANAPHONETICEXTENSIONS" "CJKUNIFIEDIDEOGRAPHSEXTENSIOND"
+           "BYZANTINE_MUSICAL_SYMBOLS" "HANGULSYLLABLES" "RUMINUMERALSYMBOLS"
+           "LOWSURROGATES" "BOPOMOFOEXTENDED"
+           "COMBININGDIACRITICALMARKSFORSYMBOLS" "DEVANAGARI"
+           "COUNTING_ROD_NUMERALS" "KANBUN" "MODIFIER TONE LETTERS"
+           "SUPPLEMENTAL_ARROWS_B" "CJKCOMPATIBILITYIDEOGRAPHSSUPPLEMENT"
+           "ETHIOPIC SUPPLEMENT" "TIBETAN" "MISCELLANEOUS TECHNICAL" "BUGINESE"
+           "KHAROSHTHI" "UNIFIEDCANADIANABORIGINALSYLLABICSEXTENDED"
+           "BRAILLE_PATTERNS" "HIGH_PRIVATE_USE_SURROGATES" "BLOCK ELEMENTS"
+           "CJKSYMBOLSANDPUNCTUATION" "MYANMAR EXTENDED-A"
+           "COMBINING_DIACRITICAL_MARKS_SUPPLEMENT"
+           "ARABIC MATHEMATICAL ALPHABETIC SYMBOLS" "KHMERSYMBOLS"
+           "LATIN EXTENDED-C" "MIAO" "SYLOTI_NAGRI" "LYCIAN" "BASIC LATIN"
+           "ANCIENT SYMBOLS" "MISCELLANEOUSMATHEMATICALSYMBOLS-B"
+           "MISCELLANEOUSMATHEMATICALSYMBOLS-A" "VARIATIONSELECTORS"
+           "INSCRIPTIONAL_PARTHIAN" "MISCELLANEOUS MATHEMATICAL SYMBOLS-B"
+           "UGARITIC" "LINEAR_B_IDEOGRAMS" "CUNEIFORMNUMBERSANDPUNCTUATION"
+           "CJK COMPATIBILITY FORMS" "CJK UNIFIED IDEOGRAPHS EXTENSION B"
+           "UNIFIEDCANADIANABORIGINALSYLLABICS" "SORA_SOMPENG"
+           "MISCELLANEOUSSYMBOLSANDARROWS" "VERTICAL FORMS"
+           "CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B" "GREEKANDCOPTIC" "SHARADA"
+           "EGYPTIANHIEROGLYPHS" "CJKSTROKES" "BYZANTINEMUSICALSYMBOLS"
+           "SPECIALS" "BRAILLE PATTERNS" "ETHIOPIC EXTENDED"
+           "ETHIOPICSUPPLEMENT" "GOTHIC" "SYLOTINAGRI" "OLCHIKI"
+           "PHONETIC_EXTENSIONS_SUPPLEMENT" "ANCIENT_GREEK_NUMBERS" "CHEROKEE"
+           "ENCLOSEDIDEOGRAPHICSUPPLEMENT" "SMALLFORMVARIANTS" "HIRAGANA"
+           "CJK_SYMBOLS_AND_PUNCTUATION" "TAI LE" "CYRILLIC_EXTENDED_A"
+           "LETTERLIKE_SYMBOLS" "EGYPTIAN_HIEROGLYPHS" "MEETEI_MAYEK"
+           "RUMI NUMERAL SYMBOLS" "CJK UNIFIED IDEOGRAPHS EXTENSION D"
+           "KATAKANA PHONETIC EXTENSIONS" "PLAYING CARDS"
+           "SUPPLEMENTAL MATHEMATICAL OPERATORS" "CJK_COMPATIBILITY"
+           "BAMUM_SUPPLEMENT" "HALFWIDTH_AND_FULLWIDTH_FORMS" "GREEK_EXTENDED"
+           "TAI_VIET" "NUMBER FORMS" "CYRILLICSUPPLEMENTARY" "KAITHI"
+           "VERTICALFORMS" "MISCELLANEOUS_TECHNICAL" "TAGS" "LETTERLIKE SYMBOLS"
+           "MAHJONGTILES" "TRANSPORTANDMAPSYMBOLS" "OLD_PERSIAN"
+           "CYRILLIC SUPPLEMENT" "CJKCOMPATIBILITYIDEOGRAPHS"
+           "GEORGIAN_SUPPLEMENT" "OSMANYA" "SPACING_MODIFIER_LETTERS"
+           "CJK UNIFIED IDEOGRAPHS EXTENSION A"
+           "CJK UNIFIED IDEOGRAPHS EXTENSION C"
+           "IDEOGRAPHICDESCRIPTIONCHARACTERS" "INSCRIPTIONALPARTHIAN" "NKO"
+           "HIGHSURROGATES" "HANGULJAMOEXTENDED-B" "OLD SOUTH ARABIAN"
+           "ARABIC PRESENTATION FORMS-B" "BATAK" "SUPPLEMENTAL PUNCTUATION"
+           "ANCIENTSYMBOLS" "SHAVIAN" "NEW TAI LUE" "DOMINOTILES"
+           "CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C" "SAURASHTRA" "HANGUL JAMO"
+           "OLD_ITALIC" "CONTROL_PICTURES" "SUPPLEMENTAL_MATHEMATICAL_OPERATORS"
+           "KAYAH LI" "CJKCOMPATIBILITYFORMS" "TAITHAM" "SAMARITAN"
+           "HANGULCOMPATIBILITYJAMO" "MATHEMATICAL_OPERATORS" "YI_RADICALS"
+           "KATAKANA_PHONETIC_EXTENSIONS" "LATIN EXTENDED-D"
+           "HANGUL_JAMO_EXTENDED_B" "LATIN_EXTENDED_C" "BOPOMOFO"
+           "ANCIENT GREEK MUSICAL NOTATION" "JAVANESE" "CYRILLICSUPPLEMENT"
+           "PLAYING_CARDS" "GLAGOLITIC" "ETHIOPIC_EXTENDED"
+           "HANGUL JAMO EXTENDED-B" "MATHEMATICAL OPERATORS"
+           "ALPHABETIC_PRESENTATION_FORMS" "SMALL_FORM_VARIANTS" "LAO"
+           "BOX DRAWING" "KATAKANA" "DEVANAGARI EXTENDED" "MEROITIC_HIEROGLYPHS"
+           "HANGUL SYLLABLES" "TRANSPORT_AND_MAP_SYMBOLS"
+           "MATHEMATICALALPHANUMERICSYMBOLS" "KAYAH_LI" "TAKRI"
+           "CURRENCY SYMBOLS" "MONGOLIAN" "CARIAN" "ANCIENT GREEK NUMBERS"
+           "CURRENCYSYMBOLS"}
+         (:block sut/character-properties))))