vendor/regex-automata/data/tests/unicode.toml

[[tests]]
name = "unicode-literal1"
pattern = '☃'
input = "☃"
matches = [[0, 3]]

[[tests]]
name = "unicode-literal2"
pattern = '☃+'
input = "☃"
matches = [[0, 3]]

[[tests]]
name = "unicode-literal3"
options = ["case-insensitive"]
pattern = '☃+'
input = "☃"
matches = [[0, 3]]

[[tests]]
name = "unicode-literal4"
options = ["case-insensitive"]
pattern = 'Δ'
input = "δ"
matches = [[0, 2]]


[[tests]]
name = "unicode-class1"
pattern = '[☃Ⅰ]+'
input = "☃"
matches = [[0, 3]]

[[tests]]
name = "unicode-class2"
pattern = '\pN'
input = "Ⅰ"
matches = [[0, 3]]

[[tests]]
name = "unicode-class3"
pattern = '\pN+'
input = "Ⅰ1Ⅱ2"
matches = [[0, 8]]

[[tests]]
name = "unicode-class4"
pattern = '\PN+'
input = "abⅠ"
matches = [[0, 2]]

[[tests]]
name = "unicode-class5"
pattern = '[\PN]+'
input = "abⅠ"
matches = [[0, 2]]

[[tests]]
name = "unicode-class6"
pattern = '[^\PN]+'
input = "abⅠ"
matches = [[2, 5]]

[[tests]]
name = "unicode-class7"
pattern = '\p{Lu}+'
input = "ΛΘΓΔα"
matches = [[0, 8]]

[[tests]]
name = "unicode-class8"
options = ["case-insensitive"]
pattern = '\p{Lu}+'
input = "ΛΘΓΔα"
matches = [[0, 10]]

[[tests]]
name = "unicode-class9"
pattern = '\pL+'
input = "ΛΘΓΔα"
matches = [[0, 10]]

[[tests]]
name = "unicode-class10"
pattern = '\p{Ll}+'
input = "ΛΘΓΔα"
matches = [[8, 10]]


[[tests]]
name = "unicode-perl1"
pattern = '\w+'
input = "dδd"
matches = [[0, 4]]

[[tests]]
name = "unicode-perl2"
pattern = '\w+'
input = "⥡"
matches = []

[[tests]]
name = "unicode-perl3"
pattern = '\W+'
input = "⥡"
matches = [[0, 3]]

[[tests]]
name = "unicode-perl4"
pattern = '\d+'
input = "1२३9"
matches = [[0, 8]]

[[tests]]
name = "unicode-perl5"
pattern = '\d+'
input = "Ⅱ"
matches = []

[[tests]]
name = "unicode-perl6"
pattern = '\D+'
input = "Ⅱ"
matches = [[0, 3]]

[[tests]]
name = "unicode-perl7"
pattern = '\s+'
input = " "
matches = [[0, 3]]

[[tests]]
name = "unicode-perl8"
pattern = '\s+'
input = "☃"
matches = []

[[tests]]
name = "unicode-perl9"
pattern = '\S+'
input = "☃"
matches = [[0, 3]]


[[tests]]
name = "unicode-class-gencat1"
pattern = '\p{Cased_Letter}'
input = "Ａ"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat2"
pattern = '\p{Close_Punctuation}'
input = "❯"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat3"
pattern = '\p{Connector_Punctuation}'
input = "⁀"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat4"
pattern = '\p{Control}'
input = "\u009F"
matches = [[0, 2]]

[[tests]]
name = "unicode-class-gencat5"
pattern = '\p{Currency_Symbol}'
input = "￡"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat6"
pattern = '\p{Dash_Punctuation}'
input = "〰"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat7"
pattern = '\p{Decimal_Number}'
input = "𑓙"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat8"
pattern = '\p{Enclosing_Mark}'
input = "\uA672"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat9"
pattern = '\p{Final_Punctuation}'
input = "⸡"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat10"
pattern = '\p{Format}'
input = "\U000E007F"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat11"
pattern = '\p{Initial_Punctuation}'
input = "⸜"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat12"
pattern = '\p{Letter}'
input = "Έ"
matches = [[0, 2]]

[[tests]]
name = "unicode-class-gencat13"
pattern = '\p{Letter_Number}'
input = "ↂ"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat14"
pattern = '\p{Line_Separator}'
input = "\u2028"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat15"
pattern = '\p{Lowercase_Letter}'
input = "ϛ"
matches = [[0, 2]]

[[tests]]
name = "unicode-class-gencat16"
pattern = '\p{Mark}'
input = "\U000E01EF"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat17"
pattern = '\p{Math}'
input = "⋿"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat18"
pattern = '\p{Modifier_Letter}'
input = "𖭃"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat19"
pattern = '\p{Modifier_Symbol}'
input = "🏿"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat20"
pattern = '\p{Nonspacing_Mark}'
input = "\U0001E94A"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat21"
pattern = '\p{Number}'
input = "⓿"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat22"
pattern = '\p{Open_Punctuation}'
input = "｟"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat23"
pattern = '\p{Other}'
input = "\u0BC9"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat24"
pattern = '\p{Other_Letter}'
input = "ꓷ"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat25"
pattern = '\p{Other_Number}'
input = "㉏"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat26"
pattern = '\p{Other_Punctuation}'
input = "𞥞"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat27"
pattern = '\p{Other_Symbol}'
input = "⅌"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat28"
pattern = '\p{Paragraph_Separator}'
input = "\u2029"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat29"
pattern = '\p{Private_Use}'
input = "\U0010FFFD"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat30"
pattern = '\p{Punctuation}'
input = "𑁍"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat31"
pattern = '\p{Separator}'
input = "\u3000"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat32"
pattern = '\p{Space_Separator}'
input = "\u205F"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat33"
pattern = '\p{Spacing_Mark}'
input = "\U00016F7E"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat34"
pattern = '\p{Symbol}'
input = "⯈"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat35"
pattern = '\p{Titlecase_Letter}'
input = "ῼ"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gencat36"
pattern = '\p{Unassigned}'
input = "\U0010FFFF"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gencat37"
pattern = '\p{Uppercase_Letter}'
input = "Ꝋ"
matches = [[0, 3]]


[[tests]]
name = "unicode-class-emoji1"
pattern = '\p{Emoji}'
input = "\u23E9"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-emoji2"
pattern = '\p{emoji}'
input = "\U0001F21A"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-emoji3"
pattern = '\p{extendedpictographic}'
input = "\U0001FA6E"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-emoji4"
pattern = '\p{extendedpictographic}'
input = "\U0001FFFD"
matches = [[0, 4]]


[[tests]]
name = "unicode-class-gcb1"
pattern = '\p{grapheme_cluster_break=prepend}'
input = "\U00011D46"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gcb2"
pattern = '\p{gcb=regional_indicator}'
input = "\U0001F1E6"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gcb3"
pattern = '\p{gcb=ri}'
input = "\U0001F1E7"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gcb4"
pattern = '\p{regionalindicator}'
input = "\U0001F1FF"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-gcb5"
pattern = '\p{gcb=lvt}'
input = "\uC989"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-gcb6"
pattern = '\p{gcb=zwj}'
input = "\u200D"
matches = [[0, 3]]


[[tests]]
name = "unicode-class-word-break1"
pattern = '\p{word_break=Hebrew_Letter}'
input = "\uFB46"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-word-break2"
pattern = '\p{wb=hebrewletter}'
input = "\uFB46"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-word-break3"
pattern = '\p{wb=ExtendNumLet}'
input = "\uFF3F"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-word-break4"
pattern = '\p{wb=WSegSpace}'
input = "\u3000"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-word-break5"
pattern = '\p{wb=numeric}'
input = "\U0001E950"
matches = [[0, 4]]


[[tests]]
name = "unicode-class-sentence-break1"
pattern = '\p{sentence_break=Lower}'
input = "\u0469"
matches = [[0, 2]]

[[tests]]
name = "unicode-class-sentence-break2"
pattern = '\p{sb=lower}'
input = "\u0469"
matches = [[0, 2]]

[[tests]]
name = "unicode-class-sentence-break3"
pattern = '\p{sb=Close}'
input = "\uFF60"
matches = [[0, 3]]

[[tests]]
name = "unicode-class-sentence-break4"
pattern = '\p{sb=Close}'
input = "\U0001F677"
matches = [[0, 4]]

[[tests]]
name = "unicode-class-sentence-break5"
pattern = '\p{sb=SContinue}'
input = "\uFF64"
matches = [[0, 3]]