icu_provider_source 2.2.0

# This file is part of ICU4X. For terms of use, please see the file
# called LICENSE at the top level of the ICU4X source tree
# (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).

# These line boundary rules are based on UAX #14, Unicode Version 15.1.0.
# https://www.unicode.org/reports/tr14/tr14-51.html

segmenter_type = "line"

# Any = non combined property (AI..ZWJ)

[[tables]]
name = "AI"

[[tables]]
name = "AK"

[[tables]]
name = "AL"

[[tables]]
# LB28a
name = "AL_DOTTED_CIRCLE"

[[tables]]
name = "AP"

[[tables]]
name = "AS"

[[tables]]
name = "B2"

[[tables]]
name = "BA"

[[tables]]
name = "BB"

[[tables]]
name = "BK"

[[tables]]
name = "CB"

[[tables]]
name = "CJ"

[[tables]]
name = "CL"

[[tables]]
name = "CM"

[[tables]]
name = "CP"

[[tables]]
name = "CR"

[[tables]]
name = "EB"

[[tables]]
name = "EM"

[[tables]]
name = "EX"

[[tables]]
name = "GL"

[[tables]]
name = "H2"

[[tables]]
name = "H3"

[[tables]]
name = "HL"

[[tables]]
name = "HY"

[[tables]]
name = "ID"

[[tables]]
# For LB30b
# Unassigned codepoints with Line_Break=ID in some blocks
# are also assigned the Extended_Pictographic property.
name = "ID_CN"

[[tables]]
name = "IN"

[[tables]]
name = "IS"

[[tables]]
name = "JL"

[[tables]]
name = "JT"

[[tables]]
name = "JV"

[[tables]]
name = "LF"

[[tables]]
name = "NL"

[[tables]]
name = "NS"

[[tables]]
name = "NU"

[[tables]]
name = "OP_EA"

[[tables]]
# LB30
name = "OP_OP30"

[[tables]]
name = "PO"

[[tables]]
name = "PO_EAW"

[[tables]]
name = "PR"

[[tables]]
name = "PR_EAW"

[[tables]]
name = "QU"

[[tables]]
# LB15b
name = "QU_PF"

[[tables]]
# LB15a
name = "QU_PI"

[[tables]]
name = "RI"

[[tables]]
name = "SA"

[[tables]]
name = "SP"

[[tables]]
name = "SY"

[[tables]]
name = "VF"

[[tables]]
name = "VI"

[[tables]]
name = "WJ"

[[tables]]
name = "XX"

[[tables]]
name = "ZW"

[[tables]]
name = "ZWJ"

[[tables]]
# LB8
name = "ZW"
left = "ZW"
right = "SP"

[[tables]]
# LB14
name = "OP_SP"
left = "OP_OP30"
right = "SP"

[[tables]]
# LB14
name = "OP_SP"
left = "OP_EA"
right = "SP"

[[tables]]
# LB14
name = "OP_SP"
left = "OP_SP"
right = "SP"

[[tables]]
# LB15a
name = "QU_SP_LB15A"
left = "QU_PI"
right = "SP"

# The following transitions are X QU_PI → QU unless X is a state that
# implies we are after (sot | BK | CR | LF | NL | OP | QU | GL | SP | ZW).
# We therefore reach the state QU_PI only if we match the initial
# context
#   (sot | BK | CR | LF | NL | OP | QU | GL | SP | ZW) [\p{Pi}&QU]
# of LB15a.
[[tables]]
# LB15a
name = "QU"
left = "AI"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AK"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AL"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AL_DOTTED_CIRCLE"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AP"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AS"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "B2"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "BA"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "BB"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "CB"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "CJ"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "CL"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "CM"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "CP"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "EB"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "EM"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "EX"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "H2"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "H3"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "HL"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "HY"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "ID"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "ID_CN"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "IN"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "IS"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "JL"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "JT"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "JV"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "NS"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "NU"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "PO"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "PO_EAW"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "PR"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "PR_EAW"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "RI"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "SA"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "SY"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "VF"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "VI"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "WJ"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "XX"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "ZWJ"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "HL_HY"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AK_VI"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "AL_DOTTED_CIRCLE_LB28A"
right = "QU_PI"

[[tables]]
# LB15a
name = "QU"
left = "RI_RI"
right = "QU_PI"
# End of the X QU_PI → QU transitions.

[[tables]]
# LB15a
name = "QU_SP_LB15A"
left = "QU_PI"
right = "SP"

[[tables]]
# LB15a
name = "QU_SP_LB15A"
left = "QU_SP_LB15A"
right = "SP"

[[tables]]
# LB15b
name = "QU_LB15B"
left = "SP"
right = "QU_PF"

[[tables]]
# LB15b / LB18
name = "QU_LB15B"
left = "B2_SP"
right = "QU_PF"
interm_break_state = true

[[tables]]
# LB15b / LB20
name = "CB_QU_LB15B"
left = "CB"
right = "QU_PF"
interm_break_state = true

[[tables]]
# LB15b / LB18
name = "QU_LB15B"
left = "CL_CP_SP"
right = "QU_PF"
interm_break_state = true

[[tables]]
# LB14 / LB15b / LB19
name = "QU_PF"
left = "OP_SP"
right = "QU_PF"

[[tables]]
# LB16
name = "CL_CP_SP"
left = "CL"
right = "SP"

[[tables]]
# LB16
name = "CL_CP_SP"
left = "CP"
right = "SP"

[[tables]]
# LB16
name = "CL_CP_SP"
left = "CL_CP_SP"
right = "SP"

[[tables]]
# LB17
name = "B2_SP"
left = "B2"
right = "SP"

[[tables]]
# LB17
name = "B2_SP"
left = "B2_SP"
right = "SP"

[[tables]]
# LB21
name = "HL_HY"
left = "HL"
right = "HY"

[[tables]]
# LB21
name = "HL_HY"
left = "HL"
right = "BA"

[[tables]]
# LB9
name = "QU_PF"
left = "QU_PF"
right = "CM"

[[tables]]
# LB9
name = "QU_PI"
left = "QU_PI"
right = "CM"

[[tables]]
# LB28a
name = "AK_VI"
left = "AK"
right = "VI"

[[tables]]
# LB28a
name = "AK_VI"
left = "AL_DOTTED_CIRCLE"
right = "VI"

[[tables]]
# LB28a
name = "AK_VI"
left = "AS"
right = "VI"

[[tables]]
# LB28a
name = "AK_AK"
left = "AK"
right = "AK"

[[tables]]
# LB28a
name = "AK_AK"
left = "AK"
right = "AL_DOTTED_CIRCLE"

[[tables]]
# LB28a
name = "AK_AK"
left = "AK"
right = "AS"

[[tables]]
# LB28a
name = "AK_AK"
left = "AL_DOTTED_CIRCLE"
right = "AK"

[[tables]]
# LB28a
name = "AL_DOTTED_CIRCLE_LB28A"
left = "AL_DOTTED_CIRCLE"
right = "AL_DOTTED_CIRCLE"

[[tables]]
# LB28a
name = "AK_AK"
left = "AL_DOTTED_CIRCLE"
right = "AS"

[[tables]]
# LB28a
name = "AK_AK"
left = "AS"
right = "AK"

[[tables]]
# LB28a
name = "AK_AK"
left = "AS"
right = "AL_DOTTED_CIRCLE"

[[tables]]
# LB28a
name = "AK_AK"
left = "AS"
right = "AS"

[[tables]]
# LB28a
name = "AK_AK"
left = "AS"
right = "AL_DOTTED_CIRCLE"

[[tables]]
# LB30a
name = "RI_RI"
left = "RI"
right = "RI"

[[rules]]
# LB3
left = [ "Any" ]
right = [ "eot" ]
break_state = true

[[rules]]
# LB3
left = [ "B2_SP", "CL_CP_SP", "HL_HY", "OP_SP", "QU_SP_LB15A", "CB_QU_LB15B", "QU_LB15B", "RI_RI" ]
right = [ "eot" ]
break_state = true

[[rules]]
# LB4
left = [ "BK" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB5
left = [ "CR" ]
right = [ "LF" ]
break_state = false

[[rules]]
# LB5
left = [ "CR", "LF", "NL" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB6
left = [ "Any" ]
right = [ "BK", "CR", "LF", "NL" ]
break_state = false

[[rules]]
# LB6
left = [ "B2_SP", "CL_CP_SP", "HL_HY", "OP_SP", "QU_SP_LB15A", "RI_RI", "AK_VI" ]
right = [ "BK", "CR", "LF", "NL" ]
break_state = false

[[rules]]
# LB7
left = [ "Any" ]
right = [ "SP", "ZW" ]
break_state = false

[[rules]]
# LB7
left = [ "HL_HY", "RI_RI", "AK_VI" ]
right = [ "SP" ]
break_state = false

[[rules]]
# LB7
left = [ "B2_SP", "CL_CP_SP", "HL_HY", "OP_SP", "QU_SP_LB15A", "RI_RI", "AK_VI" ]
right = [ "ZW" ]
break_state = false

[[rules]]
# LB8
left = [ "ZW" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB11
left = [ "WJ" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB11
left = [ "Any" ]
right = [ "WJ" ]
break_state = false

[[rules]]
# LB11
left = [ "B2_SP", "CL_CP_SP", "HL_HY", "OP_SP", "QU_SP_LB15A", "RI_RI", "AK_VI" ]
right = [ "WJ" ]
break_state = false

[[rules]]
# LB12
left = [ "GL" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB12a
left = [ "HL_HY", "OP_SP", "RI_RI", "AK_VI" ]
right = [ "GL" ]
break_state = false

[[rules]]
# LB12a
left = [ "B2_SP", "CL_CP_SP", "SP", "BA", "HY" ]
right = [ "GL" ]
break_state = true

[[rules]]
# LB12a
left = [ "Any" ]
right = [ "GL" ]
break_state = false

[[rules]]
# LB13
left = [ "Any", "RI_RI" ]
right = [ "CL", "CP", "EX", "IS", "SY" ]
break_state = false

[[rules]]
# LB13
left = [ "B2_SP", "CL_CP_SP", "QU_SP_LB15A", "AK_VI" ]
right = [ "CL", "CP", "EX", "IS", "SY" ]
break_state = false

[[rules]]
# LB14
left = [ "OP_OP30", "OP_EA", "OP_SP" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB15a
left = [ "QU_SP_LB15A" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB15b
left = [ "QU_PF", "QU_LB15B", "CB_QU_LB15B" ]
right = [ "SP", "GL", "WJ", "CL", "QU", "QU_PI", "QU_PF", "CP", "EX", "IS", "SY", "BK", "CR", "LF", "NL", "ZW" ]
break_state = false

[[rules]]
# LB15b
left = [ "QU_PF", "QU_LB15B", "CB_QU_LB15B" ]
right = [ "eot" ]
break_state = true

[[rules]]
# LB16
left = [ "CL", "CP", "CL_CP_SP" ]
right = [ "CJ", "NS" ] # NS = CJ or NS
break_state = false

[[rules]]
# LB17
left = [ "B2", "B2_SP" ]
right = [ "B2" ]
break_state = false

[[rules]]
# LB18
left = [ "SP" ]
right = [ "Any"]
break_state = true

[[rules]]
# LB18
left = [ "B2_SP", "CL_CP_SP" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB19
left = [ "Any", "RI_RI", "AK_VI" ]
right = [ "QU", "QU_PF", "QU_PI" ]
break_state = false

[[rules]]
# LB19
left = [ "QU", "QU_PF", "QU_PI" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB19
left = [ "CB_QU_LB15B" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB20
left = [ "CB" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB20
left = [ "Any", "HL_HY" ]
right = [ "CB" ]
break_state = true

[[rules]]
# LB21
left = [ "Any", "RI_RI", "AK_VI" ]
right = [ "BA", "HY", "CJ", "NS" ] # NS = CJ or NS
break_state = false

[[rules]]
# LB21
left = [ "BB" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB21a
left = [ "HL_HY" ]
right = [ "Any" ]
break_state = false

[[rules]]
# LB21b
left = [ "SY" ]
right = [ "HL" ]
break_state = false

[[rules]]
# LB22
left = [ "Any", "RI_RI", "AK_VI" ]
right = [ "IN" ]
break_state = false

[[rules]]
# LB23
# (LB1 AL = AI, AL, SA or XX)
# (LB10 AL = CM)
left = [ "AI", "AL", "AL_DOTTED_CIRCLE", "SA", "XX", "HL", "CM", "ZWJ" ]
right = [ "NU" ]
break_state = false

[[rules]]
# LB23
# (LB1 AL = AI, AL, SA or XX)
left = [ "NU" ]
right = [ "AI", "AL", "AL_DOTTED_CIRCLE", "SA", "XX", "HL" ]
break_state = false

[[rules]]
# LB23a
left = [ "PR", "PR_EAW" ]
right = [ "ID", "ID_CN", "EB", "EM" ]
break_state = false

[[rules]]
# LB23a
left = [ "ID", "ID_CN", "EB", "EM" ]
right = [ "PO", "PO_EAW" ]
break_state = false

[[rules]]
# LB24
# (LB1 AL = AI, AL, SA or XX)
# (LB10 AL = CM)
left = [ "PR", "PO", "PR_EAW", "PO_EAW" ]
right = [ "AI", "AL", "AL_DOTTED_CIRCLE", "SA", "XX", "HL" ]
break_state = false

[[rules]]
# LB24
# (LB1 AL = AI, AL, SA or XX)
# (LB10 AL = CM)
left = [ "AI", "AL", "AL_DOTTED_CIRCLE", "SA", "XX", "HL", "CM", "ZWJ" ]
right = [ "PR", "PO", "PR_EAW", "PO_EAW" ]
break_state = false

[[rules]]
# LB25
left = [ "CL", "CP", "NU" ]
right = [ "PR", "PO", "PR_EAW", "PO_EAW" ]
break_state = false

[[rules]]
# LB25
left = [ "PO", "PR", "PR_EAW", "PO_EAW" ]
right = [ "OP_EA", "OP_OP30", "NU" ]
break_state = false

[[rules]]
# LB25
left = [ "HY", "IS", "NU", "SY" ]
right = [ "NU" ]
break_state = false

[[rules]]
# LB26
left = [ "JL" ]
right = [ "JL", "JV", "H2", "H3" ]
break_state = false

[[rules]]
# LB26
left = [ "JV", "H2" ]
right = [ "JV", "JT" ]
break_state = false

[[rules]]
# LB26
left = [ "JT", "H3" ]
right = [ "JT" ]
break_state = false

[[rules]]
# LB27
left = [ "JL", "JV", "JT", "H2", "H3" ]
right = [ "PO", "PO_EAW" ]
break_state = false

[[rules]]
# LB27
left = [ "PR", "PR_EAW" ]
right = [ "JL", "JV", "JT", "H2", "H3" ]
break_state = false

[[rules]]
# LB28
# (LB1 AL = AI, AL, SA or XX)
# (LB10 AL = CM)
left = [ "AL", "HL", "AI", "XX", "SA", "CM", "ZWJ" ]
right = [ "AL", "AL_DOTTED_CIRCLE", "HL", "AI", "XX", "SA" ]
break_state = false

[[rules]]
# LB28
left = [ "AL_DOTTED_CIRCLE", "AL_DOTTED_CIRCLE_LB28A" ]
right = [ "AL", "HL", "AI", "XX", "SA" ]
break_state = false

[[rules]]
# LB28a
left = [ "AP" ]
right = [ "AK", "AL_DOTTED_CIRCLE", "AS" ]
break_state = false

[[rules]]
# LB28a
left = [ "AK", "AL_DOTTED_CIRCLE", "AS" ]
right = [ "VF" ]
break_state = false

[[rules]]
# LB28a
left = [ "AK_VI" ]
right = [ "AK", "AL_DOTTED_CIRCLE" ]
break_state = false

[[rules]]
# LB28a
left = [ "AK_AK", "AL_DOTTED_CIRCLE_LB28A" ]
right = [ "VF" ]
break_state = false

[[rules]]
# LB29
# (LB1 AL = AI, AL, SA or XX)
left = [ "IS" ]
right = [ "AL", "HL", "AI", "SA", "XX", "AL_DOTTED_CIRCLE" ]
break_state = false

[[rules]]
# LB30
# (LB1 AL = AI, AL, SA or XX)
# (LB10 AL = CM)
left = [ "AL", "HL", "NU", "AI", "XX", "SA", "CM", "ZWJ", "AL_DOTTED_CIRCLE", "AL_DOTTED_CIRCLE_LB28A" ]
right = [ "OP_OP30" ]
break_state = false

[[rules]]
# LB30
# (LB1 AL = AI, AL, SA or XX)
left = [ "CP" ]
right = [ "AL", "HL", "NU", "AI", "XX", "SA", "AL_DOTTED_CIRCLE" ]
break_state = false

[[rules]]
# LB30a
left = [ "RI_RI" ]
right = [ "RI" ]
break_state = true

[[rules]]
# LB30a
left = [ "RI_RI" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB30b
left = [ "EB", "ID_CN" ]
right = [ "EM" ]
break_state = false

[[rules]]
# LB31
left = [ "AK_VI" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB31
left = [ "AL_DOTTED_CIRCLE_LB28A" ]
right = [ "Any" ]
break_state = true

[[rules]]
# LB31
left = [ "Any" ]
right = [ "Any" ]
break_state = true