Skip to content

Commit 519402d

Browse files
agutkincopybara-github
authored andcommitted
Cleaning up Perso-Arabic presentation form rewrites.
PiperOrigin-RevId: 672607603
1 parent bc030d9 commit 519402d

File tree

1 file changed

+17
-32
lines changed

1 file changed

+17
-32
lines changed

nisaba/scripts/abjad_alphabet/data/Arab/presentation_forms.textproto

+17-32
Original file line numberDiff line numberDiff line change
@@ -14,17 +14,13 @@
1414

1515
# proto-file: nisaba/scripts/utils/unicode_strings.proto
1616
# proto-message: nisaba.UnicodeStrings
17-
17+
#
1818
# Generated from UnicodeData.txt and also from
1919
# NFKC Unicode transformations applied to Arabic Presentation Forms-A.
2020
# See:
2121
# - https://unicode.org/Public/UNIDATA/UnicodeData.txt
2222
# - https://en.wikipedia.org/wiki/Arabic_Presentation_Forms-A
2323
# - https://unicode.org/reports/tr15/
24-
#
25-
# Note:
26-
# Some forms are only supported starting from Unicode 14.0. These are
27-
# presently commented out.
2824

2925
uname_prefix: ["ARABIC", "ARABIC LETTER"]
3026

@@ -2248,24 +2244,6 @@ item { # "U+FD3D"
22482244
to_uname: ["ALEF", "FATHATAN"] to_raw: "اً"
22492245
}
22502246

2251-
# Unicode 14.0 onwards:
2252-
#LIGATURE RAHIMAHU ALLAAH U+FD40 ﵀ ﵀
2253-
#LIGATURE RADI ALLAAHU ANH U+FD41 ﵁ ﵁
2254-
#LIGATURE RADI ALLAAHU ANHAA U+FD42 ﵂ ﵂
2255-
#LIGATURE RADI ALLAAHU ANHUM U+FD43 ﵃ ﵃
2256-
#LIGATURE RADI ALLAAHU ANHUMAA U+FD44 ﵄ ﵄
2257-
#LIGATURE RADI ALLAAHU ANHUNNA U+FD45 ﵅ ﵅
2258-
#LIGATURE SALLALLAAHU ALAYHI WA-AALIH U+FD46 ﵆ ﵆
2259-
#LIGATURE ALAYHI AS-SALAAM U+FD47 ﵇ ﵇
2260-
#LIGATURE ALAYHIM AS-SALAAM U+FD48 ﵈ ﵈
2261-
#LIGATURE ALAYHIMAA AS-SALAAM U+FD49 ﵉ ﵉
2262-
#LIGATURE ALAYHI AS-SALAATU WAS-SALAAM U+FD4A ﵊ ﵊
2263-
#LIGATURE QUDDISA SIRRAH U+FD4B ﵋ ﵋
2264-
#LIGATURE SALLALLAHU ALAYHI WAAALIHEE WA-SALLAM U+FD4C ﵌ ﵌
2265-
#LIGATURE ALAYHAA AS-SALAAM U+FD4D ﵍ ﵍
2266-
#LIGATURE TABAARAKA WA-TAAALAA U+FD4E ﵎ ﵎
2267-
#LIGATURE RAHIMAHUM ALLAAH U+FD4F ﵏ ﵏
2268-
22692247
item { # "U+FD50"
22702248
uname: "LIGATURE TEH WITH JEEM WITH MEEM INITIAL FORM" raw: ""
22712249
to_uname: ["TEH", "JEEM", "MEEM"] to_raw: "تجم"
@@ -2856,9 +2834,6 @@ item { # "U+FDC7"
28562834
to_uname: ["NOON", "JEEM", "YEH"] to_raw: "نجي"
28572835
}
28582836

2859-
# Unicode 14.0 onwards.
2860-
#LIGATURE SALAAMUHU ALAYNAA U+FDCF ﷏ ﷏
2861-
28622837
item { # "U+FDF0"
28632838
uname: "LIGATURE SALLA USED AS KORANIC STOP SIGN ISOLATED FORM" raw: ""
28642839
to_uname: ["SAD", "LAM", "YEH BARREE"] to_raw: "صلے"
@@ -2909,12 +2884,22 @@ item { # "U+FDF9"
29092884
to_uname: ["SAD", "LAM", "ALEF MAKSURA"] to_raw: "صلى"
29102885
}
29112886

2912-
# TODO: Double-check.
2913-
# Unicode 14.0 onwards.
2914-
# item { # "U+FDFD"
2915-
# uname: "LIGATURE BISMILLAH AR-RAHMAN AR-RAHEEM" raw: "﷽"
2916-
# to_uname: "LIGATURE BISMILLAH AR-RAHMAN AR-RAHEEM" to_raw: "﷽"
2917-
#}
2887+
item { # "U+FDFA"
2888+
uname: "LIGATURE SALLALLAHOU ALAYHE WASALLAM" raw: ""
2889+
to_uname: [
2890+
"SAD", "LAM", "ALEF MAKSURA", "SPACE",
2891+
"ALEF", "LAM", "LAM", "HEH", "SPACE",
2892+
"AIN", "LAM", "YEH", "HEH", "SPACE",
2893+
"WAW", "SEEN", "LAM", "MEEM"
2894+
] to_raw: "صلى الله عليه وسلم"
2895+
}
2896+
2897+
item { # "U+FDFB"
2898+
uname: "LIGATURE JALLAJALALOUHOU" raw: ""
2899+
to_uname: [
2900+
"JEEM", "LAM", "SPACE", "JEEM", "LAM", "ALEF", "LAM", "HEH"
2901+
] to_raw: "جل جلاله"
2902+
}
29182903

29192904
item { # "U+FE80"
29202905
uname: "HAMZA ISOLATED FORM" raw: ""

0 commit comments

Comments
 (0)