Ababababababbababa's picture
Duplicate from arbml/Ashaar
6faf7e7
raw
history blame
No virus
1.09 kB
"""
Constants that are used by the model
"""
HARAQAT = ["ู’", "ู‘", "ูŒ", "ู", "ู", "ู‹", "ูŽ", "ู"]
ARAB_CHARS = "ู‰ุนุธุญุฑุณูŠุดุถู‚ ุซู„ุตุทูƒุขู…ุงุฅู‡ุฒุกุฃูุคุบุฌุฆุฏุฉุฎูˆุจุฐุชู†"
PUNCTUATIONS = [".", "ุŒ", ":", "ุ›", "-", "ุŸ"]
VALID_ARABIC = HARAQAT + list(ARAB_CHARS)
BASIC_HARAQAT = {
"ูŽ": "Fatha ",
"ู‹": "Fathatah ",
"ู": "Damma ",
"ูŒ": "Dammatan ",
"ู": "Kasra ",
"ู": "Kasratan ",
"ู’": "Sukun ",
"ู‘": "Shaddah ",
}
ALL_POSSIBLE_HARAQAT = {
"": "No Diacritic ",
"ูŽ": "Fatha ",
"ู‹": "Fathatah ",
"ู": "Damma ",
"ูŒ": "Dammatan ",
"ู": "Kasra ",
"ู": "Kasratan ",
"ู’": "Sukun ",
"ู‘": "Shaddah ",
"ู‘ูŽ": "Shaddah + Fatha ",
"ู‘ู‹": "Shaddah + Fathatah ",
"ู‘ู": "Shaddah + Damma ",
"ู‘ูŒ": "Shaddah + Dammatan ",
"ู‘ู": "Shaddah + Kasra ",
"ู‘ู": "Shaddah + Kasratan ",
}