scriptshifter/scriptshifter/tables/data/pushto.yml

---
general:
  name: Pushto
  case_sensitive: false

roman_to_script:
  map:
    # Punctuation marks:
    # %": "\u066A # cannot transliterate the truncation character"
    "*": "\u066D"
    ",": "\u060C"
    ";": "\u061B"
    "?": "\u061F"

    # Exceptions for specific words
    # Allah
    "Alla\u0304h": "\u0627\u0644\u0644\u0647"

    # Qur'an
    "Qur\u02BCa\u0304n": "\u0642\u0631\u0622\u0646"

    # lillah
    "lilla\u0304h": "\u0644\u0644\u0647"

    # billah
    "billa\u0304h": "\u0628\u0644\u0644\u0647"

    # Rahman
    "Rah\u0323ma\u0304n": "\u0631\u062D\u0645\u0646"

    # ta'lif

    # Ibn when it appears in the middle of a name sequence
    "ibn": "\u0628\u0646"

    # Parsing "sh[dot below] as in "Ishaq [name]"
    "%sh\u0323%": "\u0633\u062D"

    # Edition statements with Latin number
    "Cha\u0304p-i 1": "\u0686\u0627\u067E 1"
    "Cha\u0304p-i 2": "\u0686\u0627\u067E 2"
    "Cha\u0304p-i 3": "\u0686\u0627\u067E 3"
    "Cha\u0304p-i 4": "\u0686\u0627\u067E 4"
    "Cha\u0304p-i 5": "\u0686\u0627\u067E 5"
    "Cha\u0304p-i 6": "\u0686\u0627\u067E 6"
    "Cha\u0304p-i 7": "\u0686\u0627\u067E 7"
    "Cha\u0304p-i 8": "\u0686\u0627\u067E 8"
    "Cha\u0304p-i 9": "\u0686\u0627\u067E 9"

    # Numbers (Extended Arabic \u06F0-06F9 for Persian)
    # currently *not* valid MARC21 characters
    "# 0": "\u06F0"
    "# 1": "\u06F1"
    "# 2": "\u06F2"
    "# 3": "\u06F3"
    "# 4": "\u06F4"
    "# 5": "\u06F5"
    "# 6": "\u06F6"
    "# 7": "\u06F7"
    "# 8": "\u06F8"
    "# 9": "\u06F9"

    # Use Basic Arabic \u0660-0669, instead:
    "0": "\u0660"
    "1": "\u0661"
    "2": "\u0662"
    "3": "\u0663"
    "4": "\u0664"
    "5": "\u0665"
    "6": "\u0666"
    "7": "\u0667"
    "8": "\u0668"
    "9": "\u0669"

    "# prime ": " ZWNJ"
    "\u02B9a\u0304": "\u200C\u0622"
    "\u02B9a": "\u200C\u0627"
    "%\u02B9i\u0304": "\u200C\u0627\u0649"
    "\u02B9i\u0304": "\u200C\u0627\u064A"
    "i\u0304\u02B9": "\u0649\u200C"
    "a\u0323y\u02B9": "\u06D3\u200C"
    "\u02B9": "\u200C"

    # Vowel and vowel/consonant combinations
    # and hyphenated suffixes:
    # izafah here
    "%h-\u02BEi": "\u06C0"
    "%h-\u02BCi": "\u06C0 "
    "%-\u02BEi": "\u06C0"
    "%-\u02BCi": "\u06C0"

    "%h-yi": "\u0647\u200C\u0649"
    "%-yi": "\u0649"
    "%yi": "\u0649"
    "%\u02BEi\u0304": "\u0649"
    "%\u02BCi\u0304": "\u0649"
    "%i\u0304-i": "\u0649"
    "%i\u0304": "\u0649"

    "%a\u0323h": "\u06C0"
    "%ayy-i": "\u0649"
    "%a\u0304y": "\u0627\u0649"
    "%a\u0301": "\u0649\u0670"
    "%al-i": "\u0644"
    "%-i": ""

    # Hyphenated prefixes:
    "wa-": "\u0648"
    "bi-": "\u0628"
    "al-": "\u0627\u0644"
    "lil-": "\u0644\u0644"
    "li-": "\u0644"
    "ka-": "\u0643"

    # Diphthongs here
    "Ayy%": "\u0627\u064A"
    "ayy%": "\u0627\u064A"
    "%a\u0323y": "\u06D3"
    "%ay": "\u0649"
    "\u02BBAw": "\u0639\u0648"
    "\u02BBaw": "\u0639\u0648"
    "Aw": "\u0627\u0648"
    "aw": "\u0648"

    # ayn combo
    "%\u02BBa\u0304%": "\u0639\u0627"

    # alif combos
    # [final position]
    "%a\u0304\u02BE": "\u0627\u0621"
    "%a\u0304\u02BC": "\u0627\u0621"
    "%a\u0304\u02BEi\u0304": "\u0627\u0626\u0649"
    "%a\u0304\u02BCi\u0304": "\u0627\u0626\u0649"
    # [initial position]
    "A\u0304\u02BEi\u0304%": "\u0622\u0626\u064A"
    "A\u0304\u02BCi\u0304%": "\u0622\u0626\u064A"
    "a\u0304\u02BEi\u0304%": "\u0622\u0626\u064A"
    "a\u0304\u02BCi\u0304%": "\u0622\u0626\u064A"

    # [medial position]
    "a\u0304\u02BEi\u0304": "\u0627\u0626\u064A"
    "a\u0304\u02BCi\u0304": "\u0627\u0626\u064A"
    "a\u0304\u02BEi": "\u0627\u0626"
    "a\u0304\u02BCi": "\u0627\u0626"
    "a\u02BEi\u0304": "\u0626\u064A"
    "a\u02BCi\u0304": "\u0626\u064A"

    #a [macron] hamza followed by e
    "a\u0304\u02BEe": "\u0627\u0626\u064A"
    "a\u0304\u02BCe": "\u0627\u0626\u064A"

    # s[dot below]h
    "S\u0323h": "\u069A"
    "s\u0323hs\u0323h": "\u069A"
    "s\u0323h": "\u069A"

    # A
    "\u02BBA\u0304": "\u0639\u0627"
    "\u02BBa\u0304": "\u0639\u0627"
    "\u02BBA%": "\u0639"
    "\u02BBa": "\u0639"
    "A\u02BB": "\u0623\u0639"
    "a\u02BB": "\u0639"
    "a\u02BE": "\u0623"
    "a\u02BC": "\u0623"
    "\u02BEa": "\u0623"
    "\u02BCa": "\u0623"
    "a\u0304\u02BE": "\u0621"
    "a\u0304\u02BC": "\u0621"
    "A\u0304%": "\u0622"
    "a\u0304%": "\u0622"
    "A\u0304": "\u0627"
    "a\u0304": "\u0627"
    "A\u0301": "\u0649"
    "a\u0301": "\u0649"
    "ayy": "\u064A"

    # heh hamzah at end
    "a\u0323h": "\u06C0"

    "A%": "\u0627"
    "a%": "\u0627"
    "A": ""
    "a": ""

    # E
    "%e": "\u06D0"
    "E%": "\u0627\u064A"
    "e%": "\u0627\u064A"
    "e": "\u06D0"

    # I 
    "I\u02BB%": "\u0627\u0639"
    "i\u02BB%": "\u0627\u0639"
    "i\u02BB": "\u0639"
    "\u02BBI%": "\u0639"
    "I\u02BE%": "\u0627\u0626"
    "I\u02BC%": "\u0627\u0626"
    "i\u02BE": "\u0626"
    "i\u02BC": "\u0626"
    "\u02BEi": "\u0626"
    "\u02BCi": "\u0626"
    "I\u0304%": "\u0627\u064A"
    "i\u0304%": "\u0627\u064A"
    "i\u0304y": "\u064A"
    "i\u0304": "\u064A"
    "I%": "\u0627"
    "i%": "\u0627"
    "I": ""
    "i": ""

    # O
    "o%": "\u0627\u0648"
    "O%": "\u0627\u0648"
    "o": "\u0648"

    # U
    # (u [macron] hamza combos)
    "u\u0304\u02BE": "\u0624"
    "u\u0304\u02BC": "\u0624"

    "\u02BEu\u0304": "\u0626\u0648"
    "\u02BCu\u0304": "\u0626\u0648"
    "U\u02BE%": "\u0627\u0624"
    "U\u02BC%": "\u0627\u0624"
    "u\u02BE": "\u0624"
    "u\u02BC": "\u0624"
    "U\u0304w%": "\u0627\u0628"
    "u\u0304w": "\u0628"
    "U\u0304": "\u0627\u0648"
    "u\u0304": "\u0648"
    "U%": "\u0627"
    "u%": "\u0627"
    "U": ""
    "u": ""

    # Consonants:
    "B": "\u0628"
    "bb": "\u0628"
    "b": "\u0628"
    "P": "\u067E"
    "pp": "\u067E"
    "p": "\u067E"

    "T\u0324": "\u0637"
    "t\u0324t\u0324": "\u0637"
    "t\u0324": "\u0637"

    "T\u0323": "\u067C"
    "t\u0323t\u0323": "\u067C"
    "t\u0323": "\u067C"

    "T": "\u062A"
    "tt": "\u062A"
    "t": "\u062A"

    # s[dot]h used to be here

    "Sh": "\u0634"
    "shsh": "\u0634"
    "sh": "\u0634"

    "S\u0323": "\u0635"
    "s\u0323s\u0323": "\u0635"
    "s\u0323": "\u0635"

    "S\u0332": "\u062B"
    "s\u0332s\u0332": "\u062B"
    "s\u0332": "\u062B"

    "S\u0307": "\u0685"
    "s\u0307s\u0307": "\u0685"
    "s\u0307": "\u0685"

    "S": "\u0633"
    "ss": "\u0633"
    "s": "\u0633"
    "J": "\u062C"
    "jj": "\u062C"
    "j": "\u062C"
    "Ch": "\u0686"
    "chch": "\u0686"
    "ch": "\u0686"
    "H\u0323": "\u062D"
    "h\u0323h\u0323": "\u062D"
    "h\u0323": "\u062D"
    "H": "\u0647"
    "hh": "\u0647"
    "h": "\u0647"
    "Kh": "\u062E"
    "khkh": "\u062E"
    "kh": "\u062E"

    "K": "\u06A9"
    "kk": "\u06A9"
    "k": "\u06A9"

    # particle da
    "Da\u0020": "\u062F\u0020"
    "da\u0020": "\u062F\u0020"

    "D\u0323": "\u0689"
    "d\u0323d\u0323": "\u0689"
    "d\u0323": "\u0689"

    "D": "\u062F"
    "dd": "\u062F"
    "d": "\u062F"

    "Z\u0323h": "\u0696"
    "z\u0323hz\u0323h": "\u0696"
    "z\u0323h": "\u0696"

    "Zh": "\u0698"
    "zhzh": "\u0698"
    "zh": "\u0698"

    "Z\u0324": "\u0638"
    "z\u0324z\u0324": "\u0638"
    "z\u0324": "\u0638"

    "Z\u0323": "\u0636"
    "z\u0323z\u0323": "\u0636"
    "z\u0323": "\u0636"

    "Z\u0332": "\u0630"
    "z\u0332z\u0332": "\u0630"
    "z\u0332": "\u0630"

    "Z\u0307": "\u0681"
    "z\u0307z\u0307": "\u0681"
    "z\u0307": "\u0681"

    "Z": "\u0632"
    "zz": "\u0632"
    "z": "\u0632"

    "R\u0323": "\u0693"
    "r\u0323r\u0323": "\u0693"
    "r\u0323": "\u0693"

    "R": "\u0631"
    "rr": "\u0631"
    "r": "\u0631"

    "Gh": "\u063A"
    "ghgh": "\u063A"
    "gh": "\u063A"

    "G": "\u06AB"
    "gg": "\u06AB"
    "g": "\u06AB"

    "F": "\u0641"
    "ff": "\u0641"
    "f": "\u0641"

    "Q": "\u0642"
    "qq": "\u0642"
    "q": "\u0642"

    "L": "\u0644"
    "ll": "\u0644"
    "l": "\u0644"

    "M": "\u0645"
    "mm": "\u0645"
    "m": "\u0645"

    "N\u0323": "\u06BC"
    "n\u0323n\u0323": "\u06BC"
    "n\u0323": "\u06BC"

    "N": "\u0646"
    "nn": "\u0646"
    "n": "\u0646"

    "W": "\u0648"
    "ww": "\u0648"
    "w": "\u0648"

    "Y": "\u064A"
    "y": "\u064A"

    # non-Arabic consonants:

    # ain (alone)
    "\u02BB": "\u0639"

    # hamza (alone in final position)
    "%\u02BE": "\u0621"
    "%\u02BC": "\u0621"