scriptshifter/scriptshifter/tables/data/urdu.yml

general:
  name: Urdi

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"
    "alla\u0304h": "\u0627\u0644\u0644\u0647"

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

    # aur (with spaces)
    " aur ": " \u0627\u0648\u0631 "
    "Aur ": "\u0627\u0648\u0631 "

    ####
    # Abdurrahman
    "\u02BBAbdurrah\u0323ma\u0301n": "\u0639\u0628\u062F\u0627\u0644\u0631\u062D\u0645\u0670\u0646"
    "\u02BBAbdurrah\u0323ma\u0304n": "\u0639\u0628\u062F\u0627\u0644\u0631\u062D\u0645\u0670\u0646"

    # Abd names
    "\u02BBAbdul\u02BB": "\u0639\u0628\u062F\u0627\u0644\u0639"
    "\u02BBAbdula": "\u0639\u0628\u062F\u0627\u0644"
    "\u02BBAbdulb": "\u0639\u0628\u062F\u0627\u0644\u0628"
    "\u02BBAbdulf": "\u0639\u0628\u062F\u0627\u0644\u0641"
    "\u02BBAbdulg\u0332h\u0332": "\u0639\u0628\u062F\u0627\u0644\u063A"
    "\u02BBAbdulh\u0323": "\u0639\u0628\u062F\u0627\u0644\u062D"
    "\u02BBAbdulh": "\u0639\u0628\u062F\u0627\u0644\u0647 "
    "\u02BBAbdulj": "\u0639\u0628\u062F\u0627\u0644\u062C"
    "\u02BBAbdulk\u0332h\u0332": "\u0639\u0628\u062F\u0627\u0644\u062E"
    "\u02BBAbdulk": "\u0639\u0628\u062F\u0627\u0644\u0643 "
    "\u02BBAbdulm": "\u0639\u0628\u062F\u0627\u0644\u0645"
    "\u02BBAbdulq": "\u0639\u0628\u062F\u0627\u0644\u0642"
    "\u02BBAbdulv": "\u0639\u0628\u062F\u0627\u0644\u0648"
    "\u02BBAbdunn": "\u0639\u0628\u062F\u0627\u0644\u0646"
    "\u02BBAbdurr": "\u0639\u0628\u062F\u0627\u0644\u0631"
    "\u02BBAbdus\u0323s\u0323": "\u0639\u0628\u062F\u0627\u0644\u0325"
    "\u02BBAbduss": "\u0639\u0628\u062F\u0627\u0644\u0633"
    "\u02BBAbdushsh": "\u0639\u0628\u062F\u0627\u0644\u0634"
    "\u02BBAbdutt": "\u0639\u0628\u062F\u0627\u0644\u062A"
    "\u02BBAbduz\u0323z\u0323": "\u0639\u0628\u062F\u0627\u0644\u0636"
    "\u02BBAbduz\u0324z\u0324": "\u0639\u0628\u062F\u0627\u0644\u0638"

    # Abu names
    "Abu\u0304 ": "\u0627\u0628\u0648\u0020"
    "Abu\u0304": "\u0627\u0628\u0648\u200C\u0627\u0644"

    ####

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

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

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

    # Nuzhat
    "Nuzhat": "\u0646\u0632\u0647\u062A"

    # Uddin names
    "%i\u0304uddi\u0304n": "\u0649\u200C\u0627\u0644\u062F\u0651\u064A\u0646"
    "%uddi\u0304n": "\u200C\u0627\u0644\u062F\u0651\u064A\u0646"

    # ta'lif

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

    # Abbreviated name elements
    "# Ae": "\u0627\u06D2"

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

    # Numbers (\u06F0-06F9 for Persian/Urdu)
    # 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"

    # Postpositions

    # Aspirates [06BE] vs. heh [062D] combinations
    "bh\u0323": "\u0628\u062D"
    "Bh": "\u0628\u06BE"
    "bh": "\u0628\u06BE"

    "ph\u0323": "\u067E\u062D"
    "Ph": "\u067E\u06BE"
    "ph": "\u067E\u06BE"

    "th\u0323": "\u062A\u062D"
    "Th": "\u062A\u06BE"
    "th": "\u062A\u06BE"

    "t\u0323h\u0323": "\u0679\u062D"
    "T\u0323h": "\u0679\u06BE"
    "t\u0323h": "\u0679\u06BE"

    "jh\u0323": "\u062C\u062D"
    "Jh": "\u062C\u06BE"
    "jh": "\u062C\u06BE"

    "ch\u0323": "\u0686\u062D"
    "Ch": "\u0686\u06BE"
    "ch": "\u0686\u06BE"

    "dh\u0323": "\u062F\u062D"
    "Dh": "\u062F\u06BE"
    "dh": "\u062F\u06BE"

    "d\u0323h\u0323": "\u0688\u062D"
    "D\u0323h": "\u0688\u06BE"
    "d\u0323h": "\u0688\u06BE"

    "r\u0323h\u0323": "\u0691\u062D"
    "R\u0323h": "\u0691\u06BE"
    "r\u0323h": "\u0691\u06BE"

    "kh\u0323": "\u06A9\u062D"
    "Kh": "\u06A9\u06BE"
    "kh": "\u06A9\u06BE"

    "gh\u0323": "\u06AF\u062D"
    "Gh": "\u06AF\u06BE"
    "gh": "\u06AF\u06BE"

    # prime = ZWNJ"
    "\u02B9A\u0304": "\u200C\u0622"
    "\u02B9a\u0304": "\u200C\u0622"
    "a\u0304\u02BC\u02B9": "\u0627\u0621\u200C"
    "i\u0304\u02B9": "\u0649\u200C"
    "\u02B9": "\u200C"

    # Izafah here
    "%a\u0304-yi": "\u0627\u0626\u06D2"
    "%u\u0304-yi": "\u0648\u0626\u06D2"
    "%o-yi": "\u0648\u0626\u06D2"
    "%e-yi": "\u06D2"
    "%i\u0304-yi": "\u0649"
    "%h-yi": "\u06C0"
    "%-yi": "\u06C0"
    "%al-i": "\u0644"
    "%ul-i": "\u0644"
    "%-i": ""

    # Hyphenated prefixes:
    "bi-": "\u0628"
    "al-a\u0304%": "\u0627\u0644\u0627"
    "ul-a\u0304%": "\u0627\u0644\u0627"
    "al-": "\u0627\u0644"
    "ul-": "\u0627\u0644"
    "lil-i": "\u0644\u0644"
    "lil-": "\u0644\u0644"

    # al-/ul- plus sun letters
    "ar-r": "\u0627\u0644\u0631"
    "ur-r": "\u0627\u0644\u0631"
    "ar-R": "\u0627\u0644\u0631"
    "ur-R": "\u0627\u0644\u0631"
    "az\u0332-z\u0332": "\u0627\u0644\u0630"
    "uz\u0332-z\u0332": "\u0627\u0644\u0630"
    "az\u0332-Z\u0332": "\u0627\u0644\u0630"
    "uz\u0332-Z\u0332": "\u0627\u0644\u0630"
    "ad-d": "\u0627\u0644\u0627"
    "ud-d": "\u0627\u0644\u0627"
    "ad-D": "\u0627\u0644\u0627"
    "ud-D": "\u0627\u0644\u0627"
    "as\u0332-s\u0332": "\u0627\u0644\u062B"
    "us\u0332-s\u0332": "\u0627\u0644\u062B"
    "as\u0332-S\u0332": "\u0627\u0644\u062B"
    "us\u0332-S\u0332": "\u0627\u0644\u062B"
    "at-t": "\u0627\u0644\u062A"
    "ut-t": "\u0627\u0644\u062A"
    "at-T": "\u0627\u0644\u062A"
    "ut-T": "\u0627\u0644\u062A"
    "an-n": "\u0627\u0644\u0646"
    "un-n": "\u0627\u0644\u0646"
    "an-N": "\u0627\u0644\u0646"
    "un-N": "\u0627\u0644\u0646"
    "al-l": "\u0627\u0644\u0644"
    "ul-l": "\u0627\u0644\u0644"
    "al-L": "\u0627\u0644\u0644"
    "ul-L": "\u0627\u0644\u0644"
    "az\u0324-z\u0324": "\u0627\u0644\u0638"
    "uz\u0324-z\u0324": "\u0627\u0644\u0638"
    "az\u0324-Z\u0324": "\u0627\u0644\u0638"
    "uz\u0324-Z\u0324": "\u0627\u0644\u0638"
    "at\u0324-t\u0324": "\u0627\u0644\u0637"
    "ut\u0324-t\u0324": "\u0627\u0644\u0637"
    "at\u0324-T\u0324": "\u0627\u0644\u0637"
    "ut\u0324-T\u0324": "\u0627\u0644\u0637"
    "az\u0323-z\u0323": "\u0627\u0644\u0636"
    "uz\u0323-z\u0323": "\u0627\u0644\u0636"
    "az\u0323-Z\u0323": "\u0627\u0644\u0636"
    "uz\u0323-Z\u0323": "\u0627\u0644\u0636"
    "as\u0323-s\u0323": "\u0627\u0644\u0635"
    "us\u0323-s\u0323": "\u0627\u0644\u0635"
    "as\u0323-S\u0323": "\u0627\u0644\u0635"
    "us\u0323-S\u0323": "\u0627\u0644\u0635"
    "ash-sh": "\u0627\u0644\u0634"
    "ush-sh": "\u0627\u0644\u0634"
    "ash-Sh": "\u0627\u0644\u0634"
    "ush-Sh": "\u0627\u0644\u0634"
    "as-s": "\u0627\u0644\u0633"
    "us-s": "\u0627\u0644\u0633"
    "as-S": "\u0627\u0644\u0633"
    "us-S": "\u0627\u0644\u0633"
    "az-z": "\u0627\u0644\u0632"
    "uz-z": "\u0627\u0644\u0632"
    "az-Z": "\u0627\u0644\u0632"
    "uz-Z": "\u0627\u0644\u0632"

    # Diphthongs here
    "Ae": "\u0627\u06D2"
    "%ai": "\u06D2"
    "Ai": "\u0627\u064A"
    "ai%": "\u0627\u064A"
    "ai": "\u064A"
    "\u02BBAu": "\u0639\u0648"
    "\u02BBau": "\u0639\u0648"
    "Au": "\u0627\u0648"
    "au": "\u0648"

    # ayn-alif combo
    "%\u02BBa\u0304\u02BE": "\u0639\u0627\u0621"
    "%\u02BBa\u0304\u02BC": "\u0639\u0627\u0621"
    "%\u02BBa\u0304%": "\u0639\u0627"

    # hamza and vowel combo
    # [in final position]
    "%u\u0304\u02BEi\u0304": "\u0648\u0626\u0649"
    "%u\u0304\u02BCi\u0304": "\u0648\u0626\u0649"
    "%\u02BEi\u0304": "\u0626\u0649"
    "%\u02BCi\u0304": "\u0626\u0649"
    "%\u02BEe": "\u0626\u06D2"
    "%\u02BCe": "\u0626\u06D2"

    "%\u02BEu\u0304": "\u0624"
    "%\u02BCu\u0304": "\u0624"
    "%\u02BEo": "\u0624"
    "%\u02BCo": "\u0624"

    # [in medial position]
    "a\u02BEa": "\u0623"
    "a\u02BCa": "\u0623"
    "a\u0304\u02BEa": "\u0627\u0621"
    "a\u0304\u02BCa": "\u0627\u0621"
    "a\u02BEa\u0304": "\u0622"
    "a\u02BCa\u0304": "\u0622"
    "o\u02BEi\u0304": "\u0648\u0626\u064A"
    "o\u02BCi\u0304": "\u0648\u0626\u064A"
    "o\u02BEi": "\u0648\u0626"
    "o\u02BCi": "\u0648\u0626"
    "\u02BEi\u0304": "\u0626\u064A"
    "\u02BCi\u0304": "\u0626\u064A"
    "\u02BEi": "\u0626"
    "\u02BCi": "\u0626"
    "\u02BEe": "\u0626\u064A"
    "\u02BCe": "\u0626\u064A"

    "\u02BEu\u0304": "\u0624"
    "\u02BCu\u0304": "\u0624"
    "u\u0304\u02BE": "\u0624"
    "u\u0304\u02BC": "\u0624"
    "\u02BEo": "\u0624"
    "\u02BCo": "\u0624"
    "o\u02BE": "\u0624"
    "o\u02BC": "\u0624"
    "au\u02BE": "\u0624"
    "au\u02BC": "\u0624"

    "\u02BEa": "\u0626"
    "\u02BCa": "\u0626"

    "%i\u0304": "\u0649"
    "%a\u0301": "\u0649\u0670"

    # A
    "\u02BBA\u0304": "\u0639\u0627"
    "\u02BBa\u0304": "\u0639\u0627"
    "\u02BBA%": "\u0639"
    "\u02BBa": "\u0639"
    "A\u02BB": "\u0627\u0639"
    "a\u02BB%": "\u0627\u0639"
    "a\u02BB": "\u0639"
    "A\u0304%": "\u0622"
    "a\u0304%": "\u0622"
    "a\u0304": "\u0627"
    "a\u0301": "\u0649"
    "ayy": "\u064A\u0651"
    "A%": "\u0627"
    "a%": "\u0627"
    "A": ""
    "a": ""

    # E
    "%e": "\u06D2"
    "E%": "\u0627\u064A"
    "e%": "\u0627\u064A"
    "e": "\u064A"

    # I
    "\u02BBI\u0304": "\u0639\u064A"
    "\u02BBi\u0304": "\u0639\u064A"
    "I\u02BB": "\u0627\u0639"
    "i\u02BB": "\u0639"
    "\u02BBI": "\u0639"
    "I\u0304%": "\u0627\u064A"
    "i\u0304%": "\u0627\u064A"
    "i\u0304y": "\u064A"
    "i\u0304": "\u064A"
    "iyy": "\u064A\u0651"
    "I%": "\u0627"
    "i%": "\u0627"
    "I": "\u0627"
    "i": ""

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

    # U
    "\u02BBu\u0304": "\u0639\u0648"
    "\u02BBU": "\u0639"
    "\u02BBu": "\u0639"
    "U\u0304%": "\u0627\u0648"
    "u\u0304%": "\u0627\u0648"
    "u\u0304": "\u0648"
    "U%": "\u0627"
    "u%": "\u0627"
    "U": ""
    "u": ""

    # Consonants:
    "B": "\u0628"
    "bb": "\u0628\u0651"
    "b": "\u0628"
    "P": "\u067E"
    "pp": "\u067E\u0651"
    "p": "\u067E"
    "T\u0323": "\u0679"
    "t\u0323t\u0323": "\u0679\u0651"
    "t\u0323": "\u0679"
    "T\u0324": "\u0637"
    "t\u0324t\u0324": "\u0637\u0651"
    "t\u0324": "\u0637"
    "T": "\u062A"
    "tt": "\u062A\u0651"
    "t": "\u062A"
    "Sh": "\u0634"
    "shsh": "\u0634\u0651"
    "sh": "\u0634"
    "S\u0323": "\u0635"
    "s\u0323s\u0323": "\u0635\u0651"
    "s\u0323": "\u0635"
    "S\u0332": "\u062B"
    "s\u0332s\u0332": "\u062B\u0651"
    "s\u0332": "\u062B"
    "S": "\u0633"
    "ss": "\u0633\u0651"
    "s": "\u0633"
    "J": "\u062C"
    "jj": "\u062C\u0651"
    "j": "\u062C"
    "C": "\u0686"
    "cc": "\u0686\u0651"
    "c": "\u0686"
    "H\u0323": "\u062D"
    "h\u0323h\u0323": "\u062D\u0651"
    "h\u0323": "\u062D"
    "H": "\u0647"
    "hh": "\u0647\u0651"
    "h": "\u0647"
    "K\u0332h\u0332": "\u062E"
    "k\u0332h\u0332k\u0332h\u0332": "\u062E\u0651"
    "k\u0332h\u0332": "\u062E"
    "K": "\u06A9"
    "kk": "\u06A9\u0651"
    "k": "\u06A9"
    "D\u0323": "\u0688"
    "d\u0323d\u0323": "\u0688\u0651"
    "d\u0323": "\u0688"
    "D": "\u062F"
    "dd": "\u062F\u0651"
    "d": "\u062F"
    "Z\u0324": "\u0638"
    "z\u0324z\u0324": "\u0638\u0651"
    "z\u0324": "\u0638"
    "Z\u0323": "\u0636"
    "z\u0323z\u0323": "\u0636\u0651"
    "z\u0323": "\u0636"
    "Z\u0332": "\u0630"
    "z\u0332z\u0332": "\u0630\u0651"
    "z\u0332": "\u0630"
    "zz": "\u0632\u0651"
    "Zh": "\u0698"
    "zhzh": "\u0698\u0651"
    "zh": "\u0698"
    "Z": "\u0632"
    "z": "\u0632"
    "R\u0323": "\u0691"
    "r\u0323r\u0323": "\u0691\u0651"
    "r\u0323": "\u0691"
    "R": "\u0631"
    "rr": "\u0631\u0651"
    "r": "\u0631"
    "G\u0332h\u0332": "\u063A"
    "g\u0332h\u0332g\u0332h\u0332": "\u063A\u0651"
    "g\u0332h\u0332": "\u063A"
    "G": "\u06AF"
    "gg": "\u06AF\u0651"
    "g": "\u06AF"
    "F": "\u0641"
    "ff": "\u0641\u0651"
    "f": "\u0641"
    "Q": "\u0642"
    "qq": "\u0642\u0651"
    "q": "\u0642"
    "L": "\u0644"
    "ll": "\u0644\u0651"
    "l": "\u0644"
    "M": "\u0645"
    "mm": "\u0645\u0651"
    "m": "\u0645"
    "N\u0332": "\u06BA"
    "n\u0332n\u0332": "\u06BA\u0651"
    "n\u0332": "\u06BA"
    "N": "\u0646"
    "nn": "\u0646\u0651"
    "n": "\u0646"
    "V": "\u0648"
    "vv": "\u0648\u0651"
    "v": "\u0648"
    "Y": "\u064A"
    "yy": "\u064A\u0651"
    "y": "\u064A"

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

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