niveum/packages/scripts/ipa.py

#!/usr/bin/env python3
import sys

# https://www.phon.ucl.ac.uk/home/sampa/x-sampa.htm

XSAMPA_TO_IPA = {
    "!": "\uA71C",
    "!\\": "\u01C3",
    "%": "\u02CC",
    "&": "\u0276",
    "'": "\u02B2",
    "-\\": "\u203F",
    "1": "\u0268",
    "2": "\u00f8",
    "3": "\u025C",
    "3\\": "\u025E",
    "4": "\u027E",
    "5": "\u026b",
    "6": "\u0250",
    "7": "\u0264",
    "8": "\u0275",
    "9": "\u0153",
    ":": "\u02D0",
    ":\\": "\u02D1",
    "<B>": "\u02E9",
    "<F>": "\u2198",
    "<H>": "\u02E6",
    "<L>": "\u02E8",
    "<M>": "\u02E7",
    "<R>": "\u2197",
    "<T>": "\u02E5",
    "<\\": "\u02A2",
    "=": "\u0329",
    "=\\": "\u01C2",
    ">\\": "\u02A1",
    "?": "\u0294",
    "?\\": "\u0295",
    "@": "\u0259",
    "@\\": "\u0258",
    "A": "\u0251",
    "B": "\u03B2",
    "B\\": "\u0299",
    "C": "\u00E7",
    "D": "\u00F0",
    "E": "\u025B",
    "F": "\u0271",
    "G": "\u0263",
    "G\\": "\u0262",
    "G\\_<": "\u029B",
    "H": "\u0265",
    "H\\": "\u029C",
    "I": "\u026A",
    "J": "\u0272",
    "J\\_<": "\u0284",
    "K": "\u026C",
    "K\\": "\u026E",
    "L": "\u028E",
    "L\\": "\u029F",
    "M": "\u026F",
    "M\\": "\u0270",
    "N": "\u014B",
    "N\\": "\u0274",
    "O": "\u0254",
    "O\\": "\u0298",
    "P": "\u028B",
    "Q": "\u0252",
    "R": "\u0281",
    "R\\": "\u0280",
    "S": "\u0283",
    "T": "\u03B8",
    "U": "\u028A",
    "V": "\u028C",
    "W": "\u028D",
    "X": "\u03C7",
    "X\\": "\u0127",
    "Y": "\u028F",
    "Z": "\u0292",
    "^": "\uA71B",
    "_": "\u0361",
    "_+": "\u031F",
    "_-": "\u0320",
    "_0": "\u0325",
    "_=": "\u0329",
    "_>": "\u02BC",
    "_?\\": "\u02E4",
    "_A": "\u0318",
    "_B": "\u030F",
    "_B_L": "\u1DC5",
    "_F": "\u0302",
    "_F_R": "\u1dc9",
    "_G": "\u02E0",
    "_H": "\u0341",
    "_H_T": "\u1DC4",
    "_L": "\u0340",
    "_M": "\u0304",
    "_N": "\u033C",
    "_O": "\u0339",
    "_R": "\u030C",
    "_R_F": "\u1dc8",
    "_T": "\u030B",
    "_X": "\u0306",
    "_^": "\u032F",
    "_a": "\u033A",
    "_c": "\u031C",
    "_d": "\u032A",
    "_e": "\u0334",
    "_h": "\u02B0",
    "_j": "\u02B2",
    "_k": "\u0330",
    "_l": "\u02E1",
    "_m": "\u033B",
    "_n": "\u207F",
    "_o": "\u031E",
    "_q": "\u0319",
    "_r": "\u031D",
    "_t": "\u0324",
    "_v": "\u032C",
    "_w": "\u02B7",
    "_x": "\u033D",
    "_}": "\u031A",
    "_~": "\u0303",
    "`": "\u02DE",
    "b_<": "\u0253",
    "d_<": "\u0257",
    "d`": "\u0256",
    "d`_<": "\u1D91",
    "g_<": "\u0260",
    "h\\": "\u0266",
    "j\\": "\u029D",
    "l\\": "\u027A",
    "l`": "\u026D",
    "n`": "\u0273",
    "p\\": "\u0278",
    "r\\": "\u0279",
    "r\\`": "\u027B",
    "r` ": "\u027D",
    "s\\": "\u0255",
    "s`": "\u0282",
    "t`": "\u0288",
    "v\\": "\u028B",
    "x\\": "\u0267",
    "z\\": "\u0291",
    "z`": "\u0290",
    "{": "\u00E6",
    "|\\": "\u01C0",
    "|\\|\\": "\u01C1",
    "}": "\u0289",
    "~": "\u0303",
    '"': "\u02C8",
    '_"': "\u0308",
}

if __name__ == "__main__":
    text = " ".join(sys.argv[1:])

    for xsampa, ipa in sorted(XSAMPA_TO_IPA.items(), key=lambda item: -len(item[0])):
        text = text.replace(xsampa, ipa)

    sys.stdout.write(text)