Script to convert Cyrillic unicode characters to their respective LaTeX representation

unicode_to_latex.py

#!/usr/bin/python3

# original XML at http://www.w3.org/Math/characters/unicode.xml
# XSL for conversion: https://gist.github.com/798546

# only cyrillic symbols

unicode_to_latex = {
    0x0401: "\\CYRYO",
    0x0402: "\\CYRDJE",
    0x0403: "\\CYRG",
    0x0404: "\\CYRIE",
    0x0405: "\\CYRDZE",
    0x0406: "\\CYRII",
    0x0407: "\\CYRYI",
    0x0408: "\\CYRJE",
    0x0409: "\\CYRLJE",
    0x040A: "\\CYRNJE",
    0x040B: "\\CYRTSHE",
    0x040C: "\\CYRK",
    0x040E: "\\CYRUSHRT",
    0x040F: "\\CYRDZHE",
    0x0410: "\\CYRA",
    0x0411: "\\CYRB",
    0x0412: "\\CYRV",
    0x0413: "\\CYRG",
    0x0414: "\\CYRD",
    0x0415: "\\CYRE",
    0x0416: "\\CYRZH",
    0x0417: "\\CYRZ",
    0x0418: "\\CYRI",
    0x0419: "\\CYRISHRT",
    0x041A: "\\CYRK",
    0x041B: "\\CYRL",
    0x041C: "\\CYRM",
    0x041D: "\\CYRN",
    0x041E: "\\CYRO",
    0x041F: "\\CYRP",
    0x0420: "\\CYRR",
    0x0421: "\\CYRS",
    0x0422: "\\CYRT",
    0x0423: "\\CYRU",
    0x0424: "\\CYRF",
    0x0425: "\\CYRH",
    0x0426: "\\CYRC",
    0x0427: "\\CYRCH",
    0x0428: "\\CYRSH",
    0x0429: "\\CYRSHCH",
    0x042A: "\\CYRHRDSN",
    0x042B: "\\CYRERY",
    0x042C: "\\CYRSFTSN",
    0x042D: "\\CYREREV",
    0x042E: "\\CYRYU",
    0x042F: "\\CYRYA",
    0x0430: "\\cyra",
    0x0431: "\\cyrb",
    0x0432: "\\cyrv",
    0x0433: "\\cyrg",
    0x0434: "\\cyrd",
    0x0435: "\\cyre",
    0x0436: "\\cyrzh",
    0x0437: "\\cyrz",
    0x0438: "\\cyri",
    0x0439: "\\cyrishrt",
    0x043A: "\\cyrk",
    0x043B: "\\cyrl",
    0x043C: "\\cyrm",
    0x043D: "\\cyrn",
    0x043E: "\\cyro",
    0x043F: "\\cyrp",
    0x0440: "\\cyrr",
    0x0441: "\\cyrs",
    0x0442: "\\cyrt",
    0x0443: "\\cyru",
    0x0444: "\\cyrf",
    0x0445: "\\cyrh",
    0x0446: "\\cyrc",
    0x0447: "\\cyrch",
    0x0448: "\\cyrsh",
    0x0449: "\\cyrshch",
    0x044A: "\\cyrhrdsn",
    0x044B: "\\cyrery",
    0x044C: "\\cyrsftsn",
    0x044D: "\\cyrerev",
    0x044E: "\\cyryu",
    0x044F: "\\cyrya",
    0x0451: "\\cyryo",
    0x0452: "\\cyrdje",
    0x0453: "\\cyrg",
    0x0454: "\\cyrie",
    0x0455: "\\cyrdze",
    0x0456: "\\cyrii",
    0x0457: "\\cyryi",
    0x0458: "\\cyrje",
    0x0459: "\\cyrlje",
    0x045A: "\\cyrnje",
    0x045B: "\\cyrtshe",
    0x045C: "\\cyrk",
    0x045E: "\\cyrushrt",
    0x045F: "\\cyrdzhe",
    0x0460: "\\CYROMEGA",
    0x0461: "\\cyromega",
    0x0462: "\\CYRYAT",
    0x0464: "\\CYRIOTE",
    0x0465: "\\cyriote",
    0x0466: "\\CYRLYUS",
    0x0467: "\\cyrlyus",
    0x0468: "\\CYRIOTLYUS",
    0x0469: "\\cyriotlyus",
    0x046A: "\\CYRBYUS",
    0x046C: "\\CYRIOTBYUS",
    0x046D: "\\cyriotbyus",
    0x046E: "\\CYRKSI",
    0x046F: "\\cyrksi",
    0x0470: "\\CYRPSI",
    0x0471: "\\cyrpsi",
    0x0472: "\\CYRFITA",
    0x0474: "\\CYRIZH",
    0x0478: "\\CYRUK",
    0x0479: "\\cyruk",
    0x047A: "\\CYROMEGARND",
    0x047B: "\\cyromegarnd",
    0x047C: "\\CYROMEGATITLO",
    0x047D: "\\cyromegatitlo",
    0x047E: "\\CYROT",
    0x047F: "\\cyrot",
    0x0480: "\\CYRKOPPA",
    0x0481: "\\cyrkoppa",
    0x0482: "\\cyrthousands",
    0x0488: "\\cyrhundredthousands",
    0x0489: "\\cyrmillions",
    0x048C: "\\CYRSEMISFTSN",
    0x048D: "\\cyrsemisftsn",
    0x048E: "\\CYRRTICK",
    0x048F: "\\cyrrtick",
    0x0490: "\\CYRGUP",
    0x0491: "\\cyrgup",
    0x0492: "\\CYRGHCRS",
    0x0493: "\\cyrghcrs",
    0x0494: "\\CYRGHK",
    0x0495: "\\cyrghk",
    0x0496: "\\CYRZHDSC",
    0x0497: "\\cyrzhdsc",
    0x0498: "\\CYRZDSC",
    0x0499: "\\cyrzdsc",
    0x049A: "\\CYRKDSC",
    0x049B: "\\cyrkdsc",
    0x049C: "\\CYRKVCRS",
    0x049D: "\\cyrkvcrs",
    0x049E: "\\CYRKHCRS",
    0x049F: "\\cyrkhcrs",
    0x04A0: "\\CYRKBEAK",
    0x04A1: "\\cyrkbeak",
    0x04A2: "\\CYRNDSC",
    0x04A3: "\\cyrndsc",
    0x04A4: "\\CYRNG",
    0x04A5: "\\cyrng",
    0x04A6: "\\CYRPHK",
    0x04A7: "\\cyrphk",
    0x04A8: "\\CYRABHHA",
    0x04A9: "\\cyrabhha",
    0x04AA: "\\CYRSDSC",
    0x04AB: "\\cyrsdsc",
    0x04AC: "\\CYRTDSC",
    0x04AD: "\\cyrtdsc",
    0x04AE: "\\CYRY",
    0x04AF: "\\cyry",
    0x04B0: "\\CYRYHCRS",
    0x04B1: "\\cyryhcrs",
    0x04B2: "\\CYRHDSC",
    0x04B3: "\\cyrhdsc",
    0x04B4: "\\CYRTETSE",
    0x04B5: "\\cyrtetse",
    0x04B6: "\\CYRCHRDSC",
    0x04B7: "\\cyrchrdsc",
    0x04B8: "\\CYRCHVCRS",
    0x04B9: "\\cyrchvcrs",
    0x04BA: "\\CYRSHHA",
    0x04BB: "\\cyrshha",
    0x04BC: "\\CYRABHCH",
    0x04BD: "\\cyrabhch",
    0x04BE: "\\CYRABHCHDSC",
    0x04BF: "\\cyrabhchdsc",
    0x04C0: "\\CYRpalochka",
    0x04C3: "\\CYRKHK",
    0x04C4: "\\cyrkhk",
    0x04C7: "\\CYRNHK",
    0x04C8: "\\cyrnhk",
    0x04CB: "\\CYRCHLDSC",
    0x04CC: "\\cyrchldsc",
    0x04D4: "\\CYRAE",
    0x04D5: "\\cyrae",
    0x04D8: "\\CYRSCHWA",
    0x04D9: "\\cyrschwa",
    0x04E0: "\\CYRABHDZE",
    0x04E1: "\\cyrabhdze",
    0x04E8: "\\CYROTLD",
    0x04E9: "\\cyrotld"
}

import fileinput

for line in fileinput.input():
    print(line.translate(unicode_to_latex))