openmedialibrary/oml/meta/marc_countries.py

# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4

COUNTRIES = {
    "gw": "Germany", 
    "gv": "Guinea", 
    "gu": "Guam", 
    "gt": "Guatemala", 
    "gs": "Georgia (Republic)", 
    "gr": "Greece", 
    "-ge": "Germany (East)", 
    "gp": "Guadeloupe", 
    "mnu": "Minnesota", 
    "gy": "Guyana", 
    "gd": "Grenada", 
    "gb": "Kiribati", 
    "go": "Gabon", 
    "gm": "Gambia", 
    "alu": "Alabama", 
    "gi": "Gibraltar", 
    "gh": "Ghana", 
    "tz": "Tanzania", 
    "tv": "Tuvalu", 
    "tu": "Turkey", 
    "tr": "Trinidad and Tobago", 
    "ts": "United Arab Emirates", 
    "to": "Tonga", 
    "tl": "Tokelau", 
    "tk": "Turkmenistan", 
    "th": "Thailand", 
    "ti": "Tunisia", 
    "tg": "Togo", 
    "tc": "Turks and Caicos Islands", 
    "ta": "Tajikistan", 
    "-gn": "Gilbert and Ellice Islands", 
    "-us": "United States", 
    "-ajr": "Azerbaijan S.S.R.", 
    "-iu": "Israel-Syria Demilitarized Zones", 
    "-iw": "Israel-Jordan Demilitarized Zones", 
    "za": "Zambia", 
    "nbu": "Nebraska", 
    "scu": "South Carolina", 
    "bg": "Bangladesh", 
    "cau": "California", 
    "abc": "Alberta", 
    "xoa": "Northern Territory", 
    "meu": "Maine", 
    "ctu": "Connecticut", 
    "my": "Malaysia", 
    "aku": "Alaska", 
    "gl": "Greenland", 
    "-cn": "Canada", 
    "wiu": "Wisconsin", 
    "-cz": "Canal Zone", 
    "txu": "Texas", 
    "-cs": "Czechoslovakia", 
    "-cp": "Canton and Enderbury Islands", 
    "msu": "Mississippi", 
    "-ln": "Central and Southern Line Islands", 
    "nkc": "New Brunswick", 
    "it": "Italy", 
    "tnu": "Tennessee", 
    "vp": "Various places", 
    "mg": "Madagascar", 
    "mf": "Mauritius", 
    "mc": "Monaco", 
    "-ur": "Soviet Union", 
    "mm": "Malta", 
    "ml": "Mali", 
    "mo": "Montenegro", 
    "flu": "Florida", 
    "deu": "Delaware", 
    "mk": "Oman", 
    "mj": "Montserrat", 
    "mu": "Mauritania", 
    "mw": "Malawi", 
    "mv": "Moldova", 
    "mq": "Martinique", 
    "mp": "Mongolia", 
    "mr": "Morocco", 
    "-ui": "United Kingdom Misc. Islands", 
    "mx": "Mexico", 
    "-uk": "United Kingdom", 
    "mz": "Mozambique", 
    "kyu": "Kentucky", 
    "hiu": "Hawaii", 
    "enk": "England", 
    "nyu": "New York (State)", 
    "fp": "French Polynesia", 
    "fr": "France", 
    "fs": "Terres australes et antarctiques françaises", 
    "mau": "Massachusetts", 
    "snc": "Saskatchewan", 
    "fa": "Faroe Islands", 
    "fg": "French Guiana", 
    "lau": "Louisiana", 
    "fj": "Fiji", 
    "fk": "Falkland Islands", 
    "fm": "Micronesia (Federated States)", 
    "sz": "Switzerland", 
    "sy": "Syria", 
    "sx": "Namibia", 
    "ss": "Western Sahara", 
    "sr": "Surinam", 
    "sq": "Swaziland", 
    "sp": "Spain", 
    "sw": "Sweden", 
    "su": "Saudi Arabia", 
    "st": "Saint-Martin", 
    "sj": "Sudan", 
    "si": "Singapore", 
    "sh": "Spanish North Africa", 
    "so": "Somalia", 
    "sn": "Sint Maarten", 
    "sm": "San Marino", 
    "sl": "Sierra Leone", 
    "sc": "Saint-Barthélemy", 
    "sa": "South Africa", 
    "sg": "Senegal", 
    "sf": "Sao Tome and Principe", 
    "se": "Seychelles", 
    "sd": "South Sudan", 
    "-unr": "Ukraine", 
    "-kgr": "Kirghiz S.S.R.", 
    "le": "Lebanon", 
    "lb": "Liberia", 
    "-hk": "Hong Kong", 
    "lo": "Lesotho", 
    "lh": "Liechtenstein", 
    "li": "Lithuania", 
    "lv": "Latvia", 
    "lu": "Luxembourg", 
    "vtu": "Vermont", 
    "ls": "Laos", 
    "xc": "Maldives", 
    "ly": "Libya", 
    "oku": "Oklahoma", 
    "ye": "Yemen", 
    "-tkr": "Turkmen S.S.R.", 
    "nfc": "Newfoundland and Labrador", 
    "ft": "Djibouti", 
    "em": "Timor-Leste", 
    "eg": "Equatorial Guinea", 
    "ea": "Eritrea", 
    "ec": "Ecuador", 
    "-gsr": "Georgian S.S.R.", 
    "et": "Ethiopia", 
    "es": "El Salvador", 
    "er": "Estonia", 
    "ru": "Russia (Federation)", 
    "rw": "Rwanda", 
    "re": "Réunion", 
    "rb": "Serbia", 
    "rm": "Romania", 
    "rh": "Zimbabwe", 
    "-err": "Estonia", 
    "oru": "Oregon", 
    "quc": "Québec (Province)", 
    "ntc": "Northwest Territories", 
    "wlk": "Wales", 
    "xj": "Saint Helena", 
    "xk": "Saint Lucia", 
    "xh": "Niue", 
    "xn": "Macedonia", 
    "xo": "Slovakia", 
    "xl": "Saint Pierre and Miquelon", 
    "xm": "Saint Vincent and the Grenadines", 
    "xb": "Cocos (Keeling) Islands", 
    "onc": "Ontario", 
    "xa": "Christmas Island (Indian Ocean)", 
    "xf": "Midway Islands", 
    "xd": "Saint Kitts-Nevis", 
    "xe": "Marshall Islands", 
    "nhu": "New Hampshire", 
    "xx": "No place, unknown, or undetermined", 
    "fi": "Finland", 
    "xr": "Czech Republic", 
    "xs": "South Georgia and the South Sandwich Islands", 
    "xp": "Spratly Island", 
    "xv": "Slovenia", 
    "-tt": "Trust Territory of the Pacific Islands", 
    "iau": "Iowa", 
    "ncu": "North Carolina", 
    "stk": "Scotland", 
    "xra": "South Australia", 
    "miu": "Michigan", 
    "kg": "Kyrgyzstan", 
    "ke": "Kenya", 
    "ko": "Korea (South)", 
    "kn": "Korea (North)", 
    "kv": "Kosovo", 
    "ku": "Kuwait", 
    "kz": "Kazakhstan", 
    "-pt": "Portuguese Timor", 
    "ksu": "Kansas", 
    "dm": "Benin", 
    "dk": "Denmark", 
    "-ys": "Yemen (People's Democratic Republic)", 
    "-yu": "Serbia and Montenegro", 
    "-bwr": "Byelorussian S.S.R.", 
    "dr": "Dominican Republic", 
    "dq": "Dominica", 
    "qa": "Qatar", 
    "aru": "Arkansas", 
    "nuc": "Nunavut", 
    "wf": "Wallis and Futuna", 
    "wk": "Wake Island", 
    "wj": "West Bank of the Jordan River", 
    "jm": "Jamaica", 
    "vra": "Victoria", 
    "jo": "Jordan", 
    "ws": "Samoa", 
    "ji": "Johnston Atoll", 
    "-na": "Netherlands Antilles", 
    "ja": "Japan", 
    "cou": "Colorado", 
    "-wb": "West Berlin", 
    "ilu": "Illinois", 
    "-nm": "Northern Mariana Islands", 
    "ck": "Colombia", 
    "cj": "Cayman Islands", 
    "ci": "Croatia", 
    "ch": "China (Republic : 1949- )", 
    "co": "Curaçao", 
    "cm": "Cameroon", 
    "cl": "Chile", 
    "-rur": "Russian S.F.S.R.", 
    "cb": "Cambodia", 
    "ca": "Caribbean Netherlands", 
    "cg": "Congo (Democratic Republic)", 
    "cf": "Congo (Brazzaville)", 
    "-lir": "Lithuania", 
    "cd": "Chad", 
    "cy": "Cyprus", 
    "cx": "Central African Republic", 
    "cr": "Costa Rica", 
    "cq": "Comoros", 
    "cw": "Cook Islands", 
    "cv": "Cape Verde", 
    "cu": "Cuba", 
    "pr": "Puerto Rico", 
    "pp": "Papua New Guinea", 
    "pw": "Palau", 
    "py": "Paraguay", 
    "pc": "Pitcairn Island", 
    "pf": "Paracel Islands", 
    "pg": "Guinea-Bissau", 
    "pe": "Peru", 
    "pk": "Pakistan", 
    "ph": "Philippines", 
    "pn": "Panama", 
    "po": "Portugal", 
    "pl": "Poland", 
    "pic": "Prince Edward Island", 
    "xxu": "United States", 
    "gau": "Georgia", 
    "xxc": "Canada", 
    "xxk": "United Kingdom", 
    "iy": "Iraq-Saudi Arabia Neutral Zone", 
    "vb": "British Virgin Islands", 
    "vc": "Vatican City", 
    "ve": "Venezuela", 
    "iq": "Iraq", 
    "vi": "Virgin Islands of the United States", 
    "is": "Israel", 
    "ir": "Iran", 
    "vm": "Vietnam", 
    "iv": "Côte d'Ivoire", 
    "ii": "India", 
    "-ac": "Ashmore and Cartier Islands", 
    "io": "Indonesia", 
    "-ai": "Anguilla", 
    "ic": "Iceland", 
    "ie": "Ireland", 
    "pau": "Pennsylvania", 
    "-jn": "Jan Mayen", 
    "nik": "Northern Ireland", 
    "wyu": "Wyoming", 
    "-air": "Armenian S.S.R.", 
    "-sv": "Swan Islands", 
    "-mvr": "Moldavian S.S.R.", 
    "-sk": "Sikkim", 
    "riu": "Rhode Island", 
    "-sb": "Svalbard", 
    "-xi": "Saint Kitts-Nevis-Anguilla", 
    "wea": "Western Australia", 
    "cc": "China", 
    "nvu": "Nevada", 
    "mou": "Missouri", 
    "ce": "Sri Lanka", 
    "qea": "Queensland", 
    "-mh": "Macao", 
    "nju": "New Jersey", 
    "ykc": "Yukon Territory", 
    "-vs": "Vietnam, South", 
    "tma": "Tasmania", 
    "-vn": "Vietnam, North", 
    "bd": "Burundi", 
    "be": "Belgium", 
    "bf": "Bahamas", 
    "nmu": "New Mexico", 
    "ba": "Bahrain", 
    "bb": "Barbados", 
    "bl": "Brazil", 
    "bm": "Bermuda Islands", 
    "bn": "Bosnia and Hercegovina", 
    "bo": "Bolivia", 
    "bh": "Belize", 
    "bi": "British Indian Ocean Territory", 
    "bt": "Bhutan", 
    "bu": "Bulgaria", 
    "bv": "Bouvet Island", 
    "bw": "Belarus", 
    "bp": "Solomon Islands", 
    "br": "Burma", 
    "bs": "Botswana", 
    "dcu": "District of Columbia", 
    "bx": "Brunei", 
    "aca": "Australian Capital Territory", 
    "idu": "Idaho", 
    "xna": "New South Wales", 
    "ot": "Mayotte", 
    "ndu": "North Dakota", 
    "nsc": "Nova Scotia", 
    "-kzr": "Kazakh S.S.R.", 
    "mbc": "Manitoba", 
    "-lvr": "Latvia", 
    "-uzr": "Uzbek S.S.R.", 
    "wau": "Washington (State)", 
    "vau": "Virginia", 
    "sdu": "South Dakota", 
    "gz": "Gaza Strip", 
    "ht": "Haiti", 
    "hu": "Hungary", 
    "ho": "Honduras", 
    "hm": "Heard and McDonald Islands", 
    "xga": "Coral Sea Islands Territory", 
    "uy": "Uruguay", 
    "uz": "Uzbekistan", 
    "uv": "Burkina Faso", 
    "up": "United States Misc. Pacific Islands", 
    "mtu": "Montana", 
    "un": "Ukraine", 
    "utu": "Utah", 
    "ug": "Uganda", 
    "ua": "Egypt", 
    "azu": "Arizona", 
    "uc": "United States Misc. Caribbean Islands", 
    "aa": "Albania", 
    "ae": "Algeria", 
    "ag": "Argentina", 
    "af": "Afghanistan", 
    "ai": "Armenia (Republic)", 
    "inu": "Indiana", 
    "uik": "United Kingdom Misc. Islands", 
    "aj": "Azerbaijan", 
    "am": "Anguilla", 
    "ao": "Angola", 
    "an": "Andorra", 
    "aq": "Antigua and Barbuda", 
    "as": "American Samoa", 
    "au": "Austria", 
    "at": "Australia", 
    "aw": "Aruba", 
    "ay": "Antarctica", 
    "ohu": "Ohio", 
    "nl": "New Caledonia", 
    "-ry": "Ryukyu Islands, Southern", 
    "nn": "Vanuatu", 
    "no": "Norway", 
    "ne": "Netherlands", 
    "ng": "Niger", 
    "nx": "Norfolk Island", 
    "nz": "New Zealand", 
    "np": "Nepal", 
    "nq": "Nicaragua", 
    "nr": "Nigeria", 
    "mdu": "Maryland", 
    "nu": "Nauru", 
    "nw": "Northern Mariana Islands", 
    "wvu": "West Virginia", 
    "-xxr": "Soviet Union", 
    "-tar": "Tajik S.S.R.", 
    "bcc": "British Columbia"
}


if __name__ == '__main__':
    import json
    import re
    import ox
    from ox.cache import read_url

    url = "http://www.loc.gov/marc/countries/countries_code.html"
    data = read_url(url).decode('utf-8')
    countries = dict([
        [ox.strip_tags(c) for c in r]
        for r in re.compile('<tr>.*?class="code">(.*?)</td>.*?<td>(.*?)</td>', re.DOTALL).findall(data)
    ])

    data = json.dumps(countries, indent=4, ensure_ascii=False).encode('utf-8')
    with open(__file__) as f:
        pydata = f.read()
    pydata = re.sub(
        re.compile('\nCOUNTRIES = {.*?}\n\n', re.DOTALL),
        '\nCOUNTRIES = %s\n\n' % data, pydata)

    with open(__file__, 'w') as f:
        f.write(pydata)