diff --git a/country.cabal b/country.cabal --- a/country.cabal +++ b/country.cabal @@ -1,5 +1,5 @@ name: country -version: 0.1.3 +version: 0.1.4 synopsis: Country data type and functions description: The `country` library provides a data type for dealing with @@ -35,13 +35,12 @@ Country.Unsafe other-modules: Country.Unexposed.Encode.English - Country.Unexposed.Enumerate - Country.Unexposed.ExtraNames Country.Unexposed.Names Country.Unexposed.Trie Country.Unexposed.TrieByte + Country.Unexposed.Alias build-depends: - base >= 4.7 && < 4.10 + base >= 4.7 && < 4.11 , text >= 1.2 && < 1.3 , bytestring >= 0.10 && < 0.11 , primitive >= 0.6.1 && < 0.7 diff --git a/src/Country.hs b/src/Country.hs --- a/src/Country.hs +++ b/src/Country.hs @@ -11,6 +11,7 @@ -- * Name , encodeEnglish , decode + , decodeUtf8 , parser , parserUtf8 -- * Alpha-2 and Alpha-3 @@ -24,10 +25,11 @@ import Country.Unsafe (Country(..)) import Country.Unexposed.Encode.English (countryNameQuads) -import Country.Unexposed.Names (numberOfPossibleCodes,alphaTwoHashMap,alphaThreeHashMap,decodeMap,decodeNumeric,encodeEnglish) +import Country.Unexposed.Names (numberOfPossibleCodes,alphaTwoHashMap,alphaThreeHashMap,decodeMap,decodeMapUtf8,decodeNumeric,encodeEnglish) import Country.Unexposed.Trie (Trie,trieFromList,trieParser) import Country.Unexposed.TrieByte (TrieByte,trieByteFromList,trieByteParser) import Data.Text (Text) +import Data.ByteString (ByteString) import Data.Word (Word16) import Data.Primitive (indexArray,writeByteArray,indexByteArray,unsafeFreezeByteArray,newByteArray) import Data.Primitive.Array (Array(..)) @@ -84,10 +86,17 @@ timesThree x = x * 3 --- | Parse a country from its name. This function is language-agnostic. --- It can handle any source language. +-- | Parse a country from its name. This function is language-agnostic +-- and is very generous with what it accepts. It handles official +-- names, colloquial names, acroynms, and obsolete names for many +-- countries. It strives to handle any source language. Open an +-- issue on the issue tracker if their are names that are +-- missing. decode :: Text -> Maybe Country decode = flip HM.lookup decodeMap + +decodeUtf8 :: ByteString -> Maybe Country +decodeUtf8 = flip HM.lookup decodeMapUtf8 -- | Parse a country from its name using an attoparsec text parser. This -- function is language-agnostic and can handle any source language. diff --git a/src/Country/Identifier.hs b/src/Country/Identifier.hs --- a/src/Country/Identifier.hs +++ b/src/Country/Identifier.hs @@ -751,3 +751,6 @@ zimbabwe :: Country zimbabwe = Country 716 +kosovo :: Country +kosovo = Country 383 + diff --git a/src/Country/Unexposed/Alias.hs b/src/Country/Unexposed/Alias.hs new file mode 100644 --- /dev/null +++ b/src/Country/Unexposed/Alias.hs @@ -0,0 +1,521 @@ +-- This module is autogenerated. Do not edit it by hand. +module Country.Unexposed.Alias + ( aliases + ) where + +import Data.Text (Text) +import Data.Word (Word16) +import qualified Data.Text as T + +aliases :: [(Word16,Text)] +aliases = + [ (4, T.pack "Afghanistan") + , (248, T.pack "Åland Islands") + , (248, T.pack "Aaland Islands") + , (248, T.pack "Aland Islands") + , (8, T.pack "Albania") + , (8, T.pack "Arbanon") + , (8, T.pack "Republic of Albania") + , (8, T.pack "Republika e Shqipërisë") + , (12, T.pack "Algeria") + , (12, T.pack "Algérie") + , (12, T.pack "Dzayer") + , (12, T.pack "People's Democratic Republic of Algeria") + , (12, T.pack "al Jumhuriyya al Jazaa'iriyya ad-Dīmuqrāţiyya ash Sha'biyya") + , (12, T.pack "al-Jazā'ir") + , (12, T.pack "الجزائر") + , (12, T.pack "ⴷⵣⴰⵢⴻⵔ") + , (16, T.pack "American Samoa") + , (20, T.pack "Andorra") + , (24, T.pack "Angola") + , (660, T.pack "Anguilla") + , (10, T.pack "Antarctica") + , (28, T.pack "Antigua and Barbuda") + , (32, T.pack "Argentina") + , (32, T.pack "Argentine Confederation") + , (32, T.pack "Argentine Nation") + , (32, T.pack "Argentine Republic") + , (32, T.pack "La Argentina") + , (32, T.pack "The Argentine") + , (32, T.pack "United Provinces of the Río de la Plata") + , (32, T.pack "la Argentina") + , (32, T.pack "the Argentine") + , (51, T.pack "Armenia") + , (533, T.pack "Aruba") + , (36, T.pack "Australia") + , (36, T.pack "Commonwealth of Australia") + , (36, T.pack "New Holland") + , (40, T.pack "Austria") + , (40, T.pack "Republic of Austria") + , (40, T.pack "Republik Österreich") + , (40, T.pack "Österreich") + , (31, T.pack "Azerbaijan") + , (44, T.pack "Bahamas") + , (48, T.pack "Bahrain") + , (50, T.pack "Bangladesh") + , (52, T.pack "Barbados") + , (112, T.pack "Belarus") + , (56, T.pack "Belgium") + , (84, T.pack "Belize") + , (204, T.pack "Benin") + , (60, T.pack "Bermuda") + , (64, T.pack "Bhutan") + , (68, T.pack "Bolivia (Plurinational State of)") + , (535, T.pack "Bonaire, Sint Eustatius and Saba") + , (70, T.pack "Bosnia and Herzegovina") + , (72, T.pack "Botswana") + , (74, T.pack "Bouvet Island") + , (76, T.pack "Brazil") + , (76, T.pack "Empire of Brazil") + , (76, T.pack "Federative Republic of Brazil") + , (76, T.pack "Ilha de Vera Cruz") + , (76, T.pack "Pindorama") + , (76, T.pack "República Federativa do Brasil") + , (76, T.pack "Terra de Santa Cruz") + , (76, T.pack "Terra di Papaga") + , (76, T.pack "United States of Brazil") + , (86, T.pack "British Indian Ocean Territory") + , (96, T.pack "Brunei Darussalam") + , (100, T.pack "Bulgaria") + , (854, T.pack "Burkina Faso") + , (108, T.pack "Burundi") + , (116, T.pack "Cambodia") + , (116, T.pack "Democratic Kampuchea") + , (116, T.pack "Kingdom of Cambodia") + , (116, T.pack "Royaume du Cambodge") + , (120, T.pack "Cameroon") + , (124, T.pack "Canada") + , (132, T.pack "Cabo Verde") + , (132, T.pack "Cape Verde") + , (136, T.pack "Cayman Islands") + , (140, T.pack "Central African Republic") + , (148, T.pack "Chad") + , (152, T.pack "Chile") + , (152, T.pack "Capitania general de Chile") + , (152, T.pack "Chili") + , (152, T.pack "Chilli") + , (152, T.pack "Republic of Chile") + , (152, T.pack "Reyno de Chile") + , (156, T.pack "China") + , (162, T.pack "Christmas Island") + , (166, T.pack "Cocos (Keeling) Islands") + , (170, T.pack "Colombia") + , (174, T.pack "Comoros") + , (178, T.pack "Congo") + , (180, T.pack "Congo (Democratic Republic of the)") + , (184, T.pack "Cook Islands") + , (188, T.pack "Costa Rica") + , (384, T.pack "Côte d'Ivoire") + , (384, T.pack "Ivory Coast") + , (384, T.pack "Republic of Côte d'Ivoire") + , (191, T.pack "Croatia") + , (192, T.pack "Cuba") + , (531, T.pack "Curaçao") + , (196, T.pack "Cyprus") + , (203, T.pack "Czech Republic") + , (208, T.pack "Denmark") + , (208, T.pack "Danmark") + , (208, T.pack "Kingdom of Denmark") + , (208, T.pack "Kongeriget Danmark") + , (262, T.pack "Djibouti") + , (262, T.pack "Côte française des Somalis, Territoire d'Obock") + , (262, T.pack "French Somaliland") + , (262, T.pack "French Territory of the Afars and the Issas") + , (262, T.pack "Obock Territory") + , (262, T.pack "Republic of Djibouti") + , (262, T.pack "République de Djibouti") + , (262, T.pack "Territoire français des Afars et des Issas") + , (212, T.pack "Dominica") + , (214, T.pack "Dominican Republic") + , (218, T.pack "Ecuador") + , (818, T.pack "Egypt") + , (818, T.pack "Aegyptus") + , (818, T.pack "Meṣr") + , (222, T.pack "El Salvador") + , (226, T.pack "Equatorial Guinea") + , (232, T.pack "Eritrea") + , (233, T.pack "Estonia") + , (231, T.pack "Ethiopia") + , (238, T.pack "Falkland Islands (Malvinas)") + , (234, T.pack "Faroe Islands") + , (242, T.pack "Fiji") + , (246, T.pack "Finland") + , (250, T.pack "France") + , (250, T.pack "Farança") + , (250, T.pack "French Republic") + , (250, T.pack "La France") + , (250, T.pack "République Française") + , (254, T.pack "French Guiana") + , (258, T.pack "French Polynesia") + , (260, T.pack "French Southern Territories") + , (266, T.pack "Gabon") + , (270, T.pack "Gambia") + , (268, T.pack "Georgia") + , (276, T.pack "Germany") + , (276, T.pack "BRD") + , (276, T.pack "Bundesrepublik Deutschland") + , (276, T.pack "Deutschland") + , (276, T.pack "FRG") + , (276, T.pack "Federal Republic of Germany") + , (288, T.pack "Ghana") + , (292, T.pack "Gibraltar") + , (300, T.pack "Greece") + , (304, T.pack "Greenland") + , (308, T.pack "Grenada") + , (312, T.pack "Guadeloupe") + , (316, T.pack "Guam") + , (320, T.pack "Guatemala") + , (831, T.pack "Guernsey") + , (324, T.pack "Guinea") + , (624, T.pack "Guinea-Bissau") + , (328, T.pack "Guyana") + , (332, T.pack "Haiti") + , (334, T.pack "Heard Island and McDonald Islands") + , (336, T.pack "Holy See") + , (340, T.pack "Honduras") + , (344, T.pack "Hong Kong") + , (348, T.pack "Hungary") + , (352, T.pack "Iceland") + , (352, T.pack "Lýðveldið Ísland") + , (352, T.pack "Republic of Iceland") + , (352, T.pack "Ísafold") + , (352, T.pack "Ísland") + , (356, T.pack "India") + , (356, T.pack "Al-Hind") + , (356, T.pack "Aryavarta") + , (356, T.pack "Bharatham") + , (356, T.pack "Bhārat") + , (356, T.pack "Bhārat Gaṇarājya") + , (356, T.pack "Bhārat Prajatantra") + , (356, T.pack "Bhāratadēśaṁ") + , (356, T.pack "Bhāratavarṣa") + , (356, T.pack "Hind") + , (356, T.pack "Hindustan") + , (356, T.pack "Republic of India") + , (356, T.pack "Union of India") + , (356, T.pack "ഇന്ത്യ") + , (360, T.pack "Indonesia") + , (364, T.pack "Iran (Islamic Republic of)") + , (368, T.pack "Iraq") + , (368, T.pack "Assyria") + , (368, T.pack "Babylon") + , (368, T.pack "Mesopotamia") + , (372, T.pack "Ireland") + , (372, T.pack "Irish Free State") + , (372, T.pack "Republic of Ireland") + , (372, T.pack "Saorstát Éireann") + , (372, T.pack "Éire") + , (833, T.pack "Isle of Man") + , (376, T.pack "Israel") + , (376, T.pack "Dawlat Isra'il") + , (376, T.pack "Hebrew State") + , (376, T.pack "Jewish State") + , (376, T.pack "Medinat Yisra'el") + , (376, T.pack "State of Israel") + , (376, T.pack "The Hebrew State") + , (376, T.pack "The Jewish State") + , (376, T.pack "מדינת ישראל") + , (376, T.pack "دَوْلَة إِسْرَائِيل") + , (380, T.pack "Italy") + , (380, T.pack "Repubblica Italiana") + , (388, T.pack "Jamaica") + , (388, T.pack "Xamayca") + , (392, T.pack "Japan") + , (832, T.pack "Jersey") + , (400, T.pack "Jordan") + , (400, T.pack "Hashemite Kingdom of Jordan") + , (398, T.pack "Kazakhstan") + , (398, T.pack "Kazakh Soviet Socialist Republic") + , (398, T.pack "Qazaq Keñestik Socïalïstik Respwblïkası") + , (398, T.pack "Qazaqstan") + , (398, T.pack "Republic of Kazakhstan") + , (398, T.pack "Қазақ Кеңестік Социалистік Республикасы") + , (398, T.pack "Қазақстан") + , (398, T.pack "Қазақстан Республикасы") + , (404, T.pack "Kenya") + , (404, T.pack "Jamhuri ya Kenya") + , (404, T.pack "Republic of Kenya") + , (296, T.pack "Kiribati") + , (408, T.pack "Korea (Democratic People's Republic of)") + , (410, T.pack "Korea (Republic of)") + , (414, T.pack "Kuwait") + , (417, T.pack "Kyrgyzstan") + , (418, T.pack "Lao People's Democratic Republic") + , (428, T.pack "Latvia") + , (428, T.pack "Latvija") + , (428, T.pack "Letland") + , (428, T.pack "Letonija") + , (428, T.pack "Letonnie") + , (428, T.pack "Lettland") + , (422, T.pack "Lebanon") + , (426, T.pack "Lesotho") + , (426, T.pack "Basutoland") + , (426, T.pack "Kingdom of Lesotho") + , (430, T.pack "Liberia") + , (434, T.pack "Libya") + , (438, T.pack "Liechtenstein") + , (440, T.pack "Lithuania") + , (440, T.pack "Lietuva") + , (440, T.pack "Литва") + , (442, T.pack "Luxembourg") + , (442, T.pack "Grand Duchy of Luxembourg") + , (442, T.pack "Grand-Duché de Luxembourg") + , (442, T.pack "Groussherzogdem Lëtzebuerg") + , (442, T.pack "Großherzogtum Luxemburg") + , (442, T.pack "Lussemburgo") + , (442, T.pack "Luxemburg") + , (442, T.pack "Luxemburgo") + , (446, T.pack "Macao") + , (807, T.pack "Macedonia (the former Yugoslav Republic of)") + , (450, T.pack "Madagascar") + , (454, T.pack "Malawi") + , (454, T.pack "Nyasaland") + , (454, T.pack "Republic of Malawi") + , (458, T.pack "Malaysia") + , (458, T.pack "Federation of Malaysia") + , (458, T.pack "Mǎláixīyà") + , (458, T.pack "Persekutuan Malaysia") + , (458, T.pack "马来西亚") + , (462, T.pack "Maldives") + , (462, T.pack "Dhibat-al-Mahal") + , (462, T.pack "Dhivehi Raajje") + , (462, T.pack "Mahal Dvipa") + , (462, T.pack "Maléldvipa") + , (462, T.pack "Republic of the Maldives") + , (462, T.pack "The Maldive Islands") + , (462, T.pack "ދިވެހިރާއްޖެ") + , (466, T.pack "Mali") + , (470, T.pack "Malta") + , (584, T.pack "Marshall Islands") + , (474, T.pack "Martinique") + , (478, T.pack "Mauritania") + , (478, T.pack "Islamic Republic of Mauritania") + , (478, T.pack "République Islamique de Mauritanie") + , (480, T.pack "Mauritius") + , (175, T.pack "Mayotte") + , (484, T.pack "Mexico") + , (484, T.pack "Aztlán") + , (484, T.pack "Aztlān") + , (484, T.pack "Estados Unidos Mexicanos") + , (484, T.pack "MX") + , (484, T.pack "Mejico") + , (484, T.pack "Méjico") + , (484, T.pack "México") + , (484, T.pack "República Mexicana") + , (484, T.pack "United Mexican States") + , (583, T.pack "Micronesia (Federated States of)") + , (498, T.pack "Moldova (Republic of)") + , (492, T.pack "Monaco") + , (496, T.pack "Mongolia") + , (499, T.pack "Montenegro") + , (499, T.pack "Republic of Singapore") + , (499, T.pack "Singapore") + , (499, T.pack "Singapura") + , (499, T.pack "Yugoslavia") + , (500, T.pack "Montserrat") + , (504, T.pack "Morocco") + , (508, T.pack "Mozambique") + , (104, T.pack "Myanmar") + , (516, T.pack "Namibia") + , (520, T.pack "Nauru") + , (524, T.pack "Nepal") + , (528, T.pack "Netherlands") + , (540, T.pack "New Caledonia") + , (554, T.pack "New Zealand") + , (558, T.pack "Nicaragua") + , (562, T.pack "Niger") + , (566, T.pack "Nigeria") + , (570, T.pack "Niue") + , (574, T.pack "Norfolk Island") + , (580, T.pack "Northern Mariana Islands") + , (578, T.pack "Norway") + , (512, T.pack "Oman") + , (586, T.pack "Pakistan") + , (585, T.pack "Palau") + , (275, T.pack "Palestine, State of") + , (591, T.pack "Panama") + , (598, T.pack "Papua New Guinea") + , (600, T.pack "Paraguay") + , (604, T.pack "Peru") + , (608, T.pack "Philippines") + , (612, T.pack "Pitcairn") + , (616, T.pack "Poland") + , (620, T.pack "Portugal") + , (630, T.pack "Puerto Rico") + , (634, T.pack "Qatar") + , (638, T.pack "Réunion") + , (642, T.pack "Romania") + , (642, T.pack "Kingdom of Romania") + , (642, T.pack "Regatul României") + , (642, T.pack "Republica Populară Romînă") + , (642, T.pack "Republica Socialistă România") + , (642, T.pack "Romanian People's Republic") + , (642, T.pack "România") + , (642, T.pack "Roumania") + , (642, T.pack "Rumania") + , (642, T.pack "Socialist Republic of Romania") + , (643, T.pack "Russian Federation") + , (643, T.pack "Great Russia") + , (643, T.pack "R.F.") + , (643, T.pack "RF") + , (643, T.pack "RSFSR") + , (643, T.pack "Rossiya") + , (643, T.pack "Rus") + , (643, T.pack "Russia") + , (643, T.pack "Russian Socialist Federative Soviet Republic") + , (643, T.pack "Russian Soviet Federative Socialist Republic") + , (643, T.pack "Russland") + , (643, T.pack "SSSR") + , (643, T.pack "SU") + , (643, T.pack "Sovetskiy Soyuz") + , (643, T.pack "Soviet Union") + , (643, T.pack "Soyuz Sovetskikh Sotsialisticheskikh Respublik") + , (643, T.pack "USSR") + , (643, T.pack "Union of Soviet Socialist Republics") + , (643, T.pack "Российская Федерация") + , (643, T.pack "Россия") + , (643, T.pack "СССР") + , (643, T.pack "Сове́тский Сою́з") + , (643, T.pack "Союз Советских Социалистических Республик") + , (646, T.pack "Rwanda") + , (646, T.pack "Republic of Rwanda") + , (646, T.pack "Repubulika y'u Rwanda") + , (646, T.pack "Ruanda") + , (646, T.pack "Rwandese Republic") + , (646, T.pack "République du Rwanda") + , (652, T.pack "Saint Barthélemy") + , (654, T.pack "Saint Helena, Ascension and Tristan da Cunha") + , (659, T.pack "Saint Kitts and Nevis") + , (659, T.pack "Federation of Saint Kitts and Nevis") + , (659, T.pack "Saint Kitts-Nevis-Anguilla") + , (662, T.pack "Saint Lucia") + , (663, T.pack "Saint Martin (French part)") + , (666, T.pack "Saint Pierre and Miquelon") + , (670, T.pack "Saint Vincent and the Grenadines") + , (882, T.pack "Samoa") + , (674, T.pack "San Marino") + , (678, T.pack "Sao Tome and Principe") + , (682, T.pack "Saudi Arabia") + , (686, T.pack "Senegal") + , (688, T.pack "Serbia") + , (690, T.pack "Seychelles") + , (694, T.pack "Sierra Leone") + , (702, T.pack "Singapore") + , (702, T.pack "Montenegro") + , (702, T.pack "Republic of Singapore") + , (702, T.pack "Singapura") + , (702, T.pack "Yugoslavia") + , (534, T.pack "Sint Maarten (Dutch part)") + , (703, T.pack "Slovakia") + , (705, T.pack "Slovenia") + , (90, T.pack "Solomon Islands") + , (706, T.pack "Somalia") + , (710, T.pack "South Africa") + , (239, T.pack "South Georgia and the South Sandwich Islands") + , (728, T.pack "South Sudan") + , (724, T.pack "Spain") + , (724, T.pack "Espainia") + , (724, T.pack "Espanya") + , (724, T.pack "España") + , (724, T.pack "Hispania") + , (724, T.pack "Iberia") + , (724, T.pack "Reino de España") + , (724, T.pack "Spanish State") + , (144, T.pack "Sri Lanka") + , (729, T.pack "Sudan") + , (729, T.pack "Nubia") + , (729, T.pack "Republic of the Sudan") + , (729, T.pack "The Sudan") + , (740, T.pack "Suriname") + , (740, T.pack "Dutch Guiana") + , (740, T.pack "Netherlands Guiana") + , (740, T.pack "Republic of Suriname") + , (740, T.pack "Republiek Suriname") + , (740, T.pack "Surinam") + , (744, T.pack "Svalbard and Jan Mayen") + , (748, T.pack "Swaziland") + , (752, T.pack "Sweden") + , (752, T.pack "Kingdom of Sweden") + , (752, T.pack "Konungariket Sverige") + , (752, T.pack "Svea Rike") + , (752, T.pack "Sverige") + , (752, T.pack "Svitjod") + , (756, T.pack "Switzerland") + , (756, T.pack "CH") + , (756, T.pack "Confederazione Svizzera") + , (756, T.pack "Confederaziun Svizra") + , (756, T.pack "Confoederatio Helvetica") + , (756, T.pack "Confédération suisse") + , (756, T.pack "Helvetia") + , (756, T.pack "Schweiz") + , (756, T.pack "Schweizerische Eidgenossenschaft") + , (756, T.pack "Suisse") + , (756, T.pack "Svizra") + , (756, T.pack "Svizzera") + , (756, T.pack "Swiss Confederation") + , (760, T.pack "Syrian Arab Republic") + , (158, T.pack "Taiwan, Province of China") + , (158, T.pack "Chinese Taipei") + , (158, T.pack "Republic of China") + , (158, T.pack "Separate Customs Territory of Taiwan, Penghu, Kinmen and Matsu") + , (158, T.pack "Taiwan") + , (158, T.pack "Zhonggua Taipei") + , (158, T.pack "中華台北") + , (762, T.pack "Tajikistan") + , (834, T.pack "Tanzania, United Republic of") + , (764, T.pack "Thailand") + , (626, T.pack "Timor-Leste") + , (768, T.pack "Togo") + , (772, T.pack "Tokelau") + , (776, T.pack "Tonga") + , (780, T.pack "Trinidad and Tobago") + , (780, T.pack "Republic of Trinidad and Tobago") + , (780, T.pack "Trinbago") + , (788, T.pack "Tunisia") + , (792, T.pack "Turkey") + , (792, T.pack "Ottoman Empire") + , (792, T.pack "Republic of Turkey") + , (792, T.pack "Turkish Empire") + , (792, T.pack "Türkiye Cumhuriyeti") + , (795, T.pack "Turkmenistan") + , (796, T.pack "Turks and Caicos Islands") + , (798, T.pack "Tuvalu") + , (800, T.pack "Uganda") + , (804, T.pack "Ukraine") + , (804, T.pack "UA") + , (804, T.pack "UKR") + , (804, T.pack "Ukraїna") + , (784, T.pack "United Arab Emirates") + , (784, T.pack "Al Emirat al Arabbiya al Muttahida") + , (784, T.pack "U.A.E.") + , (784, T.pack "UAE") + , (826, T.pack "United Kingdom of Great Britain and Northern Ireland") + , (840, T.pack "United States of America") + , (840, T.pack "Estados Unidos de América") + , (840, T.pack "The United States") + , (840, T.pack "U.S.") + , (840, T.pack "U.S.A.") + , (840, T.pack "USA") + , (840, T.pack "United States") + , (840, T.pack "Usa") + , (840, T.pack "Usono") + , (581, T.pack "United States Minor Outlying Islands") + , (858, T.pack "Uruguay") + , (860, T.pack "Uzbekistan") + , (548, T.pack "Vanuatu") + , (548, T.pack "Ripablik blong Vanuatu") + , (548, T.pack "République de Vanuatu") + , (862, T.pack "Venezuela (Bolivarian Republic of)") + , (704, T.pack "Viet Nam") + , (92, T.pack "Virgin Islands (British)") + , (850, T.pack "Virgin Islands (U.S.)") + , (876, T.pack "Wallis and Futuna") + , (732, T.pack "Western Sahara") + , (887, T.pack "Yemen") + , (894, T.pack "Zambia") + , (894, T.pack "Republic of Zambia") + , (716, T.pack "Zimbabwe") + , (716, T.pack "Republic of Zimbabwe") + , (383, T.pack "Kosovo") + ] +{-# NOINLINE aliases #-} diff --git a/src/Country/Unexposed/Encode/English.hs b/src/Country/Unexposed/Encode/English.hs --- a/src/Country/Unexposed/Encode/English.hs +++ b/src/Country/Unexposed/Encode/English.hs @@ -260,5 +260,6 @@ , (887, T.pack "Yemen",('Y','E'),('Y','E','M')) , (894, T.pack "Zambia",('Z','M'),('Z','M','B')) , (716, T.pack "Zimbabwe",('Z','W'),('Z','W','E')) + , (383, T.pack "Kosovo",('X','K'),('X','K','X')) ] {-# NOINLINE countryNameQuads #-} diff --git a/src/Country/Unexposed/Enumerate.hs b/src/Country/Unexposed/Enumerate.hs deleted file mode 100644 --- a/src/Country/Unexposed/Enumerate.hs +++ /dev/null @@ -1,258 +0,0 @@ -module Country.Unexposed.Enumerate where - --- This module is autogenerated. Do not edit it by hand. - -import Data.Word (Word16) - -enumeratedCountries :: [Word16] -enumeratedCountries = - [ 4 - , 248 - , 8 - , 12 - , 16 - , 20 - , 24 - , 660 - , 10 - , 28 - , 32 - , 51 - , 533 - , 36 - , 40 - , 31 - , 44 - , 48 - , 50 - , 52 - , 112 - , 56 - , 84 - , 204 - , 60 - , 64 - , 68 - , 535 - , 70 - , 72 - , 74 - , 76 - , 86 - , 96 - , 100 - , 854 - , 108 - , 116 - , 120 - , 124 - , 132 - , 136 - , 140 - , 148 - , 152 - , 156 - , 162 - , 166 - , 170 - , 174 - , 178 - , 180 - , 184 - , 188 - , 384 - , 191 - , 192 - , 531 - , 196 - , 203 - , 208 - , 262 - , 212 - , 214 - , 218 - , 818 - , 222 - , 226 - , 232 - , 233 - , 231 - , 238 - , 234 - , 242 - , 246 - , 250 - , 254 - , 258 - , 260 - , 266 - , 270 - , 268 - , 276 - , 288 - , 292 - , 300 - , 304 - , 308 - , 312 - , 316 - , 320 - , 831 - , 324 - , 624 - , 328 - , 332 - , 334 - , 336 - , 340 - , 344 - , 348 - , 352 - , 356 - , 360 - , 364 - , 368 - , 372 - , 833 - , 376 - , 380 - , 388 - , 392 - , 832 - , 400 - , 398 - , 404 - , 296 - , 408 - , 410 - , 414 - , 417 - , 418 - , 428 - , 422 - , 426 - , 430 - , 434 - , 438 - , 440 - , 442 - , 446 - , 807 - , 450 - , 454 - , 458 - , 462 - , 466 - , 470 - , 584 - , 474 - , 478 - , 480 - , 175 - , 484 - , 583 - , 498 - , 492 - , 496 - , 499 - , 500 - , 504 - , 508 - , 104 - , 516 - , 520 - , 524 - , 528 - , 540 - , 554 - , 558 - , 562 - , 566 - , 570 - , 574 - , 580 - , 578 - , 512 - , 586 - , 585 - , 275 - , 591 - , 598 - , 600 - , 604 - , 608 - , 612 - , 616 - , 620 - , 630 - , 634 - , 638 - , 642 - , 643 - , 646 - , 652 - , 654 - , 659 - , 662 - , 663 - , 666 - , 670 - , 882 - , 674 - , 678 - , 682 - , 686 - , 688 - , 690 - , 694 - , 702 - , 534 - , 703 - , 705 - , 90 - , 706 - , 710 - , 239 - , 728 - , 724 - , 144 - , 729 - , 740 - , 744 - , 748 - , 752 - , 756 - , 760 - , 158 - , 762 - , 834 - , 764 - , 626 - , 768 - , 772 - , 776 - , 780 - , 788 - , 792 - , 795 - , 796 - , 798 - , 800 - , 804 - , 784 - , 826 - , 840 - , 581 - , 858 - , 860 - , 548 - , 862 - , 704 - , 92 - , 850 - , 876 - , 732 - , 887 - , 894 - , 716 - ] diff --git a/src/Country/Unexposed/ExtraNames.hs b/src/Country/Unexposed/ExtraNames.hs deleted file mode 100644 --- a/src/Country/Unexposed/ExtraNames.hs +++ /dev/null @@ -1,11 +0,0 @@ -{-# LANGUAGE OverloadedStrings #-} - -module Country.Unexposed.ExtraNames - ( ) where - -import Country.Unsafe (Country) -import Country.Identifier -import Data.Text (Text) -import Data.Primitive (indexArray,newArray,unsafeFreezeArray,writeArray, - writeByteArray,indexByteArray,unsafeFreezeByteArray,newByteArray) - diff --git a/src/Country/Unexposed/Names.hs b/src/Country/Unexposed/Names.hs --- a/src/Country/Unexposed/Names.hs +++ b/src/Country/Unexposed/Names.hs @@ -11,6 +11,7 @@ , englishIdentifierNamesText , numberOfPossibleCodes , decodeMap + , decodeMapUtf8 , alphaTwoHashMap , alphaThreeHashMap , decodeNumeric @@ -22,22 +23,26 @@ import Data.Hashable (Hashable) import Data.Primitive.Types (Prim) import Data.HashMap.Strict (HashMap) +import Data.ByteString (ByteString) import Data.Primitive (indexArray) import Data.Primitive.Array (Array(..)) import Data.Primitive.ByteArray (ByteArray(..)) import Control.Monad +import Data.Text.Encoding (encodeUtf8) +import Country.Unexposed.Alias (aliases) import qualified Data.Text as T import qualified Data.Aeson as AE import qualified Data.Aeson.Types as AET import qualified Data.HashMap.Strict as HM import qualified Data.List as L import Control.Monad.ST +import Foreign.Storable (Storable) import Data.Text (Text) import Data.Word import Data.Char (toLower,isAlpha) import Country.Unexposed.Encode.English (countryNameQuads) import Data.Primitive (Array,indexArray,newArray,unsafeFreezeArray,writeArray, - writeByteArray,indexByteArray,unsafeFreezeByteArray,newByteArray) + writeByteArray,indexByteArray,unsafeFreezeByteArray,newByteArray,sizeOf) import qualified Data.Text as T import qualified Data.Scientific as SCI @@ -45,39 +50,6 @@ encodeEnglish :: Country -> Text encodeEnglish (Country n) = indexArray englishCountryNamesText (word16ToInt n) -mexico :: Country -mexico = Country 484 - -unitedStatesOfAmerica :: Country -unitedStatesOfAmerica = Country 840 - -ålandIslands :: Country -ålandIslands = Country 248 - -venezuelaBolivarianRepublicOf :: Country -venezuelaBolivarianRepublicOf = Country 862 - -boliviaPlurinationalStateOf :: Country -boliviaPlurinationalStateOf = Country 68 - -extraNames :: [(Country,Text)] -extraNames = - [ (unitedStatesOfAmerica,"United States") - , (unitedStatesOfAmerica,"The United States") - , (unitedStatesOfAmerica,"USA") - , (unitedStatesOfAmerica,"U.S.A.") - , (unitedStatesOfAmerica,"Estados Unidos de América") - , (mexico,"Estados Unidos Mexicanos") - , (mexico,"México") - , (mexico,"Méjico") - , (ålandIslands,"Aland Islands") - , (ålandIslands,"Aaland Islands") - , (venezuelaBolivarianRepublicOf,"Venezuela") - , (venezuelaBolivarianRepublicOf,"Bolivarian Republic of Venezuela") - , (boliviaPlurinationalStateOf,"Bolivia") - , (boliviaPlurinationalStateOf,"Plurinational State of Bolivia") - ] - englishCountryNamesText :: Array Text englishCountryNamesText = runST $ do m <- newArray numberOfPossibleCodes unnamed @@ -111,17 +83,59 @@ decodeMap :: HashMap Text Country decodeMap = let baseMap = HM.union alphaTwoHashMap alphaThreeHashMap - hm1 = L.foldl' (\hm (country,name) -> HM.insert name country hm) baseMap extraNames + hm1 = L.foldl' (\hm (countryNum,name) -> HM.insert name (Country countryNum) hm) baseMap aliases hm2 = L.foldl' (\hm (countryNum,name,_,_) -> HM.insert name (Country countryNum) hm) hm1 countryNameQuads - in hm2 + hm3 = HM.foldlWithKey' (\hm name cty -> HM.insert (T.toLower name) cty $ HM.insert (T.toTitle name) cty $ hm) hm2 hm2 + in hm3 {-# NOINLINE decodeMap #-} +decodeMapUtf8 :: HashMap ByteString Country +decodeMapUtf8 = HM.foldlWithKey' (\hm k v -> HM.insert (encodeUtf8 k) v hm) HM.empty decodeMap +{-# NOINLINE decodeMapUtf8 #-} + -- | A country recognized by ISO 3166. newtype Country = Country Word16 - deriving (Eq,Ord,Prim,Hashable) + deriving (Eq,Ord,Prim,Hashable,Storable) instance Show Country where show (Country n) = T.unpack (indexArray englishIdentifierNamesText (word16ToInt n)) + +instance Enum Country where + fromEnum (Country w) = indexByteArray countryCodeToSequentialMapping (fromIntegral w) + toEnum number = if number >= 0 && number < actualNumberOfCountries + then Country (indexByteArray sequentialToCountryCodeMapping number) + else error ("toEnum: cannot convert " ++ show number ++ " to Country") + +instance Bounded Country where + minBound = Country (indexByteArray sequentialToCountryCodeMapping 0) + maxBound = Country (indexByteArray sequentialToCountryCodeMapping (actualNumberOfCountries - 1)) + +orderedCountryCodes :: [Word16] +orderedCountryCodes = L.sort $ map (\(a,_,_,_) -> a) countryNameQuads + +countryCodeToSequentialMapping :: ByteArray +countryCodeToSequentialMapping = runST $ do + numbers <- newByteArray (numberOfPossibleCodes * sizeOf (undefined :: Int)) + forM_ (zip [0 :: Int,1..] orderedCountryCodes) $ \(number,code) -> do + writeByteArray numbers (word16ToInt code) number + unsafeFreezeByteArray numbers +{-# NOINLINE countryCodeToSequentialMapping #-} + +sequentialToCountryCodeMapping :: ByteArray +sequentialToCountryCodeMapping = runST $ do + codes <- newByteArray (actualNumberOfCountries * sizeOf (undefined :: Word16)) + forM_ (zip [0 :: Int,1..] orderedCountryCodes) $ \(number,code) -> do + writeByteArray codes number (code :: Word16) + unsafeFreezeByteArray codes +{-# NOINLINE sequentialToCountryCodeMapping #-} + +actualNumberOfCountries :: Int +actualNumberOfCountries = length countryNameQuads +{-# NOINLINE actualNumberOfCountries #-} + + +codeToEnum :: Word16 -> Int +codeToEnum w = indexByteArray countryCodeToSequentialMapping (word16ToInt w) -- todo: add support for encoding directly to bytestring. -- Also, add suport for ToJSONKey and FromJSONKey once everything