langdetect-ruby 0.1.1

data/lib/lingua_ruby/profiles/id.json

@@ -0,0 +1,402 @@
+{
+  "a": 0,
+  "n": 1,
+  "e": 2,
+  "i": 3,
+  "d": 4,
+  "an": 5,
+  "u": 6,
+  "r": 7,
+  "a ": 8,
+  "s": 9,
+  "m": 10,
+  "g": 11,
+  " d": 12,
+  "k": 13,
+  "t": 14,
+  "n ": 15,
+  "an ": 16,
+  "b": 17,
+  "i ": 18,
+  "p": 19,
+  "l": 20,
+  "da": 21,
+  "ng": 22,
+  "ga": 23,
+  "er": 24,
+  "sa": 25,
+  "ra": 26,
+  "h": 27,
+  "ar": 28,
+  "me": 29,
+  "ang": 30,
+  "ne": 31,
+  "in": 32,
+  "be": 33,
+  " m": 34,
+  "ya": 35,
+  " p": 36,
+  "ah": 37,
+  "la": 38,
+  " me": 39,
+  "y": 40,
+  "ia": 41,
+  " s": 42,
+  " b": 43,
+  " da": 44,
+  "en": 45,
+  "es": 46,
+  "di": 47,
+  "du": 48,
+  "ng ": 49,
+  "g ": 50,
+  "at": 51,
+  "am": 52,
+  "o": 53,
+  "ia ": 54,
+  "si": 55,
+  "al": 56,
+  "h ": 57,
+  "ba": 58,
+  " k": 59,
+  "ke": 60,
+  " i": 61,
+  "ua": 62,
+  "em": 63,
+  "nd": 64,
+  "ah ": 65,
+  "ri": 66,
+  "pa": 67,
+  " in": 68,
+  "ara": 69,
+  "ra ": 70,
+  "ka": 71,
+  " ke": 72,
+  "ni": 73,
+  "un": 74,
+  " du": 75,
+  "yan": 76,
+  "ik": 77,
+  "a d": 78,
+  " ya": 79,
+  "ag": 80,
+  "pe": 81,
+  " sa": 82,
+  " y": 83,
+  "aga": 84,
+  " pe": 85,
+  " n": 86,
+  "ad": 87,
+  " a": 88,
+  "ta": 89,
+  "on": 90,
+  "ala": 91,
+  "do": 92,
+  " t": 93,
+  " ne": 94,
+  "neg": 95,
+  "ega": 96,
+  "gar": 97,
+  " di": 98,
+  "ha": 99,
+  "eg": 100,
+  "tu": 101,
+  "sia": 102,
+  "li": 103,
+  "nt": 104,
+  "esi": 105,
+  "j": 106,
+  " be": 107,
+  "nes": 108,
+  "dan": 109,
+  "one": 110,
+  "don": 111,
+  "ndo": 112,
+  "ind": 113,
+  "ada": 114,
+  "lah": 115,
+  "ban": 116,
+  "era": 117,
+  "nga": 118,
+  "n d": 119,
+  "ari": 120,
+  "ri ": 121,
+  "ua ": 122,
+  "atu": 123,
+  "nta": 124,
+  "a i": 125,
+  "mer": 126,
+  "kan": 127,
+  "uni": 128,
+  "at ": 129,
+  " ba": 130,
+  "sa ": 131,
+  "i d": 132,
+  "di ": 133,
+  "mem": 134,
+  "ber": 135,
+  "erb": 136,
+  "ter": 137,
+  " te": 138,
+  "aya": 139,
+  "uan": 140,
+  "na": 141,
+  "mi": 142,
+  "de": 143,
+  "rb": 144,
+  "uk": 145,
+  "t ": 146,
+  "ay": 147,
+  "k ": 148,
+  "te": 149,
+  "as": 150,
+  "dun": 151,
+  "ru": 152,
+  "up": 153,
+  "aha": 154,
+  "esa": 155,
+  "rbe": 156,
+  "mp": 157,
+  "ula": 158,
+  "pul": 159,
+  "ini": 160,
+  "ni ": 161,
+  "m ": 162,
+  "am ": 163,
+  "gan": 164,
+  " j": 165,
+  "eng": 166,
+  "den": 167,
+  " de": 168,
+  "a s": 169,
+  "ud": 170,
+  "eb": 171,
+  "a b": 172,
+  "dua": 173,
+  " r": 174,
+  "ak": 175,
+  "a m": 176,
+  "eru": 177,
+  "upa": 178,
+  "sat": 179,
+  "a p": 180,
+  "n p": 181,
+  "nia": 182,
+  "mpa": 183,
+  "ama": 184,
+  "ai ": 185,
+  " se": 186,
+  "i m": 187,
+  "gam": 188,
+  "au": 189,
+  "emi": 190,
+  "mil": 191,
+  "ya ": 192,
+  "pu": 193,
+  "ul": 194,
+  "ili": 195,
+  "ma": 196,
+  "ki": 197,
+  "u ": 198,
+  "dal": 199,
+  "se": 200,
+  " ad": 201,
+  "ki ": 202,
+  "ai": 203,
+  "pem": 204,
+  "iki": 205,
+  "lik": 206,
+  "il": 207,
+  "tia": 208,
+  "rag": 209,
+  "et": 210,
+  "i s": 211,
+  "le": 212,
+  "us ": 213,
+  "ika": 214,
+  " ju": 215,
+  "gsa": 216,
+  "ngs": 217,
+  "n a": 218,
+  "da ": 219,
+  "eda": 220,
+  "g s": 221,
+  "pen": 222,
+  "end": 223,
+  "bed": 224,
+  "uk ": 225,
+  "adi": 226,
+  "kay": 227,
+  "dar": 228,
+  "n b": 229,
+  "a r": 230,
+  "dat": 231,
+  "man": 232,
+  "dik": 233,
+  "m p": 234,
+  "lam": 235,
+  "una": 236,
+  "gun": 237,
+  "g d": 238,
+  "i y": 239,
+  "eme": 240,
+  "i n": 241,
+  "g m": 242,
+  "eri": 243,
+  "rin": 244,
+  "eka": 245,
+  "int": 246,
+  "tah": 247,
+  "asa": 248,
+  "n k": 249,
+  "has": 250,
+  "emb": 251,
+  "bah": 252,
+  "t d": 253,
+  "har": 254,
+  "r ": 255,
+  "mel": 256,
+  "si ": 257,
+  "asi": 258,
+  "h s": 259,
+  "n n": 260,
+  "aka": 261,
+  "rup": 262,
+  "i k": 263,
+  "ben": 264,
+  "is": 265,
+  "ed": 266,
+  "nj": 267,
+  "ja": 268,
+  " u": 269,
+  "i a": 270,
+  "k d": 271,
+  "bag": 272,
+  "gai": 273,
+  "us": 274,
+  "s ": 275,
+  "a y": 276,
+  " un": 277,
+  "bu": 278,
+  "a a": 279,
+  "ar ": 280,
+  "sar": 281,
+  "bes": 282,
+  "eh": 283,
+  "n t": 284,
+  "id": 285,
+  "lau": 286,
+  "gu": 287,
+  "a k": 288,
+  "n i": 289,
+  "men": 290,
+  "gs": 291,
+  "nja": 292,
+  "sam": 293,
+  "ek": 294,
+  "tua": 295,
+  "ti": 296,
+  "enj": 297,
+  "el": 298,
+  "ju": 299,
+  "nya": 300,
+  "ny": 301,
+  "nya ": 302,
+  "meny": 303,
+  "eny": 304,
+  "peny": 305,
+  "kep": 306,
+  "epa": 307,
+  "pad": 308,
+  "pada": 309,
+  " pad": 310,
+  "itu": 311,
+  " itu": 312,
+  "yang": 313,
+  " yan": 314,
+  "denga": 315,
+  "dengan": 316,
+  "ngan": 317,
+  " den": 318,
+  "juga": 319,
+  " jug": 320,
+  "uga ": 321,
+  "seb": 322,
+  "seba": 323,
+  "bagai": 324,
+  "agai": 325,
+  "telah": 326,
+  "elah": 327,
+  "tela": 328,
+  " tel": 329,
+  "untu": 330,
+  "ntuk": 331,
+  "tuk ": 332,
+  " unt": 333,
+  "ora": 334,
+  "orang": 335,
+  "rang": 336,
+  " ora": 337,
+  "dap": 338,
+  "apat": 339,
+  "dpat": 340,
+  "suda": 341,
+  "udah": 342,
+  " sud": 343,
+  "haru": 344,
+  "arus": 345,
+  " har": 346,
+  "aka ": 347,
+  "mere": 348,
+  "erek": 349,
+  "reka": 350,
+  " mer": 351,
+  "maka": 352,
+  " mak": 353,
+  "tida": 354,
+  "idak": 355,
+  "dak ": 356,
+  " tid": 357,
+  "seh": 358,
+  "sehi": 359,
+  "hing": 360,
+  "ingg": 361,
+  " seh": 362,
+  "aka": 363,
+  "masi": 364,
+  "asih": 365,
+  " mas": 366,
+  "kare": 367,
+  "aren": 368,
+  "rena": 369,
+  " kar": 370,
+  "dari": 371,
+  " dar": 372,
+  "ari ": 373,
+  "oleh": 374,
+  "leh ": 375,
+  " ole": 376,
+  "sela": 377,
+  "elam": 378,
+  " sel": 379,
+  "perl": 380,
+  "erlu": 381,
+  " per": 382,
+  "bisa": 383,
+  "isa ": 384,
+  " bis": 385,
+  "sang": 386,
+  "anga": 387,
+  "ngat": 388,
+  " san": 389,
+  "meng": 390,
+  "eng ": 391,
+  "ngg": 392,
+  "pern": 393,
+  "erna": 394,
+  "rnah": 395,
+  " per": 396,
+  "seka": 397,
+  "ekar": 398,
+  " sek": 399
+}

data/lib/lingua_ruby/profiles/ja.json

@@ -0,0 +1,302 @@
+{
+  "の": 0,
+  "で": 1,
+  "日": 2,
+  "本": 3,
+  "は": 4,
+  "す": 5,
+  "日本": 6,
+  "す ": 7,
+  "す 日": 8,
+  " 日本": 9,
+  "日本の": 10,
+  "本の": 11,
+  " 日": 12,
+  "です": 13,
+  "て": 14,
+  "世界": 15,
+  "に": 16,
+  "です ": 17,
+  "り": 18,
+  "界": 19,
+  "世": 20,
+  "り ": 21,
+  "都": 22,
+  "東": 23,
+  "ア": 24,
+  "は日本": 25,
+  "ら": 26,
+  "れ": 27,
+  "特": 28,
+  "であり": 29,
+  "あり ": 30,
+  "本は": 31,
+  "ていま": 32,
+  "います": 33,
+  "ます ": 34,
+  "文化は": 35,
+  "化は": 36,
+  "文化": 37,
+  "は日": 38,
+  "ます": 39,
+  "いま": 40,
+  "てい": 41,
+  "であ": 42,
+  "あり": 43,
+  "一": 44,
+  "化": 45,
+  "し": 46,
+  "と": 47,
+  "日本は": 48,
+  "が": 49,
+  "語": 50,
+  "文": 51,
+  "れて": 52,
+  "用": 53,
+  "あ": 54,
+  "い": 55,
+  "ま": 56,
+  "に有": 57,
+  "特に": 58,
+  "本の文": 59,
+  "の文化": 60,
+  "してい": 61,
+  "用して": 62,
+  "使用し": 63,
+  "が使用": 64,
+  "人が使": 65,
+  "万人が": 66,
+  "アジア": 67,
+  "は特": 68,
+  "化は伝": 69,
+  "らは": 70,
+  "ぷら": 71,
+  "天ぷ": 72,
+  "は伝統": 73,
+  "や天": 74,
+  "司や": 75,
+  "東ア": 76,
+  "伝統と": 77,
+  "統と現": 78,
+  "と現代": 79,
+  "現代が": 80,
+  "代が融": 81,
+  "が融合": 82,
+  "ジアに": 83,
+  "アに位": 84,
+  "に位置": 85,
+  "東アジ": 86,
+  "は東ア": 87,
+  "本は東": 88,
+  "位置す": 89,
+  "置する": 90,
+  "する島": 91,
+  "る島国": 92,
+  "島国で": 93,
+  "国です": 94,
+  "日本語": 95,
+  "本語は": 96,
+  "語は日": 97,
+  "名で": 98,
+  "名です": 99,
+  "有名": 100,
+  "本の公": 101,
+  "の公用": 102,
+  "公用語": 103,
+  "用語で": 104,
+  "語であ": 105,
+  "り 約": 106,
+  " 約一": 107,
+  "約一億": 108,
+  "一億二": 109,
+  "億二千": 110,
+  "二千万": 111,
+  "千万人": 112,
+  "らは特": 113,
+  "食文化": 114,
+  "の食文": 115,
+  "本の食": 116,
+  "れてい": 117,
+  "られて": 118,
+  "知られ": 119,
+  "に知ら": 120,
+  "的に知": 121,
+  "界的に": 122,
+  "世界的": 123,
+  "で世界": 124,
+  "業で世": 125,
+  "造業で": 126,
+  "製造業": 127,
+  "と製造": 128,
+  "新と製": 129,
+  "革新と": 130,
+  "術革新": 131,
+  "ぷらは": 132,
+  "天ぷら": 133,
+  "や天ぷ": 134,
+  "司や天": 135,
+  "寿司や": 136,
+  " 寿司": 137,
+  "り 寿": 138,
+  "おり ": 139,
+  "ており": 140,
+  "れてお": 141,
+  "されて": 142,
+  "愛され": 143,
+  "で愛さ": 144,
+  "中で愛": 145,
+  "界中で": 146,
+  "世界中": 147,
+  "は世界": 148,
+  "化は世": 149,
+  "技術革": 150,
+  "の首都": 151,
+  "本の首": 152,
+  "京は日": 153,
+  "特に有": 154,
+  "に有名": 155,
+  "有名で": 156,
+  "東京は": 157,
+  " 東京": 158,
+  "す 東": 159,
+  "のです": 160,
+  "もので": 161,
+  "のもの": 162,
+  "特のも": 163,
+  "独特の": 164,
+  "た独特": 165,
+  "した独": 166,
+  "合した": 167,
+  "融合し": 168,
+  "は技術": 169,
+  "本は技": 170,
+  "つです": 171,
+  "一つで": 172,
+  "の一つ": 173,
+  "圏の一": 174,
+  "市圏の": 175,
+  "都市圏": 176,
+  "は特に": 177,
+  "の都市": 178,
+  "大の都": 179,
+  "最大の": 180,
+  "界最大": 181,
+  "世界最": 182,
+  " 世界": 183,
+  "り 世": 184,
+  "都であ": 185,
+  "首都で": 186,
+  "は東": 187,
+  "名": 188,
+  "有": 189,
+  "ぷ": 190,
+  "天": 191,
+  "や": 192,
+  "司": 193,
+  "寿": 194,
+  "お": 195,
+  "さ": 196,
+  "愛": 197,
+  "中": 198,
+  "食": 199,
+  "知": 200,
+  "的": 201,
+  "業": 202,
+  "代": 203,
+  "約一": 204,
+  " 約": 205,
+  "語で": 206,
+  "用語": 207,
+  "公用": 208,
+  "の公": 209,
+  "語は": 210,
+  "本語": 211,
+  "国で": 212,
+  "島国": 213,
+  "る島": 214,
+  "する": 215,
+  "置す": 216,
+  "位置": 217,
+  "に位": 218,
+  "アに": 219,
+  "ジア": 220,
+  "アジ": 221,
+  "現": 222,
+  "統": 223,
+  "伝": 224,
+  "使": 225,
+  "人": 226,
+  "万": 227,
+  "千": 228,
+  "二": 229,
+  "億": 230,
+  "約": 231,
+  "公": 232,
+  "国": 233,
+  "島": 234,
+  "る": 235,
+  "置": 236,
+  "位": 237,
+  "ジ": 238,
+  "造": 239,
+  "製": 240,
+  "新": 241,
+  "革": 242,
+  "術": 243,
+  "技": 244,
+  "つ": 245,
+  "圏": 246,
+  "市": 247,
+  "大": 248,
+  "最": 249,
+  "首": 250,
+  "京": 251,
+  "も": 252,
+  "独": 253,
+  "た": 254,
+  "合": 255,
+  "融": 256,
+  "業で": 257,
+  "造業": 258,
+  "製造": 259,
+  "と製": 260,
+  "新と": 261,
+  "革新": 262,
+  "術革": 263,
+  "技術": 264,
+  "は技": 265,
+  "つで": 266,
+  "一つ": 267,
+  "の一": 268,
+  "圏の": 269,
+  "市圏": 270,
+  "都市": 271,
+  "の都": 272,
+  "大の": 273,
+  "が融": 274,
+  "寿司": 275,
+  " 寿": 276,
+  "おり": 277,
+  "てお": 278,
+  "され": 279,
+  "愛さ": 280,
+  "で愛": 281,
+  "中で": 282,
+  "界中": 283,
+  "は世": 284,
+  "食文": 285,
+  "の食": 286,
+  "られ": 287,
+  "知ら": 288,
+  "に知": 289,
+  "的に": 290,
+  "界的": 291,
+  "で世": 292,
+  "代が": 293,
+  "現代": 294,
+  "と現": 295,
+  "統と": 296,
+  "伝統": 297,
+  "は伝": 298,
+  "の文": 299
+}