langdetect-ruby 0.1.1 → 0.2.0

data/lib/lingua_ruby/profiles/te.json

@@ -0,0 +1,254 @@
+{
+  "ా": 0,
+  "ర": 1,
+  "్": 2,
+  "ు": 3,
+  "ల": 4,
+  "ద": 5,
+  "ం": 6,
+  "త": 7,
+  "ు ": 8,
+  "ి": 9,
+  "న": 10,
+  "రా": 11,
+  "్ర": 12,
+  "మ": 13,
+  "గ": 14,
+  " త": 15,
+  "ేశ": 16,
+  "దే": 17,
+  "ియ": 18,
+  " మ": 19,
+  "ే": 20,
+  "శ": 21,
+  "దేశ": 22,
+  "య": 23,
+  " తె": 24,
+  "ెల": 25,
+  "తె": 26,
+  "తెల": 27,
+  "ె": 28,
+  "ణ": 29,
+  "ష": 30,
+  "క": 31,
+  "ప్": 32,
+  "్ ": 33,
+  "మర": 34,
+  "రి": 35,
+  "యు": 36,
+  "లం": 37,
+  "ంగ": 38,
+  "గా": 39,
+  "ాష": 40,
+  "ాణ": 41,
+  "ణ ": 42,
+  " ర": 43,
+  " ఒక": 44,
+  "ఒక ": 45,
+  "లో ": 46,
+  "శం ": 47,
+  "ేశం": 48,
+  "ప్ర": 49,
+  " మర": 50,
+  "మరి": 51,
+  "రియ": 52,
+  "ియు": 53,
+  "యు ": 54,
+  "ెలం": 55,
+  "లంగ": 56,
+  "ంగా": 57,
+  "గాణ": 58,
+  "ాణ ": 59,
+  "ణ ర": 60,
+  " రా": 61,
+  "్ల": 62,
+  "ట్": 63,
+  "భ": 64,
+  "ో": 65,
+  "ఒ": 66,
+  "ధ": 67,
+  "ప": 68,
+  "ట": 69,
+  "భా": 70,
+  "ార": 71,
+  "శం": 72,
+  "ం ": 73,
+  " ద": 74,
+  "ాల": 75,
+  "లో": 76,
+  "ో ": 77,
+  "క ": 78,
+  "ఒక": 79,
+  " ఒ": 80,
+  "గరం": 81,
+  "హైద": 82,
+  "ఆంధ": 83,
+  "ంధ్": 84,
+  "ధ్ర": 85,
+  "్రప": 86,
+  "రప్": 87,
+  " హై": 88,
+  "్రద": 89,
+  "రదే": 90,
+  "ేశ్": 91,
+  "శ్ ": 92,
+  "్ మ": 93,
+  "ఖ న": 94,
+  "ుఖ ": 95,
+  "ముఖ": 96,
+  "రము": 97,
+  "్రమ": 98,
+  "ు త": 99,
+  "క ద": 100,
+  " దే": 101,
+  "ం త": 102,
+  "నగర": 103,
+  "ెలు": 104,
+  "లుగ": 105,
+  "ుగు": 106,
+  "గు ": 107,
+  "ు భ": 108,
+  " భా": 109,
+  "భాష": 110,
+  "ాషన": 111,
+  "షను": 112,
+  "ను ": 113,
+  "ు ఆ": 114,
+  " ఆం": 115,
+  " నగ": 116,
+  "ైదర": 117,
+  " మా": 118,
+  "మాట": 119,
+  "ని ": 120,
+  "ాని": 121,
+  "ధాన": 122,
+  "జధా": 123,
+  "ాజధ": 124,
+  "రాజ": 125,
+  "్ త": 126,
+  "ద్ ": 127,
+  "తార": 128,
+  "డతా": 129,
+  "ాడత": 130,
+  "ాద్": 131,
+  "ాట్": 132,
+  "ట్ల": 133,
+  "బాద": 134,
+  "్లా": 135,
+  "ాబా": 136,
+  "లాడ": 137,
+  " ప్": 138,
+  "దరా": 139,
+  "ు హ": 140,
+  "క ప": 141,
+  "ు ఒ": 142,
+  "రాబ": 143,
+  "రు ": 144,
+  "రాష": 145,
+  "ాష్": 146,
+  "ష్ట": 147,
+  "్ట్": 148,
+  "ట్ర": 149,
+  "్రా": 150,
+  "రాల": 151,
+  "ాల్": 152,
+  "ల్ల": 153,
+  "్లో": 154,
+  "ారు": 155,
+  "ో మ": 156,
+  "ి మ": 157,
+  "స": 158,
+  "న ": 159,
+  "లు": 160,
+  "ుగ": 161,
+  "గు": 162,
+  " భ": 163,
+  "షన": 164,
+  "ను": 165,
+  " ఆ": 166,
+  "ఆం": 167,
+  "ంధ": 168,
+  "ధ్": 169,
+  "రప": 170,
+  "రద": 171,
+  "శ్": 172,
+  "ష్": 173,
+  "్ట": 174,
+  "ల్": 175,
+  "మా": 176,
+  "ాట": 177,
+  "లా": 178,
+  "ాడ": 179,
+  "డత": 180,
+  "తా": 181,
+  "ఉ": 182,
+  "ఆ": 183,
+  "డ": 184,
+  "హ": 185,
+  "ై": 186,
+  "బ": 187,
+  "జ": 188,
+  "ఖ": 189,
+  "రత": 190,
+  "తద": 191,
+  "దక": 192,
+  "క్": 193,
+  "్ష": 194,
+  "షి": 195,
+  "ిణ": 196,
+  "ణా": 197,
+  "ాస": 198,
+  "సి": 199,
+  "యా": 200,
+  " ఉ": 201,
+  "ఉన": 202,
+  "న్": 203,
+  "్న": 204,
+  "రు": 205,
+  "భార": 206,
+  "ారత": 207,
+  "రతద": 208,
+  "తదే": 209,
+  "ం ద": 210,
+  " దక": 211,
+  "దక్": 212,
+  "క్ష": 213,
+  "్షి": 214,
+  "షిణ": 215,
+  "ిణా": 216,
+  "ణాస": 217,
+  "ాసి": 218,
+  "సియ": 219,
+  "ియా": 220,
+  "యాల": 221,
+  "ాలో": 222,
+  "ో ఉ": 223,
+  " ఉన": 224,
+  "ఉన్": 225,
+  "న్న": 226,
+  "్న ": 227,
+  "న ఒ": 228,
+  " హ": 229,
+  "హై": 230,
+  "ైద": 231,
+  "దర": 232,
+  "ాబ": 233,
+  "బా": 234,
+  "ాద": 235,
+  "ద్": 236,
+  "ాజ": 237,
+  "జధ": 238,
+  "ధా": 239,
+  "ాన": 240,
+  "ని": 241,
+  "ి ": 242,
+  " ప": 243,
+  "రమ": 244,
+  "ము": 245,
+  "ుఖ": 246,
+  "ఖ ": 247,
+  " న": 248,
+  "నగ": 249,
+  "గర": 250,
+  "రం": 251
+}

data/lib/lingua_ruby/profiles/th.json

@@ -0,0 +1,251 @@
+{
+  "เ": 0,
+  "ท": 1,
+  "ป": 2,
+  "ร": 3,
+  "อ": 4,
+  "ง": 5,
+  "น": 6,
+  "ย": 7,
+  "า": 8,
+  "ะ": 9,
+  "ะเ": 10,
+  "เท": 11,
+  "ระเ": 12,
+  "ะเท": 13,
+  "เทศ": 14,
+  "ปร": 15,
+  "ประ": 16,
+  "ระ": 17,
+  "ทศ": 18,
+  "ศ": 19,
+  "่": 20,
+  "ี": 21,
+  "ไท": 22,
+  "ทย": 23,
+  "เป": 24,
+  "ป็": 25,
+  "็น": 26,
+  "ไทย": 27,
+  "ป็น": 28,
+  "เป็": 29,
+  "ไ": 30,
+  "็": 31,
+  "ใ": 32,
+  "อง": 33,
+  "ห": 34,
+  "ทศไ": 35,
+  "ม": 36,
+  "นเ": 37,
+  "ก": 38,
+  "ที่": 39,
+  "ศไ": 40,
+  "ี่": 41,
+  "ศไท": 42,
+  "ที": 43,
+  "ต": 44,
+  "มือ": 45,
+  "่ใ": 46,
+  "ใน": 47,
+  "ยเป": 48,
+  "ทยเ": 49,
+  "าร": 50,
+  "ษา": 51,
+  "าษ": 52,
+  "ือง": 53,
+  "ภา": 54,
+  "ภาษ": 55,
+  "ขอ": 56,
+  "็นเ": 57,
+  "ีย": 58,
+  "นเม": 59,
+  "เมื": 60,
+  "ช": 61,
+  "ว": 62,
+  "ภ": 63,
+  "ษ": 64,
+  "ข": 65,
+  "ุ": 66,
+  "้": 67,
+  "เม": 68,
+  "ื": 69,
+  "ล": 70,
+  "มื": 71,
+  "ือ": 72,
+  "ั": 73,
+  "ยเ": 74,
+  "นป": 75,
+  "ของ": 76,
+  "องป": 77,
+  "งปร": 78,
+  "าษา": 79,
+  "นปร": 80,
+  "งป": 81,
+  "ู": 82,
+  "ษาร": 83,
+  "ใต้": 84,
+  "ต้ ": 85,
+  "้ ภ": 86,
+  " ภา": 87,
+  "นภา": 88,
+  "็นภ": 89,
+  "าไท": 90,
+  "ษาไ": 91,
+  "ในเ": 92,
+  "นเอ": 93,
+  "เอเ": 94,
+  "อเช": 95,
+  "เชี": 96,
+  "ชีย": 97,
+  "ียต": 98,
+  "ยตะ": 99,
+  "ตะว": 100,
+  "ะวั": 101,
+  "วัน": 102,
+  "ันอ": 103,
+  "นออ": 104,
+  "ออก": 105,
+  "อกเ": 106,
+  "กเฉ": 107,
+  "เฉี": 108,
+  "ฉีย": 109,
+  "ียง": 110,
+  "ยงใ": 111,
+  "งใต": 112,
+  "งหล": 113,
+  "หลว": 114,
+  "ลวง": 115,
+  "วงข": 116,
+  "งขอ": 117,
+  "ทยแ": 118,
+  "ยแล": 119,
+  "และ": 120,
+  "ละเ": 121,
+  "ะเป": 122,
+  "องท": 123,
+  "งที": 124,
+  "ี่ใ": 125,
+  "่ให": 126,
+  "ใหญ": 127,
+  "หญ่": 128,
+  "ญ่ท": 129,
+  "่ที": 130,
+  "ี่ส": 131,
+  "่สุ": 132,
+  "สุด": 133,
+  "ุดใ": 134,
+  "ดใน": 135,
+  "ในป": 136,
+  "ารา": 137,
+  "ราช": 138,
+  "าชก": 139,
+  "ชกา": 140,
+  "การ": 141,
+  "ารข": 142,
+  "รขอ": 143,
+  "ทย ": 144,
+  "ย ก": 145,
+  " กร": 146,
+  "กรุ": 147,
+  "รุง": 148,
+  "ุงเ": 149,
+  "งเท": 150,
+  "เทพ": 151,
+  "ทพม": 152,
+  "พมห": 153,
+  "มหา": 154,
+  "หาน": 155,
+  "านค": 156,
+  "นคร": 157,
+  "ครเ": 158,
+  "รเป": 159,
+  "องห": 160,
+  "ะว": 161,
+  "วั": 162,
+  "ัน": 163,
+  "นอ": 164,
+  "ออ": 165,
+  "อก": 166,
+  "กเ": 167,
+  "เฉ": 168,
+  "ฉี": 169,
+  "ยง": 170,
+  "งใ": 171,
+  "ใต": 172,
+  "ต้": 173,
+  "้ ": 174,
+  " ภ": 175,
+  "าไ": 176,
+  "นภ": 177,
+  "รา": 178,
+  "าช": 179,
+  "ชก": 180,
+  "กา": 181,
+  "รข": 182,
+  "ฉ": 183,
+  "พ": 184,
+  "ค": 185,
+  "แ": 186,
+  "ญ": 187,
+  "ส": 188,
+  "ด": 189,
+  "ศท": 190,
+  "่ต": 191,
+  "ตั": 192,
+  "ั้": 193,
+  "้ง": 194,
+  "งอ": 195,
+  "อย": 196,
+  "ยู": 197,
+  "ู่": 198,
+  "เอ": 199,
+  "อเ": 200,
+  "เช": 201,
+  "ชี": 202,
+  "ยต": 203,
+  "ตะ": 204,
+  "งท": 205,
+  "ให": 206,
+  "หญ": 207,
+  "ญ่": 208,
+  "่ท": 209,
+  "่ส": 210,
+  "สุ": 211,
+  "ุด": 212,
+  "ดใ": 213,
+  "็นป": 214,
+  "ทศท": 215,
+  "ศที": 216,
+  "ี่ต": 217,
+  "่ตั": 218,
+  "ตั้": 219,
+  "ั้ง": 220,
+  "้งอ": 221,
+  "งอย": 222,
+  "อยู": 223,
+  "ยู่": 224,
+  "ู่ใ": 225,
+  "่ใน": 226,
+  "ย ": 227,
+  " ก": 228,
+  "กร": 229,
+  "รุ": 230,
+  "ุง": 231,
+  "งเ": 232,
+  "ทพ": 233,
+  "พม": 234,
+  "มห": 235,
+  "หา": 236,
+  "าน": 237,
+  "นค": 238,
+  "คร": 239,
+  "รเ": 240,
+  "งห": 241,
+  "หล": 242,
+  "ลว": 243,
+  "วง": 244,
+  "งข": 245,
+  "ยแ": 246,
+  "แล": 247,
+  "ละ": 248
+}