@delali/narsil-react-native 0.3.1-next.155 → 0.3.1-next.157

package/cpp/core/src/languages/data/kazakh.c

@@ -0,0 +1,57 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[317] = {
+  1072, 1079, 1072, 1083, 1072, 1088, 1179, 1099, 1083, 1099, 1073, 1072, 1088, 1073, 1072, 1088, 1083, 1099, 1171,
+  1099, 1073, 1072, 1088, 1083, 1099, 1179, 1073, 1072, 1089, 1179, 1072, 1073, 1077, 1085, 1073, 1077, 1088, 1110,
+  1073, 1086, 1081, 1099, 1085, 1096, 1072, 1073, 1086, 1083, 1076, 1099, 1073, 1086, 1083, 1099, 1087, 1073, 1086,
+  1083, 1171, 1072, 1085, 1073, 1110, 1079, 1073, 1110, 1079, 1076, 1110, 1187, 1073, 1110, 1088, 1073, 1110, 1088,
+  1072, 1179, 1073, 1201, 1083, 1076, 1072, 1076, 1077, 1076, 1077, 1081, 1110, 1085, 1077, 1076, 1110, 1077, 1082,
+  1077, 1085, 1077, 1082, 1110, 1077, 1084, 1077, 1089, 1077, 1085, 1076, 1110, 1077, 1187, 1078, 1086, 1179, 1078,
+  1241, 1085, 1077, 1082, 1077, 1081, 1110, 1085, 1082, 1110, 1084, 1082, 1257, 1087, 1084, 1077, 1085, 1084, 1077,
+  1085, 1110, 1187, 1085, 1077, 1085, 1077, 1075, 1077, 1085, 1077, 1084, 1077, 1089, 1077, 1086, 1083, 1086, 1083,
+  1072, 1088, 1086, 1083, 1072, 1088, 1076, 1099, 1086, 1085, 1099, 1187, 1086, 1089, 1099, 1087, 1077, 1085, 1089,
+  1072, 1081, 1099, 1085, 1089, 1077, 1073, 1077, 1073, 1110, 1089, 1077, 1085, 1089, 1077, 1085, 1076, 1077, 1088,
+  1089, 1077, 1085, 1110, 1187, 1089, 1086, 1083, 1089, 1086, 1085, 1076, 1072, 1089, 1086, 1085, 1076, 1099, 1179,
+  1090, 1072, 1085, 1089, 1086, 1187, 1089, 1110, 1079, 1089, 1110, 1079, 1076, 1077, 1088, 1089, 1110, 1079, 1076,
+  1110, 1187, 1090, 1072, 1090, 1072, 1171, 1099, 1090, 1077, 1090, 1077, 1082, 1090, 1091, 1088, 1072, 1083, 1099,
+  1103, 1171, 1072, 1085, 1072, 1179, 1072, 1079, 1110, 1088, 1179, 1072, 1081, 1076, 1072, 1179, 1072, 1083, 1072,
+  1081, 1179, 1072, 1085, 1072, 1179, 1072, 1085, 1076, 1072, 1081, 1179, 1072, 1085, 1096, 1072, 1179, 1072, 1088,
+  1072, 1081, 1179, 1072, 1096, 1072, 1085, 1199, 1096, 1110, 1085, 1241, 1083, 1110, 1241, 1088, 1241, 1088, 1073,
+  1110, 1088, 1241, 1088, 1110, 1257, 1079, 1257, 1079, 1110, 1257, 1090, 1077,
+};
+
+static const uint32_t stop_word_offsets[79] = {
+  0, 2, 4, 10, 13, 20, 26, 31, 34, 38, 45, 50, 55, 61, 64, 70, 73, 78, 81, 83, 85, 90, 93, 97, 100, 104, 108, 110, 113,
+  117, 122, 125, 128, 131, 136, 138, 142, 148, 150, 154, 160, 164, 167, 170, 175, 181, 184, 190, 195, 198, 203, 212,
+  215, 218, 224, 230, 232, 236, 238, 241, 247, 248, 252, 257, 262, 267, 271, 277, 282, 287, 292, 296, 299, 301, 306,
+  309, 311, 314, 317,
+};
+
+static const uint32_t keep_ranges[6] = {
+  48, 57, 1024, 1279, 7296, 7303,
+};
+
+const narsil_language narsil_language_kazakh = {
+  .name = "kazakh",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 78,
+  .ascii_keep = {0x03FF000000000000ULL, 0x0000000000000000ULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 3,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/khmer.c

@@ -0,0 +1,71 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[523] = {
+  6016, 6070, 6031, 6091, 6016, 6070, 6035, 6091, 6016, 6070, 6042, 6016, 6070, 6043, 6016, 6078, 6031, 6016, 6095,
+  6016, 6098, 6031, 6072, 6017, 6070, 6020, 6017, 6075, 6047, 6017, 6098, 6025, 6075, 6086, 6017, 6098, 6043, 6087,
+  6018, 6070, 6031, 6091, 6018, 6074, 6018, 6074, 6023, 6070, 6018, 6081, 6018, 6098, 6035, 6070, 6018, 6098, 6042,
+  6070, 6019, 6078, 6025, 6021, 6020, 6091, 6021, 6070, 6036, 6091, 6021, 6070, 6086, 6021, 6075, 6020, 6021, 6076,
+  6043, 6021, 6081, 6025, 6021, 6086, 6038, 6084, 6087, 6022, 6070, 6036, 6091, 6023, 6070, 6020, 6023, 6071, 6031,
+  6023, 6076, 6035, 6023, 6077, 6041, 6023, 6078, 6020, 6024, 6036, 6091, 6024, 6040, 6025, 6073, 6016, 6025, 6082,
+  6016, 6026, 6016, 6026, 6020, 6026, 6043, 6091, 6026, 6070, 6086, 6026, 6073, 6020, 6026, 6076, 6021, 6026, 6078,
+  6042, 6026, 6082, 6042, 6026, 6082, 6043, 6026, 6084, 6041, 6026, 6086, 6026, 6095, 6026, 6095, 6033, 6083, 6030,
+  6070, 6030, 6070, 6033, 6085, 6030, 6070, 6047, 6091, 6031, 6070, 6040, 6031, 6070, 6086, 6020, 6031, 6078, 6031,
+  6082, 6032, 6070, 6032, 6098, 6040, 6072, 6033, 6033, 6081, 6033, 6043, 6091, 6033, 6070, 6035, 6091, 6033, 6070,
+  6036, 6033, 6070, 6047, 6091, 6033, 6070, 6086, 6020, 6033, 6072, 6033, 6072, 6030, 6070, 6033, 6072, 6035, 6081,
+  6087, 6033, 6072, 6035, 6084, 6087, 6033, 6075, 6016, 6033, 6076, 6033, 6085, 6033, 6080, 6031, 6033, 6081, 6033,
+  6084, 6087, 6033, 6084, 6087, 6036, 6072, 6033, 6085, 6033, 6085, 6043, 6078, 6034, 6098, 6044, 6078, 6035, 6070,
+  6020, 6035, 6070, 6041, 6035, 6071, 6020, 6035, 6073, 6016, 6035, 6073, 6020, 6035, 6076, 6044, 6035, 6081, 6087,
+  6035, 6081, 6087, 6023, 6070, 6035, 6083, 6035, 6084, 6087, 6035, 6084, 6087, 6023, 6070, 6035, 6084, 6087, 6033,
+  6081, 6035, 6085, 6035, 6085, 6031, 6082, 6035, 6085, 6033, 6072, 6035, 6085, 6043, 6078, 6036, 6072, 6036, 6072,
+  6035, 6081, 6087, 6036, 6077, 6035, 6036, 6078, 6036, 6078, 6016, 6036, 6098, 6042, 6070, 6086, 6036, 6098, 6042,
+  6078, 6036, 6098, 6042, 6082, 6037, 6020, 6037, 6075, 6031, 6038, 6072, 6038, 6072, 6042, 6038, 6072, 6043, 6078,
+  6038, 6075, 6086, 6038, 6077, 6016, 6038, 6081, 6025, 6038, 6081, 6043, 6038, 6084, 6043, 6038, 6098, 6042, 6084,
+  6087, 6039, 6070, 6038, 6039, 6098, 6043, 6074, 6040, 6016, 6040, 6070, 6035, 6040, 6071, 6035, 6040, 6076, 6043,
+  6040, 6077, 6041, 6040, 6082, 6035, 6040, 6089, 6081, 6021, 6040, 6098, 6039, 6083, 6040, 6098, 6043, 6089, 6081,
+  6087, 6041, 6016, 6041, 6078, 6020, 6041, 6089, 6070, 6020, 6042, 6041, 6088, 6042, 6070, 6043, 6091, 6042, 6074,
+  6042, 6074, 6016, 6095, 6042, 6077, 6021, 6042, 6077, 6040, 6043, 6078, 6043, 6078, 6016, 6043, 6078, 6047, 6043,
+  6082, 6020, 6043, 6084, 6016, 6043, 6098, 6050, 6044, 6070, 6044, 6071, 6025, 6044, 6082, 6020, 6047, 6084, 6087,
+  6047, 6098, 6035, 6078, 6047, 6098, 6040, 6084, 6087, 6048, 6031, 6091, 6048, 6070, 6040, 6048, 6078, 6041, 6048,
+  6085, 6048, 6090, 6070, 6035, 6049, 6078, 6020, 6049, 6078, 6041, 6050, 6047, 6091, 6050, 6070, 6021, 6050, 6072,
+  6050, 6084, 6041, 6050, 6086, 6038, 6072, 6050, 6098, 6044, 6072, 6053, 6031, 6060, 6060, 6033, 6081, 6063, 6063,
+  6016, 6084, 6063, 6020, 6063, 6030, 6070, 6065, 6098, 6041,
+};
+
+static const uint32_t stop_word_offsets[165] = {
+  0, 4, 8, 11, 14, 17, 19, 23, 26, 29, 34, 38, 42, 44, 48, 50, 54, 58, 61, 64, 68, 71, 74, 77, 80, 85, 89, 92, 95, 98,
+  101, 104, 107, 109, 112, 115, 117, 119, 122, 125, 128, 131, 134, 137, 140, 143, 145, 147, 151, 153, 157, 161, 164,
+  168, 170, 172, 174, 178, 181, 184, 188, 191, 195, 199, 201, 205, 210, 215, 218, 222, 225, 227, 230, 235, 237, 241,
+  245, 248, 251, 254, 257, 260, 263, 266, 271, 273, 276, 281, 286, 288, 292, 296, 300, 302, 307, 310, 312, 315, 320,
+  324, 328, 330, 333, 335, 338, 342, 345, 348, 351, 354, 357, 362, 365, 369, 371, 374, 377, 380, 383, 386, 390, 394,
+  400, 402, 405, 409, 412, 416, 418, 422, 425, 428, 430, 433, 436, 439, 442, 445, 447, 450, 453, 456, 460, 465, 468,
+  471, 474, 476, 480, 483, 486, 489, 492, 494, 497, 501, 505, 507, 508, 511, 512, 515, 517, 520, 523,
+};
+
+static const uint32_t keep_ranges[16] = {
+  48, 57, 65, 90, 97, 122, 383, 383, 6016, 6099, 6103, 6103, 6112, 6121, 8490, 8490,
+};
+
+const narsil_language narsil_language_khmer = {
+  .name = "khmer",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 164,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 8,
+  .splits_code_points = 1,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 2,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/kinyarwanda.c

@@ -0,0 +1,54 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[319] = {
+  97, 98, 97, 97, 98, 97, 110, 100, 105, 97, 98, 111, 97, 104, 111, 97, 107, 97, 97, 107, 111, 97, 110, 100, 105, 97,
+  114, 105, 97, 114, 105, 107, 111, 97, 121, 97, 97, 121, 111, 98, 97, 114, 105, 98, 111, 98, 111, 115, 101, 98, 117,
+  114, 105, 98, 119, 98, 121, 111, 115, 101, 99, 121, 99, 121, 97, 110, 101, 99, 121, 97, 110, 103, 119, 97, 103, 117,
+  115, 97, 104, 97, 103, 97, 116, 105, 104, 97, 110, 111, 104, 97, 114, 105, 104, 97, 114, 105, 121, 97, 104, 97, 115,
+  105, 104, 101, 106, 117, 114, 117, 104, 111, 115, 101, 105, 98, 105, 105, 98, 105, 110, 100, 105, 105, 98, 121, 111,
+  105, 99, 121, 111, 105, 107, 105, 105, 107, 105, 110, 100, 105, 105, 110, 100, 105, 105, 114, 105, 105, 114, 121, 111,
+  105, 121, 105, 105, 121, 111, 105, 122, 105, 110, 100, 105, 105, 122, 111, 106, 121, 101, 107, 97, 110, 100, 105, 107,
+  117, 107, 117, 98, 101, 114, 97, 107, 117, 107, 111, 107, 117, 114, 105, 109, 98, 101, 114, 101, 109, 117, 109, 117,
+  114, 105, 109, 119, 101, 109, 119, 101, 115, 101, 110, 110, 97, 110, 100, 105, 110, 105, 110, 105, 98, 97, 110, 106,
+  121, 101, 110, 107, 97, 110, 111, 110, 111, 110, 101, 104, 111, 110, 116, 97, 110, 121, 117, 109, 97, 114, 119, 115,
+  105, 116, 117, 114, 105, 116, 119, 101, 116, 119, 101, 115, 101, 117, 117, 98, 117, 117, 98, 119, 111, 117, 107, 117,
+  117, 107, 119, 111, 117, 110, 100, 105, 117, 114, 105, 117, 114, 117, 117, 114, 119, 111, 117, 116, 117, 117, 116,
+  119, 111, 117, 119, 111, 117, 121, 117, 119, 101, 119, 111, 119, 101, 121, 121, 97, 114, 105, 121, 111, 115, 101, 122,
+  111, 115, 101,
+};
+
+static const uint32_t stop_word_offsets[88] = {
+  0, 3, 9, 12, 15, 18, 21, 25, 28, 33, 36, 39, 43, 45, 49, 53, 55, 60, 62, 67, 74, 78, 84, 88, 92, 98, 102, 108, 112,
+  115, 121, 125, 129, 132, 138, 142, 145, 149, 152, 155, 161, 164, 167, 172, 174, 180, 184, 188, 193, 195, 199, 202,
+  207, 208, 210, 213, 215, 219, 223, 226, 228, 234, 237, 242, 244, 246, 250, 253, 258, 259, 262, 266, 269, 273, 277,
+  280, 283, 287, 290, 294, 297, 300, 302, 306, 307, 311, 315, 319,
+};
+
+static const uint32_t keep_ranges[6] = {
+  48, 57, 65, 90, 97, 122,
+};
+
+const narsil_language narsil_language_kinyarwanda = {
+  .name = "kinyarwanda",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 87,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 3,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/kirundi.c

@@ -0,0 +1,53 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[294] = {
+  97, 98, 97, 97, 98, 97, 110, 100, 105, 97, 98, 111, 97, 104, 111, 97, 107, 97, 97, 107, 111, 97, 114, 105, 97, 114,
+  105, 107, 111, 97, 121, 97, 97, 121, 111, 98, 97, 114, 105, 98, 111, 98, 111, 115, 101, 98, 119, 99, 99, 97, 110, 101,
+  99, 97, 110, 107, 101, 103, 117, 115, 97, 104, 97, 103, 97, 116, 105, 104, 97, 110, 121, 117, 109, 97, 104, 97, 114,
+  105, 104, 97, 115, 105, 104, 101, 106, 117, 114, 117, 104, 111, 115, 101, 105, 98, 105, 105, 98, 105, 110, 100, 105,
+  105, 99, 111, 105, 107, 105, 105, 107, 105, 110, 100, 105, 105, 109, 98, 101, 114, 101, 105, 110, 121, 117, 109, 97,
+  105, 114, 105, 105, 114, 121, 111, 105, 118, 121, 111, 105, 121, 105, 105, 121, 111, 105, 122, 105, 110, 100, 105,
+  105, 122, 111, 106, 101, 119, 101, 107, 97, 110, 100, 105, 107, 117, 107, 117, 98, 101, 114, 97, 107, 117, 107, 111,
+  107, 117, 114, 105, 109, 117, 109, 117, 114, 105, 109, 119, 101, 98, 119, 101, 109, 119, 101, 115, 101, 110, 110, 97,
+  110, 100, 105, 110, 105, 110, 107, 97, 110, 111, 110, 116, 97, 114, 119, 115, 105, 116, 117, 114, 105, 116, 119, 101,
+  98, 119, 101, 116, 119, 101, 115, 101, 117, 117, 98, 117, 117, 107, 117, 117, 107, 119, 111, 117, 114, 105, 117, 114,
+  117, 117, 114, 119, 111, 117, 116, 117, 117, 116, 119, 111, 117, 119, 111, 117, 119, 117, 110, 100, 105, 117, 121,
+  117, 118, 121, 118, 121, 111, 115, 101, 119, 101, 119, 101, 119, 101, 121, 121, 97, 114, 105, 121, 111, 115, 101, 122,
+  111, 115, 101,
+};
+
+static const uint32_t stop_word_offsets[81] = {
+  0, 3, 9, 12, 15, 18, 21, 24, 29, 32, 35, 39, 41, 45, 47, 48, 52, 57, 61, 67, 74, 78, 82, 88, 92, 95, 101, 104, 107,
+  113, 119, 125, 128, 132, 136, 139, 142, 148, 151, 155, 160, 162, 168, 172, 176, 178, 182, 188, 193, 194, 196, 199,
+  201, 204, 206, 209, 211, 213, 217, 223, 228, 229, 232, 235, 239, 242, 245, 249, 252, 256, 259, 265, 268, 270, 275,
+  277, 281, 282, 286, 290, 294,
+};
+
+static const uint32_t keep_ranges[6] = {
+  48, 57, 65, 90, 97, 122,
+};
+
+const narsil_language narsil_language_kirundi = {
+  .name = "kirundi",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 80,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 3,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/korean.c

@@ -0,0 +1,74 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[401] = {
+  44032, 44032, 47161, 44033, 44033, 44033, 44033, 51088, 44033, 51333, 44057, 45796, 44057, 51060, 44144, 48148, 44144,
+  51032, 44163, 44163, 46308, 44200, 50864, 44208, 44397, 44256, 47196, 44263, 44284, 44284, 50672, 44288, 54620, 44396,
+  44536, 44536, 46308, 44536, 46412, 44536, 47000, 44536, 47100, 44536, 51200, 44592, 53440, 44620, 50501, 44620, 51648,
+  44872, 45817, 45145, 45145, 45180, 51061, 45208, 45224, 46308, 45224, 51667, 45320, 45320, 55148, 45348, 45367, 45380,
+  45572, 44396, 45796, 47480, 45796, 47564, 45796, 49455, 45796, 49548, 45796, 49688, 45796, 51020, 45800, 51648, 45813,
+  45796, 45817, 49888, 45817, 51109, 45845, 44536, 46041, 50504, 46168, 46181, 46181, 46308, 46321, 46321, 46321, 46377,
+  46041, 46384, 46972, 46384, 50948, 46385, 46412, 46608, 46608, 54620, 46749, 46749, 47161, 47196, 47196, 50024, 47449,
+  47484, 47560, 51200, 47560, 52824, 47564, 50557, 47564, 51068, 47564, 53372, 47588, 47588, 48264, 47751, 47784, 47784,
+  46160, 47924, 47157, 47924, 49832, 47924, 50631, 47932, 47200, 48143, 48148, 47196, 48260, 44552, 48400, 48400, 46972,
+  48512, 53552, 48533, 48533, 48708, 47197, 49296, 44145, 49324, 49340, 49444, 47161, 49444, 47560, 49444, 49324, 49483,
+  49548, 49373, 49548, 51064, 49576, 49791, 49884, 44033, 49884, 44036, 49892, 47196, 50500, 50500, 45768, 50500, 50556,
+  50500, 51060, 50500, 54616, 50500, 54857, 50519, 50556, 50557, 44036, 50577, 51088, 50612, 50612, 45712, 50612, 46356,
+  50612, 46412, 50612, 46500, 50612, 51060, 50612, 52236, 50616, 51228, 50620, 47560, 50633, 50633, 50640, 50640, 44172,
+  50640, 49436, 50668, 50668, 44592, 50668, 45919, 50668, 48512, 50668, 49455, 50668, 52264, 50689, 50689, 52264, 50696,
+  50724, 50724, 51649, 50724, 54840, 50752, 50752, 50500, 50780, 50864, 47532, 50864, 49440, 50868, 50868, 50900, 50969,
+  50969, 50977, 51004, 47196, 51012, 51025, 51025, 45817, 51032, 51032, 54644, 51060, 51060, 44163, 51060, 44275, 51060,
+  46412, 51060, 47000, 51060, 47088, 51060, 48264, 51060, 48400, 51060, 49345, 51060, 51232, 51060, 51901, 51064, 51232,
+  51068, 51068, 44273, 51068, 45800, 51068, 46412, 51080, 45796, 51088, 51088, 44592, 51088, 49888, 51104, 44624, 51104,
+  49884, 51200, 51200, 44163, 51200, 44592, 51200, 51901, 51200, 55148, 51204, 48512, 51204, 51088, 51204, 54980, 51228,
+  51312, 44552, 51312, 52264, 51320, 51320, 51328, 51339, 50500, 51341, 51341, 51593, 51593, 49884, 51648, 47564, 52280,
+  52280, 45208, 52423, 52832, 53112, 53112, 53125, 53125, 53237, 53440, 45796, 53440, 51064, 53461, 53461, 53805, 53860,
+  53960, 53440, 54029, 54036, 54141, 54148, 47105, 54616, 54616, 45208, 54616, 54616, 54620, 45936, 54624, 46412, 54624,
+  49104, 54632, 44760, 54644, 50836, 54728, 54728, 44145, 54728, 54728, 54729, 54729, 54729, 54841, 49884, 54841, 51008,
+  54844, 51088, 55016, 50476, 55064, 51061, 55092, 55120, 55120, 55141,
+};
+
+static const uint32_t stop_word_offsets[236] = {
+  0, 1, 3, 4, 6, 8, 10, 12, 14, 16, 18, 19, 21, 23, 25, 27, 28, 29, 31, 33, 34, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53,
+  55, 57, 58, 60, 62, 63, 65, 66, 67, 68, 70, 72, 74, 76, 78, 80, 82, 84, 86, 88, 90, 92, 94, 95, 97, 98, 99, 101, 103,
+  105, 107, 108, 109, 110, 112, 114, 115, 116, 118, 119, 120, 122, 124, 126, 128, 130, 131, 133, 134, 135, 137, 139,
+  141, 143, 145, 146, 148, 150, 151, 153, 155, 157, 159, 161, 162, 163, 165, 167, 169, 170, 172, 174, 175, 176, 178,
+  180, 182, 183, 185, 187, 189, 191, 193, 194, 195, 197, 199, 200, 202, 204, 206, 208, 210, 212, 214, 216, 218, 219,
+  221, 223, 224, 226, 228, 230, 232, 234, 235, 237, 238, 239, 241, 243, 244, 246, 247, 249, 251, 253, 254, 256, 257,
+  259, 260, 261, 263, 264, 266, 267, 269, 271, 273, 275, 277, 279, 281, 283, 285, 287, 289, 290, 292, 294, 296, 298,
+  299, 301, 303, 305, 307, 308, 310, 312, 314, 316, 318, 320, 322, 323, 325, 327, 329, 330, 332, 334, 335, 337, 339,
+  340, 342, 343, 344, 346, 348, 349, 351, 353, 355, 356, 357, 359, 360, 361, 362, 364, 365, 367, 369, 371, 373, 375,
+  377, 379, 380, 382, 384, 385, 387, 389, 391, 393, 395, 397, 398, 400, 401,
+};
+
+static const uint32_t keep_ranges[80] = {
+  48, 57, 65, 90, 97, 122, 383, 383, 4352, 4607, 8490, 8490, 11904, 11929, 11931, 12019, 12032, 12245, 12293, 12293,
+  12295, 12295, 12321, 12329, 12334, 12335, 12344, 12347, 12593, 12686, 12800, 12830, 12896, 12926, 13312, 19903, 19968,
+  40959, 43360, 43388, 44032, 55203, 55216, 55238, 55243, 55291, 63744, 64109, 64112, 64217, 65440, 65470, 65474, 65479,
+  65482, 65487, 65490, 65495, 65498, 65500, 94178, 94179, 94192, 94198, 131072, 173791, 173824, 178205, 178208, 183981,
+  183984, 191456, 191472, 192093, 194560, 195101, 196608, 201546, 201552, 210041,
+};
+
+const narsil_language narsil_language_korean = {
+  .name = "korean",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 235,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 40,
+  .splits_code_points = 1,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 2,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/kurmanji.c

@@ -0,0 +1,43 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[100] = {
+  98, 105, 100, 97, 100, 101, 100, 105, 101, 109, 101, 118, 101, 119, 101, 122, 103, 101, 108, 101, 107, 104, 101, 109,
+  251, 104, 101, 114, 104, 105, 110, 104, 251, 110, 106, 105, 107, 101, 110, 103, 238, 107, 101, 115, 107, 117, 107,
+  234, 107, 238, 108, 105, 109, 101, 109, 105, 110, 116, 101, 116, 105, 351, 116, 116, 117, 118, 97, 118, 234, 118, 238,
+  119, 97, 110, 119, 101, 119, 234, 119, 238, 231, 97, 119, 97, 231, 101, 110, 100, 231, 105, 231, 105, 109, 97, 231,
+  105, 113, 97, 115, 251,
+};
+
+static const uint32_t stop_word_offsets[39] = {
+  0, 2, 4, 6, 8, 10, 12, 14, 16, 21, 25, 28, 31, 34, 36, 41, 44, 46, 48, 50, 52, 54, 57, 59, 63, 65, 67, 69, 71, 74, 76,
+  78, 80, 84, 88, 90, 94, 99, 100,
+};
+
+static const uint32_t keep_ranges[24] = {
+  48, 57, 65, 90, 97, 122, 199, 199, 202, 202, 206, 206, 219, 219, 231, 231, 234, 234, 238, 238, 251, 251, 350, 351,
+};
+
+const narsil_language narsil_language_kurmanji = {
+  .name = "kurmanji",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 38,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 12,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/kyrgyz.c

@@ -0,0 +1,65 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[457] = {
+  1072, 1075, 1072, 1072, 1076, 1072, 1084, 1072, 1081, 1090, 1090, 1099, 1072, 1081, 1090, 1099, 1084, 1099, 1085,
+  1076, 1072, 1072, 1081, 1090, 1099, 1087, 1072, 1083, 1072, 1083, 1072, 1088, 1072, 1083, 1072, 1088, 1076, 1099,
+  1085, 1072, 1083, 1075, 1072, 1085, 1072, 1083, 1091, 1091, 1072, 1083, 1099, 1087, 1072, 1085, 1076, 1072, 1072,
+  1085, 1076, 1072, 1085, 1072, 1085, 1099, 1072, 1085, 1099, 1085, 1072, 1088, 1073, 1072, 1088, 1073, 1072, 1089,
+  1084, 1072, 1073, 1072, 1096, 1073, 1072, 1096, 1082, 1072, 1073, 1072, 1096, 1082, 1099, 1073, 1072, 1096, 1095,
+  1099, 1089, 1099, 1073, 1077, 1088, 1075, 1077, 1085, 1073, 1080, 1079, 1073, 1080, 1083, 1076, 1080, 1088, 1075,
+  1077, 1085, 1073, 1080, 1083, 1076, 1080, 1088, 1076, 1080, 1073, 1080, 1088, 1073, 1080, 1088, 1080, 1085, 1095,
+  1080, 1073, 1080, 1088, 1086, 1082, 1073, 1080, 1096, 1082, 1077, 1082, 1073, 1086, 1083, 1075, 1086, 1085, 1073,
+  1086, 1083, 1086, 1090, 1073, 1086, 1083, 1089, 1086, 1073, 1086, 1083, 1091, 1087, 1073, 1086, 1102, 1085, 1095,
+  1072, 1073, 1091, 1075, 1072, 1073, 1091, 1083, 1075, 1072, 1085, 1072, 1076, 1072, 1076, 1072, 1075, 1099, 1076,
+  1077, 1075, 1077, 1085, 1076, 1077, 1076, 1080, 1076, 1077, 1087, 1078, 1072, 1085, 1072, 1078, 1072, 1090, 1072,
+  1090, 1078, 1072, 1090, 1082, 1072, 1085, 1078, 1072, 1187, 1099, 1078, 1077, 1078, 1086, 1075, 1086, 1088, 1082,
+  1091, 1078, 1086, 1082, 1078, 1086, 1083, 1078, 1086, 1083, 1091, 1082, 1072, 1073, 1099, 1083, 1082, 1072, 1083,
+  1075, 1072, 1085, 1082, 1072, 1085, 1076, 1072, 1081, 1082, 1072, 1088, 1072, 1090, 1072, 1082, 1072, 1088, 1096,
+  1099, 1082, 1072, 1090, 1072, 1088, 1099, 1082, 1077, 1083, 1075, 1077, 1085, 1082, 1077, 1088, 1077, 1082, 1082,
+  1080, 1081, 1080, 1085, 1082, 1086, 1083, 1082, 1099, 1083, 1084, 1099, 1096, 1082, 1099, 1088, 1075, 1099, 1079,
+  1082, 1199, 1085, 1199, 1082, 1257, 1087, 1084, 1072, 1072, 1083, 1099, 1084, 1072, 1090, 1084, 1072, 1084, 1083,
+  1077, 1082, 1077, 1090, 1090, 1080, 1082, 1084, 1077, 1085, 1084, 1077, 1085, 1077, 1085, 1084, 1080, 1187, 1084,
+  1091, 1088, 1076, 1072, 1075, 1099, 1084, 1099, 1081, 1079, 1072, 1084, 1084, 1099, 1085, 1076, 1072, 1081, 1084,
+  1199, 1084, 1082, 1199, 1085, 1086, 1096, 1086, 1083, 1086, 1096, 1086, 1085, 1076, 1086, 1081, 1089, 1199, 1088,
+  1257, 1090, 1089, 1257, 1079, 1090, 1072, 1088, 1072, 1073, 1099, 1085, 1072, 1085, 1090, 1091, 1088, 1075, 1072,
+  1085, 1090, 1091, 1091, 1088, 1072, 1083, 1091, 1091, 1091, 1082, 1091, 1082, 1091, 1095, 1091, 1088, 1076, 1072,
+  1095, 1077, 1081, 1080, 1085, 1095, 1077, 1082, 1101, 1082, 1077, 1085, 1080, 1085, 1101, 1082, 1080, 1101, 1083,
+  1101, 1083, 1077, 1101, 1084, 1077, 1089, 1101, 1084, 1080, 1101, 1095, 1199, 1095, 1199, 1095, 1199, 1085, 1257,
+  1079,
+};
+
+static const uint32_t stop_word_offsets[97] = {
+  0, 3, 7, 12, 21, 26, 28, 32, 39, 44, 48, 52, 56, 61, 64, 68, 70, 73, 78, 81, 86, 91, 98, 104, 107, 116, 124, 127, 134,
+  139, 145, 151, 156, 161, 166, 172, 176, 179, 183, 185, 189, 194, 198, 201, 205, 210, 216, 220, 222, 229, 232, 235,
+  239, 244, 250, 256, 262, 267, 273, 279, 284, 289, 292, 298, 304, 308, 311, 319, 330, 333, 338, 341, 348, 354, 360,
+  366, 370, 377, 382, 385, 394, 400, 408, 412, 418, 423, 426, 432, 435, 437, 440, 444, 447, 449, 451, 455, 457,
+};
+
+static const uint32_t keep_ranges[16] = {
+  48, 57, 65, 90, 97, 122, 1040, 1103, 1186, 1187, 1198, 1199, 1256, 1257, 7296, 7302,
+};
+
+const narsil_language narsil_language_kyrgyz = {
+  .name = "kyrgyz",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 96,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 8,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/lao.c

@@ -0,0 +1,46 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[137] = {
+  3713, 3761, 3737, 3713, 3761, 3738, 3713, 3789, 3714, 3766, 3785, 3737, 3714, 3789, 3716, 3761, 3785, 3719, 3716,
+  3767, 3716, 3771, 3719, 3720, 3760, 3720, 3761, 3732, 3720, 3766, 3784, 3719, 3725, 3761, 3719, 3732, 3761, 3719,
+  3733, 3761, 3785, 3719, 3733, 3789, 3784, 3734, 3767, 3713, 3734, 3785, 3762, 3735, 3761, 3719, 3735, 3765, 3784,
+  3735, 3768, 3713, 3737, 3761, 3713, 3737, 3761, 3785, 3737, 3737, 3765, 3785, 3737, 3789, 3762, 3738, 3789, 3784,
+  3740, 3771, 3737, 3742, 3771, 3738, 3745, 3762, 3745, 3765, 3745, 3767, 3785, 3746, 3769, 3784, 3749, 3771, 3719,
+  3751, 3784, 3762, 3754, 3764, 3784, 3719, 3754, 3768, 3732, 3754, 3771, 3784, 3719, 3755, 3749, 3767, 3755, 3772,
+  3767, 3757, 3765, 3713, 3758, 3761, 3738, 3777, 3733, 3784, 3779, 3737, 3779, 3755, 3785, 3780, 3732, 3785, 3780,
+  3739, 3780, 3751, 3785,
+};
+
+static const uint32_t stop_word_offsets[47] = {
+  0, 3, 6, 8, 12, 14, 18, 20, 23, 25, 28, 32, 35, 38, 42, 45, 48, 51, 54, 57, 60, 63, 67, 70, 73, 76, 79, 82, 84, 86,
+  89, 92, 95, 98, 102, 105, 109, 112, 115, 118, 121, 124, 126, 129, 132, 134, 137,
+};
+
+static const uint32_t keep_ranges[12] = {
+  48, 57, 65, 90, 97, 122, 383, 383, 3713, 3807, 8490, 8490,
+};
+
+const narsil_language narsil_language_lao = {
+  .name = "lao",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 46,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 6,
+  .splits_code_points = 1,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 2,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/latin.c

@@ -0,0 +1,46 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+#include "../rules.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[159] = {
+  97, 97, 98, 97, 99, 97, 100, 97, 116, 97, 116, 113, 117, 101, 97, 117, 116, 97, 117, 116, 101, 109, 99, 117, 109, 100,
+  101, 100, 117, 109, 101, 101, 114, 97, 110, 116, 101, 114, 97, 116, 101, 115, 116, 101, 116, 101, 116, 105, 97, 109,
+  101, 120, 104, 97, 101, 99, 104, 105, 99, 104, 111, 99, 105, 110, 105, 116, 97, 109, 101, 110, 101, 99, 110, 101, 113,
+  117, 101, 110, 111, 110, 112, 101, 114, 113, 117, 97, 113, 117, 97, 101, 113, 117, 97, 109, 113, 117, 105, 113, 117,
+  105, 98, 117, 115, 113, 117, 105, 100, 101, 109, 113, 117, 111, 113, 117, 111, 100, 114, 101, 114, 101, 98, 117, 115,
+  114, 101, 109, 114, 101, 115, 115, 101, 100, 115, 105, 115, 105, 99, 115, 117, 110, 116, 116, 97, 109, 101, 110, 116,
+  97, 110, 100, 101, 109, 116, 101, 117, 116, 118, 101, 108,
+};
+
+static const uint32_t stop_word_offsets[50] = {
+  0, 1, 3, 5, 7, 9, 14, 17, 22, 25, 27, 30, 31, 36, 40, 43, 45, 50, 52, 56, 59, 62, 64, 67, 69, 72, 77, 80, 83, 86, 90,
+  94, 97, 103, 109, 112, 116, 118, 123, 126, 129, 132, 134, 137, 141, 146, 152, 154, 156, 159,
+};
+
+static const uint32_t keep_ranges[16] = {
+  48, 57, 65, 90, 97, 122, 256, 257, 274, 275, 298, 299, 332, 333, 362, 363,
+};
+
+const narsil_language narsil_language_latin = {
+  .name = "latin",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 49,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 8,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = narsil_latin_normalise,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};

package/cpp/core/src/languages/data/latvian.c

@@ -0,0 +1,75 @@
+#include "../../analysis/languages.h"
+#include "declarations.h"
+
+#include <stddef.h>
+#include <stdint.h>
+
+static const uint16_t stop_word_units[728] = {
+  97, 105, 122, 97, 112, 97, 112, 97, 107, 353, 97, 112, 97, 107, 353, 112, 117, 115, 97, 114, 97, 114, 299, 97, 117,
+  103, 353, 112, 117, 115, 98, 101, 116, 98, 101, 122, 98, 105, 106, 97, 98, 105, 106, 105, 98, 105, 106, 117, 98, 105,
+  106, 257, 109, 98, 105, 106, 257, 116, 98, 363, 115, 98, 363, 115, 105, 98, 363, 115, 105, 101, 116, 98, 363, 115,
+  105, 109, 98, 363, 116, 98, 363, 353, 117, 99, 97, 117, 114, 100, 105, 101, 109, 382, 275, 108, 100, 105, 101, 122,
+  105, 110, 100, 114, 111, 353, 105, 100, 275, 316, 101, 115, 97, 109, 101, 115, 97, 116, 101, 115, 105, 101, 115, 109,
+  117, 103, 97, 110, 103, 97, 114, 105, 101, 107, 97, 109, 105, 101, 107, 97, 109, 115, 105, 101, 107, 257, 109, 105,
+  101, 107, 257, 109, 115, 105, 101, 107, 353, 105, 101, 107, 353, 112, 117, 115, 105, 107, 105, 114, 105, 116, 105,
+  116, 105, 110, 105, 122, 106, 97, 106, 97, 117, 106, 101, 98, 106, 101, 98, 353, 117, 106, 101, 108, 106, 111, 106,
+  257, 107, 97, 107, 97, 109, 275, 114, 107, 97, 117, 116, 107, 111, 108, 299, 100, 122, 107, 111, 112, 353, 107, 257,
+  107, 316, 117, 118, 97, 107, 316, 117, 118, 105, 107, 316, 117, 118, 117, 107, 316, 117, 118, 257, 109, 107, 316, 117,
+  118, 257, 116, 107, 316, 363, 115, 107, 316, 363, 115, 105, 107, 316, 363, 115, 105, 101, 116, 107, 316, 363, 115,
+  105, 109, 107, 316, 363, 115, 116, 107, 316, 363, 115, 116, 97, 109, 107, 316, 363, 115, 116, 97, 116, 107, 316, 363,
+  115, 116, 105, 107, 316, 363, 115, 116, 117, 107, 316, 363, 116, 107, 316, 363, 353, 117, 108, 97, 98, 97, 100, 108,
+  97, 105, 108, 101, 106, 112, 117, 115, 108, 299, 100, 122, 108, 299, 100, 122, 107, 111, 110, 101, 110, 101, 98, 363,
+  116, 110, 101, 100, 122, 110, 101, 107, 257, 110, 101, 118, 105, 115, 110, 101, 122, 105, 110, 110, 111, 110, 117,
+  110, 275, 111, 116, 114, 112, 117, 115, 112, 97, 112, 97, 114, 112, 97, 116, 112, 105, 101, 112, 105, 114, 109, 115,
+  112, 114, 101, 116, 112, 114, 105, 101, 107, 353, 112, 257, 114, 112, 275, 99, 115, 116, 97, 114, 112, 116, 97, 100,
+  116, 97, 107, 116, 97, 112, 105, 116, 97, 112, 115, 116, 97, 112, 115, 105, 116, 97, 112, 115, 105, 101, 116, 116, 97,
+  112, 115, 105, 109, 116, 97, 112, 116, 116, 97, 112, 257, 116, 116, 97, 112, 353, 117, 116, 97, 269, 117, 116, 101,
+  116, 105, 101, 99, 116, 105, 101, 107, 116, 105, 101, 107, 97, 109, 116, 105, 101, 107, 97, 116, 116, 105, 101, 107,
+  117, 116, 105, 107, 116, 105, 107, 97, 116, 105, 107, 97, 105, 116, 105, 107, 105, 116, 105, 107, 107, 111, 116, 105,
+  107, 108, 97, 98, 116, 105, 107, 108, 299, 100, 122, 116, 105, 107, 115, 116, 105, 107, 115, 105, 101, 116, 116, 105,
+  107, 115, 105, 109, 116, 105, 107, 116, 116, 105, 107, 117, 116, 105, 107, 118, 105, 101, 110, 116, 105, 107, 257,
+  109, 116, 105, 107, 257, 116, 116, 105, 107, 353, 117, 116, 111, 109, 275, 114, 116, 111, 112, 97, 116, 116, 117, 114,
+  112, 114, 101, 116, 105, 109, 116, 117, 114, 112, 114, 101, 116, 299, 116, 257, 116, 257, 100, 275, 316, 116, 257,
+  108, 97, 98, 116, 257, 112, 275, 99, 117, 110, 117, 122, 118, 97, 105, 118, 97, 114, 118, 97, 114, 97, 116, 118, 97,
+  114, 275, 106, 97, 118, 97, 114, 275, 106, 105, 118, 97, 114, 275, 106, 117, 118, 97, 114, 275, 106, 257, 109, 118,
+  97, 114, 275, 106, 257, 116, 118, 97, 114, 275, 115, 118, 97, 114, 275, 115, 105, 118, 97, 114, 275, 115, 105, 101,
+  116, 118, 97, 114, 275, 115, 105, 109, 118, 97, 114, 275, 116, 118, 97, 114, 275, 353, 117, 118, 105, 101, 110, 118,
+  105, 114, 115, 118, 105, 114, 115, 112, 117, 115, 118, 105, 115, 118, 105, 326, 112, 117, 115, 122, 101, 109, 257,
+  114, 112, 117, 115, 353, 97, 105, 112, 117, 115,
+};
+
+static const uint32_t stop_word_offsets[162] = {
+  0, 3, 5, 10, 18, 20, 23, 30, 33, 36, 40, 44, 48, 53, 58, 61, 65, 71, 76, 79, 83, 87, 94, 100, 105, 108, 112, 116, 119,
+  123, 126, 129, 134, 140, 145, 151, 155, 162, 164, 166, 168, 172, 174, 176, 179, 182, 187, 190, 192, 194, 196, 201,
+  205, 211, 215, 217, 222, 227, 232, 238, 244, 248, 253, 260, 266, 271, 278, 285, 291, 297, 301, 306, 311, 314, 320,
+  324, 330, 332, 337, 341, 345, 350, 355, 357, 359, 361, 367, 369, 372, 375, 378, 383, 387, 393, 396, 399, 404, 407,
+  410, 414, 418, 423, 430, 436, 440, 445, 450, 454, 456, 460, 464, 470, 476, 481, 484, 488, 493, 497, 502, 508, 515,
+  519, 526, 532, 536, 540, 547, 552, 557, 562, 567, 572, 581, 589, 591, 596, 601, 606, 608, 610, 613, 616, 621, 627,
+  633, 639, 646, 653, 658, 664, 672, 679, 684, 690, 694, 698, 705, 708, 714, 717, 722, 728,
+};
+
+static const uint32_t keep_ranges[28] = {
+  48, 57, 65, 90, 97, 122, 256, 257, 268, 269, 274, 275, 290, 291, 298, 299, 310, 311, 315, 316, 325, 326, 352, 353,
+  362, 363, 381, 382,
+};
+
+const narsil_language narsil_language_latvian = {
+  .name = "latvian",
+  .revision = "1",
+  .stop_word_units = stop_word_units,
+  .stop_word_offsets = stop_word_offsets,
+  .stop_word_count = 161,
+  .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
+  .keep_ranges = keep_ranges,
+  .keep_range_count = 14,
+  .splits_code_points = 0,
+  .strips_possessive = 0,
+  .normalises_diacritics = 0,
+  .min_token_length = 1,
+  .ngram_size = 0,
+  .normalise = NULL,
+  .stem_by_rules = NULL,
+  .create_snowball = NULL,
+  .snowball_stem = NULL,
+  .close_snowball = NULL,
+};