language_detection 0.0.1

Sign up to get free protection for your applications and to get access to all the features.
Files changed (100) hide show
  1. data/.gitignore +19 -0
  2. data/Gemfile +4 -0
  3. data/LICENSE.txt +22 -0
  4. data/README.md +85 -0
  5. data/Rakefile +11 -0
  6. data/ext/cld/Makefile +34 -0
  7. data/ext/cld/base/basictypes.h +348 -0
  8. data/ext/cld/base/build_config.h +124 -0
  9. data/ext/cld/base/casts.h +156 -0
  10. data/ext/cld/base/commandlineflags.h +443 -0
  11. data/ext/cld/base/crash.h +41 -0
  12. data/ext/cld/base/dynamic_annotations.h +358 -0
  13. data/ext/cld/base/global_strip_options.h +59 -0
  14. data/ext/cld/base/log_severity.h +46 -0
  15. data/ext/cld/base/logging.h +1403 -0
  16. data/ext/cld/base/macros.h +243 -0
  17. data/ext/cld/base/port.h +54 -0
  18. data/ext/cld/base/scoped_ptr.h +428 -0
  19. data/ext/cld/base/stl_decl.h +0 -0
  20. data/ext/cld/base/stl_decl_msvc.h +107 -0
  21. data/ext/cld/base/string_util.h +29 -0
  22. data/ext/cld/base/strtoint.h +93 -0
  23. data/ext/cld/base/template_util.h +96 -0
  24. data/ext/cld/base/type_traits.h +198 -0
  25. data/ext/cld/base/vlog_is_on.h +143 -0
  26. data/ext/cld/cld.so +0 -0
  27. data/ext/cld/encodings/compact_lang_det/cldutil.cc +905 -0
  28. data/ext/cld/encodings/compact_lang_det/cldutil.h +1205 -0
  29. data/ext/cld/encodings/compact_lang_det/cldutil_dbg.h +76 -0
  30. data/ext/cld/encodings/compact_lang_det/cldutil_dbg_empty.cc +76 -0
  31. data/ext/cld/encodings/compact_lang_det/compact_lang_det.cc +62 -0
  32. data/ext/cld/encodings/compact_lang_det/compact_lang_det.h +145 -0
  33. data/ext/cld/encodings/compact_lang_det/compact_lang_det_impl.cc +2574 -0
  34. data/ext/cld/encodings/compact_lang_det/compact_lang_det_impl.h +173 -0
  35. data/ext/cld/encodings/compact_lang_det/compact_lang_det_unittest_small.cc +406 -0
  36. data/ext/cld/encodings/compact_lang_det/compile.cmd +1 -0
  37. data/ext/cld/encodings/compact_lang_det/ext_lang_enc.cc +545 -0
  38. data/ext/cld/encodings/compact_lang_det/ext_lang_enc.h +119 -0
  39. data/ext/cld/encodings/compact_lang_det/generated/cld_generated_score_deltaoctachrome_0406.cc +380 -0
  40. data/ext/cld/encodings/compact_lang_det/generated/cld_generated_score_quadchrome_0406.cc +382 -0
  41. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_cjkbis_0.cc +49 -0
  42. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_ctjkvz.cc +7119 -0
  43. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_ctjkvz_0.cc +61 -0
  44. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_deltaoctachrome.cc +1263 -0
  45. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_longwords8_0.cc +53 -0
  46. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_meanscore.h +10 -0
  47. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_quads_0.cc +50 -0
  48. data/ext/cld/encodings/compact_lang_det/generated/compact_lang_det_generated_quadschrome.cc +70935 -0
  49. data/ext/cld/encodings/compact_lang_det/getonescriptspan.cc +570 -0
  50. data/ext/cld/encodings/compact_lang_det/getonescriptspan.h +131 -0
  51. data/ext/cld/encodings/compact_lang_det/letterscript_enum.cc +117 -0
  52. data/ext/cld/encodings/compact_lang_det/letterscript_enum.h +99 -0
  53. data/ext/cld/encodings/compact_lang_det/subsetsequence.cc +259 -0
  54. data/ext/cld/encodings/compact_lang_det/subsetsequence.h +44 -0
  55. data/ext/cld/encodings/compact_lang_det/subsetsequence_unittest.cc +99 -0
  56. data/ext/cld/encodings/compact_lang_det/tote.cc +299 -0
  57. data/ext/cld/encodings/compact_lang_det/tote.h +89 -0
  58. data/ext/cld/encodings/compact_lang_det/unittest_data.h +193 -0
  59. data/ext/cld/encodings/compact_lang_det/utf8propjustletter.h +1162 -0
  60. data/ext/cld/encodings/compact_lang_det/utf8propletterscriptnum.h +1222 -0
  61. data/ext/cld/encodings/compact_lang_det/utf8scannotjustletterspecial.h +1185 -0
  62. data/ext/cld/encodings/compact_lang_det/win/cld_basictypes.h +10 -0
  63. data/ext/cld/encodings/compact_lang_det/win/cld_commandlineflags.h +28 -0
  64. data/ext/cld/encodings/compact_lang_det/win/cld_google.h +18 -0
  65. data/ext/cld/encodings/compact_lang_det/win/cld_htmlutils.h +13 -0
  66. data/ext/cld/encodings/compact_lang_det/win/cld_htmlutils_google3.cc +32 -0
  67. data/ext/cld/encodings/compact_lang_det/win/cld_htmlutils_windows.cc +29 -0
  68. data/ext/cld/encodings/compact_lang_det/win/cld_logging.h +21 -0
  69. data/ext/cld/encodings/compact_lang_det/win/cld_macros.h +19 -0
  70. data/ext/cld/encodings/compact_lang_det/win/cld_strtoint.h +26 -0
  71. data/ext/cld/encodings/compact_lang_det/win/cld_unicodetext.cc +84 -0
  72. data/ext/cld/encodings/compact_lang_det/win/cld_unicodetext.h +40 -0
  73. data/ext/cld/encodings/compact_lang_det/win/cld_unilib.h +15 -0
  74. data/ext/cld/encodings/compact_lang_det/win/cld_unilib_google3.cc +18 -0
  75. data/ext/cld/encodings/compact_lang_det/win/cld_unilib_windows.cc +29 -0
  76. data/ext/cld/encodings/compact_lang_det/win/cld_utf.h +24 -0
  77. data/ext/cld/encodings/compact_lang_det/win/cld_utf8statetable.cc +224 -0
  78. data/ext/cld/encodings/compact_lang_det/win/cld_utf8statetable.h +141 -0
  79. data/ext/cld/encodings/compact_lang_det/win/cld_utf8utils.h +22 -0
  80. data/ext/cld/encodings/compact_lang_det/win/cld_utf8utils_google3.cc +18 -0
  81. data/ext/cld/encodings/compact_lang_det/win/cld_utf8utils_windows.cc +17 -0
  82. data/ext/cld/encodings/compact_lang_det/win/normalizedunicodetext.cc +172 -0
  83. data/ext/cld/encodings/compact_lang_det/win/normalizedunicodetext.h +67 -0
  84. data/ext/cld/encodings/internal/encodings.cc +12 -0
  85. data/ext/cld/encodings/lang_enc.h +254 -0
  86. data/ext/cld/encodings/proto/encodings.pb.h +169 -0
  87. data/ext/cld/encodings/public/encodings.h +301 -0
  88. data/ext/cld/extconf.rb +1 -0
  89. data/ext/cld/language_detection.cc +88 -0
  90. data/ext/cld/languages/internal/languages.cc +337 -0
  91. data/ext/cld/languages/proto/languages.pb.h +179 -0
  92. data/ext/cld/languages/public/languages.h +379 -0
  93. data/language_detection.gemspec +28 -0
  94. data/lib/language_detection/string.rb +1 -0
  95. data/lib/language_detection/version.rb +3 -0
  96. data/lib/language_detection.rb +54 -0
  97. data/test/_helper.rb +15 -0
  98. data/test/fixtures/languages.csv +80 -0
  99. data/test/language_detection_test.rb +88 -0
  100. metadata +250 -0
@@ -0,0 +1,299 @@
1
+ // Copyright (c) 2006-2009 The Chromium Authors. All rights reserved.
2
+ // Use of this source code is governed by a BSD-style license that can be
3
+ // found in the LICENSE file.
4
+
5
+ #include "encodings/compact_lang_det/tote.h"
6
+ #include <string.h> // memset
7
+
8
+ #include "encodings/compact_lang_det/win/cld_logging.h"
9
+
10
+
11
+ // Take a set of <key, value> pairs and tote them up.
12
+ // After explicitly sorting, retrieve top key, value pairs
13
+ Tote::Tote() {
14
+ gram_count_ = 0;
15
+ incr_count_ = 0;
16
+ byte_count_ = 0;
17
+ memset(key_, 0, sizeof(key_));
18
+ // No need to initialize values
19
+ }
20
+
21
+ Tote::~Tote() {
22
+ }
23
+
24
+ void Tote::Reinit() {
25
+ gram_count_ = 0;
26
+ incr_count_ = 0;
27
+ byte_count_ = 0;
28
+ memset(key_, 0, sizeof(key_));
29
+ // No need to initialize values
30
+ }
31
+
32
+ // Increment count of quadgrams/trigrams/unigrams scored
33
+ void Tote::AddGram() {
34
+ ++gram_count_;
35
+ }
36
+
37
+ // Three-way associative, guaranteeing that the largest two counts are always
38
+ // in the data structure. kMaxSize must be a multiple of 3, and is tied to the
39
+ // subscript calculations here, which are for 8 sets of 3-way associative
40
+ // buckets. The subscripts for set N are [N], [N+8], and [N+16] used in a
41
+ // slightly-weird way: The initial probe point is [N] or [N+8], whichever
42
+ // is specified by key mod 16. In most cases (nearly *all* cases except Latin
43
+ // script), this entry matches and we update/return. The second probe is
44
+ // the other of [N] and [N+8]. The third probe is only used as a fallback to
45
+ // these two, and is there only for the rare case that there are three or more
46
+ // languages with Language enum values equal mod 8, contending within the same
47
+ // bucket. This can only happen in Latin and (rarely) Cyrillic scripts, because
48
+ // the other scripts have fewer than 17 languages total.
49
+ // If you change kMaxSize, change the constants 7/8/15/16 below
50
+ void Tote::Add(uint8 ikey, int idelta) {
51
+ DCHECK(ikey != 0);
52
+ ++incr_count_;
53
+
54
+ // Look for existing entry
55
+ int sub0 = ikey & 15;
56
+ if (key_[sub0] == ikey) {
57
+ value_[sub0] += idelta;
58
+ return;
59
+ }
60
+ int sub1 = sub0 ^ 8;
61
+ if (key_[sub1] == ikey) {
62
+ value_[sub1] += idelta;
63
+ return;
64
+ }
65
+ int sub2 = (ikey & 7) + 16;
66
+ if (key_[sub2] == ikey) {
67
+ value_[sub2] += idelta;
68
+ return;
69
+ }
70
+
71
+ // Allocate new entry
72
+ int alloc = -1;
73
+ if (key_[sub0] == 0) {
74
+ alloc = sub0;
75
+ } else if (key_[sub1] == 0) {
76
+ alloc = sub1;
77
+ } else if (key_[sub2] == 0) {
78
+ alloc = sub2;
79
+ } else {
80
+ // All choices allocated, need to replace smallest one
81
+ alloc = sub0;
82
+ if (value_[sub1] < value_[alloc]) {alloc = sub1;}
83
+ if (value_[sub2] < value_[alloc]) {alloc = sub2;}
84
+ }
85
+ key_[alloc] = ikey;
86
+ value_[alloc] = idelta;
87
+ return;
88
+ }
89
+
90
+ // Return current top key
91
+ int Tote::CurrentTopKey() {
92
+ int top_key = 0;
93
+ int top_value = -1;
94
+ for (int sub = 0; sub < kMaxSize_; ++sub) {
95
+ if (key_[sub] == 0) {continue;}
96
+ if (top_value < value_[sub]) {
97
+ top_value = value_[sub];
98
+ top_key = key_[sub];
99
+ }
100
+ }
101
+ return top_key;
102
+ }
103
+
104
+
105
+ // Sort first n entries by decreasing order of value
106
+ // If key==0 other fields are not valid, treat value as -1
107
+ void Tote::Sort(int n) {
108
+ // This is n**2, but n is small
109
+ for (int sub = 0; sub < n; ++sub) {
110
+ if (key_[sub] == 0) {value_[sub] = -1;}
111
+
112
+ // Bubble sort key[sub] and entry[sub]
113
+ for (int sub2 = sub + 1; sub2 < kMaxSize_; ++sub2) {
114
+ if (key_[sub2] == 0) {value_[sub2] = -1;}
115
+ if (value_[sub] < value_[sub2]) {
116
+ // swap
117
+ uint8 tmpk = key_[sub];
118
+ key_[sub] = key_[sub2];
119
+ key_[sub2] = tmpk;
120
+ int tmpv = value_[sub];
121
+ value_[sub] = value_[sub2];
122
+ value_[sub2] = tmpv;
123
+ }
124
+ }
125
+ }
126
+ }
127
+
128
+ void Tote::Dump(FILE* f) {
129
+ for (int sub = 0; sub < kMaxSize_; ++sub) {
130
+ if (key_[sub] > 0) {
131
+ fprintf(f, "[%2d] %3d %8d\n", sub, key_[sub], value_[sub]);
132
+ }
133
+ }
134
+ fprintf(f, "%d %d %d\n", gram_count_, incr_count_, byte_count_);
135
+ }
136
+
137
+
138
+
139
+
140
+ // Take a set of <key, value> pairs and tote them up.
141
+ // After explicitly sorting, retrieve top key, value pairs
142
+ ToteWithReliability::ToteWithReliability() {
143
+ // No need to initialize score_ or value_
144
+ incr_count_ = 0;
145
+ sorted_ = 0;
146
+ memset(closepair_, 0, sizeof(closepair_));
147
+ memset(key_, 0, sizeof(key_));
148
+ }
149
+
150
+ ToteWithReliability::~ToteWithReliability() {
151
+ }
152
+
153
+ void ToteWithReliability::Reinit() {
154
+ // No need to initialize score_ or value_
155
+ incr_count_ = 0;
156
+ sorted_ = 0;
157
+ memset(closepair_, 0, sizeof(closepair_));
158
+ memset(key_, 0, sizeof(key_));
159
+ ////ss_.Init();
160
+ }
161
+
162
+ // Weight reliability by ibytes
163
+ // Also see three-way associative comments above for Tote
164
+ void ToteWithReliability::Add(uint8 ikey, int ibytes,
165
+ int score, int ireliability) {
166
+ DCHECK(ikey != 0);
167
+ CHECK(sorted_ == 0);
168
+ ++incr_count_;
169
+
170
+ // Look for existing entry
171
+ int sub0 = ikey & 15;
172
+ if (key_[sub0] == ikey) {
173
+ value_[sub0] += ibytes;
174
+ score_[sub0] += score;
175
+ reliability_[sub0] += ireliability * ibytes;
176
+ return;
177
+ }
178
+ int sub1 = sub0 ^ 8;
179
+ if (key_[sub1] == ikey) {
180
+ value_[sub1] += ibytes;
181
+ score_[sub1] += score;
182
+ reliability_[sub1] += ireliability * ibytes;
183
+ return;
184
+ }
185
+ int sub2 = (ikey & 7) + 16;
186
+ if (key_[sub2] == ikey) {
187
+ value_[sub2] += ibytes;
188
+ score_[sub2] += score;
189
+ reliability_[sub2] += ireliability * ibytes;
190
+ return;
191
+ }
192
+
193
+ // Allocate new entry
194
+ int alloc = -1;
195
+ if (key_[sub0] == 0) {
196
+ alloc = sub0;
197
+ } else if (key_[sub1] == 0) {
198
+ alloc = sub1;
199
+ } else if (key_[sub2] == 0) {
200
+ alloc = sub2;
201
+ } else {
202
+ // All choices allocated, need to replace smallest one
203
+ alloc = sub0;
204
+ if (value_[sub1] < value_[alloc]) {alloc = sub1;}
205
+ if (value_[sub2] < value_[alloc]) {alloc = sub2;}
206
+ }
207
+ key_[alloc] = ikey;
208
+ value_[alloc] = ibytes;
209
+ score_[alloc] = score;
210
+ reliability_[alloc] = ireliability * ibytes;
211
+ return;
212
+ }
213
+
214
+ // Find subscript of a given packed language, or -1
215
+ int ToteWithReliability::Find(uint8 ikey) {
216
+ DCHECK(ikey != 0);
217
+
218
+ if (sorted_) {
219
+ // Linear search if sorted
220
+ for (int sub = 0; sub < kMaxSize_; ++sub) {
221
+ if (key_[sub] == ikey) {return sub;}
222
+ }
223
+ return -1;
224
+ }
225
+
226
+ // Look for existing entry
227
+ int sub0 = ikey & 15;
228
+ if (key_[sub0] == ikey) {
229
+ return sub0;
230
+ }
231
+ int sub1 = sub0 ^ 8;
232
+ if (key_[sub1] == ikey) {
233
+ return sub1;
234
+ }
235
+ int sub2 = (ikey & 7) + 16;
236
+ if (key_[sub2] == ikey) {
237
+ return sub2;
238
+ }
239
+
240
+ return -1;
241
+ }
242
+
243
+ // Return current top key
244
+ int ToteWithReliability::CurrentTopKey() {
245
+ int top_key = 0;
246
+ int top_value = -1;
247
+ for (int sub = 0; sub < kMaxSize_; ++sub) {
248
+ if (key_[sub] == 0) {continue;}
249
+ if (top_value < value_[sub]) {
250
+ top_value = value_[sub];
251
+ top_key = key_[sub];
252
+ }
253
+ }
254
+ return top_key;
255
+ }
256
+
257
+
258
+ // Sort first n entries by decreasing order of value
259
+ // If key==0 other fields are not valid, treat value as -1
260
+ void ToteWithReliability::Sort(int n) {
261
+ // This is n**2, but n is small
262
+ for (int sub = 0; sub < n; ++sub) {
263
+ if (key_[sub] == 0) {value_[sub] = -1;}
264
+
265
+ // Bubble sort key[sub] and entry[sub]
266
+ for (int sub2 = sub + 1; sub2 < kMaxSize_; ++sub2) {
267
+ if (key_[sub2] == 0) {value_[sub2] = -1;}
268
+ if (value_[sub] < value_[sub2]) {
269
+ // swap
270
+ uint8 tmpk = key_[sub];
271
+ key_[sub] = key_[sub2];
272
+ key_[sub2] = tmpk;
273
+
274
+ int tmpv = value_[sub];
275
+ value_[sub] = value_[sub2];
276
+ value_[sub2] = tmpv;
277
+
278
+ double tmps = score_[sub];
279
+ score_[sub] = score_[sub2];
280
+ score_[sub2] = tmps;
281
+
282
+ int tmpr = reliability_[sub];
283
+ reliability_[sub] = reliability_[sub2];
284
+ reliability_[sub2] = tmpr;
285
+ }
286
+ }
287
+ }
288
+ sorted_ = 1;
289
+ }
290
+
291
+ void ToteWithReliability::Dump(FILE* f) {
292
+ for (int sub = 0; sub < kMaxSize_; ++sub) {
293
+ if (key_[sub] > 0) {
294
+ fprintf(f, "[%2d] %3d %6d %5d %4d\n",
295
+ sub, key_[sub], value_[sub], score_[sub], reliability_[sub]);
296
+ }
297
+ }
298
+ fprintf(f, " %d#\n", incr_count_);
299
+ }
@@ -0,0 +1,89 @@
1
+ // Copyright (c) 2006-2009 The Chromium Authors. All rights reserved.
2
+ // Use of this source code is governed by a BSD-style license that can be
3
+ // found in the LICENSE file.
4
+
5
+ #ifndef ENCODINGS_COMPACT_LANG_DET_TOTE_H_
6
+ #define ENCODINGS_COMPACT_LANG_DET_TOTE_H_
7
+
8
+ #include <stdio.h>
9
+ #include "encodings/compact_lang_det/win/cld_basictypes.h"
10
+
11
+ // Take a set of <key, value> pairs and tote them up.
12
+ // After explicitly sorting, retrieve top key, value pairs
13
+ class Tote {
14
+ public:
15
+ Tote();
16
+ ~Tote();
17
+ void Reinit();
18
+ void AddGram();
19
+ void Add(uint8 ikey, int idelta);
20
+ void AddBytes(int ibytes) {byte_count_ += ibytes;}
21
+ int CurrentTopKey();
22
+ void Sort(int n);
23
+ void Dump(FILE* f);
24
+ uint16 GetGramCount() const {return gram_count_;}
25
+ uint16 GetIncrCount() const {return incr_count_;}
26
+ int GetByteCount() const {return byte_count_;}
27
+ int MaxSize() const {return kMaxSize_;}
28
+ uint8 Key(int i) const {return key_[i];}
29
+ int Value(int i) const {return value_[i];}
30
+ void SetGramCount(uint16 v) {gram_count_ = v;}
31
+ void SetIncrCount(uint16 v) {incr_count_ = v;}
32
+ void SetKey(int i, int v) {key_[i] = v;}
33
+ void SetValue(int i, int v) {value_[i] = v;}
34
+
35
+ private:
36
+ static const int kMaxSize_ = 24;
37
+ uint16 gram_count_; // Number of quadgrams/etc. scored
38
+ uint16 incr_count_; // Number of Add calls (1-3 per gram)
39
+ int byte_count_; // Bytes of text scored
40
+ // Align at multiple of 8 bytes
41
+ uint8 key_[kMaxSize_]; // Lang unassigned = 0, valid = 1..255
42
+ int value_[kMaxSize_]; // Probability score sum
43
+ };
44
+
45
+
46
+ // Take a set of <key, value, reliability> triples and tote them up.
47
+ // After explicitly sorting, retrieve top key, value, reliability triples
48
+ class ToteWithReliability {
49
+ public:
50
+ ToteWithReliability();
51
+ ~ToteWithReliability();
52
+ void Reinit();
53
+ void Add(uint8 ikey, int ibytes, int score, int ireliability);
54
+ int Find(uint8 ikey);
55
+ void AddClosePair(int subscr, int val) {closepair_[subscr] += val;}
56
+ int CurrentTopKey();
57
+ void Sort(int n);
58
+ void Dump(FILE* f);
59
+
60
+ ////void AddSeq(uint8 ikey) {ss_.Add(ikey);}
61
+ ////void ExtractSeq(int n, uint8* dst) {ss_.Extract(n, dst);}
62
+
63
+ int GetIncrCount() const {return incr_count_;}
64
+ int GetClosePair(int subscr) const {return closepair_[subscr];}
65
+ int MaxSize() const {return kMaxSize_;}
66
+ uint8 Key(int i) const {return key_[i];}
67
+ int Value(int i) const {return value_[i];}
68
+ int Score(int i) const {return score_[i];}
69
+ int Reliability(int i) const {return reliability_[i];}
70
+ void SetKey(int i, int v) {key_[i] = v;}
71
+ void SetValue(int i, int v) {value_[i] = v;}
72
+ void SetScore(int i, int v) {score_[i] = v;}
73
+ void SetReliability(int i, int v) {reliability_[i] = v;}
74
+
75
+ private:
76
+ static const int kMaxSize_ = 24;
77
+ static const int kMaxClosePairSize_ = 8;
78
+ int incr_count_; // Number of Add calls
79
+ int sorted_; // Contents have been sorted, cannot Add
80
+ // Align at multiple of 8 bytes
81
+ int closepair_[kMaxClosePairSize_];
82
+ uint8 key_[kMaxSize_]; // Lang unassigned = 0, valid = 1..255
83
+ int value_[kMaxSize_]; // Bytecount this lang
84
+ int score_[kMaxSize_]; // Probability score sum
85
+ int reliability_[kMaxSize_]; // Percentage 0..100
86
+ ////SubsetSequence ss_;
87
+ };
88
+
89
+ #endif // ENCODINGS_COMPACT_LANG_DET_TOTE_H_
@@ -0,0 +1,193 @@
1
+ // Copyright (c) 2006-2009 The Chromium Authors. All rights reserved.
2
+ // Use of this source code is governed by a BSD-style license that can be
3
+ // found in the LICENSE file.
4
+
5
+ #ifndef ENCODINGS_COMPACT_LANG_DET_UNITTEST_DATA_H_
6
+ #define ENCODINGS_COMPACT_LANG_DET_UNITTEST_DATA_H_
7
+
8
+ const char* kTeststr_aa_Latn = " nagay tanito nagay tanto nagayna naharsi nahrur nake nala nammay nammay haytu nanu narig ne ni num numu o obare obe obe obisse oggole ogli olloyta ongorowe orbise othoga r rabe rade ra e rage rakub rasitte rasu reyta rog ruddi ruga s sa al bada sa ala";
9
+ const char* kTeststr_ab_Cyrl = " а зуа абзиара дақәшәоит ан лыбзиабара ахә амаӡам ауаҩы игәы иҭоу ихы иҿы ианубаалоит аҧҳәыс ҧшӡа ахацәа лышьҭоуп аҿаасҭа лара дрышьҭоуп";
10
+ const char* kTeststr_af_Latn = " aam skukuza die naam beteken hy wat skoonvee of hy wat alles onderstebo keer wysig bosveldkampe boskampe is kleiner afgeleë ruskampe wat oor min fasiliteite beskik daar is geen restaurante of winkels nie en slegs oornagbesoekers word toegelaat bateleur";
11
+ const char* kTeststr_am_Ethi = " ለመጠይቅ ወደ እስክንድርያ ላኩዋቸውና የእስክንድርያ ጳጳስ አቴናስዮስ ፍሬምንጦስን እራሳቸውን ሾመው ልከዋል ከዚያ እስከ ዓ ም ድረስ የኢትዮጵያ አቡነ";
12
+ const char* kTeststr_ar_Arab = "احتيالية بيع أي حساب";
13
+ const char* kTeststr_as_Beng = "অঞ্চল নতুন সদস্যবৃন্দ সকলোৱে ভৰ্তি হব পাৰে মুল পৃষ্ঠা জন লেখক গুগ ল দল সাৰাংশ ই পত্ৰ টা বাৰ্তা এজন";
14
+ const char* kTeststr_ay_Latn = " aru wijar aru ispañula ukaran aru witanam aru kurti aru kalis aru warani aru malta aru yatiyawi niya jakitanaka isluwiñ aru lmir phuran aru masirunan aru purtukal aru kruwat aru jakira urtu aru inklisa pirsan aru suyku aru malay aru jisk aptayma thaya";
15
+ const char* kTeststr_az_Arab = " آذربایجان دا انسان حاقلاری ائوی آچیلاجاق ب م ت ائلچيسي برمه موخاليفتي نين ليدئري ايله گؤروشه بيليب ترس شوونيسم فارس از آزادي ملتهاي تورکمن";
16
+ const char* kTeststr_az_Latn = " a az qalıb breyn rinq intellektual oyunu üzrə yarışın zona mərhələləri keçirilib miq un qalıqlarının dənizdən çıxarılması davam edir məhəmməd peyğəmbərin karikaturalarını çap edən qəzetin baş redaktoru iş otağında ölüb";
17
+ const char* kTeststr_ba_Cyrl = " арналђан бындай ђилми эш тіркињлњ тњјге тапєыр нњшер ителњ ғинуар бєхет именлектє етешлектє ауыл ўќмерџєре хеџмєт юлын ћайлаѓанда";
18
+ const char* kTeststr_be_Cyrl = " а друкаваць іх не было тэхнічна магчыма бліжэй за вільню тым самым часам нямецкае кіраўніцтва прапаноўвала апроч ўвядзення лацінкі яе";
19
+ const char* kTeststr_bg_Cyrl = " а дума попада в състояние на изпитание ключовите думи с предсказана малко под то изискване на страниците за търсене в";
20
+ const char* kTeststr_bh_Deva = " ं निमरद के घोड़ा पर मत चढ़ीं मरद मुये नाम के निमरद मुये पेट् के मरना है काशी तो खुशामद किसका मल बटुली";
21
+ const char* kTeststr_bi_Latn = " king wantaem nomo hem i sakem setan mo ol rabis enjel blong hem oli aot long heven oli kamdaon long wol taswe ol samting oli kam nogud olgeta long wol ya stat long revelesen ol faet kakae i sot ol sik mo fasin blong brekem loa oli kam antap olgeta samting";
22
+ const char* kTeststr_bn_Beng = " ংখ্যা নমুনায়ন বিন্যাস পরিসংখ্যানিক মডেল পরিসংখ্যানিক সিদ্ধান্ত ফাংশন পরিসংখ্যানিক";
23
+ const char* kTeststr_bo_Tibt = " གང ནི ཀུན ལ སྦྱར པ དང ཅན ལྡན བདག པོའི སྒྲ ག ད བ ས ན མ པ ང འ ར ལ མཐའ མེད པ བདག པོའི སྒྲ ལ པ ཉིད དོ མ མི མིན";
24
+ const char* kTeststr_br_Latn = " a chom met leuskel a ra e blas da jack irons dilabour hag aet kuit eus what is this dibab a reont da c houde michael beinhorn evit produiñ an trede pladenn kavet e vez ar ganaouennoù buhan ha buhan ganto setu stummet ar bladenn adkavet e vez enni funk";
25
+ const char* kTeststr_bs_Cyrl = "историја босне књ историја босне књ историја босне књ историја босне књ ";
26
+ const char* kTeststr_bs_Latn = " a radi bržeg rada pošto rom radi sporije nego ram izvorni rom se isključuje a dio ram a se rezerviše te se u njega ne ploča procesor ram memorija grafička kartica zvučna kartica modem mrežna kartica napojna jedinica uređaji za pohranjivanje";
27
+ const char* kTeststr_ca_Latn = "al final en un únic lloc nhorabona l correu electrònic està concebut com a eina de productivitat aleshores per què perdre el temps arxivant missatges per després intentar recordar on els veu desar i per què heu d eliminar missatges importants per l";
28
+ const char* kTeststr_chr_Cher = "ᎠᎢᏍᎩ ᎠᏟᎶᏍᏗ ᏥᏄᏍᏛᎩ ᎦᎫᏍᏛᏅᎯ ᎾᎥᎢ";
29
+ const char* kTeststr_co_Latn = " a prupusitu di risultati for utilizà a scatula per ricercà ind issi risultati servore errore u servore ha incuntratu una errore pruvisoria é ùn ha pussutu compie a vostra dumanda per piacè acimenta dinò ind una minuta tuttu listessu ligami truvà i";
30
+ const char* kTeststr_cs_Latn = " a akci opakujte film uložen vykreslit gmail tokio smazat obsah adresáře nelze načíst systémový profil jednotky smoot okud používáte pro určení polokoule značky z západ nebo v východ používejte nezáporné hodnoty zeměpisné délky nelze";
31
+ const char* kTeststr_cy_Latn = " a chofrestru eich cyfrif ymwelwch a unwaith i chi greu eich cyfrif mi fydd yn cael ei hysbysu o ch cyfeiriad ebost newydd fel eich bod yn gallu cadw mewn cysylltiad drwy gmail os nad ydych chi wedi clywed yn barod am gmail mae n gwasanaeth gwebost";
32
+ const char* kTeststr_da_Latn = " a z tallene og punktummer der er tilladte log ud angiv den ønskede adgangskode igen november gem personlige oplysninger kontrolspørgsmål det sidste tegn i dit brugernavn skal være et bogstav a z eller tal skriv de tegn du kan se i billedet nedenfor";
33
+ const char* kTeststr_de_Latn = " abschnitt ordner aktivieren werden die ordnereinstellungen im farbabschnitt deaktiviert öchten sie wirklich fortfahren eldtypen angeben optional n diesem schritt geben sie für jedesfeld aus dem datenset den typ an ieser schritt ist optional eldtypen";
34
+ const char* kTeststr_dv_Thaa = " ހިންދީ ބަހުން ވާހަކަ ދައްކާއިރު ދެވަނަ ބަހެއްގެ ގޮތުގައާއި އެނޫން ގޮތްގޮތުން ހިންދީ ބަހުން ވާހަކަ ދައްކާ މީހުންގެ އަދަދު މިލިއަނަށް";
35
+ const char* kTeststr_dz_Tibt = " རྩིས བརྐྱབ ཚུལ ལྡན དང ངེས བདེན སྦ སྟོན ནིའི དོན ལུ ཁྱོད གུག ཤད ལག ལེན འཐབ དགོ ག དང ཨིན པུཊི གྲལ ཐིག གུ";
36
+ const char* kTeststr_el_Grek = " ή αρνητική αναζήτηση λέξης κλειδιού καταστήστε τις μεμονωμένες λέξεις κλειδιά περισσότερο στοχοθετημένες με τη μετατροπή τους σε";
37
+ const char* kTeststr_en_Latn = " a backup credit card by visiting your billing preferences page or visit the adwords help centre for more details https adwords google com support bin answer py answer hl en we were unable to process the payment of for your outstanding google adwords";
38
+ const char* kTeststr_eo_Latn = " a jarcento refoje per enmetado de koncerna pastro tiam de reformita konfesio ekde refoje ekzistis luteranaj komunumanoj tamen tiuj fondis propran komunumon nur en ambaŭ apartenis ekde al la evangela eklezio en prusio resp ties rejnlanda provinceklezio en";
39
+ const char* kTeststr_es_Latn = " a continuación haz clic en el botón obtener ruta también puedes desplazarte hasta el final de la página para cambiar tus opciones de búsqueda gráfico y detalles ésta es una lista de los vídeos que te recomendamos nuestras recomendaciones se basan";
40
+ const char* kTeststr_et_Latn = " a niipea kui sinu maksimaalne igakuine krediidi limiit on meie poolt heaks kiidetud on sinu kohustuseks see krediidilimiit";
41
+ const char* kTeststr_eu_Latn = " a den eraso bat honen kontra hortaz eragiketa bakarrik behar dituen eraso batek aes apurtuko luke nahiz eta oraingoz eraso bideraezina izan gaur egungo teknologiaren mugak direla eta oraingoz kezka hauek alde batera utzi daitezke orain arteko indar";
42
+ const char* kTeststr_fa_Arab = " آب خوردن عجله می کردند به جای باز ی کتک کاری می کردند و همه چيز مثل قبل بود فقط من ماندم و يک دنيا حرف و انتظار تا عاقبت رسيد احضاريه ی ای با";
43
+ const char* kTeststr_fi_Latn = " a joilla olet käynyt tämä kerro meille kuka ä olet ei tunnistettavia käyttötietoja kuten virheraportteja käytetään google desktopin parantamiseen etsi näyttää mukautettuja uutisia google desktop keskivaihto leikkaa voit kaksoisnapsauttaa";
44
+ const char* kTeststr_fj_Latn = " i kina na i iri ka duatani na matana main a meke wesi se meke mada na meke ni yaqona oqo na meke ka dau vakayagataki ena yaqona vakaturaga e dau caka toka ga kina na vucu ka dau lagati tiko kina na ka e yaco tiko na talo ni wai ni yaqona na lewai ni wai";
45
+ const char* kTeststr_fo_Latn = " at verða átaluverdar óhóskandi ella áloypandi vit kunnu ikki garanterða at google leitanin ikki finnur naka sum er áloypandi óhóskandi ella átaluvert og google tekur onga ábyrgd yvir tær síður sum koma við í okkara leitiskipan fá tær ein";
46
+ const char* kTeststr_fr_Latn = " a accès aux collections et aux frontaux qui lui ont été attribués il peut consulter et modifier ses collections et exporter des configurations de collection toutefois il ne peut pas créer ni supprimer des collections enfin il a accès aux fonctions";
47
+ const char* kTeststr_fy_Latn = " adfertinsjes gewoan lytse adfertinsjes mei besibbe siden dy t fan belang binne foar de ynhâld fan jo berjochten wolle jo mear witte fan gmail foardat jo jo oanmelde gean dan nei wy wurkje eltse dei om gmail te ferbetterjen dêrta sille wy jo sa út en";
48
+ const char* kTeststr_ga_Latn = " a bhfuil na focail go léir i do cheist le fáil orthu ní gá ach focail breise a chur leis na cinn a cuardaíodh cheana chun an cuardach a bheachtú nó a chúngú má chuirtear focal breise isteach aimseofar fo aicme ar leith de na torthaí a fuarthas";
49
+ const char* kTeststr_gd_Latn = " air son is gum bi casg air a h uile briosgaid no gum faigh thu brath nuair a tha briosgaid a tighinn gad rannsachadh ghoogle gu ceart mura bheil briosgaidean ceadaichte cuiridh google briosgaid dha do neach cleachdaidh fa leth tha google a cleachdadh";
50
+ const char* kTeststr_gl_Latn = " a africana central local business center debe ser como mínimo taranto tendas de venda polo miúdo cociñas servizos bordado canadá viaxes parques de vehículos de recreo hotel oriental habitación recibir unha postal no enderezo indicado anteriormente";
51
+ const char* kTeststr_gn_Latn = " aháta añe ë ne mbo ehára ndive ajeruréta chupe oporandujey haĝua peëme mba épa pekaru ha áĝa oporandúvo nde eréta avei re paraguaýpe kachíke he i leúpe ndépa re úma kure tatakuápe ha leu ombohovái héë ha ujepéma kachíke he ijey";
52
+ const char* kTeststr_gu_Gujr = " આના પરિણામ પ્રમાણસર ફોન્ટ અવતરણ ચિન્હવાળા પાઠને છુપાવો બધા સમૂહો શોધાયા હાલનો જ સંદેશ વિષયની";
53
+ const char* kTeststr_gv_Latn = " and not ripe as i thought yn assyl yn shynnagh as yn lion the ass the fox and the lion va assyl as shynnagh ayns commee son nyn vendeilys as sauchys hie ad magh ayns y cheyll dy shelg cha row ad er gholl feer foddey tra veeit ad rish lion yn shynnagh";
54
+ const char* kTeststr_ha_Latn = " a cikin a kan sakamako daga sakwannin a kan sakamako daga sakwannin daga ranar zuwa a kan sakamako daga guda daga ranar zuwa a kan sakamako daga shafukan daga ranar zuwa a kan sakamako daga guda a cikin last hour a kan sakamako daga guda daga kafar";
55
+ const char* kTeststr_hi_Deva = " ं ऐडवर्ड्स विज्ञापनों के अनुभव पर आधारित हैं और इनकी मदद से आपको अपने विज्ञापनों का अधिकतम लाभ";
56
+ const char* kTeststr_hr_Latn = " venski poljski kašupski dolnolužički gornjolužički češki slovački istočnoslavenski ruski bjeloruski ukrajinski rusinski i južnoslavenski slovenski hrvatski srpski crnogorski bošnjački makedonski i bugarski južnoslavenski jezici";
57
+ const char* kTeststr_ht_Latn = " ak pitit tout sosyete a chita se pou sa leta dwe pwoteje yo nimewo leta fèt pou li pwoteje tout paran ak pitit nan peyi a menm jan kit paran yo marye kit yo pa marye tout manman ki fè pitit leta fèt pou ba yo konkoul menm jan tou pou timoun piti ak pou";
58
+ const char* kTeststr_hu_Latn = " a felhasználóim a google azonosító szöveget ikor látják a felhasználóim a google azonosító szöveget felhasználók a google azonosító szöveget fogják látni minden tranzakció után ha a vásárlását regisztrációját oldalunk";
59
+ const char* kTeststr_hy_Armn = " ա յ եվ նա հիացած աչքերով նայում է հինգհարկանի շենքի տարօրինակ փոքրիկ քառակուսի պատուհաններին դեռ մենք շատ ենք հետամնաց ասում է նա այսպես է";
60
+ const char* kTeststr_ia_Latn = " a google informationes super le sitos que tu visita isto es necessari pro render disponibile alcun functionalitates del barra de utensiles a fin que nos pote monstrar informationes ulterior super un sito le barra de utensiles google debe dicer a nos le";
61
+ const char* kTeststr_id_Latn = " account anda dana anda belum kami terima berikan kami waktu sedikit lagi kami masih menunggu penerimaan transfer bank anda jika anda belum mengirimkan dana anda silakan kunjungi lakukan transfer bank ikuti instruksi pada halaman untuk memulai proses";
62
+ const char* kTeststr_ie_Latn = " abhorre exceptiones in li derivation plu cardinal por un l i es li regularità del flexion conjugation ples comparar latino sine flexione e li antiqui projectes naturalistic queles have quasi null regules de derivation ma si on nu examina li enunciationes";
63
+ const char* kTeststr_ik_Latn = " kuubuuraqabniqsuq ataruamik colville mi aasii tavrani siku kilaabman sulukpaukkat makua niksisugrufagivut tavrani sunaimña atifa quaqqat ii quaqqat aasii ukiabmagu utiqhuta tamaufa utqiabvifñun aasiiñ tatpaaffaqapta tuvaaqatinifarufa aasiiñ";
64
+ const char* kTeststr_is_Latn = " a afköst leitarorða þinna leitarorð neikvæð leitarorð auglýsingahópa byggja upp aðallista yfir ný leitarorð fyrir auglýsingahópana og skoða ítarleg gögn um árangur leitarorða eins og samkeppni auglýsenda og leitarmagn er krafist notkun";
65
+ const char* kTeststr_it_Latn = " a causa di un intervento di manutenzione del sistema fino alle ore circa ora legale costa del pacifico del novembre le campagne esistenti continueranno a essere pubblicate come di consueto anche durante questo breve periodo di inattività ci scusiamo per";
66
+ const char* kTeststr_iu_Cans = "ᐃᑯᒪᒻᒪᑦ ᕿᓈᖏᓐᓇᓲᖑᒻᒪᑦ ᑎᑎᖅᑕᓕᒫᖅᓃᕕᑦ ᑎᑦᕆᐊᑐᓐᖏᑦᑕᑎᑦ ᑎᑎᖅᑕᑉᐱᑦ ᓯᕗᓂᖓᓂ ᑎᑎᖅᖃᖅ ᑎᑎᕆᐊᑐᓐᖏᑕᐃᑦ ᕿᓂᓲᖑᔪᒍᑦ ᑎᑎᖅᑕᓕᒫᖅᓃᕕᑦ";
67
+ const char* kTeststr_iw_Hebr = " או לערוך את העדפות ההפצה אנא עקוב אחרי השלבים הבאים כנס לחשבון האישי שלך ב";
68
+ const char* kTeststr_ja_Hani = " このペ ジでは アカウントに指定された予算の履歴を一覧にしています それぞれの項目には 予算額と特定期間のステ タスが表示されます 現在または今後の予算を設定するには";
69
+ const char* kTeststr_jw_Latn = " account ten server niki kalian username meniko tanpo judul cacahe account nggonanmu wes pol pesen mu wes diguwak pesenan mu wes di simpen sante wae pesenan mu wes ke kirim mbuh tekan ora pesenan e ke kethok pesenan mu wes ke kirim mbuh tekan ora pesenan";
70
+ const char* kTeststr_ka_Geor = " ა ბირთვიდან მიღებული ელემენტი მენდელეევის პერიოდულ სიტემაში გადაინაცვლებს ორი უჯრით";
71
+ const char* kTeststr_kha_Latn = " kaba jem jai sa sngap thuh ia ki bynta ba sharum naka sohbuin jong phi nangta sa pynhiar ia ka kti kadiang jong phi sha ka krung jong phi bad da kaba pyndonkam kumjuh ia ki shympriahti jong phi sa sngap thuh shapoh ka tohtit jong phi pyndonkam ia kajuh ka";
72
+ const char* kTeststr_kk_Arab = " ﺎ ﻗﻴﺎﻧﺎﺕ ﺑﻮﻟﻤﺎﻳﺪﻯ ﺑﯘﻝ ﭘﺮﻭﺗﺴﻪﺳﯩﻦ ﻳﺎﻋﻨﻲ ﻗﺎﻻ ﻭﻣﯩﺮﯨﻨﺪﻩ ﻗﺎﺯﺍﻕ ء ﺗﯩﻠﯩﻨﯩﯔ ﻗﻮﻟﺪﺍﻧﯩﻠﻤﺎﯞﻯ ﻗﺎﺯﺍﻕ ﺟﻪﺭﯨﻨﺪﻩ";
73
+ const char* kTeststr_kk_Cyrl = " а билердің өзіне рұқсат берілмеген егер халық талап етсе ғана хан келісім берген өздеріңіз білесіздер қр қыл мыс тық кодексінде жазаның";
74
+ const char* kTeststr_kk_Latn = " bolsa da otanyna qaityp keledi al oralmandar basqa elderde diasporasy ote az bolghandyqtan bir birine komektesip bauyrmal bolady birde men poezben oralmandardyng qazaqstangha keluin kordim monghol qazaqtary poezdan tuse sala jerdi suip jylap keletin biraq";
75
+ const char* kTeststr_kl_Latn = " at nittartakkalli uani toqqarsimasatta akornanni nittartakkanut allanut ingerlaqqittoqarsinnaavoq kanukoka tassaavoq kommuneqarfiit kattuffiat nuna tamakkerlugu kommunit nittartagaannut ingerlaqqiffiusinnaasoq kisitsiserpassuit nunatsinnut tunngasut";
76
+ const char* kTeststr_km_Khmr = " ក ខ គ ឃ ង ច ឆ ជ ឈ ញ ដ ឋ ឌ ឍ ណ ត ថ ទ ធ ន ប ផ ព ភ ម យ រ ល វ ស ហ ឡ អ ឥ ឦ ឧ ឪ ឫ ឬ ឯ ឱ ទាំងអស់";
77
+ const char* kTeststr_kn_Knda = " ಂಠಯ್ಯನವರು ತುಮಕೂರು ಜಿಲ್ಲೆಯ ಚಿಕ್ಕನಾಯಕನಹಳ್ಳಿ ತಾಲ್ಲೂಕಿನ ತೀರ್ಥಪುರ ವೆಂಬ ಸಾಧಾರಣ ಹಳ್ಳಿಯ ಶ್ಯಾನುಭೋಗರ";
78
+ const char* kTeststr_ko_Hani = " 개별적으로 리포트 액세스 권한을 부여할 수 있습니다 액세스 권한 부여사용자에게 프로필 리포트에 액세스할 수 있는 권한을 부여하시려면 가용 프로필 상자에서 프로필 이름을 선택한 다음";
79
+ const char* kTeststr_ks_Deva = "बहामास छु केरेबियन मंज अख मुलुख राजधानी नसौ सम् बद्घ विषय";
80
+ // alt const char* kTeststr_ks_Deva = "जम्मू कश्à आन्ध्र प्रदेश द अख प्रदेश राजधानी रिचमंड यि छु गोडनितिन प्रदेशन मंज अख बेय विषय जार्ज";
81
+ const char* kTeststr_ku_Arab = " بۆ به ڕێوه بردنی نامه ی که دێتن ڕاسته وخۆ ڕه وان بکه نامه کانی گ مایل بۆ حسابی پۆستێکی تر هێنانی په یوه ندکاره کان له";
82
+ const char* kTeststr_ku_Latn = " be zmaneki ter le inglis werdegeretewe em srvise heshta le cor beta daye wate hest a taqi dekrete u bashtr dekret tewawwzmanekan wernegrawnetewe u ne hemu laperakn ke eme pshtiwan dekayn be teaweti wergerawete nermwalley wergeran teksti new wene nasnatewe";
83
+ const char* kTeststr_ky_Arab = " جانا انى تانۇۇ ۇلۇتۇن تانۇۇ قىرعىزدى بئلۉۉ دەگەندىك اچىق ايتساق ماناستى تاانىعاندىق ۅزۉڭدۉ تاانىعاندىق بۉگۉن تەما جۉكتۅمۅ ق ى رع ى ز ت ى ل ى";
84
+ const char* kTeststr_ky_Cyrl = " агай эле оболу мен садыбакас аганын өзү менен эмес эмгектери менен тааныштым жылдары ташкенде өзбекстан илимдер академиясынын баяны";
85
+ const char* kTeststr_la_Latn = " a deo qui enim nocendi causa mentiri solet si iam consulendi causa mentiatur multum profecit sed aliud est quod per se ipsum laudabile proponitur aliud quod in deterioris comparatione praeponitur aliter enim gratulamur cum sanus est homo aliter cum melius";
86
+ const char* kTeststr_lb_Latn = " a gewerkschaften och hei gefuerdert dir dammen an dir häre vun de gewerkschaften denkt un déi aarm wann der äer fuerderunge formuléiert d sechst congés woch an aarbechtszäitverkierzung hëllefen hinnen net d unhiewe vun de steigerungssäz bei de";
87
+ const char* kTeststr_lg_Latn = " abaana ba bani lukaaga mu ana mu babiri abaana ba bebayi lukaaga mu abiri mu basatu abaana ba azugaadi lukumi mu ebikumi bibiri mu abiri mu babiri abaana ba adonikamu lukaaga mu nltaaga mu mukaaga abaana ba biguvaayi enkumi bbiri mu ataano mu mukaaga";
88
+ const char* kTeststr_ln_Latn = " abakisamaki ndenge esengeli moyebami abongisamaki solo mpenza kombo ya moyebami elonguamaki kombo ya bayebami elonguamaki kombo eleki molayi po na esika epesameli limbisa esika ya kotia ba kombo esuki boye esengeli olimbola ndako na yo ya mikanda kombo";
89
+ const char* kTeststr_lo_Laoo = " ກຫາທົ່ວທັງເວັບ ແລະໃນເວັບໄຮ້ສາຍ ທຳອິດໃຫ້ທຳການຊອກຫາກ່ອນ ຈາກນັ້ນ ໃຫ້ກົດປຸ່ມເມນູ ໃນໜ້າຜົນໄດ້";
90
+ const char* kTeststr_lt_Latn = " a išsijungia mano idėja dėl geriausio laiko po pastarųjų savo santykių pasimokiau penki dalykai be kurių negaliu gyventi mano miegamajame tu surasi ideali pora išsilavinimas aukštoji mokykla koledžas universitetas pagrindinis laipsnis metai";
91
+ const char* kTeststr_lv_Latn = " a gadskārtējā izpārdošana slēpošana jāņi atlaide izmaiņas trafikā kas saistītas ar sezonas izpārdošanu speciālajām atlaidēm u c ir parastas un atslēgvārdi kas ir populāri noteiktos laika posmos šajā laikā saņems lielāku klikšķu";
92
+ const char* kTeststr_mg_Latn = " amporisihin i ianao mba hijery ny dika teksta ranofotsiny an ity lahatsoratra ity tsy ilaina ny opérateur efa karohina daholo ny teny rehetra nosoratanao ampiasao anaovana dokambarotra i google telugu datin ny takelaka fikarohana sary renitakelak i";
93
+ const char* kTeststr_mi_Latn = " haere ki te kainga o o haere ki te kainga o o haere ki te kainga o te rapunga ahua o haere ki te kainga o ka tangohia he ki to rapunga kaore au mohio te tikanga whakatiki o te ra he whakaharuru te pai rapunga a te rapunga ahua a e kainga o nga awhina o te";
94
+ const char* kTeststr_mk_Cyrl = " гласовите коалицијата на вмро дпмне како партија со најмногу освоени гласови ќе добие евра а на сметката на коализијата за македонија";
95
+ const char* kTeststr_ml_Mlym = " ം അങ്ങനെ ഞങ്ങള് അവരുടെ മുമ്പില് നിന്നു ഔടും ഉടനെ നിങ്ങള് പതിയിരിപ്പില് നിന്നു എഴുന്നേറ്റു";
96
+ const char* kTeststr_mn_Cyrl = " а боловсронгуй болгох орон нутгийн ажил үйлсийг уялдуулж зохицуулах дүрэм журам боловсруулах орон нутгийн өмч хөрөнгө санхүүгийн";
97
+ const char* kTeststr_mo_Cyrl = "оперативэ а органелор ши институциилор екзекутиве ши а органелор жудичиаре але путерий де стат фиекэруй орган ал путерий де стат и се";
98
+ const char* kTeststr_mr_Deva = " ं अनंताचे मुखीं होसील गाइला अमुप विठ्ठला दास तुह्मां माझें कोठें आलें होइऩल विचारा तरीं च";
99
+ const char* kTeststr_ms_Latn = "daripada dirinya hirako shinji seorang pemuda merujuk diri mereka sebagai vizard shinji telah cuba untuk menyakinkan ichigo untuk menyertai kumpulan mereka mengatakan bahawa hanya dia sahaja yang mampu mengajar ichigo teknik untuk mengawal hollow";
100
+ const char* kTeststr_ms_Latn2 = "Secara umumnya, Rumah Melayu Tradisional Negeri Sembilan mempunyai bumbung panjang iaitu jenis yang mempunyai satu perabung yang memanjang dari kiri ke kanan atau arah hulu ke hilir selari dengan jalan atau sungai. Bumbungnya mempunyai lentik yang bervariasi";
101
+ const char* kTeststr_ms_Latn3 = "bilik sebelah berkata julai pada pm ladymariah hmm sume ni terpulang kepada individu mungkin anda bernasib baik selama ini dalam membeli hp yang bagus deli berkata julai pada pm walaupun bukan bahsa baku tp tetap bahasa melayu kan perubahan boleh dibuat";
102
+ const char* kTeststr_mt_Latn = " ata ikteb messaġġ lil indirizzi differenti billi tagħżilhom u tagħfas il buttuna ikteb żid numri tfittxijja tal kotba mur print home kotba minn pagni ghal pagna minn ghall ktieb ta aċċessa stieden habib iehor grazzi it tim tal gruppi google";
103
+ const char* kTeststr_my_Latn = " jyk ef oif gawgodcsifayvdrfhrnf bmawgrsm topf dsvj g mail tamumif avhvm atmif txjwgif yxrqhk avhvm efae m pwifavhvm ef ufkyfwdky help center odkyvmyg drsm ar avh dswjhar cgef rsm udkawdkifygw f tajzawgudk smedkifygw f jyd awmh g mail cool features rsm";
104
+ const char* kTeststr_my_Mymr = " တက္ကသုိလ္ မ္ဟ ပ္ရန္ လာ္ရပီးေနာက္ န္ဟစ္ အရ္ဝယ္ ဦးသန္ ့သည္ ပန္ းတနော္ အမ္ယုိးသား ေက္ယာင္ း";
105
+ const char* kTeststr_na_Latn = " arcol obabakaen riringa itorere ibibokiei ababaro min kuduwa airumena baoin tokin rowiowet itiket keram damadamit eigirow etoreiy row keitsito boney ibingo itsiw dorerin naoerodelaporte s nauruan dictionary a c a c d g h o p s t y aiquen ion eins aiquen";
106
+ const char* kTeststr_ne_Deva = "अरू ठाऊँबाटपनि खुलेको छ यो खाता अर अरू ठाऊँबाटपनि खुलेको छ यो खाता अर ू";
107
+ const char* kTeststr_nl_Latn = " a als volgt te werk om een configuratiebestand te maken sitemap gen py ebruik filters om de s op te geven die moeten worden toegevoegd of uitgesloten op basis van de opmaaktaal elke sitemap mag alleen de s bevatten voor een bepaalde opmaaktaal dit";
108
+ const char* kTeststr_nn_Latn = " a for verktylina til å hjelpa deg å nå oss merk at pagerank syninga ikkje automatisk kjem til å henta inn informasjon frå sider med argument dvs frå sider med eit i en dersom datamaskina di er plassert bak ein mellomtenar for vevsider kan det verka";
109
+ const char* kTeststr_no_Latn = " a er obligatorisk tidsforskyvning plassering av katalogsøk planinformasjon loggfilbane gruppenavn kontoinformasjon passord domene gruppeinformasjon alle kampanjesporing alternativ bruker grupper oppgaveplanlegger oppgavehistorikk kontosammendrag antall";
110
+ // Update to be a not-repetative nearly-French string
111
+ // const char* kTeststr_oc_Latn = " a essat enviajat lo voste messatge a essat sauvat sujet de respond a per lo voste messatge es rejetat copia copia cachada data responsa vista aquel compte orkut es estat supprimit e es pas mai disponible vòstre compte orkut es estat arrestat pendent la";
112
+ const char* kTeststr_oc_Latn = " duas menas diferentas de cartochas cartocha de ptolemeo on e s legéish p t o l m i i s la soa dusau ipotèsa que ho alavètz que lo cartocha qu èra lo signe determinatiu qu ei a díser ideografic per definir ua paraula com estant un nom reiau com aqueth";
113
+ const char* kTeststr_om_Latn = " afaan katalaa bork bork bork hiikaa jira hin argamne gareen barbaadame hin argamne gargarsa qube en gar bayee jira garee walitti firooman gareewwan walitti firooman fuula web akka tartiiba qubeetiin agarsiisi akka tartiiba qubeetiin agarsiisaa jira akka";
114
+ const char* kTeststr_or_Orya = "ଅକ୍ଟୋବର ଡିସେମ୍ବର";
115
+ const char* kTeststr_pa_Guru = " ਂ ਦਿਨਾਂ ਵਿਚ ਭਾਈ ਸਾਹਿਬ ਦੀ ਬੁੱਚੜ ਗੋਬਿੰਦ ਰਾਮ ਨਾਲ ਅੜਫਸ ਚੱਲ ਰਹੀ ਸੀ ਗੋਬਿੰਦ ਰਾਮ ਨੇ ਭਾਈ ਸਾਹਿਬ ਦੀਆਂ ਭੈਣਾ";
116
+ const char* kTeststr_pl_Latn = " a australii będzie widział inne reklamy niż użytkownik z kanady kierowanie geograficzne sprawia że reklamy są lepiej dopasowane do użytkownika twojej strony oznacza to także że możesz nie zobaczyć wszystkich reklam które są wyświetlane na";
117
+ const char* kTeststr_ps_Arab = " اتو مستقل رياست جوړ شو او د پخواني ادبي انجمن څانګې ددې رياست جز شوی او ددې انجمن د ژبې مديريت د پښتو ټولنې په لوی مديريت واوښت لوی مدير يې د";
118
+ const char* kTeststr_pt_BR = " a abit prevê que a entrada desses produtos estrangeiros no mercado têxtil e vestuário do brasil possa reduzir os preços em cerca de a partir de má notícia para os empresários que terão que lutar para garantir suas margens de lucro mas boa notícia";
119
+ const char* kTeststr_qu_Latn = " is t ipanakunatapis rikuchinankupaq qanpa simiykipi noqaykoqpa uya jllanakunamanta kunan jamoq simikunaman qelqan tiyan watukuy qpa uyata qanpa llaqtaykipi llank anakuna simimanta yanapakuna simimanta mayqen llaqtallapis kay simimanta t ijray qpa qelqa";
120
+ const char* kTeststr_rm_Latn = " access a l autostrada sviai fermamain a sanestra en l access a l autostrada a sviai fermamain a sanestra en l access a l autostrada sviai fermamain a sanestra en l access a l autostrada a sviai fermamain a sanestra en l access a l autostrada ulvai enturn";
121
+ const char* kTeststr_rn_Latn = " ishaka mu ndero y abana bawe ganira n abigisha nimba hari ingorane izo ari zo zose ushobora gusaba kubonana n umwigisha canke kuvugana nawe kuri terefone inyuma y uko babarungikira urutonde rw amanota i muhira mu bisanzwe amashure aratumira abavyeyi";
122
+ const char* kTeststr_ro_Latn = " a anunţurilor reţineţi nu plătiţi pentru clicuri sau impresii ci numai atunci când pe site ul dvs survine o acţiune dorită site urile negative nu pot avea uri de destinaţie daţi instrucţiuni societăţii dvs bancare sau constructoare să";
123
+ const char* kTeststr_ru_Cyrl = " а неправильный формат идентификатора дн назад";
124
+ const char* kTeststr_rw_Latn = " dore ibyo ukeneye kumenya ukwo watubona ibibazo byinshi abandi babaza ububonero byibibina google onjela ho izina dyikyibina kyawe onjela ho yawe mulugo kulaho ibyandiko byawe shyilaho tegula yawe tulubaka tukongeraho iyanya mishya buliko tulambula";
125
+ const char* kTeststr_sa_Deva = " ं क र्मणस् त स्य य त्कि ङ्चेह करो त्यय ं त स्माल् लोका त्पु नरै ति अस्मै लोका य क र्मण इ ति नु काम";
126
+ const char* kTeststr_sa_Latn = " brahmā tatraivāntaradhīyata tataḥ saśiṣyo vālmīkir munir vismayam āyayau tasya śiṣyās tataḥ sarve jaguḥ ślokam imaṃ punaḥ muhur muhuḥ prīyamāṇāḥ prāhuś ca bhṛśavismitāḥ samākṣaraiś caturbhir yaḥ pādair gīto";
127
+ const char* kTeststr_sco_Latn = " a gless an geordie runciman ower a gless an tamson their man preached a hale hoor aboot the glorious memories o forty three an backsliders an profane persons like esau an aboot jeroboam the son o nebat that gaed stravagin to anither kirk an made aa israel";
128
+ const char* kTeststr_sd_Arab = " اضافو ٿي ٿيو پر اها خبر عثمان کي بعد پيئي ته سگريٽ ڇڪيندڙ مسلمان نه هو بلڪ هندو هو دڪان تي پهچي عثمان ڪسبت کولي گراهڪن جي سيرب لاهڻ شروع ڪئي پر";
129
+ const char* kTeststr_sg_Latn = " atâa na âkotta zo me lâkwê angbâ gï tarrango nî âkotta zo tî koddoro nî âde agbû tenne nî na kate töngana mbênî kotta kpalle tî nzönî dutï tî halëzo pëpe atâa sô âla lü gbâ tî ândya tî mâi na sahngo asâra gbâ tî";
130
+ const char* kTeststr_si_Sinh = " අනුරාධ මිහිඳුකුල නමින් සකුරා ට ලිපියක් තැපෑලෙන් එවා තිබුණා කි ් රස්ටි ෂෙල්ටන් ප ් රනාන්දු ද";
131
+ const char* kTeststr_sit_NP = " dialekten in de roerstreek pierre bakkes oet roerstreek blz bewirk waordebook zónjig oktoeaber is t ieëste mofers waordebook oetgekaome dit waordebook is samegestèldj";
132
+ const char* kTeststr_sk_Latn = " a aktivovať reklamnú kampaň ak chcete kampaň pred spustením ešte prispôsobiť uložte ju ako šablónu a pokračujte v úprave vyberte si jednu z možností nižšie a kliknite na tlačidlo uložiť kampaň nastavenia kampane môžete ľubovoľne";
133
+ const char* kTeststr_sl_Latn = " adsense stanje prijave za google adsense google adsense račun je bil začasno zamrznjen pozdravljeni hvala za vaše zanimanje v google adsense po pregledu vaše prijavnice so naši strokovnjaki ugotovili da spletna stran ki je trenutno povezana z vašim";
134
+ const char* kTeststr_sm_Latn = " autu mea o lo totonu le e le minaomia matou te tuu i totonu i le faamatalaina o le suesuega i taimi uma mea o lo totonu fuafua i mea e tatau fa afoi tala mai le newsgroup mataupu fa afoi mai tala e ai le mataupu e ai totonu tusitala o le itu o faamatalaga";
135
+ const char* kTeststr_sn_Latn = " chete vanyori vanotevera vakabatsira kunyora zvikamu zvino kumba home tinyorere tsamba chikamu chakumbirwa hachina kuwanikwa chikamu ichi cheninge chakayiswa kuimwe nzvimbo mudhairekitori rino chimwe chikamu chopadhuze pane chinhu chatadza kushanda bad";
136
+ const char* kTeststr_so_Latn = " a oo maanta bogga koobaad ugu qoran yahey beesha caalamka laakiin si kata oo beesha caalamku ula guntato soomaaliya waxa aan shaki ku jirin in aakhirataanka dadka soomaalida oo kaliya ay yihiin ku soomaaliya ka saari kara dhibka ay ku jirto";
137
+ const char* kTeststr_sq_Latn = " a do të kërkoni nga beogradi që të njohë pavarësinë e kosovës zoti thaçi prishtina është gati ta njoh pavarësinë e serbisë ndërsa natyrisht se do të kërkohet një gjë e tillë që edhe beogradi ta njoh shtetin e pavarur dhe sovran të";
138
+ const char* kTeststr_sr_Cyrl = "балчак балчак на мапи србије уреди демографија у насељу балчак живи пунолетна становника а просечна старост становништва износи година";
139
+ const char* kTeststr_sr_Latn = " autonomnih pokrajina saveznim zakonom može se propisati poseban sastav organizacija i delokrug saveta za poslove narodne odbrane članove saveta federacije bira na predlog predsedništva savezna skupština iz reda društveno političkih i drugih javnih";
140
+ const char* kTeststr_sr_ME_Latn = "savjet pobjeda a radi bržeg rada pošto rom radi sporije nego ram izvorni rom se isključuje a dio ram a se rezerviše te se u njega ne ploča procesor ram memorija grafička kartica zvučna kartica modem mrežna kartica napojna jedinica uređaji za pohranjivanje";
141
+ const char* kTeststr_ss_Latn = " bakhokhintsela yesikhashana bafake imininingwane ye akhawunti leliciniso kulelifomu nangabe akukafakwa imininingwane leliciniso imali lekhokhiwe angeke ifakwe kumkhokhintsela lofanele imininingwane ye akhawunti ime ngalendlela lelandzelako inombolo";
142
+ const char* kTeststr_st_Latn = " bang ba nang le thahasello matshwao a sehlooho thuto e thehilweng hodima diphetho ke tsela ya ho ruta le ho ithuta e totobatsang hantle seo baithuti ba lokelang ho se fihlella ntlhatheo eo e sebetsang ka yona ke ya hore titjhere o hlakisa pele seo";
143
+ const char* kTeststr_su_Latn = " alus gampang deuih uhun im gmail obrolan ulah disimpen na koropak kuring simpen obrolan dina koropak kuring obrolan obrolan anjeun teu boga arsip obrolan slovak slovenia vietnam catalan czech estonia hindi lithuania romania tagalog thai turkish édit iber";
144
+ const char* kTeststr_sv_Latn = " a bort objekt från google desktop post äldst meny öretag dress etaljer alternativ för vad är inne yaste google skrivbord plugin program för nyheter google visa nyheter som är anpassade efter de artiklar som du läser om du till exempel läser";
145
+ const char* kTeststr_sw_Latn = " a ujumbe mpya jumla unda tafuta na angalia vikundi vya kujadiliana na kushiriki mawazo iliyopangwa kwa tarehe watumiaji wapya futa orodha hizi lugha hoja vishikanisho vilivyo dhaminiwa ujumbe sanaa na tamasha toka udhibitisho wa neno kwa haraka fikia";
146
+ const char* kTeststr_syr_Syrc = "ܐܕܪܝܣ ܓܛܘ ܫܘܪܝܐ ܡܢ ܦܪܢܣܐ ܡܢ ܐܣܦܢܝܐ ܚܐܪܘܬܐ ܒܐܕܪ ܒܢܝܣܢ ܫܛܝܚܘܬܐ ܟܠܢܝܐ ܡܝ̈ܐ ܒܥܠܡܐ";
147
+ const char* kTeststr_ta_Taml = " அங்கு ராஜேந்திர சோழனால் கட்டப்பட்ட பிரம்மாண்டமான சிவன் கோவில் ஒன்றும் உள்ளது தொகு";
148
+ const char* kTeststr_te_Telu = " ఁ దనర జయించిన తత్వ మరసి చూడఁ దాన యగును రాజయోగి యిట్లు తేజరిల్లుచు నుండు విశ్వదాభిరామ వినర వేమ";
149
+ const char* kTeststr_tg_Arab = "رادیو فردا راديوى آزادى";
150
+ const char* kTeststr_tg_Cyrl = " адолат ва инсондӯстиро бар фашизм нажодпарастӣ ва адоват тарҷеҳ додааст чоп кунед ба дигарон фиристед чоп кунед ба дигарон фиристед";
151
+ const char* kTeststr_th_Thai = " กฏในการค้นหา หรือหน้าเนื้อหา หากท่านเลือกลงโฆษณา ท่านอาจจะปรับต้องเพิ่มงบประมาณรายวันตา";
152
+ const char* kTeststr_ti_Ethi = " ሃገር ተረፎም ዘለዉ ኢትዮጵያውያን ኣብቲ ምስ ኢትዮጵያ ዝዳውብ ኣውራጃ ደቡብ ንኽነብሩ ኣይፍቀደሎምን እዩ ካብ ሃገር ንኽትወጽእ ዜጋ ኹን ወጻእተኛ ናይ";
153
+ const char* kTeststr_tk_Cyrl = " айдянларына ынанярмыка эхли боз мейданлары сурулип гутарылан тебигы ота гарып гумлукларда миллиондан да артыкмач ири шахлы малы миллиона";
154
+ const char* kTeststr_tk_Latn = " akyllylyk çyn söýgi üçin böwet däl de tebigylykdyr duýgularyň gödeňsiligi aç açanlygy bahyllygy söýgini betnyşanlyk derejesine düşürýändir söýeni söý söýmedige süýkenme özüni söýmeýändigini görmek ýigit üçin uly";
155
+ const char* kTeststr_tl_Latn = " a na ugma sa google ay nakaka bantog sa gitna nang kliks na nangyayari sa pamamagitan nang ordinaryong paggagamit at sa kliks na likha nang pandaraya o hindi tunay na paggamit bunga nito nasasala namin ang mga kliks na hindi kailangan o hindi gusto nang";
156
+ const char* kTeststr_tlh_Latn = " a ghuv bid soh naq jih lodni yisov chich wo vamvo qeylis lunge pu chah povpu vodleh a dah ghah cho ej dah wo che pujwi bommu tlhegh darinmohlahchu pu majqa horey so lom qa ip quv law may vad suvtahbogh wa sanid utlh quv pus datu pu a vitu chu pu johwi tar";
157
+ const char* kTeststr_tn_Latn = " go etela batla ditsebe tsa web tse di nang le le batla ditsebe tse di golaganya le tswang mo leka go batla web yotlhe batla mo web yotlhe go bona home page ya google batla mo a o ne o batla gore a o ne o batla ditsebe tsa bihari batla mo re maswabi ga go";
158
+ const char* kTeststr_to_Latn = " a ke kumi oku ikai ke ma u vakai ki hono hokohoko faka alafapeti api pe ko e uluaki peesi a ho o fekumi faka malatihi fekumi ki he lea oku fakaha atu pe ko ha fonua fekumi ki he fekumi ki he peesi oku ngaahi me a oku sai imisi alu ki he ki he ulu aki";
159
+ const char* kTeststr_tr_Latn = " a ayarlarınızı görmeniz ve yönetmeniz içindir eğer kampanyanız için günlük bütçenizi gözden geçirebileceğiniz yeri arıyorsanız kampanya yönetimi ne gidin kampanyanızı seçin ve kampanya ayarlarını düzenle yi tıklayın sunumu";
160
+ const char* kTeststr_ts_Latn = " a ku na timhaka leti nga ta vulavuriwa na google google yi hlonipha yi tlhela yi sirheleta vanhu hinkwavo lava tirhisaka google toolbar ku dyondza hi vusireleli eka system ya hina hi kombela u hlaya vusireleli bya hina eka toolbar mbulavulo wu tshikiwile";
161
+ const char* kTeststr_tt_Cyrl = "ачарга да бирмәде чәт чәт килеп тора безнең абыйнымы олы абыйнымы эштән";
162
+ const char* kTeststr_tt_Latn = " alarnı eşkärtü proğramnarın eşläwen däwam itü tatar söylämen buldıru wä sizep alu sistemnarın eşläwen däwat itü häm başqalar yılnıñ mayında tatar internetı ictimağıy oyışması milli ts isemle berençe däräcäle häm tat";
163
+ const char* kTeststr_tw_Latn = " amammui tumidifo no bɛtow ahyɛ atoro som so mpofirim na wɔasɛe no pasaa ma ayɛ nwonwa dɛn na ɛbɛka wɔn ma wɔayɛ saa bible no ma ho mmuae wɔ adiyisɛm nhoma no mu sɛ onyankopɔn na ɔde hyɛɛ wɔn komam sɛ wɔmma ne nsusuwii mmra mu";
164
+ const char* kTeststr_ug_Arab = " ئالەملەرنىڭ پەرۋەردىگارىدىن تىلەيمەن سىلەر بۇ يەرلەردە باغچىلاردىن بۇلاقلاردىن زىرائەتلەردىن يۇمشاق پىشقان خورمىلاردىن بەھرىمەن بولۇپ";
165
+ const char* kTeststr_ug_Cyrl = " а башлиди әмма бу қетимқи канада мәтбуатлириниң хәвәрлиридә илгирикидәк хитай һөкүмәт мәтбуатлиридин нәқил алидиған вә уни көчүрүп";
166
+ const char* kTeststr_ug_Latn = " adawet bolghachqa hazir musherrepmu bu ikki partiyining birleshme hökümet qurushta pikir birliki hasil qilalmasliqini kütüwatqan iken wehalenki pakistan xelq partiyisining rehbiri asif eli zerdari pakistandiki bashqa ushshaq partiyilerning rehberliri";
167
+ const char* kTeststr_uk_Cyrl = " а більший бюджет щоб забезпечити собі максимум прибутків від переходів відстежуйте свої об яви за датою географічним розташуванням";
168
+ const char* kTeststr_ur_Arab = " آپ کو کم سے کم ممکنہ رقم چارج کرتا ہے اس کی مثال کے طور پر فرض کریں اگر آپ کی زیادہ سے زیادہ قیمت فی کلِک امریکی ڈالر اور کلِک کرنے کی شرح ہو تو";
169
+ const char* kTeststr_uz_Arab = " آرقلی بوتون سیاسی حزب و گروه لرفعالیتیگه رخصت بیرگن اخبارات واسطه لری شو ییل مدتیده مثال سیز ترقی تاپکن و اهالی نینگ اقتصادی وضعیتی اوتمیش";
170
+ const char* kTeststr_uz_Cyrl = " а гапирадиган бўлсак бунинг иккита йўли бор биринчиси мана шу қуриган сатҳини қумликларни тўхтатиш учун экотизимни мустаҳкамлаш қумга";
171
+ const char* kTeststr_uz_Latn = " abadiylashtirildi aqsh ayol prezidentga tayyormi markaziy osiyo afg onistonga qanday yordam berishi mumkin ukrainada o zbekistonlik muhojirlar tazyiqdan shikoyat qilmoqda gruziya va ukraina hozircha natoga qabul qilinmaydi afg oniston o zbekistonni g";
172
+ const char* kTeststr_vi_Latn = " adsense cho nội dung nhà cung cấp dịch vụ di động xác minh tín dụng thay đổi nhãn kg các ô xem chi phí cho từ chối các đơn đặt hàng dạng cấp dữ liệu ác minh trang web của bạn để xem";
173
+ const char* kTeststr_vo_Latn = " brefik se volapükavol nüm balid äpubon ün dü lif lölik okas redakans älaipübons gasedi at nomöfiko äd ai mu kuratiko pläo timü koup nedäna fa ns deutän kü päproibon fa koupanef me gased at ästeifülom ad propagidön volapüki as sam ün";
174
+ const char* kTeststr_wo_Latn = " am ak dëgg dëggam ak gëm aji bind ji te gëstu ko te jëfandikoo tegtalu xel ci saxal ko sokraat nag jëfandikoo woon na xeltu ngir tas jikko yu rafet ci biir nit ñi ak dëggu ak soppante sokraat nag ñëw na mook aflaton platon sukkandiku ci ñaari";
175
+ const char* kTeststr_xh_Latn = " a naynga zonke futhi libhengezwa kwiwebsite yebond yasemzantsi afrika izinga elisebenzayo xa usenza olu tyalo mali liya kusebenza de liphele ixesha lotyalo mali lwakho inzala ihlawulwa rhoqo emva kweenyanga ezintandathu ngomhla wamashumi amathathu ananye";
176
+ const char* kTeststr_xx_Bugi = "ᨄᨛᨑᨊᨒ ᨑᨗ ᨔᨒᨗᨓᨛ ᨕᨗᨋᨗᨔᨗ ᨒᨛᨄ ᨑᨛᨔᨛᨆᨗᨊ";
177
+ const char* kTeststr_xx_Goth = "𐌰 𐌰𐌱𐍂𐌰𐌷𐌰𐌼 𐌰𐌲𐌲𐌹𐌻𐌹𐍃𐌺𐍃 𐌸𐌹𐌿𐌳𐌹𐍃𐌺𐍃 𐍆𐍂𐌰𐌲𐌺𐌹𐍃𐌺𐍃";
178
+ const char* kTeststr_yi_Hebr = "און פאנטאזיע ער איז באקאנט צים מערסטן פאר זיינע באַלאַדעס ער האָט געוווינט אין ווארשע יעס פאריס ליווערפול און לאנדאן סוף כל סוף איז ער";
179
+ const char* kTeststr_yo_Latn = " abinibi han ikawe alantakun le ni opolopo ede abinibi ti a to lesese bi eniyan to fe lo se fe lati se atunse jowo mo pe awon oju iwe itakunagbaye miran ti ako ni oniruru ede abinibi le faragba nipa atunse ninu se iwadi blogs ni ori itakun agbaye ti e ba";
180
+ const char* kTeststr_za_Hani = " 两个宾语的字数较少时 只带一个动词 否则就带两个动词 三句子类 从句子方面去谈汉 壮语结构格式相异的类型的 叫句子类 汉 壮语中 句子类结构格式有差别的自然不少";
181
+ const char* kTeststr_za_Latn = " dih yinzminz ndaej daengz bujbienq youjyau dih cingzyin caeuq cinhingz diuz daihit boux boux ma daengz lajmbwn couh miz cwyouz cinhyenz caeuq genzli bouxboux bingzdaengj gyoengq vunz miz lijsing caeuq liengzsim wngdang daih gyoengq de lumj beixnuengx";
182
+ const char* kTeststr_zh_Hani = "产品的简报和公告 提交该申请后无法进行更改 请确认您的选择是正确的 对于要提交的图书 我确认 我是版权所有者或已得到版权所有者的授权 要更改您的国家 地区 请在此表的最上端更改您的";
183
+ const char* kTeststr_zh_TW = " 之前為 帳單交易作業區 已變更 廣告內容 之前為 銷售代表 之前為 張貼日期為 百分比之前為 合約 為 目標對象條件已刪除 結束日期之前為";
184
+ const char* kTeststr_zu_Latn = " ana engu uma inkinga iqhubeka siza ubike kwi isexwayiso ngenxa yephutha lomlekeleli sikwazi ukubuyisela emuva kuphela imiphumela engaqediwe ukuthola imiphumela eqediwe zama ukulayisha kabusha leli khasi emizuzwini engu uma inkinga iqhubeka siza uthumele";
185
+ const char* kTeststr_zzb_Latn = "becoose a ve a leemit qooereees tu vurds um gesh dee bork bork nu peges vere a fuoond cunteeening is a fery cummun vurd und ves nut inclooded in yuoor seerch zee ooperetur is unnecessery ve a incloode a ell seerch terms by deffoolt um de hur de hur de hur";
186
+ const char* kTeststr_zze_Latn = " a diffewent type of seawch send feedback about google wiwewess seawch to wap google com wesuwts found on de entiwe web fow wesuwts found on de mobiwe web fow de functionawity of de toolbar up button has been expanded swightwy it now considews fow exampwe";
187
+ const char* kTeststr_zzh_Latn = " b x z un b e t und rs n a dr ss p as ry an th r a dr ss ry us n a l ss mb gu us c ti n l ke a z p c d n a dr ss nt r d pl as en r n a dr ss y ur s ar h f r n ar d d n t m tch ny l c ti n w th n m l s nd m r r at d p g s th l c ti ns b l w w r ut m t ca y";
188
+ const char* kTeststr_zzp_Latn = " a isplayda nformationia boutaa rivacypa ndaa singua heta ooglega oolbarta ilightha uttonba ipza ostalpa odeca howsa tocksa uoteqa rowseba roa earchsa ithinwa heta ooglega ebwa irectoryda ememberra astka earchsa ypeta earchsa heta ebwa ooglega magesia";
189
+
190
+ const char* kTeststr_ms_close = "sukiyaki wikipedia bahasa melayu ensiklopedia bebas sukiyaki dari wikipedia bahasa melayu ensiklopedia bebas lompat ke navigasi gelintar sukiyaki sukiyaki hirisan tipis daging lembu sayur sayuran dan tauhu di dalam periuk besi yang dimasak di atas meja makan dengan cara rebusan sukiyaki dimakan dengan mence";
191
+ const char* kTeststr_id_close = "sukiyaki wikipedia indonesia ensiklopedia bebas berbahasa bebas berbahasa indonesia langsung ke navigasi cari untuk pengertian lain dari sukiyaki lihat sukiyaki irisan tipis daging sapi sayur sayuran dan tahu di dalam panci besi yang dimasak di atas meja makan dengan cara direbus sukiyaki dimakan dengan mence";
192
+
193
+ #endif // ENCODINGS_COMPACT_LANG_DET_UNITTEST_DATA_H_