1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
|
// Copyright 2022 The Chromium Authors
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "components/segmentation_platform/internal/segmentation_ukm_helper.h"
#include <cmath>
#include <optional>
#include <string_view>
#include "base/bit_cast.h"
#include "base/strings/string_number_conversions.h"
#include "base/test/metrics/histogram_tester.h"
#include "base/test/scoped_feature_list.h"
#include "base/test/simple_test_clock.h"
#include "base/test/task_environment.h"
#include "base/time/time.h"
#include "components/prefs/testing_pref_service.h"
#include "components/segmentation_platform/internal/constants.h"
#include "components/segmentation_platform/internal/selection/segmentation_result_prefs.h"
#include "components/segmentation_platform/public/config.h"
#include "components/segmentation_platform/public/features.h"
#include "components/segmentation_platform/public/local_state_helper.h"
#include "components/segmentation_platform/public/proto/prediction_result.pb.h"
#include "components/segmentation_platform/public/proto/segmentation_platform.pb.h"
#include "components/segmentation_platform/public/segmentation_platform_service.h"
#include "components/ukm/test_ukm_recorder.h"
#include "services/metrics/public/cpp/ukm_builders.h"
#include "services/metrics/public/cpp/ukm_source_id.h"
#include "testing/gtest/include/gtest/gtest.h"
using Segmentation_ModelExecution = ukm::builders::Segmentation_ModelExecution;
namespace segmentation_platform {
namespace {
// Round errors allowed during conversion.
static const double kRoundingError = 1E-5;
float Int64ToFloat(int64_t encoded) {
return static_cast<float>(base::bit_cast<double>(encoded));
}
void CompareEncodeDecodeDifference(float tensor) {
ASSERT_LT(
std::abs(tensor -
Int64ToFloat(
segmentation_platform::SegmentationUkmHelper::FloatToInt64(
tensor))),
kRoundingError);
}
std::optional<proto::PredictionResult> GetPredictionResult(
std::optional<base::Time> prediction_time = std::nullopt) {
proto::PredictionResult result;
result.add_result(0.5);
result.add_result(0.4);
if (prediction_time.has_value()) {
result.set_timestamp_us(
prediction_time->ToDeltaSinceWindowsEpoch().InMicroseconds());
}
return result;
}
proto::SegmentInfo CreateTestSegmentInfo(proto::SegmentId segment_id,
bool upload_tensors) {
proto::SegmentInfo segment_info;
segment_info.set_segment_id(segment_id);
segment_info.mutable_model_metadata()->set_upload_tensors(upload_tensors);
return segment_info;
}
} // namespace
class SegmentationUkmHelperTest : public testing::Test {
public:
SegmentationUkmHelperTest() = default;
SegmentationUkmHelperTest(const SegmentationUkmHelperTest&) = delete;
SegmentationUkmHelperTest& operator=(const SegmentationUkmHelperTest&) =
delete;
~SegmentationUkmHelperTest() override = default;
void SetUp() override {
InitializeUkmHelper();
test_recorder_.Purge();
}
void InitializeUkmHelper() {
SegmentationUkmHelper::GetInstance()->Initialize();
}
void ExpectUkmMetrics(std::string_view entry_name,
const std::vector<std::string_view>& keys,
const std::vector<int64_t>& values,
ukm::SourceId source_id = ukm::kInvalidSourceId) {
const auto& entries = test_recorder_.GetEntriesByName(entry_name);
EXPECT_EQ(1u, entries.size());
for (const ukm::mojom::UkmEntry* entry : entries) {
if (source_id != ukm::kInvalidSourceId) {
EXPECT_EQ(entry->source_id, source_id);
}
const size_t keys_size = keys.size();
EXPECT_EQ(keys_size, values.size());
for (size_t i = 0; i < keys_size; ++i) {
test_recorder_.ExpectEntryMetric(entry, keys[i], values[i]);
}
}
}
void ExpectEmptyUkmMetrics(std::string_view entry_name) {
EXPECT_EQ(0u, test_recorder_.GetEntriesByName(entry_name).size());
}
void SetSamplingRate(int sampling_rate) {
feature_list_.InitAndEnableFeatureWithParameters(
features::kSegmentationPlatformModelExecutionSampling,
{{kModelExecutionSamplingRateKey,
base::NumberToString(sampling_rate)}});
InitializeUkmHelper();
}
protected:
base::test::TaskEnvironment task_environment_;
ukm::TestAutoSetUkmRecorder test_recorder_;
base::test::ScopedFeatureList feature_list_;
};
// Tests that basic execution results recording works properly.
TEST_F(SegmentationUkmHelperTest, TestExecutionResultReporting) {
SetSamplingRate(1);
// Allow results for OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB to be recorded.
ModelProvider::Request input_tensors = {0.1, 0.7, 0.8, 0.5};
SegmentationUkmHelper::GetInstance()->RecordModelExecutionResult(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101, input_tensors,
{0.6, 0.3});
ExpectUkmMetrics(Segmentation_ModelExecution::kEntryName,
{Segmentation_ModelExecution::kOptimizationTargetName,
Segmentation_ModelExecution::kModelVersionName,
Segmentation_ModelExecution::kInput0Name,
Segmentation_ModelExecution::kInput1Name,
Segmentation_ModelExecution::kInput2Name,
Segmentation_ModelExecution::kInput3Name,
Segmentation_ModelExecution::kPredictionResult1Name,
Segmentation_ModelExecution::kPredictionResult2Name},
{
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB,
101,
SegmentationUkmHelper::FloatToInt64(0.1),
SegmentationUkmHelper::FloatToInt64(0.7),
SegmentationUkmHelper::FloatToInt64(0.8),
SegmentationUkmHelper::FloatToInt64(0.5),
SegmentationUkmHelper::FloatToInt64(0.6),
SegmentationUkmHelper::FloatToInt64(0.3),
});
}
// Tests that execution results recording are disabled if sampling rate is 0.
TEST_F(SegmentationUkmHelperTest,
TestExecutionResultReportingwithZeroSampling) {
SetSamplingRate(0);
// Allow results for OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB to be recorded.
ModelProvider::Request input_tensors = {0.1, 0.7, 0.8, 0.5};
EXPECT_EQ(SegmentationUkmHelper::GetInstance()->RecordModelExecutionResult(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101,
input_tensors, {0.6, 0.3}),
ukm::NoURLSourceId());
}
// Tests that the training data collection recording works properly.
TEST_F(SegmentationUkmHelperTest, TestTrainingDataCollectionReporting) {
ModelProvider::Request input_tensors = {0.1};
ModelProvider::Response outputs = {1.0, 0.0};
std::vector<int> output_indexes = {2, 3};
SelectedSegment selected_segment(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 10);
selected_segment.selection_time = base::Time::Now() - base::Seconds(10);
SegmentationUkmHelper::GetInstance()->RecordTrainingData(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101,
/*ukm_source_id=*/55, input_tensors, outputs, output_indexes,
GetPredictionResult(selected_segment.selection_time), selected_segment);
ExpectUkmMetrics(Segmentation_ModelExecution::kEntryName,
{Segmentation_ModelExecution::kOptimizationTargetName,
Segmentation_ModelExecution::kModelVersionName,
Segmentation_ModelExecution::kInput0Name,
Segmentation_ModelExecution::kActualResult3Name,
Segmentation_ModelExecution::kActualResult4Name,
Segmentation_ModelExecution::kPredictionResult1Name,
Segmentation_ModelExecution::kPredictionResult2Name,
Segmentation_ModelExecution::kSelectionResultName,
Segmentation_ModelExecution::kOutputDelaySecName},
{
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB,
101,
SegmentationUkmHelper::FloatToInt64(0.1),
SegmentationUkmHelper::FloatToInt64(1.0),
SegmentationUkmHelper::FloatToInt64(0.0),
SegmentationUkmHelper::FloatToInt64(0.5),
SegmentationUkmHelper::FloatToInt64(0.4),
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB,
10,
},
/*source_id=*/55);
}
// Tests tensor uploading for default allowed list.
TEST_F(SegmentationUkmHelperTest, TestDefaultAllowedList) {
proto::SegmentInfo segment_info =
CreateTestSegmentInfo(proto::OPTIMIZATION_TARGET_UNKNOWN, false);
EXPECT_FALSE(
SegmentationUkmHelper::GetInstance()->IsUploadRequested(segment_info));
segment_info = CreateTestSegmentInfo(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, false);
EXPECT_TRUE(
SegmentationUkmHelper::GetInstance()->IsUploadRequested(segment_info));
}
// Tests that tensor uploading if default allowed list is disabled.
TEST_F(SegmentationUkmHelperTest, TestDisallowDefaultAllowedList) {
feature_list_.InitAndDisableFeature(
features::kSegmentationDefaultReportingSegments);
InitializeUkmHelper();
proto::SegmentInfo segment_info = CreateTestSegmentInfo(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, false);
EXPECT_FALSE(
SegmentationUkmHelper::GetInstance()->IsUploadRequested(segment_info));
}
// Tests that tensor uploading is enabled through metadata.
TEST_F(SegmentationUkmHelperTest, TestUploadTensorsAllowedFromMetadata) {
proto::SegmentInfo segment_info = CreateTestSegmentInfo(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, true);
EXPECT_TRUE(
SegmentationUkmHelper::GetInstance()->IsUploadRequested(segment_info));
}
// Tests that float encoding works properly.
TEST_F(SegmentationUkmHelperTest, TestFloatEncoding) {
// Compare the numbers with their IEEE754 binary representations in double.
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(0.5), 0x3FE0000000000000);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(0.25), 0x3FD0000000000000);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(0.125), 0x3FC0000000000000);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(0.75), 0x3FE8000000000000);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(1), 0x3FF0000000000000);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(0), 0);
ASSERT_EQ(SegmentationUkmHelper::FloatToInt64(10), 0x4024000000000000);
}
// Tests that floats encoded can be properly decoded later.
TEST_F(SegmentationUkmHelperTest, FloatEncodeDeocode) {
CompareEncodeDecodeDifference(0.1);
CompareEncodeDecodeDifference(0.5);
CompareEncodeDecodeDifference(0.88);
CompareEncodeDecodeDifference(0.01);
ASSERT_EQ(0, Int64ToFloat(SegmentationUkmHelper::FloatToInt64(0)));
ASSERT_EQ(1, Int64ToFloat(SegmentationUkmHelper::FloatToInt64(1)));
}
// Tests that there are too many input tensors to record.
TEST_F(SegmentationUkmHelperTest, TooManyInputTensors) {
SetSamplingRate(1);
base::HistogramTester tester;
std::string histogram_name(
"SegmentationPlatform.StructuredMetrics.TooManyTensors.Count");
ModelProvider::Request input_tensors(100, 0.1);
ukm::SourceId source_id =
SegmentationUkmHelper::GetInstance()->RecordModelExecutionResult(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101, input_tensors,
{0.6});
ASSERT_EQ(source_id, ukm::kInvalidSourceId);
tester.ExpectTotalCount(histogram_name, 1);
ASSERT_EQ(tester.GetTotalSum(histogram_name), 100);
}
// Tests output validation for |RecordTrainingData|.
TEST_F(SegmentationUkmHelperTest, OutputsValidation) {
ModelProvider::Request input_tensors{0.1};
// outputs, output_indexes size doesn't match.
ModelProvider::Response outputs{1.0, 0.0};
std::vector<int> output_indexes{0};
ukm::SourceId source_id =
SegmentationUkmHelper::GetInstance()->RecordTrainingData(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101,
ukm::kInvalidSourceId, input_tensors, outputs, output_indexes,
GetPredictionResult(), std::nullopt);
ASSERT_EQ(source_id, ukm::kInvalidSourceId);
// output_indexes value too large.
output_indexes = {100, 1000};
source_id = SegmentationUkmHelper::GetInstance()->RecordTrainingData(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101,
ukm::kInvalidSourceId, input_tensors, outputs, output_indexes,
GetPredictionResult(), std::nullopt);
ASSERT_EQ(source_id, ukm::kInvalidSourceId);
// Valid outputs.
output_indexes = {3, 0};
source_id = SegmentationUkmHelper::GetInstance()->RecordTrainingData(
proto::OPTIMIZATION_TARGET_SEGMENTATION_NEW_TAB, 101,
ukm::kInvalidSourceId, input_tensors, outputs, output_indexes,
GetPredictionResult(), std::nullopt);
ASSERT_NE(source_id, ukm::kInvalidSourceId);
}
TEST_F(SegmentationUkmHelperTest, AllowedToUploadData) {
TestingPrefServiceSimple prefs;
SegmentationPlatformService::RegisterLocalStatePrefs(prefs.registry());
LocalStateHelper::GetInstance().Initialize(&prefs);
base::SimpleTestClock clock;
clock.SetNow(base::Time::Now());
// If pref is not initialized, AllowedToUploadData() always return false.
ASSERT_FALSE(
SegmentationUkmHelper::AllowedToUploadData(base::Seconds(1), &clock));
LocalStateHelper::GetInstance().SetPrefTime(
kSegmentationUkmMostRecentAllowedTimeKey, clock.Now());
ASSERT_FALSE(
SegmentationUkmHelper::AllowedToUploadData(base::Seconds(1), &clock));
clock.Advance(base::Seconds(10));
ASSERT_TRUE(
SegmentationUkmHelper::AllowedToUploadData(base::Seconds(1), &clock));
ASSERT_FALSE(
SegmentationUkmHelper::AllowedToUploadData(base::Seconds(11), &clock));
}
} // namespace segmentation_platform
|
