• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /**
2  * Copyright 2021 Huawei Technologies Co., Ltd
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  * http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 #include "common/common.h"
17 #include "minddata/dataset/engine/datasetops/source/sampler/sampler.h"
18 #include "minddata/dataset/engine/ir/datasetops/source/samplers/distributed_sampler_ir.h"
19 #include "minddata/dataset/engine/ir/datasetops/source/samplers/pk_sampler_ir.h"
20 #include "minddata/dataset/engine/ir/datasetops/source/samplers/prebuilt_sampler_ir.h"
21 #include "minddata/dataset/engine/ir/datasetops/source/samplers/random_sampler_ir.h"
22 #include "minddata/dataset/engine/ir/datasetops/source/samplers/samplers_ir.h"
23 #include "minddata/dataset/engine/ir/datasetops/source/samplers/sequential_sampler_ir.h"
24 #include "minddata/dataset/engine/ir/datasetops/source/samplers/subset_random_sampler_ir.h"
25 #include "minddata/dataset/engine/ir/datasetops/source/samplers/subset_sampler_ir.h"
26 #include "minddata/dataset/engine/ir/datasetops/source/samplers/weighted_random_sampler_ir.h"
27 #include "minddata/dataset/core/tensor.h"
28 
29 using namespace mindspore::dataset;
30 using mindspore::dataset::Tensor;
31 
32 class MindDataTestIrSampler : public UT::DatasetOpTesting {
33  protected:
34 };
35 
TEST_F(MindDataTestIrSampler,TestCalculateNumSamples)36 TEST_F(MindDataTestIrSampler, TestCalculateNumSamples) {
37   int64_t num_rows = 30;  // dummy variable for number of rows in the dataset
38   std::shared_ptr<SamplerObj> sampl = std::make_shared<DistributedSamplerObj>(2, 1, false, 6, 1, -1, true);
39   EXPECT_NE(sampl, nullptr);
40   std::shared_ptr<SamplerRT> sampler_rt;
41   sampl->SamplerBuild(&sampler_rt);
42   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), 6);
43 
44   sampl = std::make_shared<PKSamplerObj>(3, false, 0);
45   EXPECT_NE(sampl, nullptr);
46   sampl->SamplerBuild(&sampler_rt);
47   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), -1);
48 
49   sampl = std::make_shared<RandomSamplerObj>(false, 12);
50   EXPECT_NE(sampl, nullptr);
51   sampl->SamplerBuild(&sampler_rt);
52   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), 12);
53 
54   sampl = std::make_shared<SequentialSamplerObj>(0, 10);
55   EXPECT_NE(sampl, nullptr);
56   sampl->SamplerBuild(&sampler_rt);
57   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), 10);
58 
59   std::vector<double> weights = {0.9, 0.8, 0.68, 0.7, 0.71, 0.6, 0.5, 0.4, 0.3, 0.5, 0.2, 0.1};
60   sampl = std::make_shared<WeightedRandomSamplerObj>(weights, 12);
61   EXPECT_NE(sampl, nullptr);
62   sampl->SamplerBuild(&sampler_rt);
63   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), 12);
64 
65   std::vector<int64_t> indices = {1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21};
66   sampl = std::make_shared<SubsetRandomSamplerObj>(indices, 11);
67   EXPECT_NE(sampl, nullptr);
68   sampl->SamplerBuild(&sampler_rt);
69   EXPECT_EQ(sampler_rt->CalculateNumSamples(num_rows), 11);
70 
71   // Testing chains
72   // Parent and child have num_samples
73   std::shared_ptr<SamplerObj> sampl1 = std::make_shared<WeightedRandomSamplerObj>(weights, 12);
74   EXPECT_NE(sampl1, nullptr);
75   std::shared_ptr<SamplerRT> sampler_rt1;
76   sampl1->SamplerBuild(&sampler_rt1);
77 
78   std::shared_ptr<SamplerObj> sampl2 = std::make_shared<SequentialSamplerObj>(0, 10);
79   EXPECT_NE(sampl2, nullptr);
80   std::shared_ptr<SamplerRT> sampler_rt2;
81   sampl2->SamplerBuild(&sampler_rt2);
82   sampler_rt2->AddChild(sampler_rt1);
83   EXPECT_EQ(sampler_rt2->CalculateNumSamples(num_rows), 10);
84 
85   // Parent doesn't have num_samples
86   std::shared_ptr<SamplerObj> sampl3 = std::make_shared<WeightedRandomSamplerObj>(weights, 12);
87   EXPECT_NE(sampl3, nullptr);
88   std::shared_ptr<SamplerRT> sampler_rt3;
89   sampl3->SamplerBuild(&sampler_rt3);
90 
91   std::shared_ptr<SamplerObj> sampl4 = std::make_shared<SubsetRandomSamplerObj>(indices, 0);
92   EXPECT_NE(sampl4, nullptr);
93   std::shared_ptr<SamplerRT> sampler_rt4;
94   sampl4->SamplerBuild(&sampler_rt4);
95   sampler_rt4->AddChild(sampler_rt3);
96   EXPECT_EQ(sampler_rt4->CalculateNumSamples(num_rows), 11);
97 
98   // Child doesn't have num_samples
99   std::shared_ptr<SamplerObj> sampl5 = std::make_shared<RandomSamplerObj>(false, 0);
100   EXPECT_NE(sampl5, nullptr);
101   std::shared_ptr<SamplerRT> sampler_rt5;
102   sampl5->SamplerBuild(&sampler_rt5);
103 
104   std::shared_ptr<SamplerObj> sampl6 = std::make_shared<PKSamplerObj>(3, false, 7);
105   EXPECT_NE(sampl6, nullptr);
106   std::shared_ptr<SamplerRT> sampler_rt6;
107   sampl6->SamplerBuild(&sampler_rt6);
108   sampler_rt6->AddChild(sampler_rt5);
109   EXPECT_EQ(sampler_rt6->CalculateNumSamples(num_rows), -1);
110 }
111 
TEST_F(MindDataTestIrSampler,TestSamplersMoveParameters)112 TEST_F(MindDataTestIrSampler, TestSamplersMoveParameters) {
113   std::vector<int64_t> indices = {1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 21, 23};
114   std::shared_ptr<SamplerObj> sampl1 = std::make_shared<SubsetRandomSamplerObj>(indices, 0);
115   EXPECT_FALSE(indices.empty());
116   std::shared_ptr<SamplerRT> sampler_rt = nullptr;
117   sampl1->SamplerBuild(&sampler_rt);
118   EXPECT_NE(sampler_rt, nullptr);
119   std::shared_ptr<SamplerObj> sampl2 = std::make_shared<SubsetRandomSamplerObj>(std::move(indices), 0);
120   EXPECT_TRUE(indices.empty());
121   std::shared_ptr<SamplerRT> sampler_rt2 = nullptr;
122   sampl2->SamplerBuild(&sampler_rt2);
123   EXPECT_NE(sampler_rt, nullptr);
124 }
125