xref: /aosp_15_r20/external/libtextclassifier/native/utils/normalization_test.cc (revision 993b0882672172b81d12fad7a7ac0c3e5c824a12)
1*993b0882SAndroid Build Coastguard Worker /*
2*993b0882SAndroid Build Coastguard Worker  * Copyright (C) 2018 The Android Open Source Project
3*993b0882SAndroid Build Coastguard Worker  *
4*993b0882SAndroid Build Coastguard Worker  * Licensed under the Apache License, Version 2.0 (the "License");
5*993b0882SAndroid Build Coastguard Worker  * you may not use this file except in compliance with the License.
6*993b0882SAndroid Build Coastguard Worker  * You may obtain a copy of the License at
7*993b0882SAndroid Build Coastguard Worker  *
8*993b0882SAndroid Build Coastguard Worker  *      http://www.apache.org/licenses/LICENSE-2.0
9*993b0882SAndroid Build Coastguard Worker  *
10*993b0882SAndroid Build Coastguard Worker  * Unless required by applicable law or agreed to in writing, software
11*993b0882SAndroid Build Coastguard Worker  * distributed under the License is distributed on an "AS IS" BASIS,
12*993b0882SAndroid Build Coastguard Worker  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13*993b0882SAndroid Build Coastguard Worker  * See the License for the specific language governing permissions and
14*993b0882SAndroid Build Coastguard Worker  * limitations under the License.
15*993b0882SAndroid Build Coastguard Worker  */
16*993b0882SAndroid Build Coastguard Worker 
17*993b0882SAndroid Build Coastguard Worker #include "utils/normalization.h"
18*993b0882SAndroid Build Coastguard Worker 
19*993b0882SAndroid Build Coastguard Worker #include <string>
20*993b0882SAndroid Build Coastguard Worker 
21*993b0882SAndroid Build Coastguard Worker #include "utils/base/integral_types.h"
22*993b0882SAndroid Build Coastguard Worker #include "utils/utf8/unicodetext.h"
23*993b0882SAndroid Build Coastguard Worker #include "utils/utf8/unilib.h"
24*993b0882SAndroid Build Coastguard Worker #include "gmock/gmock.h"
25*993b0882SAndroid Build Coastguard Worker #include "gtest/gtest.h"
26*993b0882SAndroid Build Coastguard Worker 
27*993b0882SAndroid Build Coastguard Worker namespace libtextclassifier3 {
28*993b0882SAndroid Build Coastguard Worker namespace {
29*993b0882SAndroid Build Coastguard Worker 
30*993b0882SAndroid Build Coastguard Worker using testing::Eq;
31*993b0882SAndroid Build Coastguard Worker 
32*993b0882SAndroid Build Coastguard Worker class NormalizationTest : public testing::Test {
33*993b0882SAndroid Build Coastguard Worker  protected:
NormalizationTest()34*993b0882SAndroid Build Coastguard Worker   NormalizationTest() : INIT_UNILIB_FOR_TESTING(unilib_) {}
35*993b0882SAndroid Build Coastguard Worker 
NormalizeTextCodepointWise(const std::string & text,const int32 codepointwise_ops)36*993b0882SAndroid Build Coastguard Worker   std::string NormalizeTextCodepointWise(const std::string& text,
37*993b0882SAndroid Build Coastguard Worker                                          const int32 codepointwise_ops) {
38*993b0882SAndroid Build Coastguard Worker     return libtextclassifier3::NormalizeTextCodepointWise(
39*993b0882SAndroid Build Coastguard Worker                unilib_, codepointwise_ops,
40*993b0882SAndroid Build Coastguard Worker                UTF8ToUnicodeText(text, /*do_copy=*/false))
41*993b0882SAndroid Build Coastguard Worker         .ToUTF8String();
42*993b0882SAndroid Build Coastguard Worker   }
43*993b0882SAndroid Build Coastguard Worker 
44*993b0882SAndroid Build Coastguard Worker   UniLib unilib_;
45*993b0882SAndroid Build Coastguard Worker };
46*993b0882SAndroid Build Coastguard Worker 
TEST_F(NormalizationTest,ReturnsIdenticalStringWhenNoNormalization)47*993b0882SAndroid Build Coastguard Worker TEST_F(NormalizationTest, ReturnsIdenticalStringWhenNoNormalization) {
48*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(NormalizeTextCodepointWise(
49*993b0882SAndroid Build Coastguard Worker                   "Never gonna let you down.",
50*993b0882SAndroid Build Coastguard Worker                   NormalizationOptions_::CodepointwiseNormalizationOp_NONE),
51*993b0882SAndroid Build Coastguard Worker               Eq("Never gonna let you down."));
52*993b0882SAndroid Build Coastguard Worker }
53*993b0882SAndroid Build Coastguard Worker 
54*993b0882SAndroid Build Coastguard Worker #if !defined(TC3_UNILIB_DUMMY)
TEST_F(NormalizationTest,DropsWhitespace)55*993b0882SAndroid Build Coastguard Worker TEST_F(NormalizationTest, DropsWhitespace) {
56*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
57*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
58*993b0882SAndroid Build Coastguard Worker           "Never gonna let you down.",
59*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_WHITESPACE),
60*993b0882SAndroid Build Coastguard Worker       Eq("Nevergonnaletyoudown."));
61*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
62*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
63*993b0882SAndroid Build Coastguard Worker           "Never\tgonna\t\tlet\tyou\tdown.",
64*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_WHITESPACE),
65*993b0882SAndroid Build Coastguard Worker       Eq("Nevergonnaletyoudown."));
66*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
67*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
68*993b0882SAndroid Build Coastguard Worker           "Never\u2003gonna\u2003let\u2003you\u2003down.",
69*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_WHITESPACE),
70*993b0882SAndroid Build Coastguard Worker       Eq("Nevergonnaletyoudown."));
71*993b0882SAndroid Build Coastguard Worker }
72*993b0882SAndroid Build Coastguard Worker 
TEST_F(NormalizationTest,DropsPunctuation)73*993b0882SAndroid Build Coastguard Worker TEST_F(NormalizationTest, DropsPunctuation) {
74*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
75*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
76*993b0882SAndroid Build Coastguard Worker           "Never gonna let you down.",
77*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_PUNCTUATION),
78*993b0882SAndroid Build Coastguard Worker       Eq("Never gonna let you down"));
79*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
80*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
81*993b0882SAndroid Build Coastguard Worker           "αʹ. Σημεῖόν ἐστιν, οὗ μέρος οὐθέν.",
82*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_PUNCTUATION),
83*993b0882SAndroid Build Coastguard Worker       Eq("αʹ Σημεῖόν ἐστιν οὗ μέρος οὐθέν"));
84*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
85*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
86*993b0882SAndroid Build Coastguard Worker           "978—3—16—148410—0",
87*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_PUNCTUATION),
88*993b0882SAndroid Build Coastguard Worker       Eq("9783161484100"));
89*993b0882SAndroid Build Coastguard Worker }
90*993b0882SAndroid Build Coastguard Worker 
TEST_F(NormalizationTest,LowercasesUnicodeText)91*993b0882SAndroid Build Coastguard Worker TEST_F(NormalizationTest, LowercasesUnicodeText) {
92*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
93*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
94*993b0882SAndroid Build Coastguard Worker           "αʹ. Σημεῖόν ἐστιν, οὗ μέρος οὐθέν.",
95*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_LOWERCASE),
96*993b0882SAndroid Build Coastguard Worker       Eq("αʹ. σημεῖόν ἐστιν, οὗ μέρος οὐθέν."));
97*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
98*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
99*993b0882SAndroid Build Coastguard Worker           "αʹ. Σημεῖόν ἐστιν, οὗ μέρος οὐθέν.",
100*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_WHITESPACE |
101*993b0882SAndroid Build Coastguard Worker               NormalizationOptions_::CodepointwiseNormalizationOp_LOWERCASE),
102*993b0882SAndroid Build Coastguard Worker       Eq("αʹ.σημεῖόνἐστιν,οὗμέροςοὐθέν."));
103*993b0882SAndroid Build Coastguard Worker }
104*993b0882SAndroid Build Coastguard Worker 
TEST_F(NormalizationTest,UppercasesUnicodeText)105*993b0882SAndroid Build Coastguard Worker TEST_F(NormalizationTest, UppercasesUnicodeText) {
106*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
107*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
108*993b0882SAndroid Build Coastguard Worker           "Κανένας άνθρωπος δεν ξέρει",
109*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_UPPERCASE),
110*993b0882SAndroid Build Coastguard Worker       Eq("ΚΑΝΈΝΑΣ ΆΝΘΡΩΠΟΣ ΔΕΝ ΞΈΡΕΙ"));
111*993b0882SAndroid Build Coastguard Worker   EXPECT_THAT(
112*993b0882SAndroid Build Coastguard Worker       NormalizeTextCodepointWise(
113*993b0882SAndroid Build Coastguard Worker           "Κανένας άνθρωπος δεν ξέρει",
114*993b0882SAndroid Build Coastguard Worker           NormalizationOptions_::CodepointwiseNormalizationOp_DROP_WHITESPACE |
115*993b0882SAndroid Build Coastguard Worker               NormalizationOptions_::CodepointwiseNormalizationOp_UPPERCASE),
116*993b0882SAndroid Build Coastguard Worker       Eq("ΚΑΝΈΝΑΣΆΝΘΡΩΠΟΣΔΕΝΞΈΡΕΙ"));
117*993b0882SAndroid Build Coastguard Worker }
118*993b0882SAndroid Build Coastguard Worker #endif
119*993b0882SAndroid Build Coastguard Worker 
120*993b0882SAndroid Build Coastguard Worker }  // namespace
121*993b0882SAndroid Build Coastguard Worker }  // namespace libtextclassifier3
122