microsoft/onnxruntime-extensions
Publicmirrored from https://github.com/microsoft/onnxruntime-extensionsAvailable
test/static_test/test_strings.cc
84lines · modecode
| 1 | // Copyright (c) Microsoft Corporation. All rights reserved. |
| 2 | // Licensed under the MIT License. |
| 3 | |
| 4 | #include "gtest/gtest.h" |
| 5 | #include "string_utils.h" |
| 6 | #include "text/re2_strings/string_regex_split_re.hpp" |
| 7 | #include "text/string_ecmaregex_split.hpp" |
| 8 | |
| 9 | TEST(strings, std_regex_test) { |
| 10 | std::regex regex("[\u2700-\u27bf\U0001f650-\U0001f67f\U0001f600-\U0001f64f\u2600-\u26ff" |
| 11 | "\U0001f300-\U0001f5ff\U0001f900-\U0001f9ff\U0001fa70-\U0001faff" |
| 12 | "\U0001f680-\U0001f6ff]"); |
| 13 | |
| 14 | std::string test =u8"abcde😀🔍🦑😁🔍🎉😂🤣"; |
| 15 | auto result = std::regex_replace(test, regex, ""); |
| 16 | std::cout << test << std::endl; |
| 17 | std::cout << result << std::endl; |
| 18 | } |
| 19 | |
| 20 | |
| 21 | TEST(strings, regex_split) { |
| 22 | std::string input = "hello world"; |
| 23 | re2::RE2 reg("(\\s)"); |
| 24 | re2::RE2 keep_reg("\\s"); |
| 25 | std::vector<std::string_view> tokens; |
| 26 | std::vector<int64_t> begin_offsets; |
| 27 | std::vector<int64_t> end_offsets; |
| 28 | RegexSplitImpl(input, reg, true, keep_reg, tokens, begin_offsets, end_offsets); |
| 29 | std::vector<std::string_view> expected_tokens{"hello", " ", " ", "world"}; |
| 30 | std::vector<int64_t> expected_begin_offsets{0, 5, 6, 7}; |
| 31 | std::vector<int64_t> expected_end_offsets{5, 6, 7, 12}; |
| 32 | EXPECT_EQ(expected_tokens, tokens); |
| 33 | EXPECT_EQ(expected_begin_offsets, begin_offsets); |
| 34 | EXPECT_EQ(expected_end_offsets, end_offsets); |
| 35 | } |
| 36 | |
| 37 | TEST(strings, regex_split_skip) { |
| 38 | std::string input = "hello world"; |
| 39 | re2::RE2 reg("(\\s)"); |
| 40 | re2::RE2 keep_reg(""); |
| 41 | std::vector<std::string_view> tokens; |
| 42 | std::vector<int64_t> begin_offsets; |
| 43 | std::vector<int64_t> end_offsets; |
| 44 | RegexSplitImpl(input, reg, true, keep_reg, tokens, begin_offsets, end_offsets); |
| 45 | std::vector<std::string_view> expected_tokens{"hello", "world"}; |
| 46 | std::vector<int64_t> expected_begin_offsets{0, 6}; |
| 47 | std::vector<int64_t> expected_end_offsets{5, 11}; |
| 48 | EXPECT_EQ(expected_tokens, tokens); |
| 49 | EXPECT_EQ(expected_begin_offsets, begin_offsets); |
| 50 | EXPECT_EQ(expected_end_offsets, end_offsets); |
| 51 | } |
| 52 | |
| 53 | TEST(strings, regex_split_no_matched) { |
| 54 | std::string input = "helloworld"; |
| 55 | std::regex reg("(\\s)"); |
| 56 | std::regex keep_reg(""); |
| 57 | std::vector<std::string_view> tokens; |
| 58 | std::vector<int64_t> begin_offsets; |
| 59 | std::vector<int64_t> end_offsets; |
| 60 | ECMARegexSplitImpl(input, reg, true, keep_reg, tokens, begin_offsets, end_offsets); |
| 61 | std::vector<std::string_view> expected_tokens{"helloworld"}; |
| 62 | std::vector<int64_t> expected_begin_offsets{0}; |
| 63 | std::vector<int64_t> expected_end_offsets{10}; |
| 64 | EXPECT_EQ(expected_tokens, tokens); |
| 65 | EXPECT_EQ(expected_begin_offsets, begin_offsets); |
| 66 | EXPECT_EQ(expected_end_offsets, end_offsets); |
| 67 | } |
| 68 | |
| 69 | TEST(strings, regex_split_begin_end_delim) { |
| 70 | std::string input = " hello world "; |
| 71 | std::regex reg("(\\s)"); |
| 72 | std::regex keep_reg("\\s"); |
| 73 | std::vector<std::string_view> tokens; |
| 74 | std::vector<int64_t> begin_offsets; |
| 75 | std::vector<int64_t> end_offsets; |
| 76 | ECMARegexSplitImpl(input, reg, true, keep_reg, tokens, begin_offsets, end_offsets); |
| 77 | std::vector<std::string_view> expected_tokens{" ", "hello"," ", "world", " "}; |
| 78 | std::vector<int64_t> expected_begin_offsets{0, 1, 6, 7, 12}; |
| 79 | std::vector<int64_t> expected_end_offsets{1, 6, 7, 12, 13}; |
| 80 | EXPECT_EQ(expected_tokens, tokens); |
| 81 | EXPECT_EQ(expected_begin_offsets, begin_offsets); |
| 82 | EXPECT_EQ(expected_end_offsets, end_offsets); |
| 83 | } |
| 84 | |
| 85 | |