From a2ee7e0315166fe38b57a4b3e477767fdbafff6d Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 08:55:56 +0200 Subject: [PATCH 1/7] fix --- src/test/embeddingsnode_test.cpp | 22 ++++++++++++++++++ src/test/llm/tokenize_endpoint_test.cpp | 30 ++++++++++++++++++++++--- src/test/reranknode_test.cpp | 22 ++++++++++++++++++ src/tokenize/tokenize_parser.cpp | 16 ++++++++----- 4 files changed, 81 insertions(+), 9 deletions(-) diff --git a/src/test/embeddingsnode_test.cpp b/src/test/embeddingsnode_test.cpp index 6443bba525..bff7ef98db 100644 --- a/src/test/embeddingsnode_test.cpp +++ b/src/test/embeddingsnode_test.cpp @@ -786,3 +786,25 @@ TEST_F(EmbeddingsTokenizeHttpTest, tokenizeBatchWithPadToMaxLen) { ovms::StatusCode::OK); AssertTokenizationResult(response, expectedTokens); } + +TEST_F(EmbeddingsTokenizeHttpTest, tokenizeEmptyNestedArray) { + std::string requestBody = R"( + { + "model": "embeddings_ov", + "text": [[]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + +TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { + std::string requestBody = R"( + { + "model": "embeddings_ov", + "text": [[], [], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index 17c390b650..4e87c76a83 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -417,9 +417,6 @@ TEST_P(LLMTokenizeTests, tokenizeArrayOfStringsWithPaddingSideLeft) { TEST_P(LLMTokenizeTests, tokenizeStringWithAddSpecialTokens) { auto params = GetParam(); - if (params.modelName == "vlm_cb_regular" || params.modelName == "vlm_legacy_regular") { - GTEST_SKIP() << "Skipping test for " << params.modelName; - } std::string requestBody = R"( { @@ -443,6 +440,33 @@ TEST_P(LLMTokenizeTests, tokenizeStringWithAddSpecialTokens) { ASSERT_GE(tokens.Size(), params.expectedTokens.size()); } +TEST_P(LLMTokenizeTests, tokenizeEmptyNestedArray) { + auto params = GetParam(); + + std::string requestBody = R"( + { + "model": ")"+ params.modelName + + R"(", + "text": [[]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + +TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArrays) { + auto params = GetParam(); + std::string requestBody = R"( + { + "model": ")"+ params.modelName + + R"(", + "text": [[], [], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + INSTANTIATE_TEST_SUITE_P( LLMTokenizeTestInstances, LLMTokenizeTests, diff --git a/src/test/reranknode_test.cpp b/src/test/reranknode_test.cpp index fa19dba86e..f92f20e42c 100644 --- a/src/test/reranknode_test.cpp +++ b/src/test/reranknode_test.cpp @@ -605,3 +605,25 @@ TEST_F(RerankTokenizeHttpTest, tokenizeIgnoreAddSpecialTokensParameter) { ovms::StatusCode::OK); AssertTokenizationResult(response, expectedTokens); } + +TEST_F(RerankTokenizeHttpTest, tokenizeEmptyNestedArray) { + std::string requestBody = R"( + { + "model": "rerank_ov", + "text": [[]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + +TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { + std::string requestBody = R"( + { + "model": "rerank_ov", + "text": [[], [], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} diff --git a/src/tokenize/tokenize_parser.cpp b/src/tokenize/tokenize_parser.cpp index d6ee9dcdd1..a98aed352e 100644 --- a/src/tokenize/tokenize_parser.cpp +++ b/src/tokenize/tokenize_parser.cpp @@ -147,28 +147,32 @@ std::variant TokenizeParser::parseI InputType input_type = InputType::NONE; for (auto& input : it->value.GetArray()) { if (input.IsArray()) { + auto array = input.GetArray(); + if (array.Size() == 0) { + return "inner arrays in " + field_name + " should not be empty"; + } if (input_type != InputType::NONE && input_type != InputType::INT_VEC && input_type != InputType::STRING_VEC) return field_name + " must be homogeneous"; - if (input.GetArray()[0].IsInt()) { + if (array[0].IsInt()) { if (input_type == InputType::STRING_VEC) return field_name + " must be homogeneous"; input_type = InputType::INT_VEC; std::vector ints; - ints.reserve(input.GetArray().Size()); - for (auto& val : input.GetArray()) { + ints.reserve(array.Size()); + for (auto& val : array) { if (val.IsInt()) ints.push_back(val.GetInt()); else return field_name + " must be homogeneous"; } input_tokens.emplace_back(std::move(ints)); - } else if (input.GetArray()[0].IsString()) { + } else if (array[0].IsString()) { if (input_type == InputType::INT_VEC) return field_name + " must be homogeneous"; input_type = InputType::STRING_VEC; std::vector strings; - strings.reserve(input.GetArray().Size()); - for (auto& val : input.GetArray()) { + strings.reserve(array.Size()); + for (auto& val : array) { if (val.IsString()) strings.push_back(val.GetString()); else From 8012f74b9cb6da29f766191d89e2ec10f65a8ee9 Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 09:02:50 +0200 Subject: [PATCH 2/7] test --- src/test/embeddingsnode_test.cpp | 11 +++++++++++ src/test/llm/tokenize_endpoint_test.cpp | 18 ++++++++++++++++-- src/test/reranknode_test.cpp | 11 +++++++++++ 3 files changed, 38 insertions(+), 2 deletions(-) diff --git a/src/test/embeddingsnode_test.cpp b/src/test/embeddingsnode_test.cpp index bff7ef98db..6bbb0fa906 100644 --- a/src/test/embeddingsnode_test.cpp +++ b/src/test/embeddingsnode_test.cpp @@ -808,3 +808,14 @@ TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } + +TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { + std::string requestBody = R"( + { + "model": "embeddings_ov", + "text": [[], ["hello world"], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index 4e87c76a83..dddc473df8 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -445,7 +445,7 @@ TEST_P(LLMTokenizeTests, tokenizeEmptyNestedArray) { std::string requestBody = R"( { - "model": ")"+ params.modelName + + "model": ")" + params.modelName + R"(", "text": [[]] } @@ -458,7 +458,7 @@ TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArrays) { auto params = GetParam(); std::string requestBody = R"( { - "model": ")"+ params.modelName + + "model": ")" + params.modelName + R"(", "text": [[], [], []] } @@ -467,6 +467,20 @@ TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArrays) { ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } +TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { + auto params = GetParam(); + std::string requestBody = R"( + { + "model": ")" + params.modelName + + R"(", + "text": [[], ["hello world"], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + + INSTANTIATE_TEST_SUITE_P( LLMTokenizeTestInstances, LLMTokenizeTests, diff --git a/src/test/reranknode_test.cpp b/src/test/reranknode_test.cpp index f92f20e42c..2791cb1a46 100644 --- a/src/test/reranknode_test.cpp +++ b/src/test/reranknode_test.cpp @@ -627,3 +627,14 @@ TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } + +TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { + std::string requestBody = R"( + { + "model": "rerank_ov", + "text": [[], ["hello world"], []] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} From f846d1f80a77c7f19a5a9472ddfca08916e5e1f0 Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 09:13:05 +0200 Subject: [PATCH 3/7] style --- src/test/llm/tokenize_endpoint_test.cpp | 1 - 1 file changed, 1 deletion(-) diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index dddc473df8..ea308be657 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -480,7 +480,6 @@ TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } - INSTANTIATE_TEST_SUITE_P( LLMTokenizeTestInstances, LLMTokenizeTests, From 840f902b9c0cc0fac06abd4fb570939f06061d5e Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 09:22:22 +0200 Subject: [PATCH 4/7] test --- src/test/embeddingsnode_test.cpp | 11 +++++++++++ src/test/llm/tokenize_endpoint_test.cpp | 15 +++++++++++++++ src/test/reranknode_test.cpp | 11 +++++++++++ 3 files changed, 37 insertions(+) diff --git a/src/test/embeddingsnode_test.cpp b/src/test/embeddingsnode_test.cpp index 6bbb0fa906..0b6644db47 100644 --- a/src/test/embeddingsnode_test.cpp +++ b/src/test/embeddingsnode_test.cpp @@ -819,3 +819,14 @@ TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmp Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } + +TEST_F(EmbeddingsTokenizeHttpTest, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { + std::string requestBody = R"( + { + "model": "embeddings_ov", + "text": [[[[[]]]]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index ea308be657..c396318688 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -480,6 +480,21 @@ TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } +TEST_P(LLMTokenizeTests, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { + auto params = GetParam(); + std::string requestBody = R"( + { + "model": ")" + params.modelName + + R"(", + "text": [[[[[]]]]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} + + + INSTANTIATE_TEST_SUITE_P( LLMTokenizeTestInstances, LLMTokenizeTests, diff --git a/src/test/reranknode_test.cpp b/src/test/reranknode_test.cpp index 2791cb1a46..04d459d8b1 100644 --- a/src/test/reranknode_test.cpp +++ b/src/test/reranknode_test.cpp @@ -638,3 +638,14 @@ TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } + +TEST_F(RerankTokenizeHttpTest, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { + std::string requestBody = R"( + { + "model": "rerank_ov", + "text": [[[[[]]]]] + } + )"; + Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +} From 4b4f5bf1870e4c323eea98bf64212bf35346b19e Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 09:24:00 +0200 Subject: [PATCH 5/7] style --- src/test/llm/tokenize_endpoint_test.cpp | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index c396318688..880dabd949 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -493,8 +493,6 @@ TEST_P(LLMTokenizeTests, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); } - - INSTANTIATE_TEST_SUITE_P( LLMTokenizeTestInstances, LLMTokenizeTests, From e75b88844b23b2fa6a021621ed70680a84fc62fd Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 09:59:42 +0200 Subject: [PATCH 6/7] copilot review --- src/tokenize/tokenize_parser.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/tokenize/tokenize_parser.cpp b/src/tokenize/tokenize_parser.cpp index a98aed352e..e5dbb54620 100644 --- a/src/tokenize/tokenize_parser.cpp +++ b/src/tokenize/tokenize_parser.cpp @@ -147,7 +147,7 @@ std::variant TokenizeParser::parseI InputType input_type = InputType::NONE; for (auto& input : it->value.GetArray()) { if (input.IsArray()) { - auto array = input.GetArray(); + const auto array = input.GetArray(); if (array.Size() == 0) { return "inner arrays in " + field_name + " should not be empty"; } From 1b1259eb32a12044f1143af08b02feaac7db0828 Mon Sep 17 00:00:00 2001 From: Pawel Rzepecki Date: Thu, 14 May 2026 12:37:09 +0200 Subject: [PATCH 7/7] extract function --- src/test/embeddingsnode_test.cpp | 36 +++------------------- src/test/llm/tokenize_endpoint_test.cpp | 41 +++---------------------- src/test/reranknode_test.cpp | 36 +++------------------- src/test/test_http_utils.hpp | 16 ++++++++++ 4 files changed, 28 insertions(+), 101 deletions(-) diff --git a/src/test/embeddingsnode_test.cpp b/src/test/embeddingsnode_test.cpp index 0b6644db47..7a74d182b4 100644 --- a/src/test/embeddingsnode_test.cpp +++ b/src/test/embeddingsnode_test.cpp @@ -788,45 +788,17 @@ TEST_F(EmbeddingsTokenizeHttpTest, tokenizeBatchWithPadToMaxLen) { } TEST_F(EmbeddingsTokenizeHttpTest, tokenizeEmptyNestedArray) { - std::string requestBody = R"( - { - "model": "embeddings_ov", - "text": [[]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "embeddings_ov", "[[]]", response, comp, responseComponents, writer, multiPartParser); } TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { - std::string requestBody = R"( - { - "model": "embeddings_ov", - "text": [[], [], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "embeddings_ov", "[[], [], []]", response, comp, responseComponents, writer, multiPartParser); } TEST_F(EmbeddingsTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { - std::string requestBody = R"( - { - "model": "embeddings_ov", - "text": [[], ["hello world"], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "embeddings_ov", R"([[], ["hello world"], []])", response, comp, responseComponents, writer, multiPartParser); } TEST_F(EmbeddingsTokenizeHttpTest, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { - std::string requestBody = R"( - { - "model": "embeddings_ov", - "text": [[[[[]]]]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "embeddings_ov", "[[[[[]]]]]", response, comp, responseComponents, writer, multiPartParser); } diff --git a/src/test/llm/tokenize_endpoint_test.cpp b/src/test/llm/tokenize_endpoint_test.cpp index 880dabd949..602961c5f5 100644 --- a/src/test/llm/tokenize_endpoint_test.cpp +++ b/src/test/llm/tokenize_endpoint_test.cpp @@ -442,55 +442,22 @@ TEST_P(LLMTokenizeTests, tokenizeStringWithAddSpecialTokens) { TEST_P(LLMTokenizeTests, tokenizeEmptyNestedArray) { auto params = GetParam(); - - std::string requestBody = R"( - { - "model": ")" + params.modelName + - R"(", - "text": [[]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), params.modelName, "[[]]", response, comp, responseComponents, writer, multiPartParser); } TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArrays) { auto params = GetParam(); - std::string requestBody = R"( - { - "model": ")" + params.modelName + - R"(", - "text": [[], [], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), params.modelName, "[[], [], []]", response, comp, responseComponents, writer, multiPartParser); } TEST_P(LLMTokenizeTests, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { auto params = GetParam(); - std::string requestBody = R"( - { - "model": ")" + params.modelName + - R"(", - "text": [[], ["hello world"], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), params.modelName, R"([[], ["hello world"], []])", response, comp, responseComponents, writer, multiPartParser); } TEST_P(LLMTokenizeTests, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { auto params = GetParam(); - std::string requestBody = R"( - { - "model": ")" + params.modelName + - R"(", - "text": [[[[[]]]]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), params.modelName, "[[[[[]]]]]", response, comp, responseComponents, writer, multiPartParser); } INSTANTIATE_TEST_SUITE_P( diff --git a/src/test/reranknode_test.cpp b/src/test/reranknode_test.cpp index 04d459d8b1..057e83e6f5 100644 --- a/src/test/reranknode_test.cpp +++ b/src/test/reranknode_test.cpp @@ -607,45 +607,17 @@ TEST_F(RerankTokenizeHttpTest, tokenizeIgnoreAddSpecialTokensParameter) { } TEST_F(RerankTokenizeHttpTest, tokenizeEmptyNestedArray) { - std::string requestBody = R"( - { - "model": "rerank_ov", - "text": [[]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "rerank_ov", "[[]]", response, comp, responseComponents, writer, multiPartParser); } TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArrays) { - std::string requestBody = R"( - { - "model": "rerank_ov", - "text": [[], [], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "rerank_ov", "[[], [], []]", response, comp, responseComponents, writer, multiPartParser); } TEST_F(RerankTokenizeHttpTest, tokenizeMultipleEmptyNestedArraysAndOneNonEmpty) { - std::string requestBody = R"( - { - "model": "rerank_ov", - "text": [[], ["hello world"], []] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "rerank_ov", R"([[], ["hello world"], []])", response, comp, responseComponents, writer, multiPartParser); } TEST_F(RerankTokenizeHttpTest, tokenizeEmptyWithArrayMultipleLevelsOfNesting) { - std::string requestBody = R"( - { - "model": "rerank_ov", - "text": [[[[[]]]]] - } - )"; - Status status = handler->dispatchToProcessor(endpointTokenize, requestBody, &response, comp, responseComponents, writer, multiPartParser); - ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); + assertTokenizeWithInvalidTextReturnsError(handler.get(), "rerank_ov", "[[[[[]]]]]", response, comp, responseComponents, writer, multiPartParser); } diff --git a/src/test/test_http_utils.hpp b/src/test/test_http_utils.hpp index 62ca393c40..b59214dbf0 100644 --- a/src/test/test_http_utils.hpp +++ b/src/test/test_http_utils.hpp @@ -14,6 +14,7 @@ // limitations under the License. //***************************************************************************** #pragma once +#include #include #include #include @@ -89,3 +90,18 @@ class V3HttpTest : public ::testing::Test { handler.reset(); } }; + +inline void assertTokenizeWithInvalidTextReturnsError( + ovms::HttpRestApiHandler* handler, + const std::string& modelName, + const std::string& textJson, + std::string& response, + ovms::HttpRequestComponents& comp, + ovms::HttpResponseComponents& responseComponents, + std::shared_ptr& writer, + std::shared_ptr& multiPartParser) { + std::string requestBody = R"({"model": ")" + modelName + R"(", "text": )" + textJson + R"(})"; + std::string endpoint = "/v3/tokenize"; + ovms::Status status = handler->dispatchToProcessor(endpoint, requestBody, &response, comp, responseComponents, writer, multiPartParser); + ASSERT_EQ(status, ovms::StatusCode::MEDIAPIPE_EXECUTION_ERROR) << status.string(); +}