Test case generation method, device and equipment, and storage medium

Abstract

The application provides a test case generation method and device, equipment and a storage medium, and relates to the technical field of testing. The test case generation method comprises: performing structural analysis on a test plan document through an intelligent model to obtain a plurality of cause-effect relationships contained in the document; classifying conditions in the plurality of cause-effect relationships to generate a classification result; retrieving historical test information matched with the test plan document from a knowledge base; generating initial test points through the intelligent model according to the historical test information, the classification result and the test plan document; performing self-inspection on the initial test points through the intelligent model to determine test points; and generating test cases through the intelligent model according to the test points and the test plan document. Short-term memory for the test plan document is constructed, the knowledge base provides long-term memory for the intelligent model, and the initial test points are self-inspected, so that the omission and redundancy of the test points are reduced, and the quality stability of the generated test cases is improved.

Description

Technical Field

[0001] This application relates to the field of testing technology, and more specifically, to a test case generation method, apparatus, device, and storage medium. Background Technology

[0002] Testing is typically based on test cases, and standardizing these test cases reduces the difficulty of execution and the cost of integration. With the development of AI (Artificial Intelligence) technology, the use of AI to generate test cases has also received widespread attention.

[0003] In related technologies, AI is used to assist humans in quickly and in batches to generate and iterate test case text, and it can generate sufficiently standardized test cases.

[0004] However, in related technologies, the inability to understand test planning documents can easily lead to redundancy and omissions in the generated test cases, resulting in unstable quality of the generated test cases. Summary of the Invention

[0005] The purpose of this application is to provide a test case generation method, apparatus, device, and storage medium to address the shortcomings of the prior art and solve the aforementioned technical problems.

[0006] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows: In a first aspect, embodiments of this application provide a test case generation method, the method comprising: The test planning document is structured and analyzed by an intelligent model to obtain multiple causal relationships contained in the document. These causal relationships are used to characterize the conditions to be tested and the corresponding results. Classify the conditions in the multiple causal relationships and generate classification results; Retrieve historical test information from the knowledge base that matches the test planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model. The intelligent model performs introspection checks on the initial test points to determine the test points. Test cases are generated using the intelligent model based on the test points and the test planning document.

[0007] Secondly, embodiments of this application also provide a test case generation apparatus, the apparatus comprising: The analysis module is used to perform structured analysis on the test planning document through an intelligent model to obtain multiple causal relationships contained in the document, wherein the causal relationships are used to characterize the conditions to be tested and the corresponding results. The classification module is used to classify the conditions in multiple causal relationships and generate classification results; The retrieval module is used to retrieve historical test information that matches the test planning document from the knowledge base; The first generation module is used to generate initial test points through the intelligent model based on the historical test information, the classification results, and the test planning document. The inspection module is used to perform introspection inspection on the initial test points through the intelligent model to determine the test points; The second generation module is used to generate test cases based on the test points and the test planning document using the intelligent model.

[0008] Thirdly, embodiments of this application also provide a test case generation device, including: a memory and a processor, wherein the memory stores a computer program executable by the processor, and the processor executes the computer program to implement the test case generation method described in any of the first aspects above.

[0009] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when read and executed, implements the test case generation method described in any of the first aspects above.

[0010] The beneficial effects of this application are as follows: This application provides a test case generation method, which may include: performing structured analysis on a test planning document using an intelligent model to obtain multiple causal relationships contained in the document, wherein the causal relationships are used to characterize the conditions to be tested and the corresponding results; classifying the conditions in the multiple causal relationships to generate classification results; retrieving historical test information matching the test planning document from a knowledge base; generating initial test points using the intelligent model based on the historical test information, classification results, and the test planning document; performing introspection checks on the initial test points using the intelligent model to determine the test points; and generating test cases using the intelligent model based on the test points and the test planning document. By performing structured analysis on the test planning document using an intelligent model, a short-term memory for the test planning document can be built. Retrieving historical test information matching the test planning document from the knowledge base provides long-term memory for the intelligent model. Furthermore, the intelligent model performs introspection checks on the initial test points, which reduces the omission and redundancy of test points, thereby avoiding redundancy and omission of test cases and improving the quality and stability of the generated test cases. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 1 ; Figure 2 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 2 ; Figure 3 A schematic diagram illustrating the retrieval of a first historical planning document and a first historical test point in a knowledge base, provided as an embodiment of this application; Figure 4 This is a schematic diagram of a test point generation process provided in an embodiment of this application; Figure 5 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 3 ; Figure 6 A schematic diagram illustrating the generation process of a second historical planning document and a first historical test case, provided for embodiments of this application; Figure 7 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 4 ; Figure 8 A schematic diagram illustrating the test case generation process provided in an embodiment of this application; Figure 9 This is a schematic diagram illustrating the process of dividing multiple test points according to an embodiment of this application. Figure 10 A schematic diagram illustrating the process of dividing multiple causal relationships as provided in an embodiment of this application; Figure 11 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 5 ; Figure 12 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 6 ; Figure 13 Historical thinking guide provided for the embodiments of this application Figure 1 A diagram illustrating the conversion of keys into standardized test cases; Figure 14A flowchart illustrating a test case generation method provided in this application embodiment. Figure 7 ; Figure 15 A schematic diagram of the input iteration content interface provided in an embodiment of this application; Figure 16 A schematic diagram of the iterative test cases provided in the embodiments of this application; Figure 17 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 8 ; Figure 18 A schematic diagram of a test planning image provided for an embodiment of this application; Figure 19 This is a schematic diagram of a character recognition result provided in an embodiment of this application; Figure 20 A schematic diagram of a test case provided in an embodiment of this application; Figure 21 A schematic diagram illustrating the training process of an intelligent model provided in an embodiment of this application; Figure 22 A schematic diagram of a third historical test case provided in an embodiment of this application; Figure 23 A schematic diagram illustrating a preloading process provided in an embodiment of this application; Figure 24 A schematic diagram of another test case provided in an embodiment of this application; Figure 25 A schematic diagram illustrating a supplementary and improved test case provided for an embodiment of this application; Figure 26 A schematic diagram of a memory list provided for an embodiment of this application; Figure 27 A schematic diagram illustrating single-line test case completion provided in an embodiment of this application; Figure 28 A schematic diagram illustrating the framework of a test case generation method provided in an embodiment of this application; Figure 29 This is a schematic diagram of the structure of a test case generation device provided in an embodiment of this application; Figure 30 This is a schematic diagram of the structure of a test case generation device provided in an embodiment of this application. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of this application, but not all embodiments.

[0014] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0015] In the description of this application, it should be noted that if the terms "upper", "lower", etc. appear to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of this application is usually placed in, it is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0016] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] It should be noted that, where there is no conflict, the features in the embodiments of this application can be combined with each other.

[0018] This application provides a test case generation method, which is applied to a test case generation device. The test case generation device can be a terminal device or a server. The terminal device can be any of the following: computer device, laptop computer, tablet computer, or smartphone.

[0019] The following explains a test case generation method provided by an embodiment of this application.

[0020] Figure 1 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 1 ;like Figure 1 As shown, the method may include: S101. The test planning document is structured and analyzed using an intelligent model to obtain multiple causal relationships contained in the document.

[0021] In this context, causal relationships are used to characterize the conditions to be tested and the corresponding results.

[0022] In some implementations, a structured document is generated based on the test planning document. An intelligent model is then used to analyze and understand the structured document by mimicking QA (Quality Assurance) thinking, sorting out the logical relationships in the structured document, and obtaining multiple causal relationships contained in the document.

[0023] It should be noted that the intelligent model can be an AI (Artificial Intelligence) model. AI uses reasoning to mimic the structured thinking of QA (Quality Assurance) systems to understand test planning documents and builds a short-term memory for understanding these documents.

[0024] S102. Classify the conditions in multiple causal relationships and generate classification results.

[0025] In this embodiment of the application, the conditions in multiple causal relationships are classified. Since each condition has a corresponding result, the classification result includes: the classified conditions and the corresponding results.

[0026] It is worth noting that conditions in multiple causal relationships can be classified sequentially or simultaneously, and this application does not impose specific limitations on this.

[0027] S103. Retrieve historical test information from the knowledge base that matches the test planning document.

[0028] It is important to note that in this embodiment, a knowledge base is incorporated to provide long-term memory for the intelligent model. When generating test points and test cases, the intelligent model not only relies on the test planning document but can also leverage the long-term memory in the knowledge base to mimic the characteristics of project test cases and find relevant and valuable historical test information. This significantly improves the performance ceiling of AI.

[0029] S104. Based on historical test information, classification results, and test planning documents, generate initial test points through an intelligent model.

[0030] The initial test points include: initial equivalence classes and initial boundary values.

[0031] In this embodiment, the classification results are traversed, and initial test points are generated through an intelligent model based on historical test information and test planning documents. Here, based on the classification results, the clarity of thought when the intelligent model generates test points and the structural orderliness when generating test cases are greatly enhanced, significantly reducing redundancy and omissions in subsequent steps.

[0032] S105. The initial test points are checked by the intelligent model to determine the test points.

[0033] In some implementations, the self-checking process is a process of identifying and filling gaps. The initial test points are deduplicated and missing test points are supplemented by an intelligent model to obtain the test points.

[0034] S106. Based on the test points and test planning documents, generate test cases using the intelligent model.

[0035] In some implementations, test cases are generated through an intelligent model based on equivalence classes and boundary values ​​in the test points, test planning documents, classification results, and a knowledge base.

[0036] In practical applications, this test case can be executed to perform the testing process.

[0037] In summary, this application provides a test case generation method, which may include: performing structured analysis on a test planning document using an intelligent model to obtain multiple causal relationships contained in the document, wherein the causal relationships are used to characterize the conditions to be tested and the corresponding results; classifying the conditions in the multiple causal relationships to generate classification results; retrieving historical test information matching the test planning document from a knowledge base; generating initial test points using the intelligent model based on the historical test information, classification results, and the test planning document; performing introspection checks on the initial test points using the intelligent model to determine the test points; and generating test cases using the intelligent model based on the test points and the test planning document. By performing structured analysis on the test planning document using an intelligent model, a short-term memory of the test planning document can be built. Retrieving historical test information matching the test planning document from the knowledge base provides long-term memory for the intelligent model. Furthermore, the intelligent model performs introspection checks on the initial test points, which reduces the omission and redundancy of test points, thereby avoiding redundancy and omission of test cases and improving the quality and stability of the generated test cases.

[0038] Optionally, Figure 2 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 2 ,like Figure 2 As shown, the process of retrieving historical test information from the knowledge base that matches the test planning document in S103 above may include: S201. Retrieve from the knowledge base the first historical planning document that matches the test planning document, and the first historical test point corresponding to the first historical planning document.

[0039] Among them, the first historical planning document and the test planning document are similar documents, and the number of first historical test points corresponding to each first historical planning document can be multiple.

[0040] Figure 3 This application provides an embodiment of a schematic diagram illustrating the retrieval of a first historical planning document and a first historical test point from a knowledge base. Figure 3 As shown, based on the test planning document, the first historical planning document and the corresponding first historical test points are retrieved from the knowledge base. The test planning document is the planning document for "Zhen Luo Dynamic Portrait," and the first historical planning documents include the planning documents for "Zhuge Dynamic Portrait" and "Liu Bei Dynamic Portrait." It should be understood that... Figure 3 The test planning document and the first historical planning document shown are merely examples, as are the number of first historical planning documents and the corresponding number of first historical test points. This application embodiment does not impose any specific limitations on these.

[0041] In addition, the knowledge base includes: multiple historical test documents, and the historical test points corresponding to each historical test document.

[0042] The process in S104 above, which generates initial test points using an intelligent model based on historical test information, classification results, and test planning documents, may include: S202. Based on the first historical planning document, the first historical test points, the classification results, and the test planning document, generate initial test points through the intelligent model.

[0043] The historical test information includes the first historical planning document and the first historical test point.

[0044] Figure 4 This is a schematic diagram of a test point generation process provided in an embodiment of this application, such as... Figure 4 As shown, an intelligent model is used to traverse the classification results, generating initial test points based not only on the test planning document but also on long-term memories such as the first historical planning document and the first historical test point in the knowledge base.

[0045] Optionally, the process in S105 above, which uses an intelligent model to perform introspection checks on the initial test points to determine the test points, may include: The intelligent model performs deduplication and supplementation on the initial test points according to the preset prompts to obtain the test points.

[0046] The preset prompt requirement can be a preset prompt requirement.

[0047] In some implementations, such as Figure 4As shown, the intelligent model removes duplicate initial test points and adds missing test points according to preset prompt requirements, resulting in the final test points. This ensures that the generated test points are neither duplicated nor missing.

[0048] In practical applications, the Prompt requirements can be fine-tuned, providing the intelligent model with the basic ability to divide test points (equivalence classes and boundary values), ensuring that no information in the document is omitted, and providing project-specific knowledge through a knowledge base to generate relevant test cases based on the private knowledge, thereby increasing the upper limit of the intelligent model's capabilities.

[0049] Optionally, Figure 5 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 3 ,like Figure 5 As shown, the process of generating test cases using an intelligent model based on test points and test planning documents in S106 above may include: S301. Retrieve from the knowledge base a second historical test plan document that matches the test plan document and test points, as well as the first historical test cases and second historical test points corresponding to the second historical test plan document.

[0050] Figure 6 A schematic diagram illustrating the generation process of a second historical planning document and a first historical test case, provided for embodiments of this application, is shown below. Figure 6 As shown, based on the test planning documents and test points, a search is performed in the knowledge base to obtain the second historical planning document and the corresponding first historical test cases and second historical test points. Among these, the second historical planning documents are the planning documents for "Zhuge Liang's Dynamic Portrait" and "Liu Bei's Dynamic Portrait". Figure 6 The test planning document and the second historical planning document shown are merely examples. The number of second historical planning documents and the corresponding number of first historical test cases and second historical test points are also merely examples. This application embodiment does not impose specific limitations on these.

[0051] S302. Based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document, generate test cases through the intelligent model.

[0052] In this embodiment of the application, when generating test cases, the test case writing ideas in the first historical test cases are found based on the test points and test planning documents, the QA thinking is imitated, and the equivalence class or boundary value partitioning experience of the second historical parameter points corresponding to the second historical planning documents is referenced to generate test cases.

[0053] Optionally, Figure 7 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 4 ,like Figure 7 As shown, the process of generating test cases through an intelligent model based on the second historical test planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document in S302 above may include: S401. Generate initial prerequisites based on the second historical planning document, the second historical test points, the first historical test cases, the classification results and test points, and the test planning document.

[0054] Among them, an intelligent model can be used to initially divide and generate initial preconditions, and there can be multiple initial preconditions.

[0055] S402. Based on the preset prompts, the initial prerequisites are deduplicated and supplemented to obtain the prerequisites.

[0056] In some implementations, the preset prompt requirement is the Prompt requirement. The intelligent model removes duplicate preconditions from the initial preconditions and adds missing preconditions based on the Prompt requirement, resulting in multiple preconditions.

[0057] In this embodiment, the Prompt requirement is a fine-tuned Prompt requirement. By fine-tuning the Prompt requirement, the capabilities of the intelligent model are further explored, and the accuracy of test case generation is improved.

[0058] S403. Generate the operation steps and expected results corresponding to the prerequisites.

[0059] It should be noted that corresponding operation steps and expected results need to be generated for each precondition. The operation steps and expected results for each precondition can be generated sequentially, or they can be generated simultaneously. This application embodiment does not impose specific limitations on this.

[0060] S404. Based on the preconditions and the corresponding operation steps and expected results, generate test cases through the intelligent model.

[0061] Figure 8 This is a schematic diagram illustrating the test case generation process provided in an embodiment of this application, as shown below. Figure 8 As shown, an intelligent model is used to initially generate multiple preconditions based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the planning document, and the test points. Among these, the test points... Figure 8 The values ​​are not shown in the diagram. Multiple initial preconditions are deduplicated and supplemented to obtain the preconditions, and the corresponding operation steps and expected results are generated.

[0062] Optionally, the process of generating test cases using an intelligent model based on test points and test planning documents in S106 above may include: Multiple test points are divided into multiple test point groups, with some overlapping test points between adjacent test point groups; test cases are generated based on the multiple test point groups and the test planning document.

[0063] In this embodiment, generating complex test cases using an intelligent model results in slow generation speed and generates numerous test case entries, leading to significant omissions. This embodiment addresses this by dividing multiple test points into multiple groups using a sliding window approach. A preset number of test points overlap between adjacent groups, and semantic connections between adjacent groups reduce redundancy while ensuring logical continuity. For example, the preset number can be 2, but this can be set based on actual needs; this embodiment does not impose specific limitations on this.

[0064] Figure 9 This application provides a schematic diagram of a process for dividing multiple test points, as shown in the embodiments of the present application. Figure 9 As shown, there are 9 test points. Group 1 includes 4 test points, Group 2 includes 3 test points, and Group 3 includes 2 test points. The 4 test points in Group 1 and 1 test point in Group 2 are used to generate one test case. The 1 test point in Group 1, 3 test points in Group 2 and 1 test point in Group 3 are used to generate one test case. The 1 test point in Group 2 and 2 test points in Group 3 are used to generate one test case.

[0065] Optionally, the process of performing structured analysis on the test planning document using an intelligent model to obtain multiple causal relationships contained in the document, as described in S101 above, may include: By using intelligent models to perform structured analysis on test planning documents, user behaviors in the test planning documents are divided into multiple causal relationships.

[0066] In this context, causality is used to characterize a user's behavior or the constraints they are subject to, and the corresponding results or outcomes. "User" refers to a player.

[0067] Figure 10 This application provides a schematic diagram of a process for dividing multiple causal relationships, as illustrated in the embodiments of this application. Figure 10 As shown, a corresponding structured document is generated based on the test planning document. This structured document includes multiple conditions and their corresponding results. Analysis and understanding of the structured document reveal multiple causal relationships. The test planning document is the relevant planning document for "Zhenluo 3D Dynamic Portrait". Figure 10The test planning document, structured document, and specific content in multiple causal relationships are examples, and this application does not impose specific limitations on them.

[0068] In summary, this embodiment of the application, by imitating the testing mindset of QA through an intelligent model, proceeds step by step. First, it builds a short-term memory of the test planning document through structured analysis. Then, it forms a long-term memory of project knowledge and general QA knowledge through knowledge base retrieval. Finally, a prompt teaches the intelligent model how to write test cases and how to identify and fill gaps in the test. This series of thought processes significantly reduces redundancy, unleashes the potential of the intelligent model, and significantly improves the accuracy of test case generation.

[0069] Figure 11 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 5 ,like Figure 11 As shown, the process in S201 above, which retrieves the first historical test point corresponding to the first historical planning document that matches the test planning document from the knowledge base, may include: S501. Use a hybrid retrieval method to retrieve multiple third-party historical planning documents that are similar to the test planning document from the knowledge base.

[0070] The knowledge base includes a public knowledge base, a private knowledge base, an incident database, and a configured checklist. These knowledge bases work together to provide the intelligent model with QA and project knowledge. The public knowledge base includes handwritten planning documents and corresponding standardized test cases for projects that have never used standardized test cases before. The private knowledge base includes project-specific planning documents (automatically generated when missing) and corresponding test cases, allowing the intelligent model to learn and imitate the project's private knowledge and habits. After generating test points, the intelligent model checks the incident database and checklist, supplementing any relevant test points that may need to be tested.

[0071] In some implementations, a hybrid retrieval method combining dense and sparse retrieval is used to retrieve multiple third-party historical planning documents similar to the test planning document from the knowledge base. Dense retrieval is based on overall semantics, while sparse retrieval is based on proper nouns.

[0072] S502. Sort the multiple third-party historical planning documents according to the similarity between each third-party historical planning document and the test planning document, as well as the quality of the test cases corresponding to each third-party historical planning document.

[0073] S503. From the sorted third historical planning documents, determine the first historical planning document.

[0074] In some implementations, multiple third-party historical planning documents are reordered. Based on the similarity between each third-party historical planning document and the test planning document, as well as the quality of the test cases corresponding to each third-party historical planning document, a comprehensive score is given to each third-party historical planning document to obtain a score corresponding to each third-party historical planning document. The third-party historical planning document with a score greater than or equal to a preset score threshold among the sorted third-party historical planning documents is placed as the first historical planning document.

[0075] The knowledge base retrieval strategy provided in this application embodiment can improve the recall rate of multiple third-party historical planning documents and ensure that low-quality test cases are filtered out, thereby enhancing the effectiveness of the knowledge base.

[0076] S504. In the knowledge base, retrieve the first historical test point corresponding to the first historical planning document.

[0077] It should be noted that each historical planning document in the knowledge base has a corresponding historical test point. Therefore, after identifying the first historical planning document in the knowledge base, the corresponding first historical test point can be retrieved.

[0078] Optionally, Figure 12 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 6 ,like Figure 12 As shown, the method may further include: S601. Obtain the second historical test case.

[0079] The second set of historical test cases were either manually written or derived from mind maps; therefore, there is no corresponding planning document for these test cases. The mind maps can be created using Xmind.

[0080] S602. Determine whether the second historical test case contains historical preconditions, historical operation steps, and historical expected results, and determine whether the second historical test case meets the preset conditions.

[0081] In this embodiment of the application, it is determined whether the second historical test cases all contain historical preconditions, historical operation steps, and historical expected results, that is, whether the content contained in the second historical test cases is complete. Simultaneously, it is determined whether the second historical test cases meet preset conditions.

[0082] S603. If the second historical test case contains historical preconditions, historical operation steps and historical expected results, and the second historical test case meets the preset conditions, then the fourth historical planning document corresponding to the second historical test case is generated according to the second historical test case and the preset synthesis prompt requirements.

[0083] If the second historical test case contains historical preconditions, historical operation steps, and historical expected results, and the second historical test case meets the preset conditions, then the second historical test case meets the conditions for automatic storage.

[0084] In some implementations, an intelligent model learns patterns from a private knowledge base for generating design document fragments based on test cases. This intelligent model then reverse-engineers the fourth historical design document fragment based on the second historical test cases. Since the test cases contain some testing information, a Prompt is used during the generation of the fourth historical design document fragment to instruct the intelligent model not to synthesize testing information from the test cases, such as GM commands (commands with specific permissions in the game, usable only by game administrators or developers), test steps, etc.

[0085] S604. Update the knowledge base based on the second historical test cases and the fourth historical planning document.

[0086] In this embodiment, second historical test cases, either manually compiled or converted from mind maps, can be updated to the knowledge base. The test planning documents and corresponding generated test cases from S101 to S106 can also be updated to the knowledge base. Specifically, when the preconditions, operation steps, and expected results of a test case are complete, and the test case meets preset conditions, the test planning documents and corresponding generated test cases from S101 to S106 are updated to the knowledge base.

[0087] Optionally, the second historical test case meets preset conditions, including: If the second historical test case passes the review, or if the execution coverage of the second historical test case is greater than the preset coverage, or if the second historical test case is a test case obtained by converting a historical mind map, then the test case is determined to meet the preset conditions.

[0088] The preset coverage rate can be set according to timing requirements; for example, the preset coverage rate can be 50%. Test cases converted from historical mind map test cases can be converted from Xmind test cases to 2.0 test cases, i.e., historical executable test cases. 2.0 test cases are standard test cases, while Xmind test cases are in 1.0 format.

[0089] It should be noted that the determination of whether the second historical test case has completed the review process is based on whether the second historical test case has passed the review.

[0090] In summary, in this embodiment of the application, by satisfying the automatic entry conditions and reverse-engineering the fourth historical planning document fragment, the private knowledge base can be kept updated and continuously learn project knowledge during the use of the testing tool.

[0091] In practical applications, when analyzing requirements and writing use cases, it is often necessary to refer to historical incidents and project-specific checklists. Previously, this typically relied on manual recall and checking, which was prone to omissions. This application's embodiment obtains incident database data and provides a checklist editing function, using an intelligent model to aid note-taking and reduce omissions.

[0092] In some implementations, the data from the incident database and check list are first preprocessed into test case templates through data engineering and Prompt engineering, and feature statements are synthesized (e.g., when encountering A, B should be considered). During retrieval, relevant entries are recalled through feature statements, and large models are used for judgment and deduplication. Finally, relevant test cases are generated.

[0093] The implementation consists of the following four steps: First, an intelligent model extracts knowledge information from the incident database and the Checklist written by QA, and constructs test case templates that can detect corresponding problems. Second, using the same method as the knowledge base retrieval strategy, test case templates related to the current planning document are recalled through feature statements during each test case generation process. Third, AI excludes test case templates similar to those generated from the planning document to avoid generating duplicate test cases. Finally, based on the test case templates and the descriptions in the planning document, test cases are generated for specific incidents or Checklists.

[0094] In summary, in this embodiment of the application, an intelligent model combined with multiple knowledge bases automatically retrieves historical valid test cases, historical incidents, and check lists, intelligently supplements test cases with high relevance to requirements, and introduces a knowledge base self-updating mechanism. By synthesizing data, the completeness and accuracy of the knowledge base are continuously optimized, enhancing the intelligent model's test knowledge and project understanding. The generated test cases are closer to the characteristics and habits of the project, reducing omissions.

[0095] The process of obtaining the second historical test case in S601 above may include: obtaining the historical mind map and the corresponding preset description characters; replacing the historical mind map with preset placeholders; and generating the second historical test case by combining the preset placeholders, preset description characters, and historical mind map.

[0096] The historical mind map is in the format of the 1.0 test case, while the standardized test cases are in the format of the 2.0 test cases. There is a significant difference in the isolation between the historical mind map and the standardized test cases.

[0097] In some implementations, the one-click test case conversion function provided by the intelligent model, based on Few-Shot Prompt, significantly improves the stability of the intelligent model's format conversion. Specifically, for historical mind maps, we first replace them with placeholders ${image:index}, allowing the intelligent model to retain these placeholders in their corresponding positions during conversion. After conversion, we restore the original images to generate a second historical test case, achieving near-lossless test case conversion.

[0098] In summary, the one-click conversion of use cases for intelligent models provided in this application embodiment can convert the format with one click while ensuring that the semantics remain unchanged, which significantly reduces migration costs.

[0099] Figure 13 Historical thinking guide provided for the embodiments of this application Figure 1 A diagram illustrating the conversion of keys into standardized test cases, as shown below. Figure 13 As shown, the mind map includes: filling in a message, specific word count conditions, and word count thresholds. In the converted standardized test cases, the format has changed, but the underlying implementation logic remains the same.

[0100] It should be noted that the intelligent model can convert historical mind map use cases into standardized use cases with one click, enhancing the reusability of historical use cases, significantly reducing the cost of manual rewriting, and assisting in the implementation of standardization and BPO (Business Process Outsourcing Testing).

[0101] Optionally, Figure 14 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 7 ,like Figure 14 As shown, the method may further include: S701. Obtain the test cases to be modified.

[0102] In practical applications, there are not only scenarios where test cases are written, but also scenarios where test cases are continuously iterated and improved based on changes in requirements.

[0103] S702. Based on the iteration content, determine the target node to be modified from multiple nodes of the test case to be modified, and add a mark to the target node.

[0104] In some implementations, an intelligent model is used to perform fine-grained difference analysis. Based on the test cases to be modified, the test cases to be modified and the iteration content are compared. Each node in the test cases to be modified is judged one by one to determine whether it needs to be modified, thereby obtaining the target node to be modified. The target node to be modified is then marked.

[0105] Among them, the target node to be modified is marked based on the intelligent tag classification method to improve the iterative test cases. Different modification methods correspond to different tags.

[0106] S703. Modify the target node after adding the mark according to the iteration content to obtain the test cases after iteration.

[0107] In some implementations, the target node after adding the tag is modified according to the iteration content and the tag type to obtain the iterated test cases.

[0108] It should be noted that in related technologies, for scenarios where game rules are frequently iterated, manually iterating historical test cases requires repeatedly spending a significant amount of time. However, in this embodiment, the system incorporates the prompt functionality of an intelligent model. The test case system provides an iteration window, allowing testers to inform the intelligent model of the latest rules or even requirements. The intelligent model can then adjust the entire set of historical test cases with a single click, based on the iteration content. This intelligent model assists manual iteration, improving the efficiency of deleting, modifying, and querying scenarios.

[0109] Optionally, the process of adding a mark to the target node in S702 above may include adding any one of the following marks to the target node: revision mark, deletion mark, and addition mark.

[0110] The process described above, S703, which modifies the target node after adding the marker based on the iteration content to obtain the iterated test cases, may include: Revise the target node after adding the revision mark according to the iteration content to obtain the test cases after the iteration; or delete the target node after adding the deletion mark according to the iteration content to obtain the test cases after the iteration; or add the target node after adding the addition mark according to the iteration content to obtain the test cases after the iteration.

[0111] In practical applications, the modifications to each target node to be modified can be displayed on the interface. The modifications will only be written into the use case once the user confirms the intelligent model's revision of the use case on the front end.

[0112] In this embodiment, three types of markers—revision marker, deletion marker, and addition marker—are set on the target node to be modified, which can efficiently and reliably achieve the iteration of test cases.

[0113] Figure 15 This is a schematic diagram of the input iteration content interface provided in the embodiments of this application, as shown below. Figure 15 As shown, the input iteration content interface displays an input box, which is used to input the iteration content. The iteration content is used to represent changing the original content to the new content. Figure 16A schematic diagram of the iterative test cases provided in the embodiments of this application, as shown below. Figure 16 As shown, all content related to the iteration in the test cases to be modified has been marked and iterated.

[0114] In summary, in this embodiment of the application, through simple prompts and documentation, the intelligent model can combine iterative content to adjust the test cases to be modified with one click, significantly reducing the workload of writing routine activity test cases and regression test cases.

[0115] Optionally, Figure 17 A flowchart illustrating a test case generation method provided in this application embodiment. Figure 8 ,like Figure 17 As shown, before analyzing the logical relationships in the test planning document and obtaining multiple causal relationships in S101 above, the method may further include: S801. Obtain the test design image. The test design image is used to represent the design content.

[0116] In practical applications, the requirements related to planning content are multimodal, ranging from relatively standard planning documents, flowcharts, and tables to non-standard meeting minutes or chat screenshots. A single model is not well-suited for integrating multimodal information.

[0117] S802. Using the optical character recognition technology in the preset hybrid model, the test planning image is used to perform text recognition to obtain the initial text recognition result.

[0118] Among them, optical character recognition technology is OCR technology. Guided by optical character recognition technology, the content in the test planning image can be fully covered, and multimodal information can be well integrated.

[0119] S803. Using the large language model in the preset hybrid model, based on the recognition prompts in the thought chain and the initial text recognition results, the text recognition results are obtained by performing text recognition on the test planning image.

[0120] S804. Generate a test planning document based on the text recognition results.

[0121] Among them, the large language model is the LLM model.

[0122] In this embodiment, an LLM model is used to perform text recognition on test planning images based on the recognition prompts in the thought process and the initial text recognition results. This can analyze the complex information in test planning images such as flowcharts, chat logs, and tables.

[0123] It's important to note that multimodal test planning images, such as meeting minutes, chat screenshots, planning documents, flowcharts, and tables, can all be recognized using optical character recognition (OCR) technology combined with a large language model and a thought chain approach. The large language model, guided by the initial text recognition results from OCR, comprehensively covers the content of the test planning images, effectively integrating multimodal information. Combined with the thought chain feature, the intelligent model can analyze complex information such as flowcharts, chat logs, and tables, resulting in more accurate test case generation.

[0124] In addition, the performance of OCR technology, large language models, and combinations of OCR technology and large language models in multiple dimensions such as semantic coherence, text recall, image semantic analysis, and information integration can be seen in Table 1.

[0125] Table 1

[0126] Figure 18 A test planning diagram provided for an embodiment of this application, such as... Figure 18 As shown, the test plan image is a descriptive image related to the growth of construction. Figure 19 This is a schematic diagram of a text recognition result provided in an embodiment of this application, such as... Figure 19 As shown, for Figure 18 The test plan image shown is identified using the process described in S803 to S803 above, resulting in... Figure 19 The text recognition results. Figure 20 A schematic diagram of a test case provided in an embodiment of this application is shown below. Figure 20 As shown, based on Figure 19 The text recognition results shown generate a test planning document, and the test cases corresponding to this test planning document are... Figure 20 The test cases shown.

[0127] In this embodiment, Sonnet-3.7 possesses strong instruction compliance capabilities, accurately generating reliable test cases based on instructions. However, its reasoning ability for complex text, especially complex Chinese text, is inferior to Deepseek-R1. Therefore, in the test case generation thought process, Deepseek-R1 is used for the structured understanding of the test planning document, leveraging its powerful Chinese text analysis and reasoning capabilities to structurally understand the causal relationships in complex documents; Sonnet is used for generating test points and test cases, utilizing its strong instruction compliance capabilities to generate compliant test cases. This hybrid model works collaboratively to improve the quality of test case generation.

[0128] Optionally, the process of generating initial test points through an intelligent model based on historical test information, classification results, and test planning documents in S104 above may include: using an intelligent model to generate test points one by one based on classification results and test planning documents; In S106 above, test cases are generated through an intelligent model based on test points and test planning documents. This includes: using an intelligent model, generating preconditions, operation steps, and expected results item by item based on test points and test planning documents to obtain test cases.

[0129] The intelligent model is a pre-trained, private intelligent model.

[0130] Optionally, Figure 21 This is a schematic diagram of the training process of an intelligent model provided in an embodiment of this application, such as... Figure 21 As shown, the intelligent model is trained using the following method: S901. Retrieve from the knowledge base the fourth historical planning document and the corresponding third historical test case that match the preset sample test planning document and the corresponding preset sample test case.

[0131] The knowledge base includes multiple historical planning documents, as well as corresponding historical test points and historical test cases.

[0132] S902. Replace some information in the third historical test case with preset information to obtain the replaced test case.

[0133] In some implementations, some information in the randomly determined third historical test cases is replaced with preset information. This can improve the model's generalization ability to different context lengths. The points in the randomly determined third historical test cases are replaced with preset information. These points include test points, preconditions, operation steps, or expected results. Here, the points in the third historical test cases can be replaced with default values ​​or semantically coherent words, which can improve the model's generalization ability to different contexts.

[0134] Figure 22 This is a schematic diagram of a third historical test case provided in an embodiment of this application, as shown below. Figure 22 As shown, replace "Server list displays normally" in the third historical test case with "List # needs to be completed here", and set the assistant information to "Server list displays normally".

[0135] S903. Iterate the fourth historical planning document through the preset large language model to obtain the iterated historical planning document.

[0136] S904. Replace some information in the iterative test cases corresponding to the iterative planning document after iteration with preset information to obtain the iterative test cases after replacement.

[0137] In this process, iterative test cases are generated based on the iterative historical planning documents. Then, some information in the iterative test cases corresponding to the iterative historical planning documents is replaced with preset information to obtain the replaced iterative test cases.

[0138] It should be noted that, in the embodiments of this application, for the scenario of iterating item by item, we use an intelligent model to automatically and randomly generate iterative requirement documents, and let the large model use the random iterative requirement documents to iterate the existing use cases to obtain iterative test cases. The replaced iterative test cases are used as the training set to train the small model, so that the small model learns the knowledge of the large model and realizes knowledge distillation.

[0139] S905. Based on the third historical test cases, the replaced test cases, the iterative test cases, and the replaced iterative test cases, train the initial model until the preset termination condition is met to obtain the intelligent model.

[0140] In this embodiment, an initial model is used to predict preset information based on the replaced test cases to obtain first predicted information. The model parameters of the initial model are then updated using the first predicted information and the replaced information from the third historical test cases. Similarly, the initial model is used to predict preset information based on the replaced iterative test cases to obtain second predicted information. The model parameters of the initial model are then updated using the second predicted information and the replaced information from the iterative test cases, until a preset termination condition is met, resulting in an intelligent model.

[0141] In this embodiment, the completion latency has been reduced to the second level during the initial model training process. However, the ideal latency for a single completion should be at the millisecond level so that the user will not perceive it. Therefore, this embodiment innovatively adopts a preloading technique to further reduce latency. When completing test points, the next test point and the first precondition of the test point (preconditions, operation steps, and expected results are also cached in advance) are pre-cached, reusing the time for human thinking to allow the AI ​​to perform reasoning, further reducing the latency of sequential completion to 150ms.

[0142] Because when manually writing test cases, the thought process is generally sequential: first, define the test points; then, define the preconditions for each test point; finally, add the operation steps and expected results; and after receiving completion suggestions, it's generally necessary to consider and decide whether to adopt them. By leveraging this premise, we can fully utilize the human thinking time to reduce latency.

[0143] Figure 23This is a schematic diagram of a preloading process provided in an embodiment of this application, as shown below. Figure 23 As shown, the "Enter Daily Tasks Interface" and "Acquire Gold Coins Through the Achievement System Interface" are preloaded. Some information in the test cases can be "Acquire Gold Coins Through Daily Tasks".

[0144] It should be noted that in related technologies, users aim to improve the efficiency of manually written test cases by completing each line of information. However, the conventional approach requires inputting all relevant information into a large model for completion, which is not only costly and has unacceptably high latency (8-10 seconds), but also yields mediocre completion results. In this embodiment, a private intelligent model is trained using synthetic data and knowledge distillation techniques, and preloading technology is used to improve the completion effect while controlling the latency to around 150 milliseconds.

[0145] Figure 24 A schematic diagram of another test case provided in an embodiment of this application, as shown below. Figure 24 As shown, the interface displays a one-click test case generation control. Selecting this control generates test cases. Simply input the requirements, and a complete test case can be generated with a single click. Previously, manual work required significant time writing preconditions and expected results; now, the intelligent model automatically organizes the text to generate each node of the test case, effectively covering equivalence classes, boundary conditions, and more. This allows QA to focus on higher-risk test cases and more valuable test content.

[0146] Figure 25 A schematic diagram illustrating a supplementary and improved test case provided for an embodiment of this application, such as... Figure 25 As shown, entering project habits in the input box will also make corresponding changes in the test cases.

[0147] This application's embodiments utilize historical test cases in the knowledge base to supplement related test points and test cases based on project habits, thereby increasing the upper limit of the intelligent model's capabilities and reducing omissions. The knowledge base is automatically updated during the use of QAStudio, ensuring that the intelligent model has the latest knowledge. For new team members and BPO third-party testing companies in the initial testing phase, this reduces omissions in test case writing due to unfamiliarity with rules, ensuring test case coverage and basic test case quality.

[0148] It's worth noting that the service integrates an incident database and a project-specific checklist, allowing the intelligent model to help with note-taking. For games with highly coupled rules, testing function A often requires associating it with testing function B. Relying on manual recall and association is prone to omissions. The intelligent model can traverse the incident database and checklist to generate associated test cases, reducing omissions.

[0149] In this embodiment of the application, the CheckList may record "involving land grade, test points of developed land need to be added". Figure 26 A schematic diagram of a memory list provided for an embodiment of this application, such as Figure 26 As shown, the memory list can be recorded in the form of branches, such as the status information of some land registration display switches, etc.

[0150] In practical applications, some scenarios still require manual writing of test cases. This application also provides intelligent line-by-line completion / iteration functions to assist in generating standardized test cases. Figure 27 This is a schematic diagram of single-line test case completion provided in an embodiment of this application, as shown below. Figure 27 As shown, the intelligent model displays the content that is completed / iterated step by step around the manually entered input box.

[0151] Figure 28 This is a schematic diagram of the framework of a test case generation method provided in an embodiment of this application, as shown below. Figure 28 As shown, test planning images are recognized using OCR technology and LLM to obtain planning document fragments. Planning documents can also be synthesized from manually written 2.0 test cases, and existing 1.0 test cases are converted using AI to obtain synthesized planning documents. Both planning document fragments and synthesized planning documents can be retrieved from the knowledge base using a hybrid retrieval method. Based on LLM, test points can be generated from the planning document fragments and historical test information retrieved from the knowledge base that matches the test planning document. LLM is also used to generate planning documents based on test points.

[0152] In summary, this application provides a test case generation method that can assist manual labor in quickly generating and iterating test case text in batches, improving writing efficiency; it can also help manual labor to sort out complex planning documents, combining knowledge bases, test case libraries, and incident libraries to reduce omissions; and it can generate sufficiently standardized test cases, reducing BPO integration costs.

[0153] The following describes the test case generation apparatus, test case generation device, and storage medium used to execute the test case generation method provided in this application. For the specific implementation process and technical effects, please refer to the relevant content of the test case generation method above, which will not be repeated below.

[0154] Figure 29 This is a schematic diagram of the structure of a test case generation device provided in an embodiment of this application, as shown below. Figure 29 As shown, the device includes: For example, the device may include: Analysis module 101 is used to perform structured analysis on test planning documents through intelligent models to obtain multiple causal relationships contained in the documents, wherein the causal relationships are used to characterize the conditions to be tested and the corresponding results. The classification module 102 is used to classify the conditions in the multiple causal relationships and generate classification results; The retrieval module 103 is used to retrieve historical test information that matches the test planning document from the knowledge base; The first generation module 104 is used to generate initial test points through the intelligent model based on the historical test information, the classification results and the test planning document. The inspection module 105 is used to perform introspection inspection on the initial test points through the intelligent model to determine the test points; The second generation module 106 is used to generate test cases through the intelligent model based on the test points and the test planning document.

[0155] Optionally, the retrieval module 103 is specifically used to retrieve from the knowledge base a first historical planning document that matches the test planning document, and a first historical test point corresponding to the first historical planning document; The first generation module 104 is specifically used to generate initial test points through the intelligent model based on the first historical planning document, the first historical test points, the classification results, and the test planning document.

[0156] Optionally, the inspection module 105 is specifically used to perform deduplication and supplementation processing on the initial test points according to preset prompts and requirements through the intelligent model to obtain the test points.

[0157] Optionally, the second generation module 106 is specifically used to retrieve from the knowledge base a second historical planning document that matches the test planning document and the test point, as well as a first historical test case and a second historical test point corresponding to the second historical planning document; and to generate the test case through the intelligent model based on the second historical planning document, the second historical test point, the first historical test case, the classification result, the test point, and the test planning document.

[0158] Optionally, the second generation module 106 is specifically configured to generate initial preconditions based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document; perform deduplication and supplementation processing on the initial preconditions according to preset prompts to obtain the preconditions; generate the operation steps and expected results corresponding to the preconditions; and generate the test cases through the intelligent model based on the preconditions and the operation steps and expected results corresponding to the preconditions.

[0159] Optionally, the second generation module 106 is specifically used to divide the multiple test points into multiple groups of test points, wherein there are some overlapping test points between adjacent groups of test points; and to generate the test cases based on the multiple groups of test points and the test planning document.

[0160] Optionally, the analysis module 101 is specifically used to perform structured analysis on the test planning document through the intelligent model, and divide the user behavior in the test planning document into multiple causal relationships. The causal relationships are used to characterize the user's behavior or the constraints they are subject to, and the corresponding results.

[0161] Optionally, the retrieval module 103 is specifically configured to use a hybrid retrieval method to retrieve multiple third historical planning documents similar to the test planning document from the knowledge base; sort the multiple third historical planning documents according to the similarity between each third historical planning document and the test planning document, and the quality of the test cases corresponding to each third historical planning document; determine the first historical planning document from the sorted multiple third historical planning documents; and retrieve the first historical test point corresponding to the first historical planning document from the knowledge base.

[0162] Optionally, the device further includes: The acquisition module is used to retrieve the second historical test cases; The judgment module is used to determine whether the second historical test case contains historical preconditions, historical operation steps and historical expected results, and to determine whether the second historical test case meets the preset conditions. The third generation module is used to generate a fourth historical planning document corresponding to the second historical test case if the second historical test case contains the historical preconditions, the historical operation steps and the historical expected results, and the second historical test case meets the preset conditions. The update module is used to update the knowledge base based on the second historical test cases and the fourth historical planning document.

[0163] Optionally, the second historical test case meets preset conditions, including: If the second historical test case passes the review, or if the execution coverage of the second historical test case is greater than the preset coverage, or if the second historical test case is a test case obtained by converting a historical mind map, then the test case is determined to meet the preset conditions.

[0164] Optionally, the acquisition module is specifically used to acquire a historical mind map and a corresponding preset description character; replace the historical mind map with a preset placeholder; and generate the second historical test case using the preset placeholder, the preset description character, and the historical mind map.

[0165] Optionally, the device further includes: The acquisition module is also used to acquire test cases to be modified; The determination module is used to determine the target node to be modified from multiple nodes of the test case to be modified based on the iteration content, and to add a mark to the target node; The modification module is used to modify the target nodes after adding tags based on the iteration content, and obtain the test cases after the iteration.

[0166] Optionally, the determining module is specifically used to add any one of the following tags to the target node: revision tag, deletion tag, and addition tag; The modification module is specifically used to revise the target node after adding the revision mark according to the iteration content to obtain the iterated test cases; or, to delete the target node after adding the deletion mark according to the iteration content to obtain the iterated test cases; or, to add the target node after adding the addition mark according to the iteration content to obtain the iterated test cases.

[0167] Optionally, the device further includes: The acquisition module is also used to acquire test planning images, which are used to represent the planning content; The recognition module is used to perform text recognition on the test planning image using optical character recognition technology in the preset hybrid model to obtain an initial text recognition result; and to perform text recognition on the test planning image using the large language model in the preset hybrid model, based on the recognition prompts in the thought chain and the initial text recognition result to obtain a text recognition result. The fourth generation module is used to generate the test planning document based on the text recognition results.

[0168] Optionally, the first generation module 104 is specifically used to generate test points one by one based on the classification results and the test planning document using the intelligent model; The second generation module 106 is specifically used to use the intelligent model to generate the preconditions, operation steps and expected results one by one according to the test points and the test planning document, so as to obtain the test cases.

[0169] Optionally, the intelligent model is trained using the following apparatus: From the knowledge base, retrieve the fourth historical planning document and the corresponding third historical test case that match the preset sample test planning document and the corresponding preset sample test case; Replace some information in the third historical test case with preset information to obtain the replaced test case; The fourth historical planning document is iterated through a preset large language model to obtain an iterated historical planning document; Replace some information in the iterative test cases corresponding to the iterative historical planning document with preset information to obtain the replaced iterative test cases. The initial model is trained based on the third historical test case, the replaced test case, the iterative test case, and the replaced iterative test case until a preset termination condition is met, thus obtaining the intelligent model.

[0170] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0171] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more digital signal processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0172] Figure 30This is a schematic diagram of the structure of a test case generation device provided in an embodiment of this application, as shown below. Figure 30 As shown, the test case generation device includes: processor 201 and memory 202.

[0173] The memory 202 is used to store programs, and the processor 201 calls the programs stored in the memory 202 to execute the above method embodiments. The specific implementation and technical effects are similar, and will not be described in detail here.

[0174] For example, the method may include: The test planning document is structured and analyzed by an intelligent model to obtain multiple causal relationships contained in the document. These causal relationships are used to characterize the conditions to be tested and the corresponding results. Classify the conditions in the multiple causal relationships and generate classification results; Retrieve historical test information from the knowledge base that matches the test planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model. The intelligent model performs introspection checks on the initial test points to determine the test points. Test cases are generated using the intelligent model based on the test points and the test planning document.

[0175] Optionally, retrieving historical test information from the knowledge base that matches the test planning document includes: From the knowledge base, retrieve a first historical planning document that matches the test planning document, and a first historical test point corresponding to the first historical planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model, including: Based on the first historical planning document, the first historical test points, the classification results, and the test planning document, initial test points are generated through the intelligent model.

[0176] Optionally, the step of performing introspection checks on the initial test points using the intelligent model to determine the test points includes: The intelligent model performs deduplication and supplementation processing on the initial test points according to preset prompts to obtain the test points.

[0177] Optionally, generating test cases using the intelligent model based on the test points and the test planning document includes: Retrieve from the knowledge base a second historical planning document that matches the test planning document and the test point, as well as the first historical test case and the second historical test point corresponding to the second historical planning document; The test cases are generated by the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document.

[0178] Optionally, generating the test cases using the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document includes: Based on the second historical planning document, the second historical test point, the first historical test case, the classification result, the test point, and the test planning document, generate the initial prerequisites; According to the preset prompts, the initial prerequisites are deduplicated and supplemented to obtain the prerequisites; Generate the operation steps and expected results corresponding to the aforementioned prerequisites; The test cases are generated by the intelligent model based on the preconditions, the corresponding operation steps, and the expected results.

[0179] Optionally, generating test cases based on the test points and the test planning document includes: The multiple test points are divided into multiple groups of test points, wherein there are some overlapping test points between adjacent groups of test points; The test cases are generated based on the multiple sets of test points and the test planning document.

[0180] Optionally, the step of performing structured analysis on the test planning document using an intelligent model to obtain multiple causal relationships contained in the document includes: The intelligent model performs structured analysis on the test planning document, dividing user behaviors in the test planning document into multiple causal relationships. These causal relationships characterize the user's behavior or the constraints they are subject to, as well as the corresponding results.

[0181] Optionally, retrieving from the knowledge base the first historical test point corresponding to the first historical planning document that matches the test planning document includes: A hybrid retrieval method is used to retrieve multiple third-party historical planning documents that are similar to the test planning document from the knowledge base; Based on the similarity between each third historical planning document and the test planning document, and the quality of the test cases corresponding to each third historical planning document, the multiple third historical planning documents are sorted. The first historical planning document is determined from the sorted plurality of third historical planning documents; Retrieve the first historical test point corresponding to the first historical planning document from the knowledge base.

[0182] Optionally, the method further includes: Obtain the second historical test case; Determine whether the second historical test case contains historical preconditions, historical operation steps, and historical expected results, and determine whether the second historical test case meets the preset conditions; If the second historical test case contains the historical preconditions, the historical operation steps, and the historical expected results, and the second historical test case meets the preset conditions, then a fourth historical planning document corresponding to the second historical test case is generated based on the second historical test case and the preset synthesis prompt requirements. The knowledge base is updated based on the second historical test cases and the fourth historical planning document.

[0183] Optionally, the second historical test case meets preset conditions, including: If the second historical test case passes the review, or if the execution coverage of the second historical test case is greater than the preset coverage, or if the second historical test case is a test case obtained by converting a historical mind map, then the test case is determined to meet the preset conditions.

[0184] Optionally, obtaining the second historical test cases includes: Retrieve historical mind maps and their corresponding preset descriptive characters; Replace the historical mind map with a preset placeholder; The preset placeholder, the preset descriptive characters, and the historical mind map are used to generate the second historical test case.

[0185] Optionally, the method further includes: Obtain the test cases to be modified; Based on the iteration content, the target node to be modified is determined from multiple nodes of the test case to be modified, and a mark is added to the target node; Modify the target node after adding the marker based on the iteration content to obtain the test cases after iteration.

[0186] Optionally, adding a marker to the target node includes: Add any of the following tags to the target node: revision tag, deletion tag, or addition tag; The step of modifying the target node after adding the marker according to the iteration content to obtain the iterated test cases includes: The target node after adding the revision mark is revised according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the deletion mark is deleted according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the addition mark is added according to the iteration content to obtain the test cases after the iteration.

[0187] Optionally, before analyzing the logical relationships in the test planning document to obtain multiple causal relationships, the method further includes: Obtain a test plan image, which is used to represent the plan content; The optical character recognition technology in the preset hybrid model is used to perform text recognition on the test planning image to obtain the initial text recognition result; Using the large language model in the preset hybrid model, based on the recognition prompts in the thought chain and the initial text recognition results, the test planning image is subjected to text recognition to obtain the text recognition results; Based on the text recognition results, the test planning document is generated.

[0188] Optionally, the step of generating initial test points using the intelligent model based on the historical test information, the classification results, and the test planning document includes: Using the intelligent model, test points are generated one by one based on the classification results and the test planning document; The step of generating test cases using the intelligent model based on the test points and the test planning document includes: Using the intelligent model, based on the test points and the test planning document, the preconditions, operation steps, and expected results are generated one by one to obtain the test cases.

[0189] Optionally, the intelligent model is trained using the following method: From the knowledge base, retrieve the fourth historical planning document and the corresponding third historical test case that match the preset sample test planning document and the corresponding preset sample test case; Replace some information in the third historical test case with preset information to obtain the replaced test case; The fourth historical planning document is iterated through a preset large language model to obtain an iterated historical planning document; Replace some information in the iterative test cases corresponding to the iterative historical planning document with preset information to obtain the replaced iterative test cases. The initial model is trained based on the third historical test case, the replaced test case, the iterative test case, and the replaced iterative test case until a preset termination condition is met, thus obtaining the intelligent model.

[0190] In summary, in this embodiment of the application, the test planning document is structured and analyzed by an intelligent model. This allows for the construction of a short-term memory for the test planning document. Historical test information matching the test planning document is retrieved from the knowledge base, providing long-term memory for the intelligent model. Furthermore, the intelligent model performs introspection checks on the initial test points, which reduces the omission and redundancy of test points. This avoids redundancy and omission of test cases and improves the quality and stability of the generated test cases.

[0191] Optionally, this application also provides a program product, such as a computer-readable storage medium, including a program that, when executed by a processor, performs the above-described method embodiments.

[0192] For example, the method may include: The test planning document is structured and analyzed by an intelligent model to obtain multiple causal relationships contained in the document. These causal relationships are used to characterize the conditions to be tested and the corresponding results. Classify the conditions in the multiple causal relationships and generate classification results; Retrieve historical test information from the knowledge base that matches the test planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model. The intelligent model performs introspection checks on the initial test points to determine the test points. Test cases are generated using the intelligent model based on the test points and the test planning document.

[0193] Optionally, retrieving historical test information from the knowledge base that matches the test planning document includes: From the knowledge base, retrieve a first historical planning document that matches the test planning document, and a first historical test point corresponding to the first historical planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model, including: Based on the first historical planning document, the first historical test points, the classification results, and the test planning document, initial test points are generated through the intelligent model.

[0194] Optionally, the step of performing introspection checks on the initial test points using the intelligent model to determine the test points includes: The intelligent model performs deduplication and supplementation processing on the initial test points according to preset prompts to obtain the test points.

[0195] Optionally, generating test cases using the intelligent model based on the test points and the test planning document includes: Retrieve from the knowledge base a second historical planning document that matches the test planning document and the test point, as well as the first historical test case and the second historical test point corresponding to the second historical planning document; The test cases are generated by the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document.

[0196] Optionally, the step of generating the test cases using the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document includes: Based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document, generate initial prerequisites; According to the preset prompts, the initial prerequisites are deduplicated and supplemented to obtain the prerequisites; Generate the operation steps and expected results corresponding to the aforementioned prerequisites; The test cases are generated by the intelligent model based on the preconditions, the corresponding operation steps, and the expected results.

[0197] Optionally, generating test cases based on the test points and the test planning document includes: The multiple test points are divided into multiple groups of test points, wherein there are some overlapping test points between adjacent groups of test points; The test cases are generated based on the multiple sets of test points and the test planning document.

[0198] Optionally, the step of performing structured analysis on the test planning document using an intelligent model to obtain multiple causal relationships contained in the document includes: The intelligent model performs structured analysis on the test planning document, dividing user behaviors in the test planning document into multiple causal relationships. These causal relationships characterize the user's behavior or the constraints they are subject to, as well as the corresponding results.

[0199] Optionally, retrieving from the knowledge base the first historical test point corresponding to the first historical planning document that matches the test planning document includes: A hybrid retrieval method is used to retrieve multiple third-party historical planning documents that are similar to the test planning document from the knowledge base; Based on the similarity between each third historical planning document and the test planning document, and the quality of the test cases corresponding to each third historical planning document, the multiple third historical planning documents are sorted. The first historical planning document is determined from the sorted plurality of third historical planning documents; Retrieve the first historical test point corresponding to the first historical planning document from the knowledge base.

[0200] Optionally, the method further includes: Obtain the second historical test case; Determine whether the second historical test case contains historical preconditions, historical operation steps, and historical expected results, and determine whether the second historical test case meets the preset conditions; If the second historical test case contains the historical preconditions, the historical operation steps, and the historical expected results, and the second historical test case meets the preset conditions, then a fourth historical planning document corresponding to the second historical test case is generated based on the second historical test case and the preset synthesis prompt requirements. The knowledge base is updated based on the second historical test cases and the fourth historical planning document.

[0201] Optionally, the second historical test case meets preset conditions, including: If the second historical test case passes the review, or if the execution coverage of the second historical test case is greater than the preset coverage, or if the second historical test case is a test case obtained by converting a historical mind map, then the test case is determined to meet the preset conditions.

[0202] Optionally, obtaining the second historical test cases includes: Retrieve historical mind maps and their corresponding preset descriptive characters; Replace the historical mind map with a preset placeholder; The preset placeholder, the preset descriptive characters, and the historical mind map are used to generate the second historical test case.

[0203] Optionally, the method further includes: Obtain the test cases to be modified; Based on the iteration content, the target node to be modified is determined from multiple nodes of the test case to be modified, and a mark is added to the target node; Modify the target node after adding the marker based on the iteration content to obtain the test cases after iteration.

[0204] Optionally, adding a marker to the target node includes: Add any of the following tags to the target node: revision tag, deletion tag, or addition tag; The step of modifying the target node after adding the marker according to the iteration content to obtain the iterated test cases includes: The target node after adding the revision mark is revised according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the deletion mark is deleted according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the addition mark is added according to the iteration content to obtain the test cases after the iteration.

[0205] Optionally, before analyzing the logical relationships in the test planning document to obtain multiple causal relationships, the method further includes: Obtain a test plan image, which is used to represent the plan content; The optical character recognition technology in the preset hybrid model is used to perform text recognition on the test planning image to obtain the initial text recognition result; Using the large language model in the preset hybrid model, based on the recognition prompts in the thought chain and the initial text recognition results, the test planning image is subjected to text recognition to obtain the text recognition results; Based on the text recognition results, the test planning document is generated.

[0206] Optionally, the step of generating initial test points using the intelligent model based on the historical test information, the classification results, and the test planning document includes: Using the intelligent model, test points are generated one by one based on the classification results and the test planning document; The step of generating test cases using the intelligent model based on the test points and the test planning document includes: Using the intelligent model, based on the test points and the test planning document, the preconditions, operation steps, and expected results are generated one by one to obtain the test cases.

[0207] Optionally, the intelligent model is trained using the following method: From the knowledge base, retrieve the fourth historical planning document and the corresponding third historical test case that match the preset sample test planning document and the corresponding preset sample test case; Replace some information in the third historical test case with preset information to obtain the replaced test case; The fourth historical planning document is iterated through a preset large language model to obtain an iterated historical planning document; Replace some information in the iterative test cases corresponding to the iterative historical planning document with preset information to obtain the replaced iterative test cases. The initial model is trained based on the third historical test case, the replaced test case, the iterative test case, and the replaced iterative test case until a preset termination condition is met, thus obtaining the intelligent model.

[0208] In summary, in this embodiment of the application, the test planning document is structured and analyzed by an intelligent model. This allows for the construction of a short-term memory for the test planning document. Historical test information matching the test planning document is retrieved from the knowledge base, providing long-term memory for the intelligent model. Furthermore, the intelligent model performs introspection checks on the initial test points, which reduces the omission and redundancy of test points. This avoids redundancy and omission of test cases and improves the quality and stability of the generated test cases.

[0209] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0210] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0211] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in a combination of hardware and software functional units.

[0212] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0213] The above are merely preferred embodiments of this application and are not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A test case generation method, characterized in that, The method includes: The test planning document is structured and analyzed by an intelligent model to obtain multiple causal relationships contained in the document. These causal relationships are used to characterize the conditions to be tested and the corresponding results. Classify the conditions in the multiple causal relationships and generate classification results; Retrieve historical test information from the knowledge base that matches the test planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model. The intelligent model performs introspection checks on the initial test points to determine the test points. Test cases are generated using the intelligent model based on the test points and the test planning document.

2. The method according to claim 1, characterized in that, The step of retrieving historical test information from the knowledge base that matches the test planning document includes: From the knowledge base, retrieve a first historical planning document that matches the test planning document, and a first historical test point corresponding to the first historical planning document; Based on the historical test information, the classification results, and the test planning document, initial test points are generated through the intelligent model, including: Based on the first historical planning document, the first historical test points, the classification results, and the test planning document, initial test points are generated through the intelligent model.

3. The method according to claim 2, characterized in that, The step of using the intelligent model to perform introspection checks on the initial test points to determine the test points includes: The intelligent model performs deduplication and supplementation processing on the initial test points according to preset prompts to obtain the test points.

4. The method according to claim 2, characterized in that, The step of generating test cases using the intelligent model based on the test points and the test planning document includes: Retrieve from the knowledge base a second historical planning document that matches the test planning document and the test point, as well as the first historical test case and the second historical test point corresponding to the second historical planning document; The test cases are generated by the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document.

5. The method according to claim 4, characterized in that, The step of generating the test cases using the intelligent model based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document includes: Based on the second historical planning document, the second historical test points, the first historical test cases, the classification results, the test points, and the test planning document, generate initial prerequisites; According to the preset prompts, the initial prerequisites are deduplicated and supplemented to obtain the prerequisites; Generate the operation steps and expected results corresponding to the aforementioned prerequisites; The test cases are generated by the intelligent model based on the preconditions, the corresponding operation steps, and the expected results.

6. The method according to claim 1, characterized in that, The step of generating test cases based on the test points and the test planning document includes: The multiple test points are divided into multiple groups of test points, wherein there are some overlapping test points between adjacent groups of test points; The test cases are generated based on the multiple sets of test points and the test planning document.

7. The method according to claim 1, characterized in that, The intelligent model is used to perform structured analysis on the test planning document, resulting in multiple causal relationships contained within the document, including: The intelligent model performs structured analysis on the test planning document, dividing user behaviors in the test planning document into multiple causal relationships. These causal relationships characterize the user's behavior or the constraints they are subject to, as well as the corresponding results.

8. The method according to claim 2, characterized in that, The step of retrieving from the knowledge base the first historical test point corresponding to the first historical planning document that matches the test planning document includes: A hybrid retrieval method is used to retrieve multiple third-party historical planning documents that are similar to the test planning document from the knowledge base; Based on the similarity between each third historical planning document and the test planning document, and the quality of the test cases corresponding to each third historical planning document, the multiple third historical planning documents are sorted. The first historical planning document is determined from the sorted plurality of third historical planning documents; Retrieve the first historical test point corresponding to the first historical planning document from the knowledge base.

9. The method according to claim 1, characterized in that, The method further includes: Obtain the second historical test case; Determine whether the second historical test case contains historical preconditions, historical operation steps, and historical expected results, and determine whether the second historical test case meets the preset conditions; If the second historical test case contains the historical preconditions, the historical operation steps, and the historical expected results, and the second historical test case meets the preset conditions, then a fourth historical planning document corresponding to the second historical test case is generated based on the second historical test case and the preset synthesis prompt requirements. The knowledge base is updated based on the second historical test cases and the fourth historical planning document.

10. The method according to claim 9, characterized in that, The second historical test case meets the preset conditions, including: If the second historical test case passes the review, or if the execution coverage of the second historical test case is greater than the preset coverage, or if the second historical test case is a test case obtained by converting a historical mind map, then the test case is determined to meet the preset conditions.

11. The method according to claim 9, characterized in that, The acquisition of the second historical test cases includes: Retrieve historical mind maps and their corresponding preset descriptive characters; Replace the historical mind map with a preset placeholder; The preset placeholder, the preset descriptive characters, and the historical mind map are used to generate the second historical test case.

12. The method according to claim 1, characterized in that, The method further includes: Obtain the test cases to be modified; Based on the iteration content, the target node to be modified is determined from multiple nodes of the test case to be modified, and a mark is added to the target node; Modify the target node after adding the marker based on the iteration content to obtain the test cases after iteration.

13. The method according to claim 12, characterized in that, Adding a marker to the target node includes: Add any of the following tags to the target node: revision tag, deletion tag, or addition tag; The step of modifying the target node after adding the marker according to the iteration content to obtain the iterated test cases includes: The target node after adding the revision mark is revised according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the deletion mark is deleted according to the iteration content to obtain the test cases after the iteration; or, the target node after adding the addition mark is added according to the iteration content to obtain the test cases after the iteration.

14. The method according to claim 1, characterized in that, Before analyzing the logical relationships in the test planning document to obtain multiple causal relationships, the method further includes: Obtain a test plan image, which is used to represent the plan content; The optical character recognition technology in the preset hybrid model is used to perform text recognition on the test planning image to obtain the initial text recognition result; Using the large language model in the preset hybrid model, based on the recognition prompts in the thought chain and the initial text recognition results, the test planning image is subjected to text recognition to obtain the text recognition results; Based on the text recognition results, the test planning document is generated.

15. The method according to claim 1, characterized in that, The step of generating initial test points using the intelligent model based on the historical test information, the classification results, and the test planning document includes: Using the intelligent model, test points are generated one by one based on the classification results and the test planning document; The step of generating test cases using the intelligent model based on the test points and the test planning document includes: Using the intelligent model, based on the test points and the test planning document, the preconditions, operation steps, and expected results are generated one by one to obtain the test cases.

16. The method according to claim 15, characterized in that, The intelligent model was trained using the following method: From the knowledge base, retrieve the fourth historical planning document and the corresponding third historical test case that match the preset sample test planning document and the corresponding preset sample test case; Replace some information in the third historical test case with preset information to obtain the replaced test case; The fourth historical planning document is iterated through a preset large language model to obtain an iterated historical planning document; Replace some information in the iterative test cases corresponding to the iterative historical planning document with preset information to obtain the replaced iterative test cases. The initial model is trained based on the third historical test case, the replaced test case, the iterative test case, and the replaced iterative test case until a preset termination condition is met, thus obtaining the intelligent model.

17. A test case generation device, characterized in that, The device includes: The analysis module is used to perform structured analysis on the test planning document through an intelligent model to obtain multiple causal relationships contained in the document, wherein the causal relationships are used to characterize the conditions to be tested and the corresponding results. The classification module is used to classify the conditions in multiple causal relationships and generate classification results; The retrieval module is used to retrieve historical test information that matches the test planning document from the knowledge base; The first generation module is used to generate initial test points through the intelligent model based on the historical test information, the classification results, and the test planning document. The inspection module is used to perform introspection inspection on the initial test points through the intelligent model to determine the test points; The second generation module is used to generate test cases based on the test points and the test planning document using the intelligent model.

18. A test case generation device, characterized in that, include: A memory and a processor, the memory storing a computer program executable by the processor, the processor executing the computer program to implement the test case generation method according to any one of claims 1-16.

19. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which, when read and executed, implements the test case generation method according to any one of claims 1-16.

Images