Auxiliary code development system and method based on ai code-assisted large models

Abstract

Disclosed in the present invention are an auxiliary code development system and method based on AI code-assisted large models. An integrated IDE plug-in is built in or connected to an advanced code-assisted large model, a required code-assisted large model is called by means of identifying a code-assisted scenario in a large model service calling request, and deep learning and natural language processing techniques in the code-assisted large model are used to implement full-chain intelligent assistance from code conception to testing, thereby significantly improving the programming efficiency, improving the code quality and shortening software development cycles.

Description

AI code-assisted large model-based code development assistance system and method TECHNICAL FIELD

[0001] The present application relates to the field of code development assistance, in particular to an AI code-assisted large model-based code development assistance system and method. BACKGROUND

[0002] With the rapid development of the software industry, the requirements for development speed and code quality are increasing. Traditional programming methods are often inefficient and prone to errors when faced with complex logic and large amounts of code. Although existing code generation and intelligent prompting functions have alleviated this problem to some extent, there is still a lack of intelligent support for the entire software development process, especially in key areas such as code writing, code debugging, and unit testing. SUMMARY

[0003] The following gives a brief summary of one or more aspects to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all contemplated aspects, and neither is it intended to identify key or critical elements of all aspects nor to delineate the scope of any or all aspects. Its sole purpose is to give a brief overview of some concepts of one or more aspects in a simplified form as a prelude to the more detailed description given later.

[0004] The present application aims to solve the above problems and provides an AI code-assisted large model-based code development assistance system. By integrating an IDE plug-in or connecting to an advanced code assistance large model, the system utilizes deep learning and natural language processing techniques in the code assistance large model to achieve intelligent assistance throughout the entire chain from code conception to testing, thereby significantly improving programming efficiency, enhancing code quality, and shortening software development cycles.

[0005] The technical solution of the present application is as follows:

[0006] The present application provides an AI code-assisted large model-based code development assistance system that obtains large model services of a code assistance large model through an integrated development IDE plug-in to assist end users in code development. The IDE plug-in includes a plug-in integration layer, a model encapsulation layer, a model calling layer, and a model implementation layer. The plug-in integration layer is used to integrate one or more IDE plug-ins developed for the code assistance large model. The integrated IDE plug-in triggers a large model service call request to obtain the required large model services. The large model service call request includes code assistance scenario information for any code assistance scenario in the entire code development cycle.

[0007] The plug-in integration layer is used to integrate one or more IDE plug-ins developed for the code assistance large model. The integrated IDE plug-in triggers a large model service call request to obtain the required large model services. The large model service call request includes code assistance scenario information for any code assistance scenario in the entire code development cycle.

[0008] The model encapsulation layer is configured to identify code assistance scenario information in a large model service call request, and call the model call layer according to the identified code assistance scenario information to obtain a corresponding large model service and feed back to the corresponding IDE plug-in.

[0009] The model call layer is configured to call a corresponding code assistance large model from the model implementation layer according to code assistance scenario information transmitted by the model encapsulation layer, to obtain a large model service required by the code assistance scenario.

[0010] The model implementation layer includes one or more code assistance large models, which are configured to generate a code assistance result required by a code assistance scenario, so as to provide a corresponding large model service for different code assistance scenarios and assist a development end user in code development.

[0011] According to an embodiment of the code development assistance system based on the AI code assistance large model, the plug-in integration layer is configured to develop a corresponding IDE plug-in according to an integrated development environment used by different IDE plug-ins when integrating the IDE plug-ins, and transmit a large model service call request to the model encapsulation model through the developed IDE plug-in, to obtain a large model service required by a code assistance scenario in the large model service call request.

[0012] According to an embodiment of the code development assistance system based on the AI code assistance large model, the model encapsulation layer is configured to uniformly encapsulate a code assistance large model to be called, so as to provide a uniformly encapsulated large model service interface for each IDE plug-in in the plug-in integration layer.

[0013] According to an embodiment of the code development assistance system based on the AI code assistance large model, the model call layer is configured to adapt a large model interface of a code assistance large model in one or more model implementation layers, and call a corresponding code assistance large model from the model implementation layer through the adapted large model interface, to obtain a large model service required by a code assistance scenario.

[0014] According to an embodiment of the code development assistance system based on the AI code assistance large model provided by the application, the code assistance large model comprises one or more target models; wherein, after the code implementation module obtains the code assistance scene information through the corresponding large model interface, the corresponding target model in the code assistance large model is selected according to the code assistance scene information to generate the corresponding code assistance result, and the generated code assistance result is fed back to the model calling layer through the corresponding large model interface, and then fed back to the corresponding IDE plug-in.

[0015] According to an embodiment of the code development assistance system based on the AI code assistance large model provided by the application, the code development assistance system based on the AI code assistance large model further comprises an extension module; wherein, the code development assistance system based on the AI code assistance large model extends the management function and the configuration function of the code development assistance system through the extension module, so as to realize the management of the code assistance large model.

[0016] According to an embodiment of the code development assistance system based on the AI code assistance large model provided by the application, the extension module comprises a model management unit, a scene management unit, a model calling configuration unit, a prompt word configuration unit and a data analysis unit; wherein,

[0017] The model management unit is used for dynamically accessing / going offline the code assistance large model connected by the IDE plug-in, so as to make the corresponding IDE plug-in update the code assistance large model in real time;

[0018] The scene management unit is used for adjusting different code assistance scenes and corresponding code assistance large models, so as to make the corresponding IDE plug-in meet the needs of different developers in real time;

[0019] The model calling configuration unit is used for adjusting the calling parameters of the code assistance large model, so as to adjust the accuracy of the code assistance result returned by the code assistance large model;

[0020] The prompt word configuration unit is used for configuring the prompt words of different code assistance scenes, so as to improve the accuracy of the reply of the code assistance large model;

[0021] The data analysis unit is used for analyzing the plug-in use data of the user, arranging different code assistance scenes, different code assistance large models and corresponding user feedback results according to the plug-in use data of the user, so as to optimize the large model assistance model.

[0022] According to an embodiment of the AI code assisted large model based code development assistance system, the AI code assisted large model based code development assistance system further comprises an interactive interface for realizing interactive operation between the development end user and the code assistance large model; wherein when the AI code assisted large model based code development assistance system assists the development end user in code development, the development end operation of the user is obtained through the loaded interactive interface, so as to generate a corresponding code assistance result in real time to assist the development end user in code development.

[0023] According to an embodiment of the AI code assisted large model based code development assistance system, the AI code assisted large model based code development assistance system performs code development assistance work in any code assistance scene except the code completion scene through the following steps:

[0024] Step C1: loading an interactive interface for interaction between the development end user and the code assistance large model, and the development end user triggering the large model service through the loaded interactive interface;

[0025] Step C2: obtaining input information of the development end user through the interactive interface, and determining a code editing position associated with the development end operation;

[0026] Step C3: determining a current code assistance scene according to the obtained input information;

[0027] Step C4: generating a corresponding code assistance result according to the determined code assistance scene and the code editing position, and displaying the code assistance result through the interactive interface to assist the development end user in code development.

[0028] According to an embodiment of the AI code assisted large model based code development assistance system, the AI code assisted large model based code development assistance system performs code development assistance work in the code completion scene through the following steps:

[0029] Step D1: loading an interactive interface for interaction between the development end user and the code assistance large model, and the development end user triggering the large model service through the loaded interactive interface;

[0030] Step D2: determining a code editing position associated with the development end operation through the interactive interface;

[0031] Step D3: obtaining code context information according to the determined code editing position;

[0032] Step D4: generating corresponding code completion information according to the obtained code context information, and displaying the code completion information through the interactive interface to prompt the development end user.

[0033] The application also provides a code development assistance method based on an AI code assistance large model, including the following steps:

[0034] Step S1: obtaining a large model service calling request, thereby triggering the start of a large model service; wherein the large model service calling request includes code assistance scene information of any code assistance scene in a code development whole cycle;

[0035] Step S2: identifying the code assistance scene information in the large model service calling request;

[0036] Step S3: calling a corresponding code assistance large model based on the identified code assistance scene information;

[0037] Step S4: the code assistance large model generating a required code assistance result based on the input code assistance scene information, providing corresponding large model services for different code assistance scenes, and assisting a development end user in code development.

[0038] According to an embodiment of the code development assistance method based on an AI code assistance large model, the code development assistance method based on an AI code assistance large model develops a corresponding IDE plug-in according to an integrated development environment used by the IDE plug-in, triggers a large model service calling request through the developed IDE plug-in, thereby obtaining a large model service required by a code assistance scene in the large model service calling request; wherein the code assistance scene includes a unit test generation scene, a complex logic explanation scene, a code annotation generation scene, a code optimization scene, a code review scene, and a code completion scene.

[0039] According to an embodiment of the code development assistance method based on an AI code assistance large model, the code development assistance method based on an AI code assistance large model uniformly encapsulates a code assistance large model to be called, thereby providing a uniformly encapsulated large model service interface for each IDE plug-in; wherein after the model encapsulation layer identifies the code assistance scene in the large model service calling request, the code assistance scene information is identified, the large model service provided by the model calling layer is called, and the uniformly encapsulated large model service interface is fed back to the corresponding IDE plug-in.

[0040] According to an embodiment of the code development assistance method based on an AI code assistance large model, when the code development assistance method based on an AI code assistance large model calls a code assistance large model required by a code assistance scene, the code assistance large model in the model implementation layer is selected according to the input code assistance scene information, and the corresponding large model service is obtained through a large model interface adapted to the code assistance large model.

[0041] According to an embodiment of the code development assistance method based on the AI code assistance large model, the code assistance large model comprises one or more target models; wherein, after the code assistance scene information is obtained through the corresponding large model interface, the code development assistance method based on the AI code assistance large model selects the corresponding target model in the code assistance large model according to the code assistance scene information to obtain the corresponding code assistance result, and feeds back the generated code assistance result to the model calling layer through the corresponding large model interface, and then feeds back to the corresponding IDE plug-in.

[0042] According to an embodiment of the code development assistance method based on the AI code assistance large model, the code development assistance method based on the AI code assistance large model is provided with an interactive interface for realizing the interactive operation between the development end user and the code assistance large model; wherein, when assisting the development end user in code development, the code development assistance method based on the AI code assistance large model obtains the development end operation of the user through the loaded interactive interface, so as to generate the corresponding code assistance result in real time to assist the development end user in code development.

[0043] According to an embodiment of the code development assistance method based on the AI code assistance large model, the code development assistance method based on the AI code assistance large model performs code development assistance work in any code assistance scene except the code completion scene through the following steps:

[0044] Step C1: loading an interactive interface for interaction between the development end user and the code assistance large model, and the development end user triggering the large model service through the loaded interactive interface;

[0045] Step C2: obtaining the input information of the development end user through the interactive interface, and determining the code editing position associated with the development end operation;

[0046] Step C3: determining the current code assistance scene according to the obtained input information;

[0047] Step C4: generating the corresponding code assistance result according to the determined code assistance scene and the code editing position, and displaying the code assistance result through the interactive interface to assist the development end user in code development.

[0048] According to an embodiment of the code development assistance method based on the AI code assistance large model, the code development assistance method based on the AI code assistance large model performs code development assistance work in the code completion scene through the following steps:

[0049] Step D1: loading an interaction interface for interaction between an end user and a code-assisted large model, and the end user triggers the start of the large model service through the loaded interaction interface;

[0050] Step D2: determining a code editing position associated with the operation of the development end through the interaction interface;

[0051] Step D3: obtaining code context information according to the determined code editing position;

[0052] Step D4: calling a corresponding AI code-assisted large model according to the obtained code context information to generate corresponding code completion information, and displaying the code completion information to the assisted development end through the interaction interface for prompting.

[0053] The application also provides a computer readable medium storing computer program code, which, when executed by a processor, implements the method as described above.

[0054] The application also provides a code development assistance device based on an AI code-assisted large model, comprising:

[0055] a memory for storing instructions executable by the processor; and

[0056] a processor for executing the instructions to implement the method as described above.

[0057] The application has the following beneficial effects compared with the prior art: the application is aimed at code assistance development work in different code assistance scenarios in the whole code development cycle, and through the integrated IDE plug-in built-in or connected to an advanced code assistance large model, the required code assistance large model is called by recognizing the large model service call request in the large model service call request, and the deep learning and natural language processing technology in the code assistance large model is used to generate the large model service through code automatic generation and optimization, intelligent unit test generation, intelligent annotation generation, real-time code interpretation, real-time error detection and repair suggestion, and collaborative development assistance. Compared with the prior art, the application can call the corresponding code assistance large model to generate the required large model service for different code assistance scenarios, realize intelligent assistance from code conception to testing, significantly improve programming efficiency, and be beneficial to improving code quality and shortening software development cycle. BRIEF DESCRIPTION OF DRAWINGS

[0058] The above features and advantages of the present application can be better understood after reading the detailed description of embodiments of the present application in conjunction with the following drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related properties or features can have the same or similar reference numerals.

[0059] FIG. 1 is a system architecture diagram illustrating an embodiment of the AI code-aided large model-based code development assistance system of the present application.

[0060] FIG. 2 is a step flow diagram illustrating an embodiment of the code development assistance work in any code assistance scenario other than the code completion scenario of the present application.

[0061] FIG. 3 is a step flow diagram illustrating an embodiment of the code development assistance work in the code completion scenario of the present application.

[0062] FIG. 4 is a step flow diagram illustrating an embodiment of the AI code-aided large model-based code development assistance method of the present application. DETAILED DESCRIPTION

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some examples or embodiments of the present application, and for those skilled in the art, the present application can also be applied to other similar scenarios without creative labor, unless it is obvious from the language environment or otherwise stated. In the drawings, the same reference numerals represent the same structures or operations, unless the context clearly indicates otherwise or otherwise stated.

[0064] As shown in the present application and claims, unless the context clearly indicates otherwise or otherwise stated, the words "one", "a", "an", and / or "the" do not specifically refer to the singular, but also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements explicitly identified, and these steps and elements do not constitute an exclusive list, and the method or device can also include other steps or elements.

[0065] Unless otherwise specifically stated, the relative arrangement of the components and steps, numerical expressions, and values set forth in these embodiments do not limit the scope of the present application. At the same time, it should be understood that the sizes of the various parts shown in the drawings are not drawn in proportion to the actual proportions. The technology, methods and devices known to those skilled in the relevant art can not be discussed in detail, but in appropriate cases, the technology, methods and devices should be considered as part of the authorized specification. In all examples shown and discussed here, any specific value should be interpreted as merely exemplary, and not as a limitation. Therefore, other examples of exemplary embodiments can have different values. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0066] In the detailed description of the embodiments of the application, the cross-sectional view of the device structure is partially enlarged without the general proportion for the convenience of illustration, and the schematic diagram is only an example, which should not limit the scope of protection of the application herein. In addition, three-dimensional spatial dimensions including length, width and depth should be included in actual production.

[0067] In the description of the present application, it should be understood that the orientation words such as "front, back, up, down, left, right", "transverse, vertical, perpendicular, horizontal" and "top, bottom" and the like indicate the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate and imply that the indicated device or element must have a specific orientation or be constructed and operated in a specific orientation, therefore it cannot be understood as a limitation on the scope of protection of the present application; the orientation words "inner, outer" refer to the inner and outer of the contour of each component itself.

[0068] An embodiment of an AI code assisted large model based code development assistance system (hereinafter sometimes referred to as code development assistance system) is disclosed herein, and Fig. 1 is a system architecture diagram of an embodiment of the AI code assisted large model based code development assistance system of the present application. As shown in Fig. 1, in the present embodiment, the AI code assisted large model based code development assistance system obtains large model services of code assistance large models through integrated development IDE plug-ins, thereby assisting the development end user to develop codes.

[0069] Specifically, in the present embodiment, the IDE plug-in includes a plug-in integration layer, a model encapsulation layer, a model calling layer and a model implementation layer. The plug-in integration layer is used to integrate one or more IDE plug-ins developed for code assistance large models, and triggers a large model service calling request through the integrated IDE plug-in to obtain the required large model service. The large model service calling request includes code assistance scene information of any code assistance scene in the whole code development cycle.

[0070] The model encapsulation layer is used to identify the code assistance scene information in the large model service calling request, and call the model calling layer through the identified code assistance scene information to obtain the corresponding large model service and feed back to the corresponding IDE plug-in.

[0071] The model calling layer and the model implementation layer provide the code-assisted large model-based large model service for the plug-in integration layer to the upper layer through the large model service interface. The model calling layer is used to call the corresponding code-assisted large model from the model implementation layer according to the code-assisted scene information transmitted by the model encapsulation layer to obtain the large model service required by the code-assisted scene. The model implementation layer includes one or more code-assisted large models, which generate the code-assisted result required by the code-assisted scene through the code-assisted large model, thereby providing the corresponding large model service for different code-assisted scenes and assisting the development end user in code development.

[0072] Specifically, in the embodiment, the plug-in integration layer develops the corresponding IDE plug-in according to the SDK development kit provided by the integrated development environment (IDE) used by different IDE plug-ins when integrating the IDE plug-in, including the IDE plug-in suitable for developing the front-end software and the IDE plug-in suitable for developing the back-end software. Wherein, one IDE can support one or more programming languages, and the plug-in integration layer internally develops or connects to the advanced AI code-assisted large model (multiple choices) through the developed IDE plug-in, so that the code-assisted development system can understand the semantic structure of multiple programming languages, and further provide context-related code suggestions and automatic generation functions. Therefore, the plug-in integration layer integrates multiple code-assisted large models and multiple IDE plug-ins to provide code assistance functions for different development teams and improve the development efficiency as a whole.

[0073] In actual application, the development end user (such as a programmer) selects and installs the development environment plug-in corresponding to the IDE in the IDE, and the development environment plug-in is docked with the IDE. When the UI (User Interface) of the IDE detects the operation behavior of the development end user, the corresponding operation information is transmitted to the ID plug-in in the plug-in integration layer, the large model service interface provided by the model encapsulation layer is called through the ID plug-in, and then the related information is formatted and encapsulated as code-assisted scene information after the specific code-assisted scene is identified in the model encapsulation layer, and the code-assisted scene information is used to access the large model service.

[0074] In addition, in the embodiment, the developed IDE plug-in adopts a layered architecture, each layer is isolated, independent of each other, low coupling between layers, and high cohesion between modules or units in the layer, so that the developed IED plug-in has high reusability, and is easy to maintain, deploy and extend. In another embodiment, the model calling layer and the model implementation layer can be merged into one level of the model layer, thereby obtaining a two-level IDE plug-in including the plug-in integration layer and the model layer. The two-level IDE plug-in provides the large model service interface of the large model service provided by the model layer.

[0075] In this embodiment, after the code assistance system triggers the large model service call request through the integrated IDE plug-in, the large model service interface wrapped by the model encapsulation layer is used to call the model calling layer, and then the model calling layer selects the large model service required by the code assistance scene in the large model service call request according to different code assistance scenes. Among them, the model encapsulation layer encapsulates the code assistance large model to be called, thereby providing a unified encapsulation of the large model service interface for each IDE plug-in in the plug-in integration layer.

[0076] Specifically, in this embodiment, after the model encapsulation layer identifies the code assistance scene in the large model service call request, it calls the large model service provided by the model calling layer according to the identified code assistance scene information, and feeds back to the corresponding IDE plug-in through the unified encapsulation of the large model service interface. Among them, the code assistance scene includes unit test generation scene, complex logic explanation scene, code comment generation scene, code optimization scene, code review scene and code completion scene. After the model encapsulation layer determines the corresponding code assistance scene according to the selection information triggered by the development end user, it calls the model calling layer to obtain the required large model service according to the code assistance scene information and the prompt content input by the development end user after determining the code assistance scene.

[0077] In this embodiment, after the model calling layer obtains the incoming code assistance scene information and the prompt content input by the development end user, it selects the required code assistance large model according to the incoming code assistance scene information and the prompt content input by the development end user. After configuring appropriate request parameters, the code assistance large model of the model implementation layer is called to obtain the code assistance result. Among them, the model calling layer adapts the large model interface of one or more code assistance large models in the model implementation layer. When the code assistance scene information is obtained, the code assistance large model in the model implementation layer is selected according to the incoming code assistance scene information, and the corresponding large model service is obtained by calling the large model interface of the code assistance large model.

[0078] Specifically, in this embodiment, the code assistance large model includes one or more target models. Among them, after the code implementation module obtains the code assistance scene information through the corresponding large model interface, the corresponding target model in the code assistance large model is selected according to the code assistance scene information to generate the corresponding code assistance result, and the generated code assistance result is fed back to the model calling layer through the corresponding large model interface, and then fed back to the corresponding IDE plug-in.

[0079] Further, in the embodiment, the code development assistance system further comprises an extension module, and the management function and the configuration function of the code development assistance system are extended through the extension module, so as to enhance the interaction between the model and the user and the future expansion of the model. The extension module comprises a model management unit, a scene management unit, a model calling configuration unit, a prompt word configuration unit, and a data analysis unit.

[0080] Specifically, in the embodiment, the model management unit can support a large model connected with a dynamically accessed / offlined plug-in, so that the plug-in can follow the latest large model in real time and continuously evolve the code assistance capability of each scene. The scene management unit supports real-time adjustment of different code assistance scenes and corresponding large models, so as to facilitate the plug-in to meet the needs of different developers at any time. The model calling configuration unit supports parameter adjustment of calling the large model, so as to flexibly control the accuracy of the assistance result returned by the large model. The prompt word configuration supports configuration of corresponding prompt words for different scenes. Good prompt engineering can improve the accuracy of the model reply. The data analysis unit analyzes the plug-in use data of the user, and sorts the results of different code assistance scenes, different large models, and corresponding user feedback. Through the data analysis unit, the performance of each large model in different code assistance scenes can be understood, which helps to continuously optimize the model selection and improve the code assistance capability.

[0081] Further, in the embodiment, the code development assistance system further comprises an interaction interface, and the interaction interface provides a conversation window for conversation between the development end user and the code assistance large model. When the code development assistance system assists the development end user in code development, the operation of the user is obtained through the loaded interaction interface, so as to generate a corresponding code assistance result in real time to assist the development end user in code development. FIG. 2 is a step flow chart of an embodiment of code development assistance work in any code assistance scene except the code completion scene. Please refer to FIG. 2, and the following is a detailed description of each step of the code development assistance work in any code assistance scene except the code completion scene.

[0082] Step C1: loading an interaction interface for interaction between the development end user and the code assistance large model, and the development end user triggers the start of the large model service through the loaded interaction interface;

[0083] Step C2: obtaining the input information of the development end user through the interaction interface, and determining the code editing position associated with the development end operation;

[0084] Step C3: determining the current code assistance scene according to the obtained input information;

[0085] Step C4: According to the determined code assistance scene and the code editing position, the corresponding AI code assistance large model is called to generate the corresponding code assistance result, and the interactive interface is displayed to assist the development end user to develop the code.

[0086] Therefore, through the above method, the code assistance system can receive the input information of the development end user through the interactive interface, and then generate the corresponding prompt information according to the code assistance scene, the code editing task and the preconfigured prompt template in response to the recognition result of the input information. Finally, based on the generated prompt information, the corresponding large model service is called to generate the corresponding code assistance result. By recognizing the code assistance scene through the interactive information, the code assistance large model can generate a code assistance result that is more in line with the code assistance needs of the development end user and has higher accuracy according to the specific code assistance scene, thereby improving the development efficiency. At the same time, the preconfigured prompt template can be passed to the code assistance large model as the prompt information of the input information according to a certain format constraint, so that the development end user and the code assistance model can interact more accurately and efficiently.

[0087] FIG. 3 is a step flow chart illustrating an embodiment of the code development assistance work of the present application in the code completion scene. The code development assistance system performs the code development assistance work in the code completion scene through the steps as shown in FIG. 3:

[0088] Step D1: An interactive interface for interaction between the development end user and the code assistance large model is loaded, and the development end user triggers the start of the large model service through the loaded interactive interface;

[0089] Step D2: The code editing position associated with the development end operation is determined through the interactive interface;

[0090] Step D3: The code context information is obtained according to the determined code editing position;

[0091] Step D4: According to the obtained code context information, the corresponding AI code assistance large model is called to generate the corresponding code completion information, and the interactive interface is displayed to assist the development end user to develop the code.

[0092] Thus, the code assistance system can obtain the code context information at the code editing position in the IDE through the above steps, and pass the code context information to the code assistance large model. Among them, the code assistance large model supports FIM (fill-in-the-middle, a method for enabling a language model to fill in the middle by modifying an existing framework), returns the code to be completed according to the context information, and displays a gray code hint block (hint code) in the plug-in, whether the user feels to adopt the completion suggestion. The user clicks tab to actually insert the hint code into the IDE, thereby completing the code completion operation. This method of obtaining the recommended code of the model by calling the code assistance large model supporting the FIM mode greatly improves the development efficiency.

[0093] The present specification also discloses an embodiment of a code development assistance method based on an AI code assistance large model. FIG. 4 is a system architecture diagram illustrating an embodiment of a code development assistance system based on an AI code assistance large model of the present application. Please refer to FIG. 4, and the following is a detailed description of each step of the code development assistance method based on the AI code assistance large model.

[0094] Step S1: Obtain a large model service call request, thereby triggering the start of the large model service; wherein the large model service call request includes code assistance scenario information of any code assistance scenario in the entire code development cycle.

[0095] In this embodiment, corresponding IDE plug-ins are developed according to the integrated development environment used by different IDE plug-ins, and the large model service call request is triggered by the developed IDE plug-in, thereby obtaining the large model service required by the code assistance scenario in the large model service call request. Among them, when integrating the development IDE plug-in, the corresponding IDE plug-in is developed according to the SDK development kit provided by the integrated development environment (IDE) used by different IDE plug-ins, including IDE plug-ins suitable for developing front-end software and IDE plug-ins suitable for developing back-end software. Among them, an IDE can support one or more programming languages, and the plug-in integration layer is built-in or connected to an advanced AI code assistance large model (multiple choices) through the developed IDE plug-in, so that the code assistance development system can understand the semantic structure of multiple programming languages, and further provide context-related code suggestions and automatic generation functions. Thus, by integrating multiple code assistance large models and multiple IDE plug-ins, code assistance functions are provided for development teams with different development teams, and overall development efficiency is improved.

[0096] In actual application, a development end user (such as a programmer) selects and installs a development environment plug-in corresponding to the IDE in the IDE, and the development environment plug-in is docked with the IDE. When the UI (User Interface) of the IDE detects the operation behavior of the development end user, the corresponding operation information is transmitted to the ID plug-in, the large model service interface is called through the ID plug-in, the specific code assistance scene is identified, the related information is formatted and packaged as code assistance scene information, and the large model service is accessed through the code assistance scene information.

[0097] Step S2: identifying the code assistance scene information in the large model service calling request.

[0098] Step S3: calling the corresponding code assistance large model based on the identified code assistance scene information.

[0099] In the embodiment, after triggering the large model service calling request through the integrated IDE plug-in, the code assistance scene information is identified according to the obtained large model service calling request, and then the required large model service is selected for different code assistance scenes. The code assistance scene includes a unit test generation scene, a complex logic explanation scene, a code annotation generation scene, a code optimization scene, a code review scene, and a code completion scene. After determining the corresponding code assistance scene according to the selection information triggered by the development end user, the required large model service is obtained according to the code assistance scene information and the prompt content input by the development end user after determining the code assistance scene.

[0100] Step S4: the code assistance large model generates the required code assistance result based on the transmitted code assistance scene information, provides the corresponding large model service for different code assistance scenes, and assists the development end user in code development.

[0101] In the embodiment, after obtaining the transmitted code assistance scene information and the prompt content input by the development end user through the above steps, the required code assistance large model is selected according to the transmitted code assistance scene information and the prompt content input by the development end user. After configuring appropriate request parameters, the code assistance large model is called to obtain the code assistance result. When transmitting information, the code assistance scene information and the prompt content input by the development end user are obtained through the large model interface, the corresponding code assistance large model is selected according to the transmitted code assistance scene information, the corresponding large model service is obtained through the large model interface of the code assistance large model, and then the corresponding IDE plug-in is fed back.

[0102] Further, in the embodiment, when the end-user develops the code with the assistance of the above method, an interactive interface can be provided to provide a conversation window for the end-user to have a conversation with the code-assisted large model. When the end-user develops the code, the end-user's operation is obtained through the loaded interactive interface, and the corresponding code-assisted result is generated in real time to assist the end-user in developing the code. FIG. 2 is a step flow chart of an embodiment of the code development assistance in any code-assisted scenario other than the code completion scenario. Please refer to FIG. 2, and the following is a detailed description of each step of the code development assistance in any code-assisted scenario other than the code completion scenario.

[0103] Step C1: loading an interactive interface for interaction between the end-user and the code-assisted large model, and the end-user triggers the start of the large model service through the loaded interactive interface.

[0104] Step C2: obtaining the input information of the end-user through the interactive interface, and determining the code editing position associated with the end-user's operation.

[0105] Step C3: determining the current code-assisted scenario according to the obtained input information.

[0106] Step C4: generating a corresponding code-assisted result according to the determined code-assisted scenario and code editing position, and displaying the code-assisted result through the interactive interface to assist the end-user in developing the code.

[0107] Therefore, through the above method, the input information of the end-user can be received through the interactive interface, and then the corresponding prompt information is generated according to the code-assisted scenario, the code editing task, and the preconfigured prompt template in response to the recognition result of the input information. Finally, the corresponding large model service is called based on the generated prompt information to generate the corresponding code-assisted result. By recognizing the code-assisted scenario through the interactive information, the code-assisted large model can generate a code-assisted result that is more in line with the code-assisted needs of the end-user and has higher accuracy according to the specific code-assisted scenario, thereby improving the development efficiency. At the same time, the preconfigured prompt template can be delivered to the code-assisted large model as prompt information according to a certain format constraint, so that the end-user and the code-assisted model can have more accurate and efficient interaction.

[0108] FIG. 3 is a step flow chart of an embodiment of the code development assistance in the code completion scenario. The code development assistance in the code completion scenario is performed through the steps shown in FIG. 3.

[0109] Step D1: loading an interactive interface for interaction between an end user and a code-assisted large model, and the end user triggers the start of the large model service through the loaded interactive interface;

[0110] Step D2: determining a code editing position associated with the development end operation through the interactive interface;

[0111] Step D3: obtaining code context information according to the determined code editing position;

[0112] Step D4: calling a corresponding AI code-assisted large model according to the obtained code context information to generate corresponding code completion information, and displaying the code completion information to the assisted development end user through the interactive interface for prompting.

[0113] Therefore, through the above steps, the code context information can be obtained at the code editing position in the IDE, and the code context information is transmitted to the code-assisted large model. The code-assisted large model supports FIM (fill-in-the-middle, a method for enabling a language model to fill in the middle by modifying an existing framework), returns the code to be completed according to the context information, and displays a gray code hint block (hint code) in the plug-in, and the user decides whether to adopt the completion suggestion. After the user clicks tab, the hint code is actually inserted into the IDE, thereby completing the code completion operation. This method of obtaining the recommended code of the model by calling the code-assisted large model supporting the FIM mode greatly improves the development efficiency.

[0114] The present specification also provides a computer-readable medium storing computer program code, which, when executed by a processor, implements the AI code-assisted large model-based code development assistance method as described above.

[0115] The present specification also provides an AI code-assisted large model-based code development assistance system device, including a memory storing instructions executable by a processor, and a processor for executing the instructions in the instruction memory to implement the AI code-assisted large model-based code development assistance method as described above.

[0116] The provision of the foregoing description of the disclosure is to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein can be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the broadest scope consistent with the principles and novel features disclosed herein.

[0117] Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.

[0118] The various illustrative logical blocks, modules, and circuits described in connection with the embodiments disclosed herein can be implemented or performed with a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor can be a microprocessor, but in the alternative, the processor can be any conventional processor, controller, microcontroller, or state machine. A processor can also be implemented as a combination of computing devices, e.g., a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a DSP core, or any other such configuration.

[0119] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software layer executed by a processor, or in a combination of the two. A software layer can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

[0120] In one or more exemplary embodiments, the functions described can be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functions can be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. Storage media can be any available media that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. Disk and disc, as used herein, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

Claims

1. An AI code-based large model code development assistance system, characterized by, The large model service of the code assistance large model is obtained through an IDE plug-in developed by integration, so as to assist the code development of the development end user; wherein, the IDE plug-in comprises a plug-in integration layer, a model encapsulation layer, a model calling layer and a model implementation layer; wherein, The plug-in integration layer is used to integrate one or more IDE plug-ins for code assistance large model development, and triggers a large model service calling request through the integrated IDE plug-in to obtain the required large model service; wherein, the code assistance scene information of any code assistance scene in the whole code development cycle is included in the large model service calling request; The model encapsulation layer is used to identify the code assistance scene information in the large model service calling request, and call the model calling layer through the identified code assistance scene information to obtain the corresponding large model service and feedback to the corresponding IDE plug-in; The model calling layer is used to call the corresponding code assistance large model from the model implementation layer according to the code assistance scene information transmitted by the model encapsulation layer, so as to obtain the large model service required by the code assistance scene; The model implementation layer comprises one or more code assistance large models, which generate the code assistance results required by the code assistance scene through the code assistance large models, so as to provide corresponding large model services for different code assistance scenes and assist the code development of the development end user. 2.The AI code-based large model code development assistance system of claim 1, wherein When integrating the IDE plug-in, the plug-in integration layer develops the corresponding IDE plug-in according to the integrated development environment used by different IDE plug-ins, and sends a large model service calling request to the model encapsulation model through the developed IDE plug-in, so as to obtain the large model service required by the code assistance scene in the large model service calling request; wherein, the code assistance scene includes a unit test generation scene, a complex logic explanation scene, a code annotation generation scene, a code optimization scene, a code review scene and a code completion scene. 3.The AI code-based large model code development assistance system of claim 1, wherein, The model encapsulation layer uniformly encapsulates the code assistance large models to be called, so as to provide a uniformly encapsulated large model service interface for each IDE plug-in in the plug-in integration layer; wherein, after the model encapsulation layer identifies the code assistance scene in the large model service calling request, it calls the large model service provided by the model calling layer according to the identified code assistance scene information, and feeds back to the corresponding IDE plug-in through the uniformly encapsulated large model service interface. 4.The AI code-based large model-assisted code development assistance system of claim 3, characterized in that, The model calling layer is used to adapt the large model interface of one or more code assistance large models in the model implementation layer, and call the corresponding code assistance large model from the model implementation layer through the adapted large model interface, so as to obtain the large model service required by the code assistance scene; wherein, after the model calling layer obtains the code assistance scene information, it selects the code assistance large model in the model implementation layer according to the transmitted code assistance scene information, and obtains the corresponding large model service through the large model interface of the code assistance large model. 5.The AI code-based large model code development assistance system of claim 4, wherein, The code-assisted large model comprises one or more target models; wherein, after the code implementation module obtains the code-assisted scene information through the corresponding large model interface, the corresponding target model in the code-assisted large model is selected according to the code-assisted scene information to generate the corresponding code-assisted result, and the generated code-assisted result is fed back to the model calling layer through the corresponding large model interface, and then fed back to the corresponding IDE plug-in. 6.The AI code-based large model code development assistance system of claim 1, wherein The code development assistance system based on the AI code-assisted large model further comprises an extension module; wherein, the code development assistance system based on the AI code-assisted large model extends the management function and configuration function of the code development assistance system through the extension module, so as to realize the management of the code-assisted large model. 7.The AI code-based large model-assisted code development assistance system of claim 6, wherein, The extension module comprises a model management unit, a scene management unit, a model calling configuration unit, a prompt word configuration unit and a data analysis unit; wherein, The model management unit is used for dynamically accessing / going offline the code-assisted large model connected by the IDE plug-in, so as to make the corresponding IDE plug-in update the code-assisted large model in real time; The scene management unit is used for adjusting different code-assisted scenes and corresponding code-assisted large models, so as to make the corresponding IDE plug-in meet the needs of different developers in real time; The model calling configuration unit is used for adjusting the calling parameters of the code-assisted large model, so as to adjust the accuracy of the code-assisted result returned by the code-assisted large model; The prompt word configuration unit is used for configuring the prompt words of different code-assisted scenes, so as to improve the accuracy of the reply of the code-assisted large model; The data analysis unit is used for analyzing the plug-in use data of the user, arranging different code-assisted scenes, different code-assisted large models and corresponding user feedback results according to the plug-in use data of the user, so as to optimize the large model assistance model. 8.The AI code-based large model code development assistance system of claim 1, wherein, The code development assistance system based on the AI code-assisted large model further comprises an interactive interface for realizing the interactive operation between the development end user and the code-assisted large model; wherein, when the code development assistance system based on the AI code-assisted large model assists the development end user to develop the code, the development end operation of the user is obtained through the loaded interactive interface, so as to generate the corresponding code-assisted result in real time to assist the development end user to develop the code. 9.The AI code-based large model code development assistance system of claim 8, wherein, The code development assistance system based on the AI code-assisted large model performs the code development assistance work in any code-assisted scene except the code completion scene through the following steps: Step C1: loading an interactive interface for the interactive operation between the development end user and the code-assisted large model, and the development end user triggers the large model service through the loaded interactive interface; Step C2: obtaining the input information of the development end user through the interactive interface, and determining the code editing position associated with the development end operation; Step C3: determining the current code-assisted scene according to the obtained input information; Step C4: calling the corresponding AI code-assisted large model to generate the corresponding code-assisted result according to the determined code-assisted scene and code editing position, and displaying the code-assisted result through the interactive interface to assist the development end user to develop the code. 10.The AI code-based large model code development assistance system of claim 8, wherein, The AI code-assisted large model-based code development assistance system performs code development assistance in a code completion scenario through the following steps: Step D1: load an interaction interface for interaction between a development end user and a code assistance large model, and the development end user triggers the large model service through the loaded interaction interface; Step D2: determine the code editing position associated with the development end operation through the interaction interface; Step D3: obtain code context information according to the determined code editing position; Step D4: generate corresponding code completion information according to the obtained code context information, and display the code completion information to the assistance development end user through the interaction interface for prompting. 11.A code development assistance method based on an AI code assisted large model, characterized in that, The steps include: Step S1: obtain a large model service calling request to trigger the large model service; wherein the large model service calling request includes code assistance scene information in any code assistance scene in the code development whole cycle; Step S2: identify the code assistance scene information in the large model service calling request; Step S3: call the corresponding code assistance large model based on the identified code assistance scene information; Step S4: the code assistance large model generates the required code assistance result based on the input code assistance scene information, provides corresponding large model services for different code assistance scenes, and assists the development end user in code development. 12.The AI code-based large model code development assistance method of claim 11, wherein, The AI code-assisted large model-based code development assistance method develops corresponding IDE plug-ins according to the integrated development environments used by different IDE plug-ins, triggers a large model service calling request through the developed IDE plug-in, and obtains the large model service required by the code assistance scene in the large model service calling request; wherein the code assistance scene includes a unit test generation scene, a complex logic explanation scene, a code comment generation scene, a code optimization scene, a code review scene, and a code completion scene. 13.The AI code-based large model code development assistance method of claim 11, wherein, The AI code-assisted large model-based code development assistance method block encapsulates the code assistance large model to be called, thereby providing a uniformly encapsulated large model service interface for each IDE plug-in; wherein after the model encapsulation layer identifies the code assistance scene in the large model service calling request, it calls the large model service provided by the model calling layer according to the identified code assistance scene information, and feeds back to the corresponding IDE plug-in through the uniformly encapsulated large model service interface. 14.The AI code-based large model code development assistance method of claim 13, wherein, When the AI code-assisted large model-based code development assistance method calls the code assistance large model required by the code assistance scene, it selects the code assistance large model in the model implementation layer according to the input code assistance scene information, and obtains the corresponding large model service through the large model interface adapted to the code assistance large model. 15.The AI code-based large model code development assistance method of claim 11, wherein, The code-assisted large model comprises one or more target models; wherein the code development assistance method based on the AI code-assisted large model obtains code assistance scene information through a corresponding large model interface, selects a corresponding target model in the code assistance large model according to the code assistance scene information to obtain a corresponding code assistance result, and feeds back the generated code assistance result to the model calling layer through the corresponding large model interface, and then feeds back to the corresponding IDE plug-in. 16.The AI code-based large model code development assistance method of claim 11, wherein, The code development assistance method based on the AI code-assisted large model is provided with an interactive interface for realizing the interactive operation between the development end user and the code assistance large model; wherein when the code development assistance method based on the AI code-assisted large model assists the development end user in code development, the development end user's operation is obtained through the loaded interactive interface, so as to generate corresponding code assistance results in real time to assist the development end user in code development. 17.The AI code-based large model code development assistance method of claim 16, wherein, The code development assistance method based on the AI code-assisted large model performs code development assistance work in any code assistance scene except the code completion scene through the following steps: Step C1: load an interactive interface for interaction between the development end user and the code assistance large model, and the development end user triggers the start of large model service through the loaded interactive interface; Step C2: obtain the input information of the development end user through the interactive interface, and determine the code editing position associated with the development end operation; Step C3: determine the current code assistance scene according to the obtained input information; Step C4: generate corresponding code assistance results according to the determined code assistance scene and code editing position, and display through the interactive interface to assist the development end user in code development. 18.The AI code-based large model code development assistance method of claim 16, wherein, The code development assistance method based on the AI code-assisted large model performs code development assistance work in the code completion scene through the following steps: Step D1: load an interactive interface for interaction between the development end user and the code assistance large model, and the development end user triggers the start of large model service through the loaded interactive interface; Step D2: determine the code editing position associated with the development end operation through the interactive interface; Step D3: obtain the code context information according to the determined code editing position; Step D4: generate corresponding code completion information according to the obtained code context information, and display to the development end user through the interactive interface for prompting.

19. A computer readable medium having stored computer program code, the computer program code comprising instructions for causing a computer to perform the method of any one of claims 1 to 18.

19. The computer program code, when executed by the processor, implements the method of any one of claims 11-18. 20.A code development assistance apparatus based on an AI code assisted large model, characterized by Comprise: a memory for storing instructions executable by the processor; and a processor for executing the instructions to implement the method of any one of claims 11-18.

Images