Intelligent DevOps assembly line generation system based on natural language description

Abstract

The invention relates to the technical field of artificial intelligence and software development, in particular to a DevOps assembly line intelligent generation system based on natural language description. According to the intelligent generation system for the DevOps assembly line based on the natural language description, an interaction layer is responsible for visually editing an interaction interface and outputting a generated executable script and a flow chart to a user; the intelligent processing layer takes Python and Go as development languages, fuses a knowledge graph technology, supports a dynamic code generation capability and realizes intelligent business process management and automatic execution; the infrastructure layer comprises three core modules, namely a metadatabase, a sandbox environment and an execution engine, and is responsible for providing basic support and resource guarantee for system operation. According to the intelligent generation system for the DevOps assembly line based on the natural language description, the configuration threshold of the DevOps assembly line is remarkably reduced, the cross-platform compatibility is ensured, the safety and reliability of the assembly line are greatly improved, and production accidents caused by manual configuration errors can be effectively avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence and software development, and particularly relates to a DevOps pipeline intelligent generation system based on natural language description. BACKGROUND

[0002] Currently, the configuration and maintenance of the DevOps pipeline highly depend on manual scripting (such as Jenkinsfile, etc.), which requires developers to have professional syntax knowledge and tool chain experience, and the learning cost is high and errors are easily introduced due to human negligence. Although low-code / visual tools (such as Gitee DevOps) can simplify part of the configuration process, their flexibility is limited and it is difficult to support complex logic (such as multi-environment differentiated deployment, dynamic condition judgment, etc.).

[0003] In recent years, automatic solutions based on natural language processing (NLP) have begun to explore the direct conversion of user requirements into executable code, but the existing technology has the following limitations:

[0004] (1) Insufficient semantic understanding: general large models (such as GPT-4) lack precise analysis of DevOps field terms (such as "blue-green deployment" "SonarQube scanning"), resulting in a disconnection between the generated content and the actual tool chain;

[0005] (2) Tool chain adaptation rigidity: most solutions only support a single platform and cannot dynamically adjust according to the actual enterprise infrastructure (such as private Kubernetes clusters, self-built Harbor repositories);

[0006] (3) Lack of verification mechanism: the generated pipeline has not been tested in a sandbox or confirmed by the user, which may directly cause production environment failures due to permission errors, missing dependencies, etc.

[0007] Therefore, an automatic pipeline generation method that combines large model intelligent understanding, dynamic tool chain binding, and safety verification is urgently needed to reduce the technical threshold and improve reliability.

[0008] Based on the above situation, the present application provides a DevOps pipeline intelligent generation system based on natural language description. SUMMARY

[0009] The present application provides a simple and efficient DevOps pipeline intelligent generation system based on natural language description to overcome the shortcomings of the prior art.

[0010] The present application is implemented by the following technical solutions:

[0011] A DevOps pipeline intelligent generation system based on natural language description includes an interaction layer, an intelligent processing layer, and an infrastructure layer.

[0012] The interaction layer is responsible for providing a natural language and visual editing interactive interface through a web front end (React / Vue) and a chat robot interface (such as a Slack Bot) to output a generated executable script (such as a Jenkinsfile) and a flowchart to a user;

[0013] The intelligent processing layer uses Python and Go as development languages, integrates knowledge graph technology, and supports dynamic code generation capabilities to achieve intelligent business process management and automated execution.

[0014] The intelligent processing layer is configured with a semantic parsing engine, a workflow orchestrator, and a tool chain adapter.

[0015] The semantic parsing engine is used to provide a large model and a rule engine, the workflow orchestrator is responsible for generating and optimizing a directed acyclic graph (DAG), and the tool chain adapter is responsible for providing plug-in integration of various DevOps tools.

[0016] The infrastructure layer includes a metadata database, a sandbox environment, and an execution engine, which are responsible for providing basic support and resource guarantees for system operation.

[0017] The metadata database is used to store enterprise tool chain configuration information, such as Kubernetes cluster addresses, cloud platform account permissions, and other key parameters, to achieve unified management and rapid invocation.

[0018] The sandbox environment is used to provide an isolated testing space designed for high-risk operations such as code testing and vulnerability verification to ensure that the main system is not affected.

[0019] The execution engine provides an automated task scheduling component responsible for receiving instructions from the upper layer and driving specific tasks to be executed.

[0020] The isolated testing space is implemented based on a container or a Kubernetes orchestration system.

[0021] The Jenkins agent node (Agent) is used as the automated task scheduling component to achieve distributed and scalable automated task execution.

[0022] A DevOps pipeline intelligent generation method based on natural language description includes the following steps:

[0023] Step S1: Natural language requirement input and intent analysis

[0024] The user inputs an unstructured pipeline requirement description through an interactive interface. A fine-tuned large language model (such as GPT-4 or Claude 3) is used to identify the core intent of the user's requirements. Through entity extraction, key parameters are extracted, including trigger conditions, execution actions, and constraint rules.

[0025] In step S1, if the fine-tuned large language model identifies the core intent of the user's requirements as ambiguous, an intelligent agent is called to ask the user for additional information through dialogue.

[0026] Step S2: Abstract workflow logic generation

[0027] The parsed requirements are converted into abstract workflow logic, represented as a directed acyclic graph (DAG) to represent task dependency relationships. Conflict detection is performed to check for logical contradictions. Performance optimization is achieved by merging parallel tasks.

[0028] Step S3: Tool chain adaptation and script generation

[0029] The enterprise meta-database is queried to match the enterprise tool chain, and the corresponding specific tools are dynamically bound.

[0030] A standardized configuration file is output based on the enterprise tool chain, and enterprise-level parameters are injected to generate executable code.

[0031] Step S4: Sandbox verification and user confirmation

[0032] The generated pipeline is simulated in an isolated sandbox to automatically detect syntax errors, permission deficiencies, and dependency conflicts.

[0033] The user is shown a flowchart and generated code, and can customize visual interaction for correction.

[0034] In step S4, two types of visual interaction correction methods are supported: natural language instruction correction and direct editing of DAG nodes to adjust the execution order.

[0035] Step S5: Deployment and monitoring

[0036] The verified pipeline configuration file is pushed to the target platform for one-click deployment.

[0037] Real-time execution logs are collected. If the execution fails, the user is automatically notified and repair suggestions are provided.

[0038] The beneficial effects of the present application are: the DevOps pipeline intelligent generation system based on natural language description significantly reduces the configuration threshold of the DevOps pipeline, supports dynamic adaptation of enterprise heterogeneous tool chains, ensures cross-platform compatibility, and greatly improves the safety and reliability of the pipeline through the built-in sandbox verification and visual correction mechanism, which can effectively avoid production accidents caused by manual configuration errors. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0040] ATTACHMENT Figure 1 The schematic diagram of the DevOps pipeline intelligent generation system based on natural language description of the present application is shown in the figure. DETAILED DESCRIPTION

[0041] In order to make the person skilled in the art better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be described in detail below. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0042] The DevOps pipeline intelligent generation system based on natural language description comprises an interaction layer, an intelligent processing layer and an infrastructure layer;

[0043] The interaction layer is responsible for providing a natural language and visual editing interactive interface through a Web front end (React / Vue) and a chat robot interface (such as a Slack Bot), outputting generated executable scripts (such as Jenkinsfile) and flowcharts to users;

[0044] The intelligent processing layer uses Python and Go as development languages, integrates knowledge graph technology, and supports dynamic code generation capability to realize intelligent business process management and automatic execution;

[0045] The intelligent processing layer is configured with a semantic parsing engine, a workflow orchestrator and a tool chain adapter;

[0046] Among them, the semantic analysis engine is used to provide a large model and a rule engine, the workflow arranger is responsible for realizing directed acyclic graph (DAG) generation and optimization, and the tool chain adapter is responsible for providing plug-in integration of various DevOps tools.

[0047] The infrastructure layer includes three core modules, namely, a metadata database, a sandbox environment and an execution engine, which are responsible for providing basic support and resource guarantee for system operation.

[0048] The metadata database is used to store enterprise tool chain configuration information, such as Kubernetes cluster address, cloud platform account permission and other key parameters, to realize unified management and rapid calling.

[0049] The sandbox environment is used to provide an isolated test space, which is designed for high-risk operations such as code testing and vulnerability verification, to ensure that the main system is not affected.

[0050] The execution engine provides an automated task scheduling component, which is responsible for receiving instructions from the upper layer and driving specific tasks to be executed.

[0051] The isolated test space is implemented based on containers or Kubernetes orchestration system.

[0052] The agent node (Agent) of Jenkins is used as the automated task scheduling component to realize distributed and scalable automated task execution.

[0053] The metadata database serves as the configuration hub, the sandbox environment guarantees the security boundary, and the execution engine relies on containerization technology to realize elastic resource allocation and efficient task processing. The three work together to form a solid foundation for the enterprise-level DevOps system.

[0054] The natural language description-based DevOps pipeline intelligent generation method includes the following steps:

[0055] Step S1: Natural language requirement input and intent analysis

[0056] Users input unstructured pipeline requirement descriptions through interactive interfaces (such as Web forms and chat robots), such as: "Run unit tests every time the code is committed to the main branch, and if the coverage is greater than or equal to 90%, build a Docker image and deploy it to the pre-production environment." "Perform database backup every Friday at 3 a.m., compress and upload to S3 storage bucket, and keep the last 5 versions."

[0057] Fine-tuned large language models (e.g., GPT-4, Claude 3) identify the core intent of user requirements (e.g., "CI / CD build" "scheduled backup") through entity extraction, extracting key parameters, including trigger conditions (e.g., "code commit to main branch" "every Friday at 3 a.m."), execution actions (e.g., "run unit tests" "build Docker image"), and constraint rules (e.g., "coverage ≥ 90%" "keep the last 5 versions").

[0058] In step S1, if the fine-tuned large language model identifies the core intent of user requirements, it is found that the user's requirements are ambiguous (e.g., "deploy to test environment" does not specify the specific environment), and the agent is called to ask the user for additional information through dialogue.

[0059] Step S2: Abstract workflow logic generation

[0060] Convert the parsed requirements into abstract workflow logic, and represent the task dependency relationship in the form of a directed acyclic graph (DAG), for example: [code commit]→[unit test]→(coverage ≥ 90%)→[build image]→[deploy pre-release];

[0061] Perform conflict detection to check for logical contradictions (e.g., "scheduled task" and "code trigger" coexist, prefer event-driven), and merge parallel tasks (e.g., "unit test" and "code scanning" are executed simultaneously) to achieve performance optimization;

[0062] Step S3: Tool chain adaptation and script generation

[0063] Query the enterprise metadata database, match the enterprise tool chain, and dynamically bind the corresponding specific tools, for example: code repository→GitLab; build tool→Docker / Jenkins; deployment platform→Kubernetes;

[0064] Output a standardized configuration file (e.g., Jenkinsfile) according to the enterprise tool chain, and inject enterprise-level parameters (e.g., API key, cluster name) to generate executable code;

[0065] Step S4: Sandbox verification and user confirmation

[0066] Simulate the execution of the generated pipeline in an isolated sandbox, automatically detect syntax errors (e.g., YAML format errors), permission deficiencies (e.g., insufficient AWS S3 upload permissions), and dependency conflicts (e.g., Dockerfile does not define the base image);

[0067] Show the user the flowchart and generated code, and customize the visual interaction for correction;

[0068] In the step S4, two visualization interaction correction methods are supported, including natural language instruction (such as "change the deployment to blue-green mode") correction and direct editing of the directed acyclic graph (DAG) node and drag adjustment of the execution order for correction.

[0069] Step S5: deployment and monitoring

[0070] The pipeline configuration file that passes the verification is pushed to the target platform (such as the.github / workflows directory of the GitHub repository) to realize one-key deployment.

[0071] Real-time collection of execution logs, if the execution fails (such as test failure), the user is automatically notified and repair suggestions are provided (such as "coverage is only 85%, whether to adjust the threshold?").

[0072] In summary, the DevOps pipeline intelligent generation system based on natural language description significantly reduces the configuration threshold of the DevOps pipeline through natural language interaction and intelligent automation technology, enabling non-professionals to quickly generate standardized and executable CI / CD pipelines, while supporting dynamic adaptation to enterprise heterogeneous tool chains to ensure cross-platform compatibility. The built-in sandbox verification and visual correction mechanism significantly improves the safety and reliability of the pipeline, effectively avoiding production accidents caused by manual configuration errors.

[0073] The DevOps pipeline intelligent generation system based on natural language description can be widely applied to enterprise-level CI / CD rapid construction, multi-cloud / hybrid cloud deployment, and compliance automation scenarios, and can shorten the traditional pipeline configuration work of several hours to minutes. At the same time, through continuous learning optimization of the generation strategy, it can long-term improve the software delivery efficiency, promote the transformation of DevOps to intelligence and self-service, save a lot of manpower and time cost for enterprises, and significantly reduce the technical risk.

[0074] The above-described embodiments are only one of the specific implementations of the present application, and the usual changes and replacements made by those skilled in the art within the scope of the technical solutions of the present application should be included in the protection scope of the present application.

Claims

1. A system for intelligent generation of DevOps pipeline based on natural language description, characterized in that: It comprises an interaction layer, an intelligent processing layer and an infrastructure layer. The interaction layer is responsible for providing a natural language and visualized editing interactive interface through a web front end and a chat robot interface, and outputting generated executable scripts and flowcharts to users. The intelligent processing layer uses Python and Go as development languages, integrates knowledge graph technology, and supports dynamic code generation capabilities to realize intelligent business process management and automated execution. The infrastructure layer includes a metadata database, a sandbox environment and an execution engine, which are responsible for providing basic support and resource guarantee for system operation. The intelligent processing layer is configured with a semantic parsing engine, a workflow orchestrator and a tool chain adapter. 2.The natural language description based DevOps pipeline intelligent generation system of claim 1, wherein: The semantic parsing engine is used to provide large models and rule engines, the workflow orchestrator is responsible for generating and optimizing directed acyclic graphs (DAGs), and the tool chain adapter is responsible for providing plug-in integration of various DevOps tools. The metadata database is used to store enterprise tool chain configuration information. 3.The natural language description based DevOps pipeline intelligent generation system of claim 2, wherein: The sandbox environment is used to provide an isolated test space. The execution engine provides an automated task scheduling component, which is responsible for receiving instructions from the upper layer and driving specific tasks to be executed. The isolated test space is implemented based on containers or Kubernetes orchestration systems.

4. The natural language description based DevOps pipeline intelligent generation system according to claim 3, characterized in that: The automated task scheduling component uses Jenkins agent nodes to achieve distributed and scalable automated task execution.

5. The natural language description based DevOps pipeline intelligent generation system according to claim 3, wherein: The system implementation based on any one of claims 1-5 comprises the following steps: 6.A method for intelligent generation of a DevOps pipeline based on a natural language description, characterized in that: Step S1: natural language requirement input and intent analysis Users input unstructured pipeline requirement descriptions through the interactive interface, use fine-tuned large language models to identify the core intent of user requirements, extract key parameters including trigger conditions, execution actions and constraint rules through entity extraction, and perform conflict detection, check logical contradictions, and perform performance optimization by merging parallel tasks. Step S2: abstract workflow logic generation Step S3: tool chain adaptation and script generation Query the enterprise metadata database, match the enterprise tool chain, and dynamically bind the corresponding specific tools. Output standardized configuration files according to the enterprise tool chain, and inject enterprise-level parameters to generate executable code. Step S4: sandbox verification and user confirmation Simulate the execution of the generated pipeline in an isolated sandbox, automatically detect syntax errors, permission omissions and dependency conflicts, and show the flowchart and generated code to the user for visual interaction and correction. Step S5: deployment and monitoring Push the verified pipeline configuration file to the target platform to realize one-click deployment. Real-time collection of execution logs, if the execution fails, automatically notify the user and provide repair suggestions. In step S1, if the fine-tuned large language model is used to identify the core intent of user requirements and finds that the user's requirements are ambiguous, the intelligent agent will ask the user for additional information through a dialogue. In step S4, two visual interaction correction methods are supported, namely natural language instruction correction and direct editing of directed acyclic graph (DAG) nodes, and drag-and-drop adjustment of execution order for correction. ​ 7. The method of claim 6, wherein: ​ 8.The method for natural language description based DevOps pipeline intelligent generation according to claim 6, characterized in that: ​

Images