RAGflow+AI: Realizing Intelligent "Test Requirements Analyst"

Introduction to RAGflow 

RAGflow is a powerful open source retrieval-augmented generation (RAG) engine that is based on deep document understanding technology and can accurately extract key information from unstructured data in various complex formats. The engine supports a variety of heterogeneous data sources and combines large language models and embedding representation technology to provide users with accurate and reliable answers and well-reasoned citations.

Installation and configuration of RAGflow 

The installation of RAGflow is relatively simple. Here are the specific steps:

Environmental preparation :

• Make sure Docker is installed on your system. Docker simplifies the deployment, scaling, and management of applications.

Pull the RAGflow Docker image :

 Open a terminal or command line interface and execute the following command to pull the RAGflow image from Docker Hub:

docker pull infiniflow/ragflow

Run the RAGflow container :

 Use the following command to start the RAGflow container and specify the port mapping (assuming that it is mapped to port 8080):

docker run -d -p 8080:8080 --name ragflow infiniflow/ragflow

After startup, you can access it through the browserhttp://localhost:8080to use the RAGflow web interface.

Upload requirement documents and build a knowledge base 

RAGflow uploads requirement documents through API and builds a knowledge base:

Prepare requirements document :

Organize the requirement documents into JSON, Markdown, PDF and other formats to ensure the content is clear and accurate.

Call the upload interface :

import  requests

# RAGflow upload interface URL (need to be modified according to actual situation)
upload_url =  'http://localhost:8080/api/documents/upload'
file_path =  '/path/to/your/document.pdf'
headers = { 'Content-Type' :  'multipart/form-data' }
response = requests.post(upload_url, headers=headers, files={ 'file' : open(file_path,  'rb' )})

if  response.status_code ==  200 :
    upload_result = response.json()
    print(upload_result)
else :
    print( f"File upload failed, status code: {response.status_code} " )
    print(response.text)

Will/path/to/your/document.pdfReplace with the actual document path. RAGflow will parse the document content and automatically classify and store it.

Verify the upload result :

Verify whether the document has been successfully uploaded to the knowledge base through the RAGflow web interface or API interface.

Generate test scenarios corresponding to requirements 

Although RAGflow itself does not directly provide the function of AI generating test cases, it can be integrated with third-party AI services to achieve automatic generation of test cases. The following are the steps to call AI to generate test scenarios through the interface:

Preparation requirements description :

The requirement description is extracted from the RAGflow knowledge base through the interface as the input of the AI ​​service.

Call the AI ​​service interface :

Use HTTP POST request to call the AI ​​service interface, take the requirement description as input, and request to generate test cases:

import  openai
import  requests

# Set OpenAI API key
openai.api_key =  'your_openai_api_key'

# Get the requirement description from RAGflow (example, the RAGflow API should be called in practice)
ragflow_url =  'http://localhost:8080/api/requirements/{requirement_id}'
requirement_id =  'your_requirement_id'
response = requests.get(ragflow_url.format(requirement_id=requirement_id))
requirement_desc = response.json()[ 'description' ]

# Construct the prompt for input to AI
prompt =  f"Generate test cases based on the following requirement description:\n {requirement_desc} "

# Call OpenAI GPT-3 to generate test cases
completion = openai.Completion.create(
    engine = "davinci" ,
    prompt=prompt,
    max_tokens = 150 ,
    n= 1 ,
    stop= None ,
    temperature = 0.7 ,
)

# Print the generated test cases
print(completion.choices[ 0 ].text)

Note: The above code is only an example. When actually used, it needs to be adjusted according to the OpenAI API documentation, and the actual requirement description needs to be obtained through the RAGflow API.

Integration test cases : 

Integrate AI-generated test cases into the test plan and adjust and supplement them according to actual needs.