TestGPT

A command-line tool for generating unit tests for your files automatically using OpenAI GPT-3.5-turbo model (gpt-4 also supported for developers who have it).

If you have access to GPT-4 API, you can now pass --model/-m option, see below for an example.

Now there is a VScode extension the process even faster, check the extension here (You have to install the latest version of testgpt for it to work)

NOTE: From version 3.0.0 and upwards, testgpt.config.json was replaced with testgpt.config.yaml, and a new custom property (examples) is added.

Installation

To use TestGPT, follow these steps:

Install TestGPT by running one of these commands:

# Install globally
npm install -g testgpt@latest

# OR install locally in your project
npm install testgpt@latest

Get your OpenAI API key by requesting access to the OpenAI API and obtaining your API key. Then, export your OpenAI key based on your operating system:

You then need to export your OpenAI key. Follow the steps to export for your OS:
- macOS or Linux (zsh): Add the following line to .zshrc in your home or ~ directory:
```
export OPENAI_API_KEY="Your OpenAI API Key."
```
  Then run the command:
```
source ~/.zshrc
```
- Linux (bash): Add the following line to ~/.bashrc:
```
export OPENAI_API_KEY="Your OpenAI API Key."
```
  Then run the command:
```
source ~/.bashrc
```
- Windows: Go to System -> Settings -> Advanced -> Environment Variables, click New under System Variables, and create a new entry with the key OPENAI_API_KEY and your OpenAI API Key as the value.
If you have testgpt.config.yaml you can Generate unit tests by running this command from your project's root directory (where testgpt.config.yaml is located):
```
testgpt -i <path to your input file> -o <path to your test output file>
```
Note that --inputFile/-i and --outputFile/-o can be either a file path or directory, and in the later case, a tests will be generated for each file inside the directory.

If you don't provide an --outputFile/-o, the generated test file will be saved in the same directory as the input file.
```
testgpt -i ./src/component.tsx
# Output file will be ./src/component.test.tsx
```

Providing custom techs using a config file

You can create a testgpt.config.yaml file in your project's root directory to specify technologies and tips for each file extension. Example:

.ts:
  techs:
    - jest
  tips:
    - use 2 spaces for indentation
    - wrap each group of tests in a describe block
    - Don't forget to import the things you need.

.tsx:
  techs:
    - jest
    - react-testing-library
    - userEvent
  tips:
    - use 2 spaces for indentation
    - use screen
    - wrap each group of tests in a describe block
    - when using user event, use an async function and await the user event
    - prefer not to use getByTestId
    - Don't forget to import the things you need.

Then you can run the following command from the same directory as the config file:

testgpt -i ./src/component.tsx

You can also pass the config file path using the --config/-c argument.

testgpt -i ./src/component.tsx -c `./testgpt.config.yaml`

Providing examples

The file testgpt.config.yaml supports the examples property for each file extension:

.tsx
   techs:
      - jest
      - react-testing-library
   tips:
      - wrap each group of tests in a describe block
   examples:
      - fileName: file1.tsx
        code: <code content for file1.tsx>
        tests: <tests content for file1.tsx>
      - fileName: file2.tsx
        code: <code content for file2.tsx>
        tests: <tests content for file2.tsx>

Providing custom techs directly

You can pass --techs/-t and --tips/-p directly for each command

testgpt -i ./src/component.tsx --techs react,jest --tips "Don't forget to import what you need"`

Providing custom model

You can pass --model/-m option to use a custom model other than the default (currently gpt-3.5-turbo):

testgpt -i ./src/component.tsx -m gpt-4

License

TestGPT is released under the MIT License. Feel free to use it and contribute to it!

Thanks to OpenAI for generating this README and my unit tests for this project :)