DTJ (Directory-to-JSON)

Overview

DTJ (Directory-to-JSON) is a Python command-line tool, particularly useful for quickly generating and sharing representations of directory structures and their contents in a structured and token-efficient manner, making it ideal for interactions with language models like ChatGPT.

Example output

dtj

{
  "my_project": {
    "__init__.py": "",
    "main.py": "# Main application file\nimport app\n\napp.run()",
    "app.py": "# App module\n\ndef run():\n    print('Running the app')",
    "utils": {
      "helper.py": "# Utility functions\n\ndef helper():\n    return 'Helper function'"
    }
  }
}

Features

• Convert directory contents to JSON format.
• Include or exclude files using patterns (supports fnmatch style, e.g., *.py, data*).
• Recursive directory parsing.
• Options for output: printing to console, saving to a file, or copying to the clipboard.

Updates

• new JSON format reduces token usage ~30%

Installation

pip install dtj

Usage

Run DTJ from the command line with the following options:

dtj <target-directory> [options]

If no target directory is specified, DTJ will default to the current working directory.

Options:

• -t or --target-file: Target a single file. This option is mutually exclusive with -i, -e, and -r.
• -i or --include: Patterns to include files. Enclose patterns in quotes to avoid shell expansion (e.g., '*.py', 'data*').
• -e or --exclude: Patterns to exclude files. Enclose patterns in quotes to avoid shell expansion (e.g., '*.xml', 'temp*').
• -o or --output-file: Set the output JSON file name.
• -r or --recursive: Enable recursive search in directories. Not valid when targeting a single file.
• -p or --print: Print the output using rich formatting.
• -c or --clipboard: Copy the output to the clipboard.

Example

simple

dtj

Generates an output.json with the content

advanced

dtj myfolder -i '*.py' '*.html' -o output.json -r

This command will parse all .py and .html files in myfolder recursively and save the JSON output to output.json.

Using it alongside gptwc

dtj | gptwc

outputs the token count 1337

Authors

• Adrian Galilea - Initial work

Acknowledgments

• Hat tip to ChatGPT for assistance with project setup and documentation.