Shell commands are esoteric. Most commands were invented decades ago and it almost feels like a language only the nerdiest of hackers are fluent in. I always end up googling reference materials even for the most basic of shell commands.<\/p>\n\n
That changes today. The future is now. What a time to be alive.<\/p>\n\n
Let's build a REPL CLI tool using Python that receives our intent in plain English and employs the OpenAI API to translate that into a
\nShell command which is then executed as part of the Python process.<\/p>\n
OpenAI requires an account and payment method with a minimum of $10 to use their API. However, new users receive a $10 credit upon sign up and the cost is based on the amount of tokens in the prompt and completion, and the model used. See here for additional billing information<\/a>.<\/p>\n\n Setup your Personal account and copy your API key somewhere safe. Remember not to commit your API key to any public code repository (or any repository for that matter) as someone could use it to steal your credits and run up your bill. Keep your API key secret.<\/strong><\/p>\n Create a development directory and name it gibson<\/strong> in honor of the writer William Gibson who pioneered the AI and Cyberpunk Science Fiction subgenres. Install the openai<\/strong> package and create a text file named gibson.py<\/strong>. Add the following to gibson.py<\/strong> and replace the placeholder The OpenAI Create Completion request will take the full prompt<\/strong> as an argument along with the name of the model<\/strong> to process it with. Use the OpenAI playground to try out the different models. The \"dumber\" the model, the cheaper it is to use. The davincii<\/em> model is the most \"intelligent\" model at this time so it is also the most expensive but will give us the best results. It's great for creating proofs of concept and demos but you'll want to fine-tune a new model for a specific use-case in a production environment. The ada<\/em> model is the cheapest and might be able to solve your use-case for a much lower price and faster execution time. <\/p>\n\n The temperature<\/strong> value refers to what sampling temperature<\/em> to use. Higher values means the model will take more risks. Try 0.9 for more creative applications, and 0 (argmax sampling) for ones with a well-defined answer. In this case, a low temperature<\/strong> is best since the completion is expected to be constant given the same prompt.<\/p>\n\n The max_tokens<\/strong> value sets a limit on the prompt and completion size. You can think of tokens as words but more complex words may have multiple tokens. This can help with rate limiting but also runs the risk that the completion will be missing text if the prompt takes up too many tokens. See this article for more detail on how to count tokens<\/a>. In this case, 256 is more than enough.<\/p>\n\n The Spring Boot framework includes a simple, server-side caching solution which is easy to enable and see immediate results and even easier to extend using robust caching solutions all through an Annotation based interface that makes it all feel like configuration rather than implementation.<\/p>\n\n When you work with different sources of data, your cache control rules may vary and conditions may need to be evaluated before performing caching operations. With this requirement, we risk polluting our business code with caching logic. Spring Boot's The Demo application serves as Backend-For-Frontend service for Typicode's JSONPlaceholder<\/a> which serves RESTful JSON samples which can then be used for demos and tests concerning external APIs.<\/p>\n\n For the sake of this tutorial, the JSONPlaceholder API has a significant network latency of 3 seconds in every call to read collections and records. We could cache every single endpoint but the stakeholders have stated an additional requirement. We can only cache the \n \n \n OpenAI Python Library\n<\/h2>\n\n
\n<\/p>\n\nmkdir <\/span>gibson\ncd <\/span>gibson\npip install<\/span> --upgrade<\/span> openai\ntouch <\/span>gibson.py\n<\/code><\/pre>\n\n<\/div>\n\n\n\n\n \n \n Gibson REPL CLI\n<\/h2>\n\n
OPENAI_API_KEY<\/code> with the API key from the previous steps. Change the OS<\/strong> variable to reflect whatever operating system the script will be executing the commands on.
\n<\/p>\n\n#! \/c\/Python310\/python\n<\/span>import<\/span> os<\/span>\nimport<\/span> openai<\/span>\n\n# Set your OS to get the correct Shell commands. Windows, Mac, Linx, etc.\n<\/span>OS<\/span> =<\/span> '<\/span>Windows<\/span>'<\/span>\n\n# Set the OpenAI API Key\n<\/span>openai<\/span>.<\/span>api_key<\/span> =<\/span> '<\/span>OPENAI_API_KEY<\/span>'<\/span>\n\n# Keep prompting for more input.\n<\/span>while<\/span> True<\/span>:<\/span>\n # Capture the user's intent in plain English\n<\/span> intent<\/span> =<\/span> input<\/span>(<\/span>'<\/span>$ <\/span>'<\/span>)<\/span>\n # Exit the Python process if the user types 'exit'\n<\/span> if<\/span> intent<\/span> ==<\/span> '<\/span>exit<\/span>'<\/span>:<\/span>\n exit<\/span>()<\/span>\n # Interpolate the user's intent and OS into the OpenAI prompt.\n<\/span> prompt<\/span> =<\/span> f<\/span>\"\"\"<\/span>\n Translate this English statement to a Shell command for the OS.\n OS: <\/span>{<\/span>OS<\/span>}<\/span>\n English: <\/span>{<\/span>intent<\/span>}<\/span>\n Shell:\n <\/span>\"\"\"<\/span>\n # Fetch the prompt completion using the OpenAI library.\n<\/span> completion<\/span> =<\/span> openai<\/span>.<\/span>Completion<\/span>.<\/span>create<\/span>(<\/span>\n prompt<\/span>=<\/span>prompt<\/span>.<\/span>strip<\/span>(),<\/span> # Always remove surrounding whitespace for performance reasons on OpenAI.\n<\/span> engine<\/span>=<\/span>'<\/span>text-davinci-003<\/span>'<\/span>,<\/span> # The OpenAI model.\n<\/span> temperature<\/span>=<\/span>0<\/span>,<\/span> # How \"creative\". Lower means less creative.\n<\/span> max_tokens<\/span>=<\/span>256<\/span> # Max of 256 tokens including prompt and completion.\n<\/span> )<\/span>\n # Execute the completion text as a Shell statement\n<\/span> os<\/span>.<\/span>system<\/span>(<\/span>completion<\/span>.<\/span>choices<\/span>[<\/span>0<\/span>].<\/span>text<\/span>.<\/span>strip<\/span>())<\/span>\n<\/code><\/pre>\n\n<\/div>\n\n\n\n\n \n \n Limitations\n<\/h2>\n\n
\n
@Cacheable<\/code><\/a> interface handles this elegantly by employing Java Annotations and the Spring Expression Language (SpEL)<\/a>.<\/p>\n\n\n \n \n Use Case\n<\/h1>\n\n
\/users\/<\/code>, \/albums<\/code> and \/photos<\/code> endpoints. All the others endpoints contain data that cannot be cached for reasons. The Front-End team will implement a loading screen for those endpoints instead.<\/p>\n\n\n \n \n Starter Template\n<\/h1>\n\n