go-rename

A small Go utility that uses the OpenAI API to infer the correct ordering of page image files in a directory (for example, magazine pages), based on their filenames.

The program:

1. Reads all file names from a working directory.
2. Asks an OpenAI model to guess the correct order of those files based on page numbers in the names.
3. Parses the JSON array of ordered file names returned by the model.

This repository is a minimal example meant to be edited and extended in GoLand.

Prerequisites

• Go 1.22+ (or the version specified in go.mod)
• An OpenAI API key with access to the model you configure in main.go

Configuration

The program is configured entirely via environment variables:

• OPENAI_API_KEY (required): Your OpenAI API key.
• WORKING_DIR (required): Absolute or relative path to the directory whose files you want to analyze.

In GoLand, you can set these in Run �| Edit Configurations... under Environment variables.

Installation

Clone the repository and download Go dependencies:

git clone <your-repo-url> go-rename
cd go-rename
go mod tidy

Running the program

From the command line:

export OPENAI_API_KEY="your-api-key-here"
export WORKING_DIR="/path/to/your/files"

go run ./...

From GoLand:

1. Create or edit a Run/Debug Configuration for main.go.
2. Set OPENAI_API_KEY and WORKING_DIR in Environment variables.
3. Run the configuration.

How it works

The core logic is in main.go:

1. Read OPENAI_API_KEY and WORKING_DIR from the environment.
2. Use os.ReadDir to list all files in WORKING_DIR and print them.
3. Build a natural-language prompt enumerating those filenames.
4. Call the OpenAI Chat Completions API via the official Go SDK (github.com/openai/openai-go/v3) with that prompt.
5. Expect the model to respond with a JSON array (e.g. ["page_01.png", "page_02.png", ...]).
6. Use encoding/json to unmarshal the JSON string into a []string slice (orderedFiles).

You can then extend the program to:

• Actually rename or move files according to orderedFiles.
• Extract additional metadata (e.g., publication month/year) from the cover page.
• Handle errors and edge cases more robustly.

License

Add your preferred license here (e.g., MIT).