Docx Proofreader with OpenAI

This repository provides a Python script for proofreading .docx documents using OpenAI models. The script checks for spelling, grammar, and agreement errors in any language, splits large texts into manageable chunks, and saves a detailed error report in plain text format.

---

Features

• Proofreads .docx files for spelling, typo, grammar, and agreement errors using advanced OpenAI models.
• Handles large documents by automatically splitting them into model-sized chunks.
• Error findings are clearly listed and explained in the output.
• Model selection is easy: choose your preferred model with a command-line flag.
• Each run produces a report with the model name, date, and source file included in the header and filename.
• Robust API error handling: Retries API calls on rate limits and network issues.
• Easy to use: Just specify your .docx input file, and receive a full report as a .txt file.

---

Requirements

1. Python 3.7 or higher

2. Required libraries:

   • openai
   • python-docx
   • tiktoken

3. A valid OpenAI API key.

To install the requirements, run:

pip install openai python-docx tiktoken

---

Setup

Configuration File

Create a config.json file in the same directory as the script and add your OpenAI API key:

{
    "OPENAI_API_KEY": "your-api-key-here"
}

---

Usage

1. Place your input .docx file (e.g., input.docx) in the script directory.

2. Run the script from the command line. Basic usage:

   python proofreader.py

   Specify a different model or input/output file if you want:

   python proofreader.py --model o3 --input myfile.docx

3. The script will create an output file with a name like:

   proofreading_report_gpt-4o_20250702_1334.txt
   

   containing a detailed error report, the model name, the date, and the name of the original file.

---

Output

• The report is saved as a plain text .txt file.
• The file header shows the date, model name, and source file.
• Each section corresponds to a chunk of your document, with errors listed and explained.
• If no errors are found, the report will state so.

---

Customization

• To change which model is used by default, edit the DEFAULT_MODEL variable at the top of the script.
• To adjust chunk size or API retry behavior, modify the CHUNK_TOKENS or RETRIES variables.
• You can further refine the proofreading prompt for other languages or criteria in the SYSTEM_PROMPT variable in the script.

---

Example Command

python proofreader.py --model o3 --input document_to_check.docx --output my_report.txt

---