Arabic PDF to Text Converter

A Rust CLI tool that converts Arabic PDFs to text using Google's Gemini API.

Features

• ✅ Uploads full PDF once using resumable upload protocol
• ✅ Processes PDFs in page ranges with MapReduce pattern
• ✅ Concurrent processing with rate limit handling
• ✅ Preserves Arabic text formatting exactly
• ✅ Progress tracking for each page range
• ✅ Automatic retry on rate limit errors

How It Works

1. Single Upload: The entire PDF is uploaded once to Gemini API
2. Page Range Processing: The tool requests text extraction for specific page ranges (e.g., pages 1-5, 6-10, etc.)
3. MapReduce Pattern: Multiple page ranges are processed concurrently
4. Rate Limiting: Automatic delays and retries handle API rate limits
5. Result Aggregation: All extracted text is combined in the correct order

Current Implementation

The tool successfully:

• Uploads PDFs using Gemini's resumable upload API
• Processes page ranges concurrently (default: 5 pages per chunk)
• Handles rate limiting with 6-second delays between requests
• Retries failed requests up to 3 times with 30-second delays
• Combines results maintaining page order

Installation

From crates.io

cargo install arabic_pdf_to_text

From Source

# Clone the repository
git clone https://github.com/RustSandbox/arabic_pdf_to_text.git
cd arabic_pdf_to_text

# Build the project
cargo build --release

# The binary will be at target/release/arabic_pdf_to_text

Prerequisites

• Rust 1.70 or later
• Google Gemini API key (Get one here)

Configuration

1. Copy .env.example to .env:

   cp .env.example .env

2. Add your Gemini API key:

   export GEMINI_API_KEY="your-api-key"

Usage

# Process a PDF
./arabic_pdf_to_text "path/to/arabic.pdf" -o output.txt

# With custom chunk size
./arabic_pdf_to_text "path/to/arabic.pdf" --chunk-size 524288 -o output.txt

# See all options
./arabic_pdf_to_text --help

Recommendations

For production use, consider:

1. Using a PDF library to split PDFs at page boundaries instead of byte boundaries
2. Implementing a queue system to respect rate limits
3. Using a paid API tier for higher quotas
4. Caching processed chunks to avoid reprocessing

Building

cargo build --release

Testing

cargo test

Contributing

We welcome contributions! Please see our Contributing Guide for details.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Attribution for Commercial Use

While this software is free for any use including commercial, if you use it in a commercial product or service, we kindly request (but do not require) that you include the following attribution:

This product includes software developed by the arabic_pdf_to_text project
(https://github.com/RustSandbox/arabic_pdf_to_text)

Acknowledgments

• Google Gemini API for providing the PDF processing capabilities
• The Rust community for excellent libraries and tools