Skip to content

README.md

Latest commit

 

History

History
78 lines (64 loc) · 3.4 KB

README.md

File metadata and controls

78 lines (64 loc) · 3.4 KB

TrazbanG

A high-fidelity document translation toolkit designed for large PDFs and text files. This tool supports a granular "Extract -> Refine -> Translate" workflow, ensuring the highest quality by enabling manual text cleanup before AI translation.

Features

  • Smart Chunking: Automatically slices PDFs based on Bookmarks (TOC) or Chapter headings.
  • Two-Pass Translation: Performs a literal technical draft followed by an eloquent scholarly refinement.
  • Dual Outputs: Saves both _draft.txt (literal) and _final.txt (eloquent) for every section.
  • Refinement-First Workflow: Extracts raw text so you can fix OCR errors or remove headers/footers before translating.
  • Resumable: Skips already processed segments to save time and API quota.
  • Flexible Filters: Process a single section, a range of indices, or the entire book.
  • Systematic Organization: Automatically creates dedicated folders for segments and translations based on the input filename.

Installation

  1. Ensure you have Python 3.10+ installed.
  2. Create and activate a virtual environment: 
    python3 -m venv venv
source venv/bin/activate
  3. Install dependencies: 
    pip install google-genai pymupdf
  4. Set your Gemini API Key: 
    export GEMINI_API_KEY="your_api_key_here"

Usage

1. Preparation (Slicing & Extraction)

Break the PDF into manageable chunks and extract the text for refinement.

python main.py --input MyBook.pdf --mode bookmark --action prepare
  • Result: A new folder sections_MyBook/ is created containing .pdf segments and .txt extractions.

2. Manual Refinement (Recommended)

Open the .txt files in sections_MyBook/ and clean up any "noise" like page numbers, running headers, or bad line breaks.

3. Translation

Submit your refined text to Gemini for a two-pass translation.

# Translate with a custom glossary (.json or .txt)
python main.py --input MyBook.pdf --action translate --glossary glossary.json

## Translate a single section
python main.py --input MyBook.pdf --action translate --index 7

# Translate a range of sections
python main.py --input MyBook.pdf --action translate --start 10 --end 20
  • Result: Final translations are saved in translations_MyBook/.
    • ..._draft.txt: The initial technical draft.
    • ..._final.txt: The polished, natural translation.

Note: To re-translate a section, you must first remove or rename its existing _final.txt file.

Command-Line Arguments

Flag Description Options
--input (Required) Path to your PDF or TXT file. File path
--mode How the book should be divided. bookmark (default), chapter, full
--action What step to perform. slice, extract, translate, prepare (slice+extract)
--lang The target language for translation. e.g., Persian, Spanish, French
--glossary Path to glossary file. .json or .txt
--index Process only this specific section number. Integer
--start / --end Process a range of sections. Integers
--key Your Google Gemini API Key. String

Folders Created

  • sections_<filename>/: Contains the sliced PDF parts and extracted text files.
  • translations_<filename>/: Contains the final dual-pass output from Gemini.

Acknowledgments

This project was entirely authored and architected by Google Gemini.