# This requirements.txt file is intended for use with Hugging Face Spaces (Python 3.10), and so is not intended as the file to use to install packages for a local install. Please refer to the README.md file for install instructions for the app.

# --- Core and data packages ---
numpy<=2.4.4
pandas<=2.3.3
bleach<=6.3.0
polars<=1.38.1
pyarrow<=23.0.1
openpyxl<=3.1.5
boto3<=1.42.91
python-dotenv<=1.0.1
defusedxml<=0.7.1
Faker<=40.8.0
python-levenshtein<=0.27.3
rapidfuzz<=3.14.5
markdown<=3.10.2
tabulate<=0.10.0

# --- PDF / OCR / Redaction tools ---
pdfminer.six<=20260107
pdf2image<=1.17.0
pymupdf<=1.27.1
pikepdf<=10.3.0
opencv-python<=4.13.0.92
presidio_analyzer<=2.2.362
presidio_anonymizer<=2.2.362
presidio-image-redactor<=0.0.58

# --- Document generation ---
python-docx<=1.2.0

# --- Gradio and apps ---
gradio[mcp]<=6.10.0
gradio_image_annotation_redaction==0.5.4 # Custom annotator version with rotation, zoom, labels, and box IDs
spaces<=0.48.3

# --- AWS Lambda runtime ---
awslambdaric<=3.1.1

# --- Machine learning / NLP ---
scikit-learn<=1.8.0
spacy<=3.8.14
spaczz<=0.6.1
# en_core_web_lg is not on PyPI; it is downloaded on first NLP use (see tools/load_spacy_model_custom_recognisers.py) or run: python -m spacy download en_core_web_lg
transformers<=5.3.0
accelerate<=1.13.0
bitsandbytes<=0.49.2
sentencepiece<=0.2.1
optimum<=2.1.0

# --- Testing ---
pytest<=9.0.3
pytest-cov<=7.1.0

# --- LLM libraries ---
google-genai<=1.73.0
openai<=2.31.0

# --- PyTorch (CUDA 12.6) ---
--extra-index-url https://download.pytorch.org/whl/cu128
torch<=2.8.0
torchvision<=0.23.0
flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.8cxx11abiTRUE-cp310-cp310-linux_x86_64.whl
# GPTQModel @ https://github.com/ModelCloud/GPTQModel/releases/download/v5.8.0/gptqmodel-5.8.0+cu128torch2.8-cp310-cp310-linux_x86_64.whl # Need to install from wheel as requirements-based compile fails on Hugging Face spaces