Makefile

PYTHON = python3
PIP = pip install
SRC_DIR = src
DATA_DIR = data
TEST_DIR = tests
MODELS_DIR = models

.PHONY: setup preprocess train tune clean

setup:
	$(PIP) -r requirements.txt
	$(PYTHON) -m nltk.downloader stopwords

preprocess:
	$(PYTHON) -m $(SRC_DIR).preprocess \
		--data_dir $(DATA_DIR) \
		--min_density 35 \
		--max_length 128

train:
	$(PYTHON) -m $(SRC_DIR).train \
		--hparams configs/best_hparams.yml \
		--train_data $(DATA_DIR)/processed_train.parquet \
		--test_data $(DATA_DIR)/processed_test.parquet \
		--save_path models/final_model.pth

tune:
	$(PYTHON) -m $(SRC_DIR).optimize \
		--config configs/hparams_search.yml \
		--train_data $(DATA_DIR)/processed_train.parquet \
		--val_data $(DATA_DIR)/processed_val.parquet \
		--save_path $(MODELS_DIR)/best_model_state.pth \
		--train_subset 100000 \
        --val_subset 10000

export_onnx:
	$(PYTHON) -m $(SRC_DIR).export_onnx \
		--hparams configs/best_hparams.yml \
		--model_path models/final_model.pth \
		--output_path models/arabic_diacritizer.onnx

evaluate_onnx: 
	$(PYTHON) -m src.evaluate \
		--model models/arabic_diacritizer.onnx \
		--test_data data/processed_test.parquet \
		--max_samples 10000
		
test:
	$(PYTHON) -m pytest $(TEST_DIR) -v

test-fast:
	$(PYTHON) -m pytest $(TEST_DIR) -x -v

test-cov:
	$(PYTHON) -m pytest $(TEST_DIR) --cov=$(SRC_DIR) --cov-report=term-missing

clean:
	find . -type d -name "__pycache__" -exec rm -rf {} +

help:
	@echo "Available commands:"
	@echo "  make setup      : Install dependencies"
	@echo "  make preprocess : Merge MSA with CA and Clean data"
	@echo "  make train      : train on a single run"
	@echo "  make tune       : Run Optuna optimization"
	@echo "  make test       : Run all tests"
	@echo "  make test-fast  : Stop on first failure"
	@echo "  make test-cov   : Run tests with coverage"
	@echo "  make clean      : Clean the repo"