all: install data train eval

city_distances.csv: generate_data.py
	@echo "Generating distance data..."
	@bash -c 'time python generate_data.py --country US --workers 8 --chunk-size 4200'
	@echo "Calculating range of generated data..."
	@cat city_distances.csv | tail -n +2 | sort -t',' -k3n | head -n1
	@cat city_distances.csv | tail -n +2 | sort -t',' -k3nr | head -n1

data: city_distances.csv

train: check train.py
	@echo "Training embeddings..."
	@bash -c 'time python train.py'

eval: check eval.py
	@echo "Evaluating trained model..."
	@bash -c 'time python eval.py'

lint:
	@echo "Auto-linting files and performing final style checks..."
	@isort --profile=black .
	@black .
	@flake8 --max-line-length=88 --ignore E203 .

check: lint
	@echo "Checking for unstaged or untracked changes..."
	@git diff-index --quiet HEAD -- || { echo "Unstaged or untracked changes detected!"; exit 1; }

clean:
	@echo "Removing outputs/ and checkpoints/ directories"
	@rm -rf output/
	@rm -rf checkpoints/

compress: plots/progress_12474_sm.png

plots/progress_12474_sm.png: plots/progress_12474.png
	@convert -resize 33% plots/progress_12474.png progress_sample.png

install: .requirements_installed

.requirements_installed: requirements.txt
	pip install -r requirements.txt
	@echo "installed requirements" > .requirements_installed

.PHONY: data train eval lint check clean all