Snakefile.txt

rule all:
	input: 
		"3_plot/out/choropleth_plot.html"
		

rule get_hu_metadata:
	input:
		csv_file = "1_fetch/src/in/huc_inputs.csv"
	output:
		out_file = "1_fetch/out/tmp/hu_metadata.pkl"
	script:
		"1_fetch/src/huc_extraction.py"

rule get_data:
	params:
		param_code = "00010",
		stat_code  = "00003",
		start	   = "2022-01-01",
		end 	   = "2022-12-31"	
	input:
		in_file = "1_fetch/out/tmp/hu_metadata.pkl"
	output:
		out_file = "1_fetch/out/tmp/timeseries_info.pkl"
	script:
		"1_fetch/src/data_extraction.py"

rule clean_data:
	input:
		in_file = "1_fetch/out/tmp/timeseries_info.pkl"
	output:
		out_file = "2_process/out/tmp/cleaned_timeseries_info.pkl"
	script:
		"2_process/src/data_cleaning.py"

rule process_data:
	input:
		in_file = "2_process/out/tmp/cleaned_timeseries_info.pkl"
	output:
		out_file = "2_process/out/tmp/processed_cleaned_timeseries_info.pkl"
	script:
		"2_process/src/data_processing.py"

rule plot_data:
	input:
		in_file = "2_process/out/tmp/processed_cleaned_timeseries_info.pkl"
	output:
		out_file = "3_plot/out/choropleth_plot.html"
	script:
		"3_plot/src/plotting.py"