import pandas as pd

# CONFIG
input_file_w_nuls = "data/af_daily.csv"
file_wo_nuls = "data/af_daily_nn.csv"
start_date = "9/11/2001" # Next time give me dates in the CSV, please
resample_type = "M" # Resample to monthly periods

def remove_nul(f_in, f_out):
    input = open(f_in, "rb")
    data = input.read()
    input.close()
    output = open(f_out, "wb")
    output.write(data.replace(b"\x00", b""))
    output.close()

remove_nul(input_file_w_nuls, file_wo_nuls) # Don't fuck with me, NULs

af_input = pd.read_csv(file_wo_nuls, header=None) # Fuck the headers

date_index = pd.date_range(start_date, periods=len(af_input.index), freq="D") # We have daily data, hence the daily data range
af_input=af_input.as_matrix(columns=None).ravel() # Convert DataFrame to numpy array

airflow_data = pd.Series(af_input, index=date_index) # Do the magic

airflow_data = airflow_data.resample(resample_type)

# ... and a whole bunch of code below here, not really important