lire csv pandas nrow

read_csv(..., skiprows=1000000, nrows=999999)
Naughty Nightingale