Currently, as soon as there are broken lines (random binary data) within the CSV file, it cannot be imported at all. E.g., those lines can appear whenever the sensor isn't shut down correctly, leading to corrupt data being written.
I suggest just ignoring corrupt lines and extracting measurements from the valid lines. This can be done fairly easily in read_csv(), but I wonder whether this isn't done on purpose or if there is any reason to not do this?
Currently, as soon as there are broken lines (random binary data) within the CSV file, it cannot be imported at all. E.g., those lines can appear whenever the sensor isn't shut down correctly, leading to corrupt data being written.
I suggest just ignoring corrupt lines and extracting measurements from the valid lines. This can be done fairly easily in read_csv(), but I wonder whether this isn't done on purpose or if there is any reason to not do this?