Feature Request: Option to convert Unix/Linux Epoch Timestamp into Date/Time

[jpummill]

Thank you for creating and sharing tabulate.

I have a feature request that I think would be useful.

Add an option to review the first couple rows of data to determine if any columns contain Unix/Linux Epoch timestamp values (either in seconds or milliseconds) and convert then automatically convert those columns into Date/Time values.

I am providing an example function that just detects whether a column is likely an epoch timestamp but it doesn't actually change the timestamp to a datetime value, but that is pretty easy to add.

You could also enhance the code by allowing the user to set the following:

• The size of the window to allow (min_dt & max_dt). I just used +/- 10 years from current but this could be passed in.
• Format of the datetime data.

`
def is_likely_epoch(n, unit="s"):

if not isinstance(n, (int, float)):
    return False

# Convert milliseconds to seconds if needed
if unit == "ms":
    n = n / 1000.0

try:
    now = datetime.datetime.now(datetime.timezone.utc)

    # Define dynamic bounds: now ± 10 years
    min_dt = now - datetime.timedelta(days=365 * 10)
    max_dt = now + datetime.timedelta(days=365 * 10)

    MIN_EPOCH_S = min_dt.timestamp()
    MAX_EPOCH_S = max_dt.timestamp()

    # Check if value falls within bounds
    if not (MIN_EPOCH_S <= n <= MAX_EPOCH_S):
        return False

    # Attempt conversion to datetime
    datetime.datetime.fromtimestamp(n, tz=datetime.timezone.utc)
    return True

except (ValueError, OSError, OverflowError):
    return False  

`

Thank you for considering this request.