Help replicating Excel PITimeFilterVal in Python.

[NPhuthiFQML]

Hi. I am in need of replicating the PITimeFilterVal Excel PIDataLink function in Python, with the purpose of calculating the duration over which the value of a PIPoint is above a given limit.

So far, I recognise that I can pull the PI series recorded values. I then have the series, but am stumped as how to calculate durations the same way that PITimeFilterVal() function does:

def get_duration_by_limit(start_time, end_time, tag, greater_than_filter=None, less_than_filter=None): with PI.PIServer() as server: points = server.search(f'*{tag}') values = [point.recorded_values(start_time, end_time) for point in points][0] if greater_than_filter: values[values < greater_than_filter] = 0 if less_than_filter: values[values > less_than_filter] = 0 print(values)

Any help is greatly appreciated. Thanks.

[NPhuthiFQML]

I have since found a workaround. I generate a dataframe from the PISeries object, then create a "flags" column based on a rolling mean with a window of 1, which will evaluate each item against the filter.:
val_df['flags'] = val_df['values'].rolling(1, center=True).mean() >= filter

From there another column is created, that allocates a group number to each of the regions in the dataframes where the 'flag' is either true or false:
val_df["group"] = val_df["flag"].diff().cumsum().bfill().astype(int)

I can then split the dataframe by 'group' column into a list of dataframes
for i in range(val_df["group"].max() + 1): if (val_df["flag"][0] and (i % 2 == 0)) or ((not val_df["flag"][0]) and (i % 2 == 1)): ls_out.append(val_df[val_df["group"] == i])
Then, the maxes and mins of the indices (Timeseries objects) of the dataframes can be subtracted from each other, then the deltas summed and the total duration returned in seconds:
total_hours = pd.Series([max(i.index) - min(i.index) for i in ls_out]).sum().total_seconds()

As a note, the builtin sum() of python is meant for integer and float iterables, so a solution is to convert the list to a pandas series again, and then use the series' sum function.

Credit to the following links:
https://stackoverflow.com/questions/64679467/splitting-a-dataframe-based-on-threshold-value
https://stackoverflow.com/questions/47014765/use-center-in-pandas-rolling-when-using-a-time-series