Find First and Second Occurrence of Value Across Two Columns by Group

I have a df that looks like the one below. It is sorted by Ref1 and Seq.

I am trying to work out how to do two things:

1. count the first occurrences of a 1 in either PP1 or PP2 grouped by Ref1 and EvNo. There may be no occurrences or there may be multiple occurrences but there will never be a 1 in both columns on the same row.

2. after the first occurrence (if any) count if there is a 1 in the other of PP1 or PP2 in the same group. Eg if the first 1 in a group was in PP1 count if the next occurrence of 1 is in PP2. If the next 1 is also in PP1 it should not be counted. There may be no further occurrences of a 1 in either column.

Output:

I managed to get the result by applying a separate function to groups of the dataframe, turning the result into a new dataframe and summarizing it.

In the function the main trick is to use np.where. I used it on a sum of columns PP1 and PP2 to find just any occurences, and then checked which column provided the occurence just by checking if the value in PP1 column is 1 (if yes - then the occurence is in PP1, if not - in PP2, as you said that the occurences can not happen simultaneously).

Although, I am not sure why your output doesn't have 1 for P1 Second Occ, because the first group (Ref1 == "aaaa" and EvnNo == 0) shows exactly that, if I understood the question correctly.

import numpy as np

def count_occurences(group):
    result = [0] * 4

    occurences = np.where(group.sum(axis=1) == 1)[0]
    
    # track first occurence
    if len(occurences) > 0 and group.iloc[occurences[0]]["PP1"] == 1:
        result[0] += 1
    elif len(occurences) > 0:
        result[1] += 1
    
    # track second occurence
    if len(occurences) > 1 and group.iloc[occurences[1]]["PP1"] == 1:
        if result[0] != 1:
            result[2] += 1
    elif len(occurences) > 1:
        if result[1] != 1:
            result[3] += 1
        
    return result


occurences_df = pd.DataFrame(
    df \
        .groupby(["Ref1", "EvnNo"]) \
        [["PP1", "PP2"]] \
        .apply(count_occurences) \
        .to_list(),
    columns = ["P1 First Occ", "P2 First Occ", "P1 Second Occ", "P2 Second Occ"]
)

print(occurences_df.sum())

Output:

P1 First Occ     2
P2 First Occ     1
P1 Second Occ    1
P2 Second Occ    1