Resample pandas df with multiple groupbys so each condition has the same number of total days of data

I have been going round in circles with this and haven't been able to figure it out.

Suppose I have the following dataframe:

df = pd.DataFrame({
    "person_id": ["1", "1", "1", "1", "2", "2", "2", "2", "3", "3", "3", "3"],
    "event": ["Alert1", "Alert1", "Alert1", "Alert2", "Alert1", "Alert1", "Alert1", "Alert2", "Alert2", "Alert2", "Alert2", "Alert2"],
    "mode": ["Manual", "Manual", "Auto", "Manual", "Auto", "Auto", "Auto", "Manual", "Manual", "Manual", "Auto", "Manual"],
    "date": ["2020-01-01", "2020-01-01", "2020-01-03", "2020-01-03", "2020-01-03", "2020-01-03", "2020-01-04", "2020-01-04", "2020-01-04", "2020-01-04", "2020-01-05", "2020-01-05"]
}
)

df

What I want is the count of each possible combination per possible day (the minimum date would be the first date appearing in the dataset, in this case 2020-01-01 and the maximum date would be the last date appearing in the dataset, in this case 2020-01-05). For example, in the case of the df above, the output would look like this:

Importantly, each combination should have the exact same number of unique datetimes at the end, so if I run the following line of code:

df_summarized.groupby(['person_id', 'event', 'mode'])['date'].nunique().reset_index()

The result should clearly show that each combination has 5 unique days of data.

How could I achieve this?

Thanks in advance

IIUC, what you need to do is first create a finite set of all possible combinations, and then count their occurences.

import pandas as pd
import numpy as np
from itertools import product

# Create the original DataFrame
df = pd.DataFrame({
    "person_id": ["1", "1", "1", "1", "2", "2", "2", "2", "3", "3", "3", "3"],
    "event": ["Alert1", "Alert1", "Alert1", "Alert2", "Alert1", "Alert1", "Alert1", "Alert2", "Alert2", "Alert2", "Alert2", "Alert2"],
    "mode": ["Manual", "Manual", "Auto", "Manual", "Auto", "Auto", "Auto", "Manual", "Manual", "Manual", "Auto", "Manual"],
    "date": ["2020-01-01", "2020-01-01", "2020-01-03", "2020-01-03", "2020-01-03", "2020-01-03", "2020-01-04", "2020-01-04", "2020-01-04", "2020-01-04", "2020-01-05", "2020-01-05"]
})

df['date'] = pd.to_datetime(df['date'])

person_ids = df['person_id'].unique()
events = df['event'].unique()
modes = df['mode'].unique()
dates = pd.date_range(df['date'].min(), df['date'].max(), freq='D')

all_combinations = pd.DataFrame(list(product(person_ids, events, modes, dates)), columns=['person_id', 'event', 'mode', 'date'])

count_df = df.groupby(['person_id', 'event', 'mode', 'date']).size().reset_index(name='count')

result = all_combinations.merge(count_df, on=['person_id', 'event', 'mode', 'date'], how='left').fillna(0)
result.reset_index(drop=True, inplace