If 'W' is in 'X' append 'Y' to 'Z'

I have a dataframe which is similar to this:

I'm trying to make it so if the value in 'Col A' is in 'Col C' then is takes 'Col B' and creates a list in a new column like below:

The Values in 'Col C' is currently an object (I can change but this is what the import from a text file and textfsm formatting set it as.

The Values in the new column should be a string. (I should also set 'Col C' to be a string to make this easier on myself)

Nothing I've tried seems to get it to work the way I'm hoping.
I'm still pretty new with Pandas and I've been struggling to find a way to get this to work.
I feel the answer is super simple and right in front of me too but I can't seem to get it.

With the below attempt I get "unhashable type: 'Series'" however the data isn't created as a series itself within my code, it's loaded from a text file and formatted with TextFSM.

for i in df.iterrows():
    if df['Column C'].str.contains(df['Column A']).any():
        results.append(df['Column B'][i])
df['VLANS on Port'] = results

Another possible solution :

conn = (
    df.assign(tmp = df["Column C"].str.split(",\s*"))
        .explode("tmp").groupby("tmp")["Column B"].agg(", ".join)
)
                          
df["Column E"] = df["Column A"].map(conn)

Output :

print(df)

  Column A Column B Column C Column E
0       P1        A   P3, P4     B, D
1       P2        B   P1, P3        C
2       P3        C   P2, P3  A, B, C
3       P4        D   P1, P4     A, D

You can define a look-up function to find the indices of the rows and so construct the string. In the code below: .map sends each value in col_A to the mapper function. This forms an index list of the rows in which col_C contains the value.
This index List is then used to form another List of the values from col_B at those indices. Then all those values are added to the
string using a comma separator. I have assumed string types throughout which seems to be your intention.

import pandas as pd

df = pd.DataFrame({ 'col_A': ['P1', 'P2', 'P3', 'P4'],
                    'col_B': ['A', 'B', 'C', 'D'],
                    'col_C': [ 'P3, P4', 'P1, P3', 'P2, P3', 'P1, P4']
                    })

def mapper(val):
    res = ','.join(df.loc[df[df['col_C'].str.contains(val)].index, 'col_B'])
    return res

df['col_D'] = df['col_A'].map(mapper)

print(df)

gives:

  col_A col_B   col_C  col_D
0    P1     A  P3, P4    B,D
1    P2     B  P1, P3      C
2    P3     C  P2, P3  A,B,C
3    P4     D  P1, P4    A,D

# input
import pandas as pd
df = pd.read_clipboard()
df["Column C"] = df["Column C"].str.split(", ")  # assuming this is a column of lists

One option to then recreate your output is then:

# find which Px are in each row, then transpose (so which rows contain Px)
# reindex so that if any do Px do not occur then wrong values not returned
# fillna with 0 (as reindexing will return NaNs)
# rename columns with Column B values
dummies = df["Column C"].apply(pd.Series).stack().str\
    .get_dummies().sum(level=0).T\
        .reindex(df["Column A"]).fillna(0)\
            .rename(columns=df["Column B"].to_dict())

#           A  B  C  D
# Column A            
# P1        0  1  0  1
# P2        0  0  1  0
# P3        1  1  1  0
# P4        1  0  0  1

# for all non-zero values in each row, convert values to column names
# the output for each row will be a list
# reset the index so that it matches df (assuming that index is standard 0,...)
df["Column D"] = dummies.apply(
    lambda row: list(dummies.columns[row.values.nonzero()]), axis=1)\
    .reset_index(drop=True)

#   Column A Column B  Column C   Column D
# 0       P1        A  [P3, P4]     [B, D]
# 1       P2        B  [P1, P3]        [C]
# 2       P3        C  [P2, P3]  [A, B, C]
# 3       P4        D  [P1, P4]     [A, D]