How can i implement this recursive function?

i am trying to do a recursive function that allows me to do the following algorithm.

I have a dataframe like this:

+----+---+
|Prod| No|
+----+---+
|   A|  1|
|   A|  2|
|   B|  1|
|   B|  5|
|   1|  x|
|   B|  2|
|   2|  z|
|   x|  w|
+----+---+

And the output I need is:

+----+---+
|Prod| No|
+----+---+
|   A|  A|
|   A|  1|
|   A|  2|
|   A|  x|
|   A|  w|
|   B|  B|
|   B|  1|
|   B|  5|
|   B|  x|
|   B|  w|
|   B|  2|
|   B|  z|
|   1|  1|
|   1|  x|
|   1|  w|
|   2|  2|
|   2|  z|
|   x|  x|
|   x|  w|
+----+---+

That is, for each element on Prod column, i need all dependencies it has, directly or indirectly. Each element also has a dependence on each own.

I am trying a recursive function, but i'm loosing some elements. I have converted dataframes to list for easy operation. Can anyone help? Thank you.

The code i have is like this:

def check_dependencies(value):
    next_list = df.where(df.Prod == value).select('No').collect()
    return next_list


def recursive (dependencia):
    global new_iter
    dep_dinam = dependencia
    if(new_iter):
        k=i
    else:
        k=0
    newiter= False
    j=i
    comprimento =  len(dep_dinam)
    #for k in range(len(dependencia)):
    while k < comprimento:
        sub = check_dependencies(dep_dinam[k][0])
        list.extend(sub)
        if sub == []:
            lista_final.extend(list)
            return
        else:
            dep_dinam = sub
            comprimento = len(sub)
        k = k+1
        recursive(dep_dinam)

for i in range(len_unicos): 
    print(lista_unicos[i][0])
    new_iter=True
    recursive(lista_unicos)

Before, i have defined

data = [('A','1'),('A','2'),('A','3'),('B','1'),('B','5'),('1','x'),('1','y'),('B','2'),('2','z'),('x','w')]
columns = ["Prod","No"]
df = spark.createDataFrame(data=data, schema = columns)

unicos = df.select(df.Prod).distinct()
lista_unicos = unicos.collect()

In the example i have a smaller dataframe just for presentation purpose.