duplicate rows and swap clumns based on condition - python-3.x

Question

I have the following table:
name a0 a1 type val
0 name1 1 0 AB 100
1 name1 2 0 AB 105
2 name2 1 2 BB 110
3 name3 1 0 AN 120
and I want to do this.
For every type I see where the type name does not contain the same 2 letters, I want to duplicate the row and swap the a0 and a1 columns and the letters of the type column. So, my result will be:
name a0 a1 type val
0 name1 1 0 AB 100
1 name1 0 1 BA 100
2 name1 2 0 AB 105
3 name1 0 2 BA 105
4 name2 1 2 BB 110
5 name3 1 0 AN 120
6 name3 0 1 NA 120
Note that for example for the same name we can have the same type but different a0 and a1 (and hence val).
So, we can have name1 and type AB as in the first two lines of the original table.
I tried:
df1 = pd.DataFrame({'name':['name1', 'name1', 'name2', 'name3'], 'a0':[1, 2, 1, 1], 'a1':[0, 0, 2, 0], 'type':['AB', 'AB', 'BB', 'AN'], 'val':[100,105, 110, 120]})
for idx in df1.index:
a1 = df1.loc[idx, 'a0']
a0 = df1.loc[idx, 'a1']
val = df1.loc[idx, 'val']
name = df1.loc[idx, 'name']
if df1.loc[idx, 'type'] == 'AB':
new_type = 'BA'
elif df1.loc[idx, 'type'] == 'AN':
new_type = 'NA'
row = pd.DataFrame({'name':name, 'a0':a0 , 'a1':a1 , 'type':new_type, 'val':val}, index=np.arange(idx))
df1 = df1.append(row, ignore_index=False)
df1 = df1.sort_index().reset_index(drop=True)
but it gives me:
name a0 a1 type val
0 name1 1 0 AB 100
1 name1 2 0 BA 105
2 name1 0 2 BA 105
3 name1 2 0 BA 105
4 name1 0 2 BA 105
5 name1 2 0 BA 105
6 name1 0 2 BA 105
7 name1 2 0 AB 105
8 name2 1 2 BB 110
9 name3 1 0 AN 120

Answer 1

First create mask for identify values with 2 different letters, create new DataFrame by DataFrame.assign, swap values in columns, join together and sorting by index, last create default index values:
mask = df['type'].apply(set).str.len() == 2
df1 = df[mask].assign(type=lambda x: df['type'].str[1] + df['type'].str[0])
df1[['a0','a1']] = df1[['a1','a0']].to_numpy()
#pandas below
#df1[['a0','a1']] = df1[['a1','a0']].values
df = pd.concat([df, df1]).sort_index().reset_index(drop=True)
print (df)
name a0 a1 type val
0 name1 1 0 AB 100
1 name1 0 1 BA 100
2 name1 2 0 AB 105
3 name1 0 2 BA 105
4 name2 1 2 BB 110
5 name3 1 0 AN 120
6 name3 0 1 NA 120

Answer 2

You can use:
def myfunc(x):
x['type']=x['type'][::-1]
x[['a0','a1']]=x[['a1','a0']].values
return x
m=df.type.apply(set).str.len().gt(1)
pd.concat([df,df.loc[m].apply(myfunc,axis=1)],ignore_index=True).sort_values(['name','val'])
name a0 a1 type val
0 name1 1 0 AB 100
4 name1 0 1 BA 100
1 name1 2 0 AB 105
5 name1 0 2 BA 105
2 name2 1 2 BB 110
3 name3 1 0 AN 120
6 name3 0 1 NA 120