I have a text file as shown below with "|" delimiters, the number of values for an attribute can be any. Need new columns based equal to number of values for an attribute.

Attribute1|6
Attribute2|10|15|27
Attribute3|3|7

Required output in DataFrame should be:

Attribute1         6
Attribute2_val_1  10
Attribute2_val_2  15
Attribute3_val_3  27
Attribute3_val_1   3
Attribute3_val_2   7

>Solution :

You can use the following:

MAX_COL = 10
df = (pd.read_csv('filename.txt', sep=r'\|', names=range(MAX_COL+1))
        .dropna(how='all', axis=1)
        .set_index(0).rename_axis('id').stack()
        .reset_index(name='value')
        .assign(id=lambda d: (d['id'].where(d.groupby('id')['id'].transform('size').le(1),
                                            d['id']+'_val_'+d['level_1'].astype(str))
                             )
               )
        .drop(columns=['level_1'])
      )

output:

>>> df
                 id  value
0        Attribute1    6.0
1  Attribute2_val_1   10.0
2  Attribute2_val_2   15.0
3  Attribute2_val_3   27.0
4  Attribute3_val_1    3.0
5  Attribute3_val_2    7.0

alternative:

with open('filename.txt') as f:
    df = pd.DataFrame([[s[0], s[1:]] for s in
                        map(lambda s: s.strip().split('|'), f.readlines())],
                      columns=['id', 'value']
                     )

df = (df.explode('value')
        .assign(id=lambda d: (d['id'].where(d.groupby('id')['id'].transform('size').le(1),
                                            d['id']+'_val_'+d.groupby('id')['id'].cumcount().add(1).astype(str))
                             )
               )
      )

output:

                 id value
0        Attribute1     6
1  Attribute2_val_1    10
1  Attribute2_val_2    15
1  Attribute2_val_3    27
2  Attribute3_val_1     3
2  Attribute3_val_2     7