Creating a nested dictionary with for loop

You can use pd.read_csv() and to_dict():

import pandas as pd

contacts = pd.read_csv('test.csv', sep=';').set_index('key').to_dict(orient='index')

Yields:

{'Tom': {'name': 'Tom Techie', 'phone': '123 123546', 'email': '[email protected]', 'skype': 'skypenick'}, 'Mike': {'name': 'Mike Mechanic', 'phone': '000 123456', 'email': '[email protected]', 'skype': '-Mike-M-'}, 'Archie': {'name': 'Archie Architect', 'phone': '050 987654', 'email': 'archie@archie', 'skype': nan}}

I like the pandas answer, but if you don't want a 3rd party library, use the built-in csv module:

import csv
from pprint import pprint

D = {}
with open('csv',newline='') as f:
    r = csv.DictReader(f,delimiter=';')
    for line in r:
        name = line['key']
        del line['key']
        D[name] = dict(line)

pprint(D)

Output:

{'Archie': {'email': 'archie@archie',
            'name': 'Archie Architect',
            'phone': '050 987654',
            'skype': None},
 'Mike': {'email': '[email protected]',
          'name': 'Mike Mechanic',
          'phone': '000 123456',
          'skype': '-Mike-M-'},
 'Tom': {'email': '[email protected]',
         'name': 'Tom Techie',
         'phone': '123 123546',
         'skype': 'skypenick'}}

You can use zip() to achieve your goal:

file = """key;name;phone;email;skype
Tom;Tom Techie;123 123546;[email protected];skypenick
Mike;Mike Mechanic;000 123456;[email protected];-Mike-M-
Archie;Archie Architect;050 987654;archie@archie""".splitlines()

d = {}
h = None
for i in file: # works the same for your csv-file
    # first row == header, store in h
    if h is None:
        h = i.strip().split(";")[1:]
        continue # done for first row

    x = i.strip().split(";")

    # zip pairs the read in line with the header line to get tuples
    # wich are fed into the dict constructor that creates the inner dict
    d[x[0]] = dict(zip(h,x[1:]+[""])) # no default for skype

    # use this instead if you want the skype key always present with empty default
    # d[x[0]] = dict(zip(h,x[1:]+[""]))  

print(d)

zip() discards the elements of the longer list - you won't need any checks for that.

Output:

{'Tom':   {'name': 'Tom Techie', 'phone': '123 123546', 
           'email': '[email protected]', 'skype': 'skypenick'}, 
 'Mike':   {'name': 'Mike Mechanic', 'phone': '000 123456', 
            'email': '[email protected]', 'skype': '-Mike-M-'}, 
 'Archie': {'name': 'Archie Architect', 'phone': '050 987654', 
            'email': 'archie@archie'}}

If you use the commented line, the data will get a default value of '' for the skype - works only b/c skype is the last element of the splitted line

You can use a dict comprehension! Assuming the data is something like

with open("df.csv", "r") as file:
  d = {x.split(";")[0]:{
      "name": x.split(";")[2], 
      "phone": x.split(";")[3],
      "email": x.split(";")[1], 
      "skype": x.split(";")[4][:-1] # Slice off trailing newline
  } for x in file}
  d.pop("")

We want to open files using with whenever possible to benefit from Python's context management. See https://www.python.org/dev/peps/pep-0343/ for fundamental understanding of the with statement.

Since the key "" only appears once at the head of the csv, we can pop it at the end and avoid performing a comparison at every iteration. A dict comprehension accomplishes the same thing you wanted to achieve with d.update.

More about comprehensions: https://docs.python.org/3/tutorial/datastructures.html#list-comprehensions

Edit: refactoring to remove the repetitive calls to .split can look something like this:

def line_to_dict(x, d):
  x = x.split(";")
  d[x[0]] = {
      "name": x[2], 
      "phone": x[3],
      "email": x[1], 
      "skype": x[4][:-1] # Slice off trailing newline
  }

with open("df.csv", "r") as file:
  d = {}
  for x in file:
    line_to_dict(x, d)
  d.pop("")