Multiple keys

Dictionary (data structures)

Multi-key combinations to access values

a) Table/matrix representation using tupels

Tupels are 'hashable' objects and hence can be used as a key in python dictionaries.

{(key_part_1, key_part_2): value}

Example

d={}

d[('Antje','place')]='Barcelona'

d[('Antje','year')]='1987'

d[('Mike','place')]='Berlin'

key=('Antje','year')

if key in d:

print( d[key] )

1987

# Problem: when key-pair does not exist: key error for missing entries

print( d[('Mike','year')] )

Traceback (most recent call last):

File "<stdin>", line 1, in <module>

KeyError: ('Mikes', 'year')

# Solution: provide values for all key-pairs (a), or use sparse nested dictionary (b)

b) Sparse matrix or tree like representation using dict in dict

# Nested dictionaries in python

from collections import defaultdict

d = defaultdict(dict)

d['Antje']['place']='Barcelona'

d['Antje']['year']='1987'

defaultdict(<type 'dict'>, {'Antje': {'place': 'Barcelona', 'year': '1987'}})

d['Mike']['place']='Berlin'

# print if key pair exist

keyA='Antje'

keyB='year'

if ( keyA in d ) & ( keyB in d[keyA] ):

print( d[keyA][keyB] )

1987

# key error for missing entried

print( d['Mike']['year'] )

Traceback (most recent call last):

File "<stdin>", line 1, in <module>

KeyError: 'year'

# use get, to print default value if second key ('year') is missing

print( d['Mike'].get('year', 'NaN') )

NaN

# check if both key's are present

if 'year' in d.get('Antje', {}):

print( d['Antje']['year'] )

# check if second key is present

if d['Antje'].get('year', False):

print( d['Antje']['year'] )

# check if first key is present

'Mike' in d

True

# get all first level keys

sorted(d.keys())

['Antje', 'Mike']

# get second level keys of selected first level key 'Mike'

sorted([ k for k in d['Mike'] ])

['place']

# get all possible second level keys of all first level keys together (set comprehension)

sorted(list({k2 for v in d.values() for k2 in v}))

['place', 'year']

# get values of all second keys of a first key identifier (name)

name='Antje'

[ d[name][k] for k in d[name] ]

['Barcelona', '1987']

# get all possible values of a selected second key: 'place' or 'year' (NaN if key is missing)

[ d[k].get('place', 'NaN') for k in d]

['Barcelona', 'Berlin']

[ d[k].get('year', 'NaN') for k in d]

['1987', 'NaN']

# sorted unique list of values of second level key 'place'

sorted(set([ d[k].get('place', 'NaN') for k in d]))

['Barcelona', 'Berlin']

# get all keys with same value 'Berlin' in second key 'place' (get all people born in Berlin)

[k for k in d.keys() if d[k].get('place', 'NaN') == 'Berlin']

['Mike']

# get all keys for which the second key 'year' exist

[k for k in d.keys() if d[k].get('year', False)]

['Antje']

# get all possible pairs of (first, second) keys

[ (k1,k2) for (k1, v) in d.items() for k2 in v ]

[('Antje', 'place'), ('Antje', 'year'), ('Mike', 'place')]

# exchange/replace/reorder key with value: place -> name mapping

newdict = dict((d[k].get('place', 'NaN'), k) for k in d.keys())

{'Berlin': 'Mike', 'Barcelona': 'Antje'}

https://gist.github.com/hrldcpr/2012250

http://sahandsaba.com/thirty-python-language-features-and-tricks-you-may-not-know.html

http://blog.amir.rachum.com/blog/2013/01/02/python-the-dictionary-playbook/