计算两个Python字典中包含的键的差异

Suppose I have two Python dictionaries – dictA and dictB. I need to find out if there are any keys which are present in dictB but not in dictA. What is the fastest way to go about it?

Should I convert the dictionary keys into a set and then go about?

Interested in knowing your thoughts…

Thanks for your responses.

Apologies for not stating my question properly. My scenario is like this – I have a dictA which can be the same as dictB or may have some keys missing as compared to dictB or else the value of some keys might be different which has to be set to that of dictA key’s value.

Problem is the dictionary has no standard and can have values which can be dict of dict.

Say

dictA={'key1':a, 'key2':b, 'key3':{'key11':cc, 'key12':dd}, 'key4':{'key111':{....}}}
dictB={'key1':a, 'key2:':newb, 'key3':{'key11':cc, 'key12':newdd, 'key13':ee}.......

So ‘key2’ value has to be reset to the new value and ‘key13’ has to be added inside the dict. The key value does not have a fixed format. It can be a simple value or a dict or a dict of dict.

You can use set operations on the keys:

diff = set(dictb.keys()) - set(dicta.keys())

Here is a class to find all the possibilities: what was added, what was removed, which key-value pairs are the same, and which key-value pairs are changed.

class DictDiffer(object):
    """
    Calculate the difference between two dictionaries as:
    (1) items added
    (2) items removed
    (3) keys same in both but changed values
    (4) keys same in both and unchanged values
    """
    def __init__(self, current_dict, past_dict):
        self.current_dict, self.past_dict = current_dict, past_dict
        self.set_current, self.set_past = set(current_dict.keys()), set(past_dict.keys())
        self.intersect = self.set_current.intersection(self.set_past)
    def added(self):
        return self.set_current - self.intersect 
    def removed(self):
        return self.set_past - self.intersect 
    def changed(self):
        return set(o for o in self.intersect if self.past_dict[o] != self.current_dict[o])
    def unchanged(self):
        return set(o for o in self.intersect if self.past_dict[o] == self.current_dict[o])

Here is some sample output:

>>> a = {'a': 1, 'b': 1, 'c': 0}
>>> b = {'a': 1, 'b': 2, 'd': 0}
>>> d = DictDiffer(b, a)
>>> print "Added:", d.added()
Added: set(['d'])
>>> print "Removed:", d.removed()
Removed: set(['c'])
>>> print "Changed:", d.changed()
Changed: set(['b'])
>>> print "Unchanged:", d.unchanged()
Unchanged: set(['a'])

Available as a github repo: https://github.com/hughdbrown/dictdiffer

In case you want the difference recursively, I have written a package for python: https://github.com/seperman/deepdiff

Installation

Install from PyPi:

pip install deepdiff

Example usage

Importing

>>> from deepdiff import DeepDiff
>>> from pprint import pprint
>>> from __future__ import print_function # In case running on Python 2

Same object returns empty

>>> t1 = {1:1, 2:2, 3:3}
>>> t2 = t1
>>> print(DeepDiff(t1, t2))
{}

Type of an item has changed

>>> t1 = {1:1, 2:2, 3:3}
>>> t2 = {1:1, 2:"2", 3:3}
>>> pprint(DeepDiff(t1, t2), indent=2)
{ 'type_changes': { 'root[2]': { 'newtype': <class 'str'>,
                                 'newvalue': '2',
                                 'oldtype': <class 'int'>,
                                 'oldvalue': 2}}}

Value of an item has changed

>>> t1 = {1:1, 2:2, 3:3}
>>> t2 = {1:1, 2:4, 3:3}
>>> pprint(DeepDiff(t1, t2), indent=2)
{'values_changed': {'root[2]': {'newvalue': 4, 'oldvalue': 2}}}

Item added and/or removed

>>> t1 = {1:1, 2:2, 3:3, 4:4}
>>> t2 = {1:1, 2:4, 3:3, 5:5, 6:6}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff)
{'dic_item_added': ['root[5]', 'root[6]'],
 'dic_item_removed': ['root[4]'],
 'values_changed': {'root[2]': {'newvalue': 4, 'oldvalue': 2}}}

String difference

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":"world"}}
>>> t2 = {1:1, 2:4, 3:3, 4:{"a":"hello", "b":"world!"}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{ 'values_changed': { 'root[2]': {'newvalue': 4, 'oldvalue': 2},
                      "root[4]['b']": { 'newvalue': 'world!',
                                        'oldvalue': 'world'}}}

String difference 2

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":"world!\nGoodbye!\n1\n2\nEnd"}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":"world\n1\n2\nEnd"}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{ 'values_changed': { "root[4]['b']": { 'diff': '--- \n'
                                                '+++ \n'
                                                '@@ -1,5 +1,4 @@\n'
                                                '-world!\n'
                                                '-Goodbye!\n'
                                                '+world\n'
                                                ' 1\n'
                                                ' 2\n'
                                                ' End',
                                        'newvalue': 'world\n1\n2\nEnd',
                                        'oldvalue': 'world!\n'
                                                    'Goodbye!\n'
                                                    '1\n'
                                                    '2\n'
                                                    'End'}}}

>>> 
>>> print (ddiff['values_changed']["root[4]['b']"]["diff"])
--- 
+++ 
@@ -1,5 +1,4 @@
-world!
-Goodbye!
+world
 1
 2
 End

Type change

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, 3]}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":"world\n\n\nEnd"}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{ 'type_changes': { "root[4]['b']": { 'newtype': <class 'str'>,
                                      'newvalue': 'world\n\n\nEnd',
                                      'oldtype': <class 'list'>,
                                      'oldvalue': [1, 2, 3]}}}

List difference

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, 3, 4]}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2]}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{'iterable_item_removed': {"root[4]['b'][2]": 3, "root[4]['b'][3]": 4}}

List difference 2:

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, 3]}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 3, 2, 3]}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{ 'iterable_item_added': {"root[4]['b'][3]": 3},
  'values_changed': { "root[4]['b'][1]": {'newvalue': 3, 'oldvalue': 2},
                      "root[4]['b'][2]": {'newvalue': 2, 'oldvalue': 3}}}

List difference ignoring order or duplicates: (with the same dictionaries as above)

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, 3]}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 3, 2, 3]}}
>>> ddiff = DeepDiff(t1, t2, ignore_order=True)
>>> print (ddiff)
{}

List that contains dictionary:

>>> t1 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, {1:1, 2:2}]}}
>>> t2 = {1:1, 2:2, 3:3, 4:{"a":"hello", "b":[1, 2, {1:3}]}}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (ddiff, indent = 2)
{ 'dic_item_removed': ["root[4]['b'][2][2]"],
  'values_changed': {"root[4]['b'][2][1]": {'newvalue': 3, 'oldvalue': 1}}}

Sets:

>>> t1 = {1, 2, 8}
>>> t2 = {1, 2, 3, 5}
>>> ddiff = DeepDiff(t1, t2)
>>> pprint (DeepDiff(t1, t2))
{'set_item_added': ['root[3]', 'root[5]'], 'set_item_removed': ['root[8]']}

Named Tuples:

>>> from collections import namedtuple
>>> Point = namedtuple('Point', ['x', 'y'])
>>> t1 = Point(x=11, y=22)
>>> t2 = Point(x=11, y=23)
>>> pprint (DeepDiff(t1, t2))
{'values_changed': {'root.y': {'newvalue': 23, 'oldvalue': 22}}}

Custom objects:

>>> class ClassA(object):
...     a = 1
...     def __init__(self, b):
...         self.b = b
... 
>>> t1 = ClassA(1)
>>> t2 = ClassA(2)
>>> 
>>> pprint(DeepDiff(t1, t2))
{'values_changed': {'root.b': {'newvalue': 2, 'oldvalue': 1}}}

Object attribute added:

>>> t2.c = "new attribute"
>>> pprint(DeepDiff(t1, t2))
{'attribute_added': ['root.c'],
 'values_changed': {'root.b': {'newvalue': 2, 'oldvalue': 1}}}

not sure whether its “fast” or not, but normally, one can do this

dicta = {"a":1,"b":2,"c":3,"d":4}
dictb = {"a":1,"d":2}
for key in dicta.keys():
    if not key in dictb:
        print key

As Alex Martelli wrote, if you simply want to check if any key in B is not in A, any(True for k in dictB if k not in dictA) would be the way to go.

To find the keys that are missing:

diff = set(dictB)-set(dictA) #sets

C:\Dokumente und Einstellungen\thc>python -m timeit -s "dictA =    
dict(zip(range(1000),range
(1000))); dictB = dict(zip(range(0,2000,2),range(1000)))" "diff=set(dictB)-set(dictA)"
10000 loops, best of 3: 107 usec per loop

diff = [ k for k in dictB if k not in dictA ] #lc

C:\Dokumente und Einstellungen\thc>python -m timeit -s "dictA = 
dict(zip(range(1000),range
(1000))); dictB = dict(zip(range(0,2000,2),range(1000)))" "diff=[ k for k in dictB if
k not in dictA ]"
10000 loops, best of 3: 95.9 usec per loop

So those two solutions are pretty much the same speed.

If you really mean exactly what you say (that you only need to find out IF “there are any keys” in B and not in A, not WHICH ONES might those be if any), the fastest way should be:

if any(True for k in dictB if k not in dictA): ...

If you actually need to find out WHICH KEYS, if any, are in B and not in A, and not just “IF” there are such keys, then existing answers are quite appropriate (but I do suggest more precision in future questions if that’s indeed what you mean;-).

Use set():

set(dictA.keys()).intersection(dictB.keys())

The top answer by hughdbrown suggests using set difference, which is definitely the best approach:

diff = set(dictb.keys()) - set(dicta.keys())

The problem with this code is that it builds two lists just to create two sets, so it’s wasting 4N time and 2N space. It’s also a bit more complicated than it needs to be.

Usually, this is not a big deal, but if it is:

diff = dictb.keys() - dicta

• You don’t need to convert the right dict to a set; set difference takes any iterable (and a dict is an iterable of its keys).
• You also don’t need to convert the left dict to a set, because anything that complies with collections.abc.Mapping has a KeysView that acts like a Set.

Python 2

In Python 2, keys() returns a list of the keys, not a KeysView. So you have to ask for viewkeys() directly.

diff = dictb.viewkeys() - dicta

For dual-version 2.7/3.x code, you’re hopefully using six or something similar, so you can use six.viewkeys(dictb):

diff = six.viewkeys(dictb) - dicta

In 2.4-2.6, there is no KeysView. But you can at least cut the cost from 4N to N by building your left set directly out of an iterator, instead of building a list first:

diff = set(dictb) - dicta

Items

I have a dictA which can be the same as dictB or may have some keys missing as compared to dictB or else the value of some keys might be different

So you really don’t need to compare the keys, but the items. An ItemsView is only a Set if the values are hashable, like strings. If they are, it’s easy:

diff = dictb.items() - dicta.items()

Recursive diff

Although the question isn’t directly asking for a recursive diff, some of the example values are dicts, and it appears the expected output does recursively diff them. There are already multiple answers here showing how to do that.

There is an other question in stackoverflow about this argument and i have to admit that there is a simple solution explained: the datadiff library of python helps printing the difference between two dictionaries.

Here’s a way that will work, allows for keys that evaluate to False, and still uses a generator expression to fall out early if possible. It’s not exceptionally pretty though.

any(map(lambda x: True, (k for k in b if k not in a)))

EDIT:

THC4k posted a reply to my comment on another answer. Here’s a better, prettier way to do the above:

any(True for k in b if k not in a)

Not sure how that never crossed my mind…

This is an old question and asks a little bit less than what I needed so this answer actually solves more than this question asks. The answers in this question helped me solve the following:

1. (asked) Record differences between two dictionaries
2. Merge differences from #1 into base dictionary
3. (asked) Merge differences between two dictionaries (treat dictionary #2 as if it were a diff dictionary)
4. Try to detect item movements as well as changes
5. (asked) Do all of this recursively

All this combined with JSON makes for a pretty powerful configuration storage support.

The solution (also on github):

from collections import OrderedDict
from pprint import pprint


class izipDestinationMatching(object):
    __slots__ = ("attr", "value", "index")

    def __init__(self, attr, value, index):
        self.attr, self.value, self.index = attr, value, index

    def __repr__(self):
        return "izip_destination_matching: found match by '%s' = '%s' @ %d" % (self.attr, self.value, self.index)


def izip_destination(a, b, attrs, addMarker=True):
    """
    Returns zipped lists, but final size is equal to b with (if shorter) a padded with nulls
    Additionally also tries to find item reallocations by searching child dicts (if they are dicts) for attribute, listed in attrs)
    When addMarker == False (patching), final size will be the longer of a, b
    """
    for idx, item in enumerate(b):
        try:
            attr = next((x for x in attrs if x in item), None)  # See if the item has any of the ID attributes
            match, matchIdx = next(((orgItm, idx) for idx, orgItm in enumerate(a) if attr in orgItm and orgItm[attr] == item[attr]), (None, None)) if attr else (None, None)
            if match and matchIdx != idx and addMarker: item[izipDestinationMatching] = izipDestinationMatching(attr, item[attr], matchIdx)
        except:
            match = None
        yield (match if match else a[idx] if len(a) > idx else None), item
    if not addMarker and len(a) > len(b):
        for item in a[len(b) - len(a):]:
            yield item, item


def dictdiff(a, b, searchAttrs=[]):
    """
    returns a dictionary which represents difference from a to b
    the return dict is as short as possible:
      equal items are removed
      added / changed items are listed
      removed items are listed with value=None
    Also processes list values where the resulting list size will match that of b.
    It can also search said list items (that are dicts) for identity values to detect changed positions.
      In case such identity value is found, it is kept so that it can be re-found during the merge phase
    @param a: original dict
    @param b: new dict
    @param searchAttrs: list of strings (keys to search for in sub-dicts)
    @return: dict / list / whatever input is
    """
    if not (isinstance(a, dict) and isinstance(b, dict)):
        if isinstance(a, list) and isinstance(b, list):
            return [dictdiff(v1, v2, searchAttrs) for v1, v2 in izip_destination(a, b, searchAttrs)]
        return b
    res = OrderedDict()
    if izipDestinationMatching in b:
        keepKey = b[izipDestinationMatching].attr
        del b[izipDestinationMatching]
    else:
        keepKey = izipDestinationMatching
    for key in sorted(set(a.keys() + b.keys())):
        v1 = a.get(key, None)
        v2 = b.get(key, None)
        if keepKey == key or v1 != v2: res[key] = dictdiff(v1, v2, searchAttrs)
    if len(res) <= 1: res = dict(res)  # This is only here for pretty print (OrderedDict doesn't pprint nicely)
    return res


def dictmerge(a, b, searchAttrs=[]):
    """
    Returns a dictionary which merges differences recorded in b to base dictionary a
    Also processes list values where the resulting list size will match that of a
    It can also search said list items (that are dicts) for identity values to detect changed positions
    @param a: original dict
    @param b: diff dict to patch into a
    @param searchAttrs: list of strings (keys to search for in sub-dicts)
    @return: dict / list / whatever input is
    """
    if not (isinstance(a, dict) and isinstance(b, dict)):
        if isinstance(a, list) and isinstance(b, list):
            return [dictmerge(v1, v2, searchAttrs) for v1, v2 in izip_destination(a, b, searchAttrs, False)]
        return b
    res = OrderedDict()
    for key in sorted(set(a.keys() + b.keys())):
        v1 = a.get(key, None)
        v2 = b.get(key, None)
        #print "processing", key, v1, v2, key not in b, dictmerge(v1, v2)
        if v2 is not None: res[key] = dictmerge(v1, v2, searchAttrs)
        elif key not in b: res[key] = v1
    if len(res) <= 1: res = dict(res)  # This is only here for pretty print (OrderedDict doesn't pprint nicely)
    return res

what about standart (compare FULL Object)

PyDev->new PyDev Module->Module: unittest

import unittest


class Test(unittest.TestCase):


    def testName(self):
        obj1 = {1:1, 2:2}
        obj2 = {1:1, 2:2}
        self.maxDiff = None # sometimes is usefull
        self.assertDictEqual(d1, d2)

if __name__ == "__main__":
    #import sys;sys.argv = ['', 'Test.testName']

    unittest.main()

If on Python ≥ 2.7:

# update different values in dictB
# I would assume only dictA should be updated,
# but the question specifies otherwise

for k in dictA.viewkeys() & dictB.viewkeys():
    if dictA[k] != dictB[k]:
        dictB[k]= dictA[k]

# add missing keys to dictA

dictA.update( (k,dictB[k]) for k in dictB.viewkeys() - dictA.viewkeys() )

Here is a solution for deep comparing 2 dictionaries keys:

def compareDictKeys(dict1, dict2):
  if type(dict1) != dict or type(dict2) != dict:
      return False

  keys1, keys2 = dict1.keys(), dict2.keys()
  diff = set(keys1) - set(keys2) or set(keys2) - set(keys1)

  if not diff:
      for key in keys1:
          if (type(dict1[key]) == dict or type(dict2[key]) == dict) and not compareDictKeys(dict1[key], dict2[key]):
              diff = True
              break

  return not diff

here’s a solution that can compare more than two dicts:

def diff_dict(dicts, default=None):
    diff_dict = {}
    # add 'list()' around 'd.keys()' for python 3 compatibility
    for k in set(sum([d.keys() for d in dicts], [])):
        # we can just use "values = [d.get(k, default) ..." below if 
        # we don't care that d1[k]=default and d2[k]=missing will
        # be treated as equal
        if any(k not in d for d in dicts):
            diff_dict[k] = [d.get(k, default) for d in dicts]
        else:
            values = [d[k] for d in dicts]
            if any(v != values[0] for v in values):
                diff_dict[k] = values
    return diff_dict

usage example:

import matplotlib.pyplot as plt
diff_dict([plt.rcParams, plt.rcParamsDefault, plt.matplotlib.rcParamsOrig])

My recipe of symmetric difference between two dictionaries:

def find_dict_diffs(dict1, dict2):
    unequal_keys = []
    unequal_keys.extend(set(dict1.keys()).symmetric_difference(set(dict2.keys())))
    for k in dict1.keys():
        if dict1.get(k, 'N\A') != dict2.get(k, 'N\A'):
            unequal_keys.append(k)
    if unequal_keys:
        print 'param', 'dict1\t', 'dict2'
        for k in set(unequal_keys):
            print str(k)+'\t'+dict1.get(k, 'N\A')+'\t '+dict2.get(k, 'N\A')
    else:
        print 'Dicts are equal'

dict1 = {1:'a', 2:'b', 3:'c', 4:'d', 5:'e'}
dict2 = {1:'b', 2:'a', 3:'c', 4:'d', 6:'f'}

find_dict_diffs(dict1, dict2)

And result is:

param   dict1   dict2
1       a       b
2       b       a
5       e       N\A
6       N\A     f

As mentioned in other answers, unittest produces some nice output for comparing dicts, but in this example we don’t want to have to build a whole test first.

Scraping the unittest source, it looks like you can get a fair solution with just this:

import difflib
import pprint

def diff_dicts(a, b):
    if a == b:
        return ''
    return '\n'.join(
        difflib.ndiff(pprint.pformat(a, width=30).splitlines(),
                      pprint.pformat(b, width=30).splitlines())
    )

so

dictA = dict(zip(range(7), map(ord, 'python')))
dictB = {0: 112, 1: 'spam', 2: [1,2,3], 3: 104, 4: 111}
print diff_dicts(dictA, dictB)

Results in:

{0: 112,
-  1: 121,
-  2: 116,
+  1: 'spam',
+  2: [1, 2, 3],
   3: 104,
-  4: 111,
?        ^

+  4: 111}
?        ^

-  5: 110}

Where:

• ‘-‘ indicates key/values in the first but not second dict
• ‘+’ indicates key/values in the second but not the first dict

Like in unittest, the only caveat is that the final mapping can be thought to be a diff, due to the trailing comma/bracket.

@Maxx has an excellent answer, use the unittest tools provided by Python:

import unittest


class Test(unittest.TestCase):
    def runTest(self):
        pass

    def testDict(self, d1, d2, maxDiff=None):
        self.maxDiff = maxDiff
        self.assertDictEqual(d1, d2)

Then, anywhere in your code you can call:

try:
    Test().testDict(dict1, dict2)
except Exception, e:
    print e

The resulting output looks like the output from diff, pretty-printing the dictionaries with + or - prepending each line that is different.

Not sure if it is still relevant but I came across this problem, my situation i just needed to return a dictionary of the changes for all nested dictionaries etc etc. Could not find a good solution out there but I did end up writing a simple function to do this. Hope this helps,

If you want a built-in solution for a full comparison with arbitrary dict structures, @Maxx’s answer is a good start.

import unittest

test = unittest.TestCase()
test.assertEqual(dictA, dictB)

Based on ghostdog74’s answer,

dicta = {"a":1,"d":2}
dictb = {"a":5,"d":2}

for value in dicta.values():
    if not value in dictb.values():
        print value

will print differ value of dicta

Try this to find de intersection, the keys that is in both dictionarie, if you want the keys not found on second dictionarie, just use the not in…

intersect = filter(lambda x, dictB=dictB.keys(): x in dictB, dictA.keys())