Assume I have this:
[
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
and by searching "Pam" as name, I want to retrieve the related dictionary: {name: "Pam", age: 7}
How to achieve this ?
You can use a generator expression:
>>> dicts = [
... { "name": "Tom", "age": 10 },
... { "name": "Mark", "age": 5 },
... { "name": "Pam", "age": 7 },
... { "name": "Dick", "age": 12 }
... ]
>>> next(item for item in dicts if item["name"] == "Pam")
{'age': 7, 'name': 'Pam'}
If you need to handle the item not being there, then you can do what user Matt suggested in his comment and provide a default using a slightly different API:
next((item for item in dicts if item["name"] == "Pam"), None)
And to find the index of the item, rather than the item itself, you can enumerate() the list:
next((i for i, item in enumerate(dicts) if item["name"] == "Pam"), None)
This looks to me the most pythonic way:
people = [
{'name': "Tom", 'age': 10},
{'name': "Mark", 'age': 5},
{'name': "Pam", 'age': 7}
]
filter(lambda person: person['name'] == 'Pam', people)
result (returned as a list in Python 2):
[{'age': 7, 'name': 'Pam'}]
Note: In Python 3, a filter object is returned. So the python3 solution would be:
list(filter(lambda person: person['name'] == 'Pam', people))
#Frédéric Hamidi's answer is great. In Python 3.x the syntax for .next() changed slightly. Thus a slight modification:
>>> dicts = [
{ "name": "Tom", "age": 10 },
{ "name": "Mark", "age": 5 },
{ "name": "Pam", "age": 7 },
{ "name": "Dick", "age": 12 }
]
>>> next(item for item in dicts if item["name"] == "Pam")
{'age': 7, 'name': 'Pam'}
As mentioned in the comments by #Matt, you can add a default value as such:
>>> next((item for item in dicts if item["name"] == "Pam"), False)
{'name': 'Pam', 'age': 7}
>>> next((item for item in dicts if item["name"] == "Sam"), False)
False
>>>
You can use a list comprehension:
def search(name, people):
return [element for element in people if element['name'] == name]
I tested various methods to go through a list of dictionaries and return the dictionaries where key x has a certain value.
Results:
Speed: list comprehension > generator expression >> normal list iteration >>> filter.
All scale linear with the number of dicts in the list (10x list size -> 10x time).
The keys per dictionary does not affect speed significantly for large amounts (thousands) of keys. Please see this graph I calculated: https://imgur.com/a/quQzv (method names see below).
All tests done with Python 3.6.4, W7x64.
from random import randint
from timeit import timeit
list_dicts = []
for _ in range(1000): # number of dicts in the list
dict_tmp = {}
for i in range(10): # number of keys for each dict
dict_tmp[f"key{i}"] = randint(0,50)
list_dicts.append( dict_tmp )
def a():
# normal iteration over all elements
for dict_ in list_dicts:
if dict_["key3"] == 20:
pass
def b():
# use 'generator'
for dict_ in (x for x in list_dicts if x["key3"] == 20):
pass
def c():
# use 'list'
for dict_ in [x for x in list_dicts if x["key3"] == 20]:
pass
def d():
# use 'filter'
for dict_ in filter(lambda x: x['key3'] == 20, list_dicts):
pass
Results:
1.7303 # normal list iteration
1.3849 # generator expression
1.3158 # list comprehension
7.7848 # filter
people = [
{'name': "Tom", 'age': 10},
{'name': "Mark", 'age': 5},
{'name': "Pam", 'age': 7}
]
def search(name):
for p in people:
if p['name'] == name:
return p
search("Pam")
Have you ever tried out the pandas package? It's perfect for this kind of search task and optimized too.
import pandas as pd
listOfDicts = [
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
# Create a data frame, keys are used as column headers.
# Dict items with the same key are entered into the same respective column.
df = pd.DataFrame(listOfDicts)
# The pandas dataframe allows you to pick out specific values like so:
df2 = df[ (df['name'] == 'Pam') & (df['age'] == 7) ]
# Alternate syntax, same thing
df2 = df[ (df.name == 'Pam') & (df.age == 7) ]
I've added a little bit of benchmarking below to illustrate pandas' faster runtimes on a larger scale i.e. 100k+ entries:
setup_large = 'dicts = [];\
[dicts.extend(({ "name": "Tom", "age": 10 },{ "name": "Mark", "age": 5 },\
{ "name": "Pam", "age": 7 },{ "name": "Dick", "age": 12 })) for _ in range(25000)];\
from operator import itemgetter;import pandas as pd;\
df = pd.DataFrame(dicts);'
setup_small = 'dicts = [];\
dicts.extend(({ "name": "Tom", "age": 10 },{ "name": "Mark", "age": 5 },\
{ "name": "Pam", "age": 7 },{ "name": "Dick", "age": 12 }));\
from operator import itemgetter;import pandas as pd;\
df = pd.DataFrame(dicts);'
method1 = '[item for item in dicts if item["name"] == "Pam"]'
method2 = 'df[df["name"] == "Pam"]'
import timeit
t = timeit.Timer(method1, setup_small)
print('Small Method LC: ' + str(t.timeit(100)))
t = timeit.Timer(method2, setup_small)
print('Small Method Pandas: ' + str(t.timeit(100)))
t = timeit.Timer(method1, setup_large)
print('Large Method LC: ' + str(t.timeit(100)))
t = timeit.Timer(method2, setup_large)
print('Large Method Pandas: ' + str(t.timeit(100)))
#Small Method LC: 0.000191926956177
#Small Method Pandas: 0.044392824173
#Large Method LC: 1.98827004433
#Large Method Pandas: 0.324505090714
To add just a tiny bit to #FrédéricHamidi.
In case you are not sure a key is in the the list of dicts, something like this would help:
next((item for item in dicts if item.get("name") and item["name"] == "Pam"), None)
Simply using list comprehension:
[i for i in dct if i['name'] == 'Pam'][0]
Sample code:
dct = [
{'name': 'Tom', 'age': 10},
{'name': 'Mark', 'age': 5},
{'name': 'Pam', 'age': 7}
]
print([i for i in dct if i['name'] == 'Pam'][0])
> {'age': 7, 'name': 'Pam'}
You can achieve this with the usage of filter and next methods in Python.
filter method filters the given sequence and returns an iterator.
next method accepts an iterator and returns the next element in the list.
So you can find the element by,
my_dict = [
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
next(filter(lambda obj: obj.get('name') == 'Pam', my_dict), None)
and the output is,
{'name': 'Pam', 'age': 7}
Note: The above code will return None incase if the name we are searching is not found.
One simple way using list comprehensions is , if l is the list
l = [
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
then
[d['age'] for d in l if d['name']=='Tom']
def dsearch(lod, **kw):
return filter(lambda i: all((i[k] == v for (k, v) in kw.items())), lod)
lod=[{'a':33, 'b':'test2', 'c':'a.ing333'},
{'a':22, 'b':'ihaha', 'c':'fbgval'},
{'a':33, 'b':'TEst1', 'c':'s.ing123'},
{'a':22, 'b':'ihaha', 'c':'dfdvbfjkv'}]
list(dsearch(lod, a=22))
[{'a': 22, 'b': 'ihaha', 'c': 'fbgval'},
{'a': 22, 'b': 'ihaha', 'c': 'dfdvbfjkv'}]
list(dsearch(lod, a=22, b='ihaha'))
[{'a': 22, 'b': 'ihaha', 'c': 'fbgval'},
{'a': 22, 'b': 'ihaha', 'c': 'dfdvbfjkv'}]
list(dsearch(lod, a=22, c='fbgval'))
[{'a': 22, 'b': 'ihaha', 'c': 'fbgval'}]
This is a general way of searching a value in a list of dictionaries:
def search_dictionaries(key, value, list_of_dictionaries):
return [element for element in list_of_dictionaries if element[key] == value]
dicts=[
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
from collections import defaultdict
dicts_by_name=defaultdict(list)
for d in dicts:
dicts_by_name[d['name']]=d
print dicts_by_name['Tom']
#output
#>>>
#{'age': 10, 'name': 'Tom'}
names = [{'name':'Tom', 'age': 10}, {'name': 'Mark', 'age': 5}, {'name': 'Pam', 'age': 7}]
resultlist = [d for d in names if d.get('name', '') == 'Pam']
first_result = resultlist[0]
This is one way...
You can try this:
''' lst: list of dictionaries '''
lst = [{"name": "Tom", "age": 10}, {"name": "Mark", "age": 5}, {"name": "Pam", "age": 7}]
search = raw_input("What name: ") #Input name that needs to be searched (say 'Pam')
print [ lst[i] for i in range(len(lst)) if(lst[i]["name"]==search) ][0] #Output
>>> {'age': 7, 'name': 'Pam'}
Put the accepted answer in a function to easy re-use
def get_item(collection, key, target):
return next((item for item in collection if item[key] == target), None)
Or also as a lambda
get_item_lambda = lambda collection, key, target : next((item for item in collection if item[key] == target), None)
Result
key = "name"
target = "Pam"
print(get_item(target_list, key, target))
print(get_item_lambda(target_list, key, target))
#{'name': 'Pam', 'age': 7}
#{'name': 'Pam', 'age': 7}
In case the key may not be in the target dictionary use dict.get and avoid KeyError
def get_item(collection, key, target):
return next((item for item in collection if item.get(key, None) == target), None)
get_item_lambda = lambda collection, key, target : next((item for item in collection if item.get(key, None) == target), None)
My first thought would be that you might want to consider creating a dictionary of these dictionaries ... if, for example, you were going to be searching it more a than small number of times.
However that might be a premature optimization. What would be wrong with:
def get_records(key, store=dict()):
'''Return a list of all records containing name==key from our store
'''
assert key is not None
return [d for d in store if d['name']==key]
Most (if not all) implementations proposed here have two flaws:
They assume only one key to be passed for searching, while it may be interesting to have more for complex dict
They assume all keys passed for searching exist in the dicts, hence they don't deal correctly with KeyError occuring when it is not.
An updated proposition:
def find_first_in_list(objects, **kwargs):
return next((obj for obj in objects if
len(set(obj.keys()).intersection(kwargs.keys())) > 0 and
all([obj[k] == v for k, v in kwargs.items() if k in obj.keys()])),
None)
Maybe not the most pythonic, but at least a bit more failsafe.
Usage:
>>> obj1 = find_first_in_list(list_of_dict, name='Pam', age=7)
>>> obj2 = find_first_in_list(list_of_dict, name='Pam', age=27)
>>> obj3 = find_first_in_list(list_of_dict, name='Pam', address='nowhere')
>>>
>>> print(obj1, obj2, obj3)
{"name": "Pam", "age": 7}, None, {"name": "Pam", "age": 7}
The gist.
Here is a comparison using iterating throuhg list, using filter+lambda or refactoring(if needed or valid to your case) your code to dict of dicts rather than list of dicts
import time
# Build list of dicts
list_of_dicts = list()
for i in range(100000):
list_of_dicts.append({'id': i, 'name': 'Tom'})
# Build dict of dicts
dict_of_dicts = dict()
for i in range(100000):
dict_of_dicts[i] = {'name': 'Tom'}
# Find the one with ID of 99
# 1. iterate through the list
lod_ts = time.time()
for elem in list_of_dicts:
if elem['id'] == 99999:
break
lod_tf = time.time()
lod_td = lod_tf - lod_ts
# 2. Use filter
f_ts = time.time()
x = filter(lambda k: k['id'] == 99999, list_of_dicts)
f_tf = time.time()
f_td = f_tf- f_ts
# 3. find it in dict of dicts
dod_ts = time.time()
x = dict_of_dicts[99999]
dod_tf = time.time()
dod_td = dod_tf - dod_ts
print 'List of Dictionries took: %s' % lod_td
print 'Using filter took: %s' % f_td
print 'Dict of Dicts took: %s' % dod_td
And the output is this:
List of Dictionries took: 0.0099310874939
Using filter took: 0.0121960639954
Dict of Dicts took: 4.05311584473e-06
Conclusion:
Clearly having a dictionary of dicts is the most efficient way to be able to search in those cases, where you know say you will be searching by id's only.
interestingly using filter is the slowest solution.
I would create a dict of dicts like so:
names = ["Tom", "Mark", "Pam"]
ages = [10, 5, 7]
my_d = {}
for i, j in zip(names, ages):
my_d[i] = {"name": i, "age": j}
or, using exactly the same info as in the posted question:
info_list = [{"name": "Tom", "age": 10}, {"name": "Mark", "age": 5}, {"name": "Pam", "age": 7}]
my_d = {}
for d in info_list:
my_d[d["name"]] = d
Then you could do my_d["Pam"] and get {"name": "Pam", "age": 7}
Ducks will be a lot faster than a list comprehension or filter. It builds an index on your objects so lookups don't need to scan every item.
pip install ducks
from ducks import Dex
dicts = [
{"name": "Tom", "age": 10},
{"name": "Mark", "age": 5},
{"name": "Pam", "age": 7}
]
# Build the index
dex = Dex(dicts, {'name': str, 'age': int})
# Find matching objects
dex[{'name': 'Pam', 'age': 7}]
Result: [{'name': 'Pam', 'age': 7}]
You have to go through all elements of the list. There is not a shortcut!
Unless somewhere else you keep a dictionary of the names pointing to the items of the list, but then you have to take care of the consequences of popping an element from your list.
I found this thread when I was searching for an answer to the same
question. While I realize that it's a late answer, I thought I'd
contribute it in case it's useful to anyone else:
def find_dict_in_list(dicts, default=None, **kwargs):
"""Find first matching :obj:`dict` in :obj:`list`.
:param list dicts: List of dictionaries.
:param dict default: Optional. Default dictionary to return.
Defaults to `None`.
:param **kwargs: `key=value` pairs to match in :obj:`dict`.
:returns: First matching :obj:`dict` from `dicts`.
:rtype: dict
"""
rval = default
for d in dicts:
is_found = False
# Search for keys in dict.
for k, v in kwargs.items():
if d.get(k, None) == v:
is_found = True
else:
is_found = False
break
if is_found:
rval = d
break
return rval
if __name__ == '__main__':
# Tests
dicts = []
keys = 'spam eggs shrubbery knight'.split()
start = 0
for _ in range(4):
dct = {k: v for k, v in zip(keys, range(start, start+4))}
dicts.append(dct)
start += 4
# Find each dict based on 'spam' key only.
for x in range(len(dicts)):
spam = x*4
assert find_dict_in_list(dicts, spam=spam) == dicts[x]
# Find each dict based on 'spam' and 'shrubbery' keys.
for x in range(len(dicts)):
spam = x*4
assert find_dict_in_list(dicts, spam=spam, shrubbery=spam+2) == dicts[x]
# Search for one correct key, one incorrect key:
for x in range(len(dicts)):
spam = x*4
assert find_dict_in_list(dicts, spam=spam, shrubbery=spam+1) is None
# Search for non-existent dict.
for x in range(len(dicts)):
spam = x+100
assert find_dict_in_list(dicts, spam=spam) is None
Related
I need to parse the flatten structure and create nested structure using the list of keys provided. I have solved the problem but I am looking for an improvement and I would like to learn what I can change in my code. Can somebody review it and refactor using better knowledge?
src_data = [
{
"key1": "XX",
"key2": "X111",
"key3": "1aa",
"key4": 1
},
{
"key1": "YY",
"key2": "Y111",
"key3": "1bb",
"key4": 11
},
{
"key1": "ZZ",
"key2": "Z111",
"key3": "1cc",
"key4": 2.4
},
{
"key1": "AA",
"key2": "A111",
"key3": "1cc",
"key4": 33333.2122
},
{
"key1": "BB",
"key2": "B111",
"key3": "1bb",
"key4": 2
},
]
this is my code I developed so far creating the final result.
def plant_tree(ll):
master_tree = {}
for i in ll:
tree = master_tree
for n in i:
if n not in tree:
tree[n] = {}
tree = tree[n]
return master_tree
def make_nested_object(tt, var):
elo = lambda l: reduce(lambda x, y: {y: x}, l[::-1], var)
return {'n_path': tt, 'n_structure': elo(tt)}
def getFromDict(dataDict, mapList):
return reduce(operator.getitem, mapList, dataDict)
def set_nested_item(dataDict, mapList, val):
"""Set item in nested dictionary"""
reduce(getitem, mapList[:-1], dataDict)[mapList[-1]] = val
return dataDict
def update_tree(data_tree):
# MAKE NESTED OBJECT
out = (make_nested_object(k, v) for k,v, in res_out.items())
for dd in out:
leaf_data = dd['n_structure']
leaf_path = dd['n_path']
data_tree = set_nested_item(data_tree, leaf_path, getFromDict(leaf_data, leaf_path))
return data_tree
this is the customed itemgeter function from this question
def customed_itemgetter(*args):
# this handles the case when one key is provided
f = itemgetter(*args)
if len(args) > 2:
return f
return lambda obj: (f(obj),)
define the nesting level
nesting_keys = ['key1', 'key3', 'key2']
grouper = customed_itemgetter(*nesting_keys)
ii = groupby(sorted(src_data, key=grouper), grouper)
res_out = {key: [{k:v for k,v in i.items() if k not in nesting_keys} for i in group] for key,group in ii}
#
ll = ([dd[x] for x in nesting_keys] for dd in src_data)
data_tree = plant_tree(ll)
get results
result = update_tree(data_tree)
How can I improve my code?
If the itemgetter [Python-doc] is given a single element, it returns that single element, and does not wrap it in a singleton-tuple.
We can however construct a function for that, like:
from operator import itemgetter
def itemgetter2(*args):
f = itemgetter(*args)
if len(args) > 2:
return f
return lambda obj: (f(obj),)
then we can thus use the new itemgetter2, like:
grouper = itemgetter2(*ll)
ii = groupby(sorted(src_data, key=grouper), grouper)
EDIT: Based on your question however, you want to perform multilevel grouping, we can make a function for that, like:
def multigroup(groups, iterable, index=0):
if len(groups) <= index:
return list(iterable)
else:
f = itemgetter(groups[index])
i1 = index + 1
return {
k: multigroup(groups, vs, index=i1)
for k, vs in groupby(sorted(iterable, key=f), f)
}
For the data_src in the question, this then generates:
>>> multigroup(['a', 'b'], src_data)
{1: {2: [{'a': 1, 'b': 2, 'z': 3}]}, 2: {3: [{'a': 2, 'b': 3, 'e': 2}]}, 4: {3: [{'a': 4, 'x': 3, 'b': 3}]}}
You can post-process the values in the list(..) call however. We can for example generate dictionaries without the elements in the grouping columns:
def multigroup(groups, iterable):
group_set = set(groups)
fs = [itemgetter(group) for group in groups]
def mg(iterable, index=0):
if len(groups) <= index:
return [
{k: v for k, v in item.items() if k not in group_set}
for item in iterable
]
else:
i1 = index + 1
return {
k: mg(vs, index=i1)
for k, vs in groupby(sorted(iterable, key=fs[index]), fs[index])
}
return mg(iterable)
For the given sample input, we get:
>>> multigroup(['a', 'b'], src_data)
{1: {2: [{'z': 3}]}, 2: {3: [{'e': 2}]}, 4: {3: [{'x': 3}]}}
or for the new sample data:
>>> pprint(multigroup(['key1', 'key3', 'key2'], src_data))
{'AA': {'1cc': {'A111': [{'key4': 33333.2122}]}},
'BB': {'1bb': {'B111': [{'key4': 2}]}},
'XX': {'1aa': {'X111': [{'key4': 1}]}},
'YY': {'1bb': {'Y111': [{'key4': 11}]}},
'ZZ': {'1cc': {'Z111': [{'key4': 2.4}]}}}
What is the best package to easy filter ? like django-filter but i outside the django object.
for example
>>dict = {
{'name':'apple','amount':10},
{'name':'apple','amount':5},
{'name':'banana','amount':10},
{'name':'green-apple','amount':20},
}
>>filter(dict,name='apple')
{'name':'apple','amount':10},{'name':'apple','amount':10}
>>filter(dict,name='apple',amount=10)
{'name':'apple','amount':10}
>>filter(dict,amount_gt=5)
{'name':'apple','amount':10},{'name':'banana','amount':10}
>>filter(dict,amount_gte=5)
{'name':'apple','amount':10},{'name':'apple','amount':5},{'name':'banana','amount':10}
>>filter(dict,amount_lt=10)
{'name':'apple','amount':5}
>>filter(dict,name_regex=".*apple")
{'name':'apple','amount':5},{'name':'apple','amount':10},{'name':'green-apple','amount':20}
You can use a function like this:
def filter_dicts(dict_list, **kwargs):
for item in dict_list:
if all(item[k] == v for k, v in kwargs.items()):
yield item
usage:
dict_list = [
{"name": "apple", "amount": 10},
{"name": "apple", "amount": 15},
{"name": "orange", "amount": 5}
]
for d in filter_dicts(dict_list, name="apple", amount=10):
print(d)
Or, alternatively you can do this:
filtered_dicts = [item for item in dict_list if item["amount"] > 10 and item["name"] == "apple"]
you can user items to convert dict to a list
and use filter on that list, namely
filter(lambda x : x[0] == 'amount', dict.items())
an you can convert it back too
Is there any way in Python 2.6 to supply a custom key or cmp function to JSON's sort_keys?
I've got a list of dicts coming from JSON like so:
[
{
"key": "numberpuzzles1",
"url": "number-puzzle-i.html",
"title": "Number Puzzle I",
"category": "nestedloops",
"points": "60",
"n": "087"
},
{
"key": "gettingindividualdigits",
"url": "getting-individual-digits.html",
"title": "Getting Individual Digits",
"category": "nestedloops",
"points": "80",
"n": "088"
}
]
...which I've stored into the list variable assigndb. I'd like to be able to load in the JSON, modify it, and serialized it back out with dumps (or whatever), keeping the orders of the keys intact.
So far, I've tried something like this:
ordering = {'key': 0, 'url': 1, 'title': 2, 'category': 3,
'flags': 4, 'points': 5, 'n': 6}
def key_func(k):
return ordering[k]
# renumber assignments sequentially
for (i, a) in enumerate(assigndb):
a["n"] = "%03d" % (i+1)
s = json.dumps(assigndb, indent=2, sort_keys=True, key=key_func)
...but of course dumps doesn't support a custom key like list.sort() does. Something with a custom JSONEncoder maybe? I can't seem to get it going.
An idea (tested with 2.7):
import json
import collections
json.encoder.c_make_encoder = None
d = collections.OrderedDict([("b", 2), ("a", 1)])
json.dumps(d)
# '{"b": 2, "a": 1}'
See: OrderedDict + issue6105. The c_make_encoder hack seems only to be needed for Python 2.x. Not a direct solution because you have to change dicts for OrderedDicts, but it may be still usable. I checked the json library (encode.py) and the ordered is hardcoded:
if _sort_keys:
items = sorted(dct.items(), key=lambda kv: kv[0])
This is kind of ugly, but in case tokland's solution does not work for you:
data = [{'category': 'nestedloops', 'title': 'Number Puzzle I', 'url': 'number-puzzle-i.html', 'n': '087', 'points': '60', 'key': 'numberpuzzles1'}, {'category': 'nestedloops', 'title': 'Getting Individual Digits', 'url': 'getting-individual-digits.html', 'n': '088', 'points': '80', 'key': 'gettingindividualdigits'}]
ordering = {'key': 0, 'url': 1, 'title': 2, 'category': 3,
'flags': 4, 'points': 5, 'n': 6}
outlist = []
for d in data:
outlist.append([])
for k in sorted(d.keys(), key=lambda k: ordering[k]):
outlist[-1].append(json.dumps({k: d[k]}))
for i, l in enumerate(outlist):
outlist[i] = "{" + ",".join((s[1:-1] for s in outlist[i])) + "}"
s = "[" + ",".join(outlist) + "]"
Compact yet powerful recursive implementation with "prepended" and "appended" keys: https://gist.github.com/jeromerg/91f73d5867c5fa04ee7dbc0c5a03d611
def sort_recursive(node, first_keys, last_keys):
""" Sort the dictionary entries in a whole JSON object tree"""
fixed_placements = {
**{key: (0, idx) for idx, key in enumerate(first_keys)},
**{key: (2, idx) for idx, key in enumerate(last_keys)},
}
return _sort_recursive(node, lambda key: fixed_placements.get(key, (1, key)))
def _sort_recursive(node, key_fn):
if isinstance(node, list):
return [_sort_recursive(val, key_fn) for val in node]
elif isinstance(node, dict):
sorted_keys = sorted(node.keys(), key=key_fn)
return {k:_sort_recursive(node[k], key_fn) for k in sorted_keys}
else:
return node
I had the same problem and collections.OrderedDict was just not fit for the task because it ordered everything alphabetically. So I wrote something similar to Andrew Clark's solution:
def json_dumps_sorted(data, **kwargs):
sorted_keys = kwargs.get('sorted_keys', tuple())
if not sorted_keys:
return json.dumps(data)
else:
out_list = []
for element in data:
element_list = []
for key in sorted_keys:
if key in element:
element_list.append(json.dumps({key: element[key]}))
out_list.append('{{{}}}'.format(','.join((s[1:-1] for s in element_list))))
return '[{}]'.format(','.join(out_list))
You use it like this:
json_string = json_dumps_sorted([
{
"key": "numberpuzzles1",
"url": "number-puzzle-i.html",
"title": "Number Puzzle I",
"category": "nestedloops",
"points": "60",
"n": "087"
}, {
"key": "gettingindividualdigits",
"url": "getting-individual-digits.html",
"title": "Getting Individual Digits",
"category": "nestedloops",
"points": "80",
"n": "088"
}
], sorted_keys=(
'key',
'url',
'title',
'category',
'flags',
'points',
'n'
))
Thanks. I needed to put a timestamp key:value at the top of my JSON object no matter what. Obviously sorting the keys screwed this up as it starts with "t".
Using something like this, while putting the timestamp key in the dict_data right away worked:
d = collections.OrderedDict(dict_data)
I have the following object in python:
{
name: John,
age: {
years:18
},
computer_skills: {
years:4
},
mile_runner: {
years:2
}
}
I have an array with 100 people with the same structure.
What is the best way to go through all 100 people and make it such that there is no more "years"? In other words, each object in the 100 would look something like:
{
name: John,
age:18,
computer_skills:4,
mile_runner:2
}
I know I can do something in pseudocode:
for(item in list):
if('years' in (specific key)):
specifickey = item[(specific key)][(years)]
But is there a smarter/more efficent way?
Your pseudo-code is already pretty good I think:
for person in persons:
for k, v in person.items():
if isinstance(v, dict) and 'years' in v:
person[k] = v['years']
This overwrites every property which is a dictionary that has a years property with that property’s value.
Unlike other solutions (like dict comprehensions), this will modify the object in-place, so no new memory to keep everything is required.
def flatten(d):
ret = {}
for key, value in d.iteritems():
if isinstance(value, dict) and len(value) == 1 and "years" in value:
ret[key] = value["years"]
else:
ret[key] = value
return ret
d = {
"name": "John",
"age": {
"years":18
},
"computer_skills": {
"years":4
},
"mile_runner": {
"years":2
}
}
print flatten(d)
Result:
{'age': 18, 'mile_runner': 2, 'name': 'John', 'computer_skills': 4}
Dictionary comprehension:
import json
with open("input.json") as f:
cont = json.load(f)
print {el:cont[el]["years"] if "years" in cont[el] else cont[el] for el in cont}
prints
{u'age': 18, u'mile_runner': 2, u'name': u'John', u'computer_skills': 4}
where input.json contains
{
"name": "John",
"age": {
"years":18
},
"computer_skills": {
"years":4
},
"mile_runner": {
"years":2
}
}
Linear with regards to number of elements, you can't really hope for any lower.
As people said in the comments, it isn't exactly clear what your "object" is, but assuming that you actually have a list of dicts like this:
list = [{
'name': 'John',
'age': {
'years': 18
},
'computer_skills': {
'years':4
},
'mile_runner': {
'years':2
}
}]
Then you can do something like this:
for item in list:
for key in item:
try:
item[key] = item[key]['years']
except (TypeError, KeyError):
pass
Result:
list = [{'age': 18, 'mile_runner': 2, 'name': 'John', 'computer_skills': 4}]
Is there any way in Python 2.6 to supply a custom key or cmp function to JSON's sort_keys?
I've got a list of dicts coming from JSON like so:
[
{
"key": "numberpuzzles1",
"url": "number-puzzle-i.html",
"title": "Number Puzzle I",
"category": "nestedloops",
"points": "60",
"n": "087"
},
{
"key": "gettingindividualdigits",
"url": "getting-individual-digits.html",
"title": "Getting Individual Digits",
"category": "nestedloops",
"points": "80",
"n": "088"
}
]
...which I've stored into the list variable assigndb. I'd like to be able to load in the JSON, modify it, and serialized it back out with dumps (or whatever), keeping the orders of the keys intact.
So far, I've tried something like this:
ordering = {'key': 0, 'url': 1, 'title': 2, 'category': 3,
'flags': 4, 'points': 5, 'n': 6}
def key_func(k):
return ordering[k]
# renumber assignments sequentially
for (i, a) in enumerate(assigndb):
a["n"] = "%03d" % (i+1)
s = json.dumps(assigndb, indent=2, sort_keys=True, key=key_func)
...but of course dumps doesn't support a custom key like list.sort() does. Something with a custom JSONEncoder maybe? I can't seem to get it going.
An idea (tested with 2.7):
import json
import collections
json.encoder.c_make_encoder = None
d = collections.OrderedDict([("b", 2), ("a", 1)])
json.dumps(d)
# '{"b": 2, "a": 1}'
See: OrderedDict + issue6105. The c_make_encoder hack seems only to be needed for Python 2.x. Not a direct solution because you have to change dicts for OrderedDicts, but it may be still usable. I checked the json library (encode.py) and the ordered is hardcoded:
if _sort_keys:
items = sorted(dct.items(), key=lambda kv: kv[0])
This is kind of ugly, but in case tokland's solution does not work for you:
data = [{'category': 'nestedloops', 'title': 'Number Puzzle I', 'url': 'number-puzzle-i.html', 'n': '087', 'points': '60', 'key': 'numberpuzzles1'}, {'category': 'nestedloops', 'title': 'Getting Individual Digits', 'url': 'getting-individual-digits.html', 'n': '088', 'points': '80', 'key': 'gettingindividualdigits'}]
ordering = {'key': 0, 'url': 1, 'title': 2, 'category': 3,
'flags': 4, 'points': 5, 'n': 6}
outlist = []
for d in data:
outlist.append([])
for k in sorted(d.keys(), key=lambda k: ordering[k]):
outlist[-1].append(json.dumps({k: d[k]}))
for i, l in enumerate(outlist):
outlist[i] = "{" + ",".join((s[1:-1] for s in outlist[i])) + "}"
s = "[" + ",".join(outlist) + "]"
Compact yet powerful recursive implementation with "prepended" and "appended" keys: https://gist.github.com/jeromerg/91f73d5867c5fa04ee7dbc0c5a03d611
def sort_recursive(node, first_keys, last_keys):
""" Sort the dictionary entries in a whole JSON object tree"""
fixed_placements = {
**{key: (0, idx) for idx, key in enumerate(first_keys)},
**{key: (2, idx) for idx, key in enumerate(last_keys)},
}
return _sort_recursive(node, lambda key: fixed_placements.get(key, (1, key)))
def _sort_recursive(node, key_fn):
if isinstance(node, list):
return [_sort_recursive(val, key_fn) for val in node]
elif isinstance(node, dict):
sorted_keys = sorted(node.keys(), key=key_fn)
return {k:_sort_recursive(node[k], key_fn) for k in sorted_keys}
else:
return node
I had the same problem and collections.OrderedDict was just not fit for the task because it ordered everything alphabetically. So I wrote something similar to Andrew Clark's solution:
def json_dumps_sorted(data, **kwargs):
sorted_keys = kwargs.get('sorted_keys', tuple())
if not sorted_keys:
return json.dumps(data)
else:
out_list = []
for element in data:
element_list = []
for key in sorted_keys:
if key in element:
element_list.append(json.dumps({key: element[key]}))
out_list.append('{{{}}}'.format(','.join((s[1:-1] for s in element_list))))
return '[{}]'.format(','.join(out_list))
You use it like this:
json_string = json_dumps_sorted([
{
"key": "numberpuzzles1",
"url": "number-puzzle-i.html",
"title": "Number Puzzle I",
"category": "nestedloops",
"points": "60",
"n": "087"
}, {
"key": "gettingindividualdigits",
"url": "getting-individual-digits.html",
"title": "Getting Individual Digits",
"category": "nestedloops",
"points": "80",
"n": "088"
}
], sorted_keys=(
'key',
'url',
'title',
'category',
'flags',
'points',
'n'
))
Thanks. I needed to put a timestamp key:value at the top of my JSON object no matter what. Obviously sorting the keys screwed this up as it starts with "t".
Using something like this, while putting the timestamp key in the dict_data right away worked:
d = collections.OrderedDict(dict_data)