Related
What is the recommended way of serializing a namedtuple to json with the field names retained?
Serializing a namedtuple to json results in only the values being serialized and the field names being lost in translation. I would like the fields also to be retained when json-ized and hence did the following:
class foobar(namedtuple('f', 'foo, bar')):
__slots__ = ()
def __iter__(self):
yield self._asdict()
The above serializes to json as I expect and behaves as namedtuple in other places I use (attribute access etc.,) except with a non-tuple like results while iterating it (which fine for my use case).
What is the "correct way" of converting to json with the field names retained?
If it's just one namedtuple you're looking to serialize, using its _asdict() method will work (with Python >= 2.7)
>>> from collections import namedtuple
>>> import json
>>> FB = namedtuple("FB", ("foo", "bar"))
>>> fb = FB(123, 456)
>>> json.dumps(fb._asdict())
'{"foo": 123, "bar": 456}'
This is pretty tricky, since namedtuple() is a factory which returns a new type derived from tuple. One approach would be to have your class also inherit from UserDict.DictMixin, but tuple.__getitem__ is already defined and expects an integer denoting the position of the element, not the name of its attribute:
>>> f = foobar('a', 1)
>>> f[0]
'a'
At its heart the namedtuple is an odd fit for JSON, since it is really a custom-built type whose key names are fixed as part of the type definition, unlike a dictionary where key names are stored inside the instance. This prevents you from "round-tripping" a namedtuple, e.g. you cannot decode a dictionary back into a namedtuple without some other a piece of information, like an app-specific type marker in the dict {'a': 1, '#_type': 'foobar'}, which is a bit hacky.
This is not ideal, but if you only need to encode namedtuples into dictionaries, another approach is to extend or modify your JSON encoder to special-case these types. Here is an example of subclassing the Python json.JSONEncoder. This tackles the problem of ensuring that nested namedtuples are properly converted to dictionaries:
from collections import namedtuple
from json import JSONEncoder
class MyEncoder(JSONEncoder):
def _iterencode(self, obj, markers=None):
if isinstance(obj, tuple) and hasattr(obj, '_asdict'):
gen = self._iterencode_dict(obj._asdict(), markers)
else:
gen = JSONEncoder._iterencode(self, obj, markers)
for chunk in gen:
yield chunk
class foobar(namedtuple('f', 'foo, bar')):
pass
enc = MyEncoder()
for obj in (foobar('a', 1), ('a', 1), {'outer': foobar('x', 'y')}):
print enc.encode(obj)
{"foo": "a", "bar": 1}
["a", 1]
{"outer": {"foo": "x", "bar": "y"}}
It looks like you used to be able to subclass simplejson.JSONEncoder to make this work, but with the latest simplejson code, that is no longer the case: you have to actually modify the project code. I see no reason why simplejson should not support namedtuples, so I forked the project, added namedtuple support, and I'm currently waiting for my branch to be pulled back into the main project. If you need the fixes now, just pull from my fork.
EDIT: Looks like the latest versions of simplejson now natively support this with the namedtuple_as_object option, which defaults to True.
I wrote a library for doing this: https://github.com/ltworf/typedload
It can go from and to named-tuple and back.
It supports quite complicated nested structures, with lists, sets, enums, unions, default values. It should cover most common cases.
edit: The library also supports dataclass and attr classes.
It's impossible to serialize namedtuples correctly with the native python json library. It will always see tuples as lists, and it is impossible to override the default serializer to change this behaviour. It's worse if objects are nested.
Better to use a more robust library like orjson:
import orjson
from typing import NamedTuple
class Rectangle(NamedTuple):
width: int
height: int
def default(obj):
if hasattr(obj, '_asdict'):
return obj._asdict()
rectangle = Rectangle(width=10, height=20)
print(orjson.dumps(rectangle, default=default))
=>
{
"width":10,
"height":20
}
There is a more convenient solution is to use the decorator (it uses the protected field _fields).
Python 2.7+:
import json
from collections import namedtuple, OrderedDict
def json_serializable(cls):
def as_dict(self):
yield OrderedDict(
(name, value) for name, value in zip(
self._fields,
iter(super(cls, self).__iter__())))
cls.__iter__ = as_dict
return cls
#Usage:
C = json_serializable(namedtuple('C', 'a b c'))
print json.dumps(C('abc', True, 3.14))
# or
#json_serializable
class D(namedtuple('D', 'a b c')):
pass
print json.dumps(D('abc', True, 3.14))
Python 3.6.6+:
import json
from typing import TupleName
def json_serializable(cls):
def as_dict(self):
yield {name: value for name, value in zip(
self._fields,
iter(super(cls, self).__iter__()))}
cls.__iter__ = as_dict
return cls
# Usage:
#json_serializable
class C(NamedTuple):
a: str
b: bool
c: float
print(json.dumps(C('abc', True, 3.14))
It recursively converts the namedTuple data to json.
print(m1)
## Message(id=2, agent=Agent(id=1, first_name='asd', last_name='asd', mail='2#mai.com'), customer=Customer(id=1, first_name='asd', last_name='asd', mail='2#mai.com', phone_number=123123), type='image', content='text', media_url='h.com', la=123123, ls=4512313)
def reqursive_to_json(obj):
_json = {}
if isinstance(obj, tuple):
datas = obj._asdict()
for data in datas:
if isinstance(datas[data], tuple):
_json[data] = (reqursive_to_json(datas[data]))
else:
print(datas[data])
_json[data] = (datas[data])
return _json
data = reqursive_to_json(m1)
print(data)
{'agent': {'first_name': 'asd',
'last_name': 'asd',
'mail': '2#mai.com',
'id': 1},
'content': 'text',
'customer': {'first_name': 'asd',
'last_name': 'asd',
'mail': '2#mai.com',
'phone_number': 123123,
'id': 1},
'id': 2,
'la': 123123,
'ls': 4512313,
'media_url': 'h.com',
'type': 'image'}
The jsonplus library provides a serializer for NamedTuple instances. Use its compatibility mode to output simple objects if needed, but prefer the default as it is helpful for decoding back.
This is an old question. However:
A suggestion for all those with the same question, think carefully about using any of the private or internal features of the NamedTuple because they have before and will change again over time.
For example, if your NamedTuple is a flat value object and you're only interested in serializing it and not in cases where it is nested into another object, you could avoid the troubles that would come up with __dict__ being removed or _as_dict() changing and just do something like (and yes this is Python 3 because this answer is for the present):
from typing import NamedTuple
class ApiListRequest(NamedTuple):
group: str="default"
filter: str="*"
def to_dict(self):
return {
'group': self.group,
'filter': self.filter,
}
def to_json(self):
return json.dumps(self.to_dict())
I tried to use the default callable kwarg to dumps in order to do the to_dict() call if available, but that didn't get called as the NamedTuple is convertible to a list.
Here is my take on the problem. It serializes the NamedTuple, takes care of folded NamedTuples and Lists inside of them
def recursive_to_dict(obj: Any) -> dict:
_dict = {}
if isinstance(obj, tuple):
node = obj._asdict()
for item in node:
if isinstance(node[item], list): # Process as a list
_dict[item] = [recursive_to_dict(x) for x in (node[item])]
elif getattr(node[item], "_asdict", False): # Process as a NamedTuple
_dict[item] = recursive_to_dict(node[item])
else: # Process as a regular element
_dict[item] = (node[item])
return _dict
simplejson.dump() instead of json.dump does the job. It may be slower though.
I have a Python dict-of-dict structure with a large number of outer-dict keys (millions to billions). The inner dicts are mostly empty, but can store key-value pairs. Currently I create a separate dict as each of the inner dicts. But it uses a lot of memory that I don't end up using. Each empty dict is small, but I have a lot of them. I'd like to delay creating the inner dict until needed.
Ideally, I'd like to even delay creating the inner dict until a key-value pair is set in the inner dict. I envision using a single DelayDict object for ALL outer-dict values. This object would act like an empty dict for get and getitem calls, but as soon as a setitem or update call comes in it would create an empty dict to take its place. I run into trouble having the delaydict object know how to connect the new empty dict with the dict-of-dict structure.
class DelayDict(object): % can do much more - only showing get/set
def __init__(self, dod):
self.dictofdict = dod % the outer dict
def __getitem__(self, key):
raise KeyError(key)
def __setitem__(self, key, value):
replacement = {key: value}
% replace myself in the outer dict!!
self.dict-of-dict[?????] = replacement
I can't think of how to store the new replacement dict in the dict-of-dict structure so that it replaces the DelayDict class as the inner dict. I know properties can do similar things, but I believe the same fundamental trouble arises when I try to replace myself inside the outer dict.
Old question, but I came across a similar problem. I'm not sure that it's a
good idea to try to spare some memory, but if you really need to do that, you should try to build your own data structure.
If you are stuck with the dict of dict, here's a solution.
First, you need a way to create keys in the OuterDict without value (value is {}) by default). if OuterDict is a wrapper around a dict __d:
def create(self, key):
self.__d[key] = None
How much memory will you spare?
>>> import sys
>>> a = {}
>>> sys.getsizeof(a)
136
As you pointed out, None is created only once, but you have to keep a reference on it. In Cpython (64 bits), it's 8 bytes. For 1 billion elements, you spare (136-8)* 10**9 bytes = 128 Gb (and not Mb, thanks!). You need to give a
placeholder when someone ask for the value. The placeholder keeps track of the outer dict and the key in the outer dict. It wraps a dict and assigns this dict to outer[key] when you assign a value.
No more talking, code:
class OuterDict():
def __init__(self):
self.__d = {}
def __getitem__(self, key):
v = self.__d[key]
if v is None: # an orphan key
v = PlaceHolder(self.__d, key)
return v
def create(self, key):
self.__d[key] = None
class PlaceHolder():
def __init__(self, parent, key):
self.__parent = parent
self.__key = key
self.__d = {}
def __getitem__(self, key):
return self.__d[key]
def __setitem__(self, key, value):
if not self.__d:
self.__parent[self.__key] = self.__d # copy me in the outer dict
self.__d[key] = value
def __repr__(self):
return repr("PlaceHolder for "+str(self.__d))
# __len__, ...
A test:
o = OuterDict()
o.create("a") # a is empty
print (o["a"])
try:
o["a"]["b"] # Key Error
except KeyError as e:
print ("KeyError", e)
o["a"]["b"] = 2
print (o["a"])
# output:
# 'PlaceHolder for {}'
# KeyError 'b'
# {'b': 2}
Why it doesn't use much memory? Because you are not building billions of placeholders. You release them when you don't need them anymore. Maybe you will need just one at a time.
Possible improvements: you can create a pool of PlaceHolders. A stack may be a good data structure: recently created placeholder are likely to be released soon. When you need a new PlaceHolder, you
look into the stack, and if a placeholder has only one ref (sys.getrefcount(ph) == 1), you can use it. To fasten the process, when you are looking for
a free placeholder, you can remember the placeholder with the maximum refcount. You switch the free placeholder with this "max refcount" placeholder. Hence, the placeholders with the maximum
refcount are sent to the bottom of the stack.
I'm imitating the behavior of the ConfigParser module to write a highly specialized parser that exploits some well-defined structure in the configuration files for a particular application I work with. Several sections of the config file contain hundreds of variable and routine mappings prefixed with either Variable_ or Routine_, like this:
[Map.PRD]
Variable_FOO=LOC1
Variable_BAR=LOC2
Routine_FOO=LOC3
Routine_BAR=LOC4
...
[Map.SHD]
Variable_FOO=LOC1
Variable_BAR=LOC2
Routine_FOO=LOC3
Routine_BAR=LOC4
...
I'd like to maintain the basic structure of ConfigParser where each section is stored as a single dictionary, so users would still have access to the classic syntax:
config.content['Mappings']['Variable_FOO'] = 'LOC1'
but also be able to use a simplified API that drills down to this section:
config.vmapping('PRD')['FOO'] = 'LOC1'
config.vmapping('PRD')['BAR'] = 'LOC2'
config.rmapping('PRD')['FOO'] = 'LOC3'
config.rmapping('PRD')['BAR'] = 'LOC4'
Currently I'm implementing this by storing the section in a special subclass of dict to which I've added a prefix attribute. The variable and routine properties of the parser set the prefix attribute of the dict-like object to 'Variable_' or 'Routine_' and then modified __getitem__ and __setitem__ attributes of the dict handle gluing the prefix together with the key to access the appropriate item. It's working, but involves a lot of boilerplate to implement all the associated niceties like supporting iteration.
I suppose my ideal solution would be do dispense with the subclassed dict and have have the variable and routine properties somehow present a "view" of the plain dict object underneath without the prefixes.
Update
Here's the solution I implemented, largely based on #abarnet's answer:
class MappingDict(object):
def __init__(self, prefix, d):
self.prefix, self.d = prefix, d
def prefixify(self, name):
return '{}_{}'.format(self.prefix, name)
def __getitem__(self, name):
name = self.prefixify(name)
return self.d.__getitem__(name)
def __setitem__(self, name, value):
name = self.prefixify(name)
return self.d.__setitem__(name, value)
def __delitem__(self, name):
name = self.prefixify(name)
return self.d.__delitem__(name)
def __iter__(self):
return (key.partition('_')[-1] for key in self.d
if key.startswith(self.prefix))
def __repr__(self):
return 'MappingDict({})'.format(dict.__repr__(self))
class MyParser(object):
SECTCRE = re.compile(r'\[(?P<header>[^]]+)\]')
def __init__(self, filename):
self.filename = filename
self.content = {}
lines = [x.strip() for x in open(filename).read().splitlines()
if x.strip()]
for line in lines:
match = re.match(self.SECTCRE, line)
if match:
section = match.group('header')
self.content[section] = {}
else:
key, sep, value = line.partition('=')
self.content[section][key] = value
def write(self, filename):
fp = open(filename, 'w')
for section in sorted(self.content, key=sectionsort):
fp.write("[%s]\n" % section)
for key in sorted(self.content[section], key=cpfsort):
value = str(self.content[section][key])
fp.write("%s\n" % '='.join([key,value]))
fp.write("\n")
fp.close()
def vmapping(self, nsp):
section = 'Map.{}'.format(nsp)
return MappingDict('Variable', self.content[section])
def rmapping(self, nsp):
section = 'Map.{}'.format(nsp)
return MappingDict('Routine', self.content[section])
It's used like this:
config = MyParser('myfile.cfg')
vmap = config.vmapping('PRD')
vmap['FOO'] = 'LOC5'
vmap['BAR'] = 'LOC6'
config.write('newfile.cfg')
The resulting newfile.cfg reflects the LOC5 and LOC6 changes.
I don't think you want inheritance here. You end up with two separate dict objects which you have to create on load and then paste back together on save…
If that's acceptable, you don't even need to bother with the prefixing during normal operations; just do the prefixing while saving, like this:
class Config(object):
def save(self):
merged = {'variable_{}'.format(key): value for key, value
in self.variable_dict.items()}
merged.update({'routine_{}'.format(key): value for key, value
in self.routine_dict.items()}
# now save merged
If you want that merged object to be visible at all times, but don't expect to be called on that very often, make it a #property.
If you want to access the merged dictionary regularly, at the same time you're accessing the two sub-dictionaries, then yes, you want a view:
I suppose my ideal solution would be do dispense with the subclassed dict and have have the global and routine properties somehow present a "view" of the plain dict object underneath without the prefixes.
This is going to be very hard to do with inheritance. Certainly not with inheritance from dict; inheritance from builtins.dict_items might work if you're using Python 3, but it still seems like a stretch.
But with delegation, it's easy. Each sub-dictionary just holds a reference to the parent dict:
class PrefixedDict(object):
def __init__(self, prefix, d):
self.prefix, self.d = prefix, d
def prefixify(self, key):
return '{}_{}'.format(self.prefix, key)
def __getitem__(self, key):
return self.d.__getitem__(self.prefixify(key))
def __setitem__(self, key, value):
return self.d.__setitem__(self.prefixify(key), value)
def __delitem__(self, key):
return self.d.__delitem__(self.prefixify(key))
def __iter__(self):
return (key[len(self.prefix):] for key in self.d
if key.startswith(self.prefix)])
You don't get any of the dict methods for free that way—but that's a good thing, because they were mostly incorrect anyway, right? Explicitly delegate the ones you want. (If you do have some you want to pass through as-is, use __getattr__ for that.)
Besides being conceptually simpler and harder to screw up through accidentally forgetting to override something, this also means that PrefixDict can work with any type of mapping, not just a dict.
So, no matter which way you go, where and how do these objects get created?
The easy answer is that they're attributes that you create when you construct a Config:
def __init__(self):
self.d = {}
self.variable = PrefixedDict('Variable', self.d)
self.routine = PrefixedDict('Routine', self.d)
If this needs to be dynamic (e.g., there can be an arbitrary set of prefixes), create them at load time:
def load(self):
# load up self.d
prefixes = set(key.split('_')[0] for key in self.d)
for prefix in prefixes:
setattr(self, prefix, PrefixedDict(prefix, self.d)
If you want to be able to create them on the fly (so config.newprefix['foo'] = 3 adds 'Newprefix_foo'), you can do this instead:
def __getattr__(self, name):
return PrefixedDict(name.title(), self.d)
But once you're using dynamic attributes, you really have to question whether it isn't cleaner to use dictionary (item) syntax instead, like config['newprefix']['foo']. For one thing, that would actually let you call one of the sub-dictionaries 'global', as in your original question…
Or you can first build the dictionary syntax, use what's usually referred to as an attrdict (search ActiveState recipes and PyPI for 3000 implementations…), which lets you automatically make config.newprefix mean config['newprefix'], so you can use attribute syntax when you have valid identifiers, but fall back to dictionary syntax when you don't.
There are a couple of options for how to proceed.
The simplest might be to use nested dictionaries, so Variable_FOO becomes config["variable"]["FOO"]. You might want to use a defaultdict(dict) for the outer dictionary so you don't need to worry about initializing the inner ones when you add the first value to them.
Another option would be to use tuple keys in a single dictionary. That is, Variable_FOO would become config[("variable", "FOO")]. This is easy to do with code, since you can simply assign to config[tuple(some_string.split("_"))]. Though, I suppose you could also just use the unsplit string as your key in this case.
A final approach allows you to use the syntax you want (where Variable_FOO is accessed as config.Variable["FOO"]), by using __getattr__ and a defaultdict behind the scenes:
from collections import defaultdict
class Config(object):
def __init__(self):
self._attrdicts = defaultdict(dict)
def __getattr__(self, name):
return self._attrdicts[name]
You could extend this with behavior for __setattr__ and __delattr__ but it's probably not necessary. The only serious limitation to this approach (given the original version of the question), is that the attributes names (like Variable) must be legal Python identifiers. You can't use strings with leading numbers, Python keywords (like global) or strings containing whitespace characters.
A downside to this approach is that it's a bit more difficult to use programatically (by, for instance, your config-file parser). To read a value of Variable_FOO and save it to config.Variable["FOO"] you'll probably need to use the global getattr function, like this:
name, value = line.split("=")
prefix, suffix = name.split("_")
getattr(config, prefix)[suffix] = value
What is the recommended way of serializing a namedtuple to json with the field names retained?
Serializing a namedtuple to json results in only the values being serialized and the field names being lost in translation. I would like the fields also to be retained when json-ized and hence did the following:
class foobar(namedtuple('f', 'foo, bar')):
__slots__ = ()
def __iter__(self):
yield self._asdict()
The above serializes to json as I expect and behaves as namedtuple in other places I use (attribute access etc.,) except with a non-tuple like results while iterating it (which fine for my use case).
What is the "correct way" of converting to json with the field names retained?
If it's just one namedtuple you're looking to serialize, using its _asdict() method will work (with Python >= 2.7)
>>> from collections import namedtuple
>>> import json
>>> FB = namedtuple("FB", ("foo", "bar"))
>>> fb = FB(123, 456)
>>> json.dumps(fb._asdict())
'{"foo": 123, "bar": 456}'
This is pretty tricky, since namedtuple() is a factory which returns a new type derived from tuple. One approach would be to have your class also inherit from UserDict.DictMixin, but tuple.__getitem__ is already defined and expects an integer denoting the position of the element, not the name of its attribute:
>>> f = foobar('a', 1)
>>> f[0]
'a'
At its heart the namedtuple is an odd fit for JSON, since it is really a custom-built type whose key names are fixed as part of the type definition, unlike a dictionary where key names are stored inside the instance. This prevents you from "round-tripping" a namedtuple, e.g. you cannot decode a dictionary back into a namedtuple without some other a piece of information, like an app-specific type marker in the dict {'a': 1, '#_type': 'foobar'}, which is a bit hacky.
This is not ideal, but if you only need to encode namedtuples into dictionaries, another approach is to extend or modify your JSON encoder to special-case these types. Here is an example of subclassing the Python json.JSONEncoder. This tackles the problem of ensuring that nested namedtuples are properly converted to dictionaries:
from collections import namedtuple
from json import JSONEncoder
class MyEncoder(JSONEncoder):
def _iterencode(self, obj, markers=None):
if isinstance(obj, tuple) and hasattr(obj, '_asdict'):
gen = self._iterencode_dict(obj._asdict(), markers)
else:
gen = JSONEncoder._iterencode(self, obj, markers)
for chunk in gen:
yield chunk
class foobar(namedtuple('f', 'foo, bar')):
pass
enc = MyEncoder()
for obj in (foobar('a', 1), ('a', 1), {'outer': foobar('x', 'y')}):
print enc.encode(obj)
{"foo": "a", "bar": 1}
["a", 1]
{"outer": {"foo": "x", "bar": "y"}}
It looks like you used to be able to subclass simplejson.JSONEncoder to make this work, but with the latest simplejson code, that is no longer the case: you have to actually modify the project code. I see no reason why simplejson should not support namedtuples, so I forked the project, added namedtuple support, and I'm currently waiting for my branch to be pulled back into the main project. If you need the fixes now, just pull from my fork.
EDIT: Looks like the latest versions of simplejson now natively support this with the namedtuple_as_object option, which defaults to True.
I wrote a library for doing this: https://github.com/ltworf/typedload
It can go from and to named-tuple and back.
It supports quite complicated nested structures, with lists, sets, enums, unions, default values. It should cover most common cases.
edit: The library also supports dataclass and attr classes.
It's impossible to serialize namedtuples correctly with the native python json library. It will always see tuples as lists, and it is impossible to override the default serializer to change this behaviour. It's worse if objects are nested.
Better to use a more robust library like orjson:
import orjson
from typing import NamedTuple
class Rectangle(NamedTuple):
width: int
height: int
def default(obj):
if hasattr(obj, '_asdict'):
return obj._asdict()
rectangle = Rectangle(width=10, height=20)
print(orjson.dumps(rectangle, default=default))
=>
{
"width":10,
"height":20
}
There is a more convenient solution is to use the decorator (it uses the protected field _fields).
Python 2.7+:
import json
from collections import namedtuple, OrderedDict
def json_serializable(cls):
def as_dict(self):
yield OrderedDict(
(name, value) for name, value in zip(
self._fields,
iter(super(cls, self).__iter__())))
cls.__iter__ = as_dict
return cls
#Usage:
C = json_serializable(namedtuple('C', 'a b c'))
print json.dumps(C('abc', True, 3.14))
# or
#json_serializable
class D(namedtuple('D', 'a b c')):
pass
print json.dumps(D('abc', True, 3.14))
Python 3.6.6+:
import json
from typing import TupleName
def json_serializable(cls):
def as_dict(self):
yield {name: value for name, value in zip(
self._fields,
iter(super(cls, self).__iter__()))}
cls.__iter__ = as_dict
return cls
# Usage:
#json_serializable
class C(NamedTuple):
a: str
b: bool
c: float
print(json.dumps(C('abc', True, 3.14))
It recursively converts the namedTuple data to json.
print(m1)
## Message(id=2, agent=Agent(id=1, first_name='asd', last_name='asd', mail='2#mai.com'), customer=Customer(id=1, first_name='asd', last_name='asd', mail='2#mai.com', phone_number=123123), type='image', content='text', media_url='h.com', la=123123, ls=4512313)
def reqursive_to_json(obj):
_json = {}
if isinstance(obj, tuple):
datas = obj._asdict()
for data in datas:
if isinstance(datas[data], tuple):
_json[data] = (reqursive_to_json(datas[data]))
else:
print(datas[data])
_json[data] = (datas[data])
return _json
data = reqursive_to_json(m1)
print(data)
{'agent': {'first_name': 'asd',
'last_name': 'asd',
'mail': '2#mai.com',
'id': 1},
'content': 'text',
'customer': {'first_name': 'asd',
'last_name': 'asd',
'mail': '2#mai.com',
'phone_number': 123123,
'id': 1},
'id': 2,
'la': 123123,
'ls': 4512313,
'media_url': 'h.com',
'type': 'image'}
The jsonplus library provides a serializer for NamedTuple instances. Use its compatibility mode to output simple objects if needed, but prefer the default as it is helpful for decoding back.
This is an old question. However:
A suggestion for all those with the same question, think carefully about using any of the private or internal features of the NamedTuple because they have before and will change again over time.
For example, if your NamedTuple is a flat value object and you're only interested in serializing it and not in cases where it is nested into another object, you could avoid the troubles that would come up with __dict__ being removed or _as_dict() changing and just do something like (and yes this is Python 3 because this answer is for the present):
from typing import NamedTuple
class ApiListRequest(NamedTuple):
group: str="default"
filter: str="*"
def to_dict(self):
return {
'group': self.group,
'filter': self.filter,
}
def to_json(self):
return json.dumps(self.to_dict())
I tried to use the default callable kwarg to dumps in order to do the to_dict() call if available, but that didn't get called as the NamedTuple is convertible to a list.
Here is my take on the problem. It serializes the NamedTuple, takes care of folded NamedTuples and Lists inside of them
def recursive_to_dict(obj: Any) -> dict:
_dict = {}
if isinstance(obj, tuple):
node = obj._asdict()
for item in node:
if isinstance(node[item], list): # Process as a list
_dict[item] = [recursive_to_dict(x) for x in (node[item])]
elif getattr(node[item], "_asdict", False): # Process as a NamedTuple
_dict[item] = recursive_to_dict(node[item])
else: # Process as a regular element
_dict[item] = (node[item])
return _dict
simplejson.dump() instead of json.dump does the job. It may be slower though.
I'm new to Python, and am sort of surprised I cannot do this.
dictionary = {
'a' : '123',
'b' : dictionary['a'] + '456'
}
I'm wondering what the Pythonic way to correctly do this in my script, because I feel like I'm not the only one that has tried to do this.
EDIT: Enough people were wondering what I'm doing with this, so here are more details for my use cases. Lets say I want to keep dictionary objects to hold file system paths. The paths are relative to other values in the dictionary. For example, this is what one of my dictionaries may look like.
dictionary = {
'user': 'sholsapp',
'home': '/home/' + dictionary['user']
}
It is important that at any point in time I may change dictionary['user'] and have all of the dictionaries values reflect the change. Again, this is an example of what I'm using it for, so I hope that it conveys my goal.
From my own research I think I will need to implement a class to do this.
No fear of creating new classes -
You can take advantage of Python's string formating capabilities
and simply do:
class MyDict(dict):
def __getitem__(self, item):
return dict.__getitem__(self, item) % self
dictionary = MyDict({
'user' : 'gnucom',
'home' : '/home/%(user)s',
'bin' : '%(home)s/bin'
})
print dictionary["home"]
print dictionary["bin"]
Nearest I came up without doing object:
dictionary = {
'user' : 'gnucom',
'home' : lambda:'/home/'+dictionary['user']
}
print dictionary['home']()
dictionary['user']='tony'
print dictionary['home']()
>>> dictionary = {
... 'a':'123'
... }
>>> dictionary['b'] = dictionary['a'] + '456'
>>> dictionary
{'a': '123', 'b': '123456'}
It works fine but when you're trying to use dictionary it hasn't been defined yet (because it has to evaluate that literal dictionary first).
But be careful because this assigns to the key of 'b' the value referenced by the key of 'a' at the time of assignment and is not going to do the lookup every time. If that is what you are looking for, it's possible but with more work.
What you're describing in your edit is how an INI config file works. Python does have a built in library called ConfigParser which should work for what you're describing.
This is an interesting problem. It seems like Greg has a good solution. But that's no fun ;)
jsbueno as a very elegant solution but that only applies to strings (as you requested).
The trick to a 'general' self referential dictionary is to use a surrogate object. It takes a few (understatement) lines of code to pull off, but the usage is about what you want:
S = SurrogateDict(AdditionSurrogateDictEntry)
d = S.resolve({'user': 'gnucom',
'home': '/home/' + S['user'],
'config': [S['home'] + '/.emacs', S['home'] + '/.bashrc']})
The code to make that happen is not nearly so short. It lives in three classes:
import abc
class SurrogateDictEntry(object):
__metaclass__ = abc.ABCMeta
def __init__(self, key):
"""record the key on the real dictionary that this will resolve to a
value for
"""
self.key = key
def resolve(self, d):
""" return the actual value"""
if hasattr(self, 'op'):
# any operation done on self will store it's name in self.op.
# if this is set, resolve it by calling the appropriate method
# now that we can get self.value out of d
self.value = d[self.key]
return getattr(self, self.op + 'resolve__')()
else:
return d[self.key]
#staticmethod
def make_op(opname):
"""A convience class. This will be the form of all op hooks for subclasses
The actual logic for the op is in __op__resolve__ (e.g. __add__resolve__)
"""
def op(self, other):
self.stored_value = other
self.op = opname
return self
op.__name__ = opname
return op
Next, comes the concrete class. simple enough.
class AdditionSurrogateDictEntry(SurrogateDictEntry):
__add__ = SurrogateDictEntry.make_op('__add__')
__radd__ = SurrogateDictEntry.make_op('__radd__')
def __add__resolve__(self):
return self.value + self.stored_value
def __radd__resolve__(self):
return self.stored_value + self.value
Here's the final class
class SurrogateDict(object):
def __init__(self, EntryClass):
self.EntryClass = EntryClass
def __getitem__(self, key):
"""record the key and return"""
return self.EntryClass(key)
#staticmethod
def resolve(d):
"""I eat generators resolve self references"""
stack = [d]
while stack:
cur = stack.pop()
# This just tries to set it to an appropriate iterable
it = xrange(len(cur)) if not hasattr(cur, 'keys') else cur.keys()
for key in it:
# sorry for being a duche. Just register your class with
# SurrogateDictEntry and you can pass whatever.
while isinstance(cur[key], SurrogateDictEntry):
cur[key] = cur[key].resolve(d)
# I'm just going to check for iter but you can add other
# checks here for items that we should loop over.
if hasattr(cur[key], '__iter__'):
stack.append(cur[key])
return d
In response to gnucoms's question about why I named the classes the way that I did.
The word surrogate is generally associated with standing in for something else so it seemed appropriate because that's what the SurrogateDict class does: an instance replaces the 'self' references in a dictionary literal. That being said, (other than just being straight up stupid sometimes) naming is probably one of the hardest things for me about coding. If you (or anyone else) can suggest a better name, I'm all ears.
I'll provide a brief explanation. Throughout S refers to an instance of SurrogateDict and d is the real dictionary.
A reference S[key] triggers S.__getitem__ and SurrogateDictEntry(key) to be placed in the d.
When S[key] = SurrogateDictEntry(key) is constructed, it stores key. This will be the key into d for the value that this entry of SurrogateDictEntry is acting as a surrogate for.
After S[key] is returned, it is either entered into the d, or has some operation(s) performed on it. If an operation is performed on it, it triggers the relative __op__ method which simple stores the value that the operation is performed on and the name of the operation and then returns itself. We can't actually resolve the operation because d hasn't been constructed yet.
After d is constructed, it is passed to S.resolve. This method loops through d finding any instances of SurrogateDictEntry and replacing them with the result of calling the resolve method on the instance.
The SurrogateDictEntry.resolve method receives the now constructed d as an argument and can use the value of key that it stored at construction time to get the value that it is acting as a surrogate for. If an operation was performed on it after creation, the op attribute will have been set with the name of the operation that was performed. If the class has a __op__ method, then it has a __op__resolve__ method with the actual logic that would normally be in the __op__ method. So now we have the logic (self.op__resolve) and all necessary values (self.value, self.stored_value) to finally get the real value of d[key]. So we return that which step 4 places in the dictionary.
finally the SurrogateDict.resolve method returns d with all references resolved.
That'a a rough sketch. If you have any more questions, feel free to ask.
If you, just like me wandering how to make #jsbueno snippet work with {} style substitutions, below is the example code (which is probably not much efficient though):
import string
class MyDict(dict):
def __init__(self, *args, **kw):
super(MyDict,self).__init__(*args, **kw)
self.itemlist = super(MyDict,self).keys()
self.fmt = string.Formatter()
def __getitem__(self, item):
return self.fmt.vformat(dict.__getitem__(self, item), {}, self)
xs = MyDict({
'user' : 'gnucom',
'home' : '/home/{user}',
'bin' : '{home}/bin'
})
>>> xs["home"]
'/home/gnucom'
>>> xs["bin"]
'/home/gnucom/bin'
I tried to make it work with the simple replacement of % self with .format(**self) but it turns out it wouldn't work for nested expressions (like 'bin' in above listing, which references 'home', which has it's own reference to 'user') because of the evaluation order (** expansion is done before actual format call and it's not delayed like in original % version).
Write a class, maybe something with properties:
class PathInfo(object):
def __init__(self, user):
self.user = user
#property
def home(self):
return '/home/' + self.user
p = PathInfo('thc')
print p.home # /home/thc
As sort of an extended version of #Tony's answer, you could build a dictionary subclass that calls its values if they are callables:
class CallingDict(dict):
"""Returns the result rather than the value of referenced callables.
>>> cd = CallingDict({1: "One", 2: "Two", 'fsh': "Fish",
... "rhyme": lambda d: ' '.join((d[1], d['fsh'],
... d[2], d['fsh']))})
>>> cd["rhyme"]
'One Fish Two Fish'
>>> cd[1] = 'Red'
>>> cd[2] = 'Blue'
>>> cd["rhyme"]
'Red Fish Blue Fish'
"""
def __getitem__(self, item):
it = super(CallingDict, self).__getitem__(item)
if callable(it):
return it(self)
else:
return it
Of course this would only be usable if you're not actually going to store callables as values. If you need to be able to do that, you could wrap the lambda declaration in a function that adds some attribute to the resulting lambda, and check for it in CallingDict.__getitem__, but at that point it's getting complex, and long-winded, enough that it might just be easier to use a class for your data in the first place.
This is very easy in a lazily evaluated language (haskell).
Since Python is strictly evaluated, we can do a little trick to turn things lazy:
Y = lambda f: (lambda x: x(x))(lambda y: f(lambda *args: y(y)(*args)))
d1 = lambda self: lambda: {
'a': lambda: 3,
'b': lambda: self()['a']()
}
# fix the d1, and evaluate it
d2 = Y(d1)()
# to get a
d2['a']() # 3
# to get b
d2['b']() # 3
Syntax wise this is not very nice. That's because of us needing to explicitly construct lazy expressions with lambda: ... and explicitly evaluate lazy expression with ...(). It's the opposite problem in lazy languages needing strictness annotations, here in Python we end up needing lazy annotations.
I think with some more meta-programmming and some more tricks, the above could be made more easy to use.
Note that this is basically how let-rec works in some functional languages.
The jsbueno answer in Python 3 :
class MyDict(dict):
def __getitem__(self, item):
return dict.__getitem__(self, item).format(self)
dictionary = MyDict({
'user' : 'gnucom',
'home' : '/home/{0[user]}',
'bin' : '{0[home]}/bin'
})
print(dictionary["home"])
print(dictionary["bin"])
Her ewe use the python 3 string formatting with curly braces {} and the .format() method.
Documentation : https://docs.python.org/3/library/string.html