-
Notifications
You must be signed in to change notification settings - Fork 1
/
records.py
213 lines (150 loc) · 6 KB
/
records.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
"""Create type-safe dynamic namedtuples by abusing ordered kwargs!
Example usage:
>>> r_orig = record(foo='ayy', bar='lmao')
>>> r_same = record(foo='ayy', bar='lmao')
>>> r_orig
record(foo='ayy', bar='lmao')
>>> r_same
record(foo='ayy', bar='lmao')
>>> assert r_orig == r_same
Records with the same fields, in the same order, have the same type:
>>> assert type(r_orig) is type(r_same)
Records are hashable:
>>> assert r_orig in {r_orig}
>>> assert r_orig in {r_same}
Records with different values are not equal, but still have the same type:
>>> r_diff = record(foo='ayy', bar='wat')
>>> assert r_orig != r_diff and r_same != r_diff
>>> assert r_diff not in (r_orig, r_same)
>>> assert type(r_orig) is type(r_diff)
Different field orders result in different, inequivalent types:
>>> r_forward = record(foo='ayy', bar='lmao')
>>> r_reverse = record(bar='lmao', foo='ayy')
>>> assert r_forward.foo == r_reverse.foo and r_forward.bar == r_reverse.bar
>>> assert r_forward != r_reverse
>>> assert type(r_reverse) is not type(r_forward)
Their type names indicate this difference:
>>> print(type(r_forward))
<class 'records.record['foo', 'bar']'>
>>> print(type(r_reverse))
<class 'records.record['bar', 'foo']'>
Instances are cached:
>>> assert r_orig is r_same
Caching is implemented via weakrefs, so instances are destroyed when no
non-weak references remain:
>>> id1 = id(record(a=1, b=2))
>>> id2 = id(record(a=1, b=2))
>>> assert id1 != id2
>>> ref = record(a=1, b=2)
>>> id1 = id(record(a=1, b=2))
>>> id2 = id(record(a=1, b=2))
>>> assert id1 == id2
>>> del ref
>>> id1 = id(record(a=1, b=2))
>>> id2 = id(record(a=1, b=2))
>>> assert id1 != id2
Note, however, that implicit references may persist during the
evaluation of certain expressions:
>>> assert record(a=1, b=2) is record(a=1, b=2)
Derivative records can be created by calling instances:
>>> r_derived = r_orig(bar='wat')
>>> r_derived
record(foo='ayy', bar='wat')
Derivative records subclass the base record class, not each other:
>>> for cls in record(a=1)(b=2)(c=3).__class__.__mro__: print(cls)
<class 'records.record['a', 'b', 'c']'>
<class 'records.record'>
<class 'object'>
Caching still applies to new records created this way:
>>> assert r_orig() is r_orig
>>> assert r_orig(foo='ayy') is r_orig
>>> assert r_orig(foo='ayy', bar='lmao') is r_orig
>>> assert r_orig(bar='lmao') is r_orig
Even when the fields are provided in a different order in the call:
>>> assert r_orig(bar='lmao', foo='ayy') is r_orig
(TODO: This does not seem ideal, but I'm not sure what to do instead.)
New fields can also be added this way:
>>> r_orig(baz='new')
record(foo='ayy', bar='lmao', baz='new')
Iterating over a record yields its key-value pairs:
>>> list(r_orig)
[('foo', 'ayy'), ('bar', 'lmao')]
>>> fields, values = zip(*r_orig)
>>> fields
('foo', 'bar')
>>> values
('ayy', 'lmao')
Records have no __dict__, but can easily create one:
>>> vars(r_orig)
Traceback (most recent call last):
...
TypeError: vars() argument must have __dict__ attribute
>>> dict(r_orig)
{'foo': 'ayy', 'bar': 'lmao'}
"""
from weakref import WeakValueDictionary
class VintageRecordStore(dict):
"""Simpler implementation of records using namedtuples.
Tuple subclasses, including namedtuples, don't support weakrefs; so
records created this way cannot be instance-cached.
See https://docs.python.org/3/reference/datamodel.html#notes-on-using-slots
"""
def __call__(self, **attrs):
return self[tuple(attrs)](**attrs)
def __missing__(self, key):
from collections import namedtuple
self[key] = cls = namedtuple('record', key)
return cls
class RecordMeta(type):
"""Metaclass which dynamically creates subclasses and caches instances."""
_subclasses = WeakValueDictionary()
_instances = WeakValueDictionary()
def __call__(cls, **kwargs):
"""Return a record instance with the given fields."""
key = tuple(kwargs.items())
try:
instance = cls._instances[key]
except KeyError:
instance = cls._instances[key] = cls.create_instance(kwargs)
return instance
def __getitem__(cls, fields):
"""Return a record subclass with the given fields."""
if not isinstance(fields, tuple):
fields = (fields,)
try:
subclass = cls._subclasses[fields]
except KeyError:
subclass = cls._subclasses[fields] = cls.create_subclass(fields)
return subclass
def create_instance(cls, kwargs):
subclass = cls[tuple(kwargs)]
subclass_super = super(type(subclass), subclass)
return subclass_super.__call__(**kwargs)
def create_subclass(cls, fields):
if fields:
fields_repr = ', '.join(map(repr, fields))
name = f"{cls.__name__}[{fields_repr}]"
else:
name = cls.__name__
return type(name, (cls,), {'__slots__': fields})
class record(metaclass=RecordMeta):
"""Auto-cached, slots-based data class.
Similar to namedtuple, but with less nonsense and weakref support.
"""
__slots__ = ['__weakref__']
def __init__(self, **attrs):
for name, value in attrs.items():
setattr(self, name, value)
def __call__(self, **updates):
"""Construct a new record based on this instance, with some updates."""
return record(**{**dict(self), **updates}) # 😚👌
def __iter__(self):
"""Iterate over key-value PAIRS, as G-d intended."""
for name in self.__class__.__slots__:
if not name.startswith('_'):
yield name, getattr(self, name)
def __repr__(self):
return 'record({})'.format(
', '.join(f'{name}={value!r}' for name, value in self)
)
# Instances are cached, so equivalence is just identity!