|
1
|
|
|
""" |
|
2
|
|
|
dataset.py |
|
3
|
|
|
""" |
|
4
|
|
|
from functools import total_ordering |
|
5
|
|
|
from six import ( |
|
6
|
|
|
iteritems, |
|
7
|
|
|
with_metaclass, |
|
8
|
|
|
) |
|
9
|
|
|
|
|
10
|
|
|
from zipline.pipeline.term import Term, AssetExists |
|
11
|
|
|
from zipline.utils.input_validation import ensure_dtype |
|
12
|
|
|
from zipline.utils.preprocess import preprocess |
|
13
|
|
|
|
|
14
|
|
|
|
|
15
|
|
|
class Column(object): |
|
16
|
|
|
""" |
|
17
|
|
|
An abstract column of data, not yet associated with a dataset. |
|
18
|
|
|
""" |
|
19
|
|
|
|
|
20
|
|
|
@preprocess(dtype=ensure_dtype) |
|
21
|
|
|
def __init__(self, dtype): |
|
22
|
|
|
self.dtype = dtype |
|
23
|
|
|
|
|
24
|
|
|
def bind(self, name): |
|
25
|
|
|
""" |
|
26
|
|
|
Bind a `Column` object to its name. |
|
27
|
|
|
""" |
|
28
|
|
|
return _BoundColumnDescr(dtype=self.dtype, name=name) |
|
29
|
|
|
|
|
30
|
|
|
|
|
31
|
|
|
class _BoundColumnDescr(object): |
|
32
|
|
|
""" |
|
33
|
|
|
Intermediate class that sits on `DataSet` objects and returns memoized |
|
34
|
|
|
`BoundColumn` objects when requested. |
|
35
|
|
|
""" |
|
36
|
|
|
def __init__(self, dtype, name): |
|
37
|
|
|
self.dtype = dtype |
|
38
|
|
|
self.name = name |
|
39
|
|
|
|
|
40
|
|
|
def __get__(self, instance, owner): |
|
41
|
|
|
return BoundColumn( |
|
42
|
|
|
dtype=self.dtype, |
|
43
|
|
|
dataset=owner, |
|
44
|
|
|
name=self.name, |
|
45
|
|
|
) |
|
46
|
|
|
|
|
47
|
|
|
|
|
48
|
|
|
class BoundColumn(Term): |
|
49
|
|
|
""" |
|
50
|
|
|
A Column of data that's been concretely bound to a particular dataset. |
|
51
|
|
|
""" |
|
52
|
|
|
mask = AssetExists() |
|
53
|
|
|
extra_input_rows = 0 |
|
54
|
|
|
inputs = () |
|
55
|
|
|
|
|
56
|
|
|
def __new__(cls, dtype, dataset, name): |
|
57
|
|
|
return super(BoundColumn, cls).__new__( |
|
58
|
|
|
cls, |
|
59
|
|
|
domain=dataset.domain, |
|
60
|
|
|
dtype=dtype, |
|
61
|
|
|
dataset=dataset, |
|
62
|
|
|
name=name, |
|
63
|
|
|
) |
|
64
|
|
|
|
|
65
|
|
|
def _init(self, dataset, name, *args, **kwargs): |
|
66
|
|
|
self._dataset = dataset |
|
67
|
|
|
self._name = name |
|
68
|
|
|
return super(BoundColumn, self)._init(*args, **kwargs) |
|
69
|
|
|
|
|
70
|
|
|
@classmethod |
|
71
|
|
|
def static_identity(cls, dataset, name, *args, **kwargs): |
|
72
|
|
|
return ( |
|
73
|
|
|
super(BoundColumn, cls).static_identity(*args, **kwargs), |
|
74
|
|
|
dataset, |
|
75
|
|
|
name, |
|
76
|
|
|
) |
|
77
|
|
|
|
|
78
|
|
|
@property |
|
79
|
|
|
def dataset(self): |
|
80
|
|
|
return self._dataset |
|
81
|
|
|
|
|
82
|
|
|
@property |
|
83
|
|
|
def name(self): |
|
84
|
|
|
return self._name |
|
85
|
|
|
|
|
86
|
|
|
@property |
|
87
|
|
|
def qualname(self): |
|
88
|
|
|
""" |
|
89
|
|
|
Fully qualified of this column. |
|
90
|
|
|
""" |
|
91
|
|
|
return '.'.join([self.dataset.__name__, self.name]) |
|
92
|
|
|
|
|
93
|
|
|
@property |
|
94
|
|
|
def latest(self): |
|
95
|
|
|
from zipline.pipeline.factors import Latest |
|
96
|
|
|
return Latest(inputs=(self,), dtype=self.dtype) |
|
97
|
|
|
|
|
98
|
|
|
def __repr__(self): |
|
99
|
|
|
return "{qualname}::{dtype}".format( |
|
100
|
|
|
qualname=self.qualname, |
|
101
|
|
|
dtype=self.dtype.name, |
|
102
|
|
|
) |
|
103
|
|
|
|
|
104
|
|
|
def short_repr(self): |
|
105
|
|
|
return self.qualname |
|
106
|
|
|
|
|
107
|
|
|
|
|
108
|
|
|
@total_ordering |
|
109
|
|
|
class DataSetMeta(type): |
|
110
|
|
|
""" |
|
111
|
|
|
Metaclass for DataSets |
|
112
|
|
|
|
|
113
|
|
|
Supplies name and dataset information to Column attributes. |
|
114
|
|
|
""" |
|
115
|
|
|
|
|
116
|
|
|
def __new__(mcls, name, bases, dict_): |
|
117
|
|
|
newtype = super(DataSetMeta, mcls).__new__(mcls, name, bases, dict_) |
|
118
|
|
|
# collect all of the column names that we inherit from our parents |
|
119
|
|
|
column_names = set().union( |
|
120
|
|
|
*(getattr(base, '_column_names', ()) for base in bases) |
|
121
|
|
|
) |
|
122
|
|
|
for maybe_colname, maybe_column in iteritems(dict_): |
|
123
|
|
|
if isinstance(maybe_column, Column): |
|
124
|
|
|
# add column names defined on our class |
|
125
|
|
|
bound_column_descr = maybe_column.bind(maybe_colname) |
|
126
|
|
|
setattr(newtype, maybe_colname, bound_column_descr) |
|
127
|
|
|
column_names.add(maybe_colname) |
|
128
|
|
|
|
|
129
|
|
|
newtype._column_names = frozenset(column_names) |
|
130
|
|
|
return newtype |
|
131
|
|
|
|
|
132
|
|
|
@property |
|
133
|
|
|
def columns(self): |
|
134
|
|
|
return frozenset( |
|
135
|
|
|
getattr(self, colname) for colname in self._column_names |
|
136
|
|
|
) |
|
137
|
|
|
|
|
138
|
|
|
def __lt__(self, other): |
|
139
|
|
|
return id(self) < id(other) |
|
140
|
|
|
|
|
141
|
|
|
def __repr__(self): |
|
142
|
|
|
return '<DataSet: %r>' % self.__name__ |
|
143
|
|
|
|
|
144
|
|
|
|
|
145
|
|
|
class DataSet(with_metaclass(DataSetMeta, object)): |
|
146
|
|
|
domain = None |
|
147
|
|
|
|
quantopian /
zipline
