|
1
|
|
|
import six |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
class DataIterator(six.Iterator): |
|
5
|
|
|
"""An iterator over data, representing a single epoch. |
|
6
|
|
|
|
|
7
|
|
|
Parameters |
|
8
|
|
|
---------- |
|
9
|
|
|
data_stream : :class:`DataStream` |
|
10
|
|
|
The data stream over which to iterate. |
|
11
|
|
|
request_iterator : iterator |
|
12
|
|
|
An iterator which returns the request to pass to the data stream |
|
13
|
|
|
for each step. |
|
14
|
|
|
as_dict : bool, optional |
|
15
|
|
|
If `True`, return dictionaries mapping source names to data |
|
16
|
|
|
from each source. If `False` (default), return tuples in the |
|
17
|
|
|
same order as `data_stream.sources`. |
|
18
|
|
|
|
|
19
|
|
|
""" |
|
20
|
|
|
def __init__(self, data_stream, request_iterator=None, as_dict=False): |
|
21
|
|
|
self.data_stream = data_stream |
|
22
|
|
|
self.request_iterator = request_iterator |
|
23
|
|
|
self.as_dict = as_dict |
|
24
|
|
|
|
|
25
|
|
|
def __iter__(self): |
|
26
|
|
|
return self |
|
27
|
|
|
|
|
28
|
|
|
def __next__(self): |
|
29
|
|
|
if self.request_iterator is not None: |
|
30
|
|
|
data = self.data_stream.get_data(next(self.request_iterator)) |
|
31
|
|
|
else: |
|
32
|
|
|
data = self.data_stream.get_data() |
|
33
|
|
|
if self.as_dict: |
|
34
|
|
|
return dict(zip(self.data_stream.sources, data)) |
|
35
|
|
|
else: |
|
36
|
|
|
return data |
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
class TransformerDataIterator(six.Iterator): |
|
40
|
|
|
"""An iterator over transformed data, representing a single epoch. |
|
41
|
|
|
|
|
42
|
|
|
Parameters |
|
43
|
|
|
---------- |
|
44
|
|
|
transformer : :class:`Transformer` |
|
45
|
|
|
The transformer over which to iterate. |
|
46
|
|
|
child_iterator : :class:`DataIterator` |
|
47
|
|
|
The iterator of the transformer's data stream. |
|
48
|
|
|
request_iterator : iterator |
|
49
|
|
|
An iterator which returns the request to pass to the transformer |
|
50
|
|
|
for each step. |
|
51
|
|
|
as_dict : bool, optional |
|
52
|
|
|
If `True`, return dictionaries mapping source names to data |
|
53
|
|
|
from each source. If `False` (default), return tuples in the |
|
54
|
|
|
same order as `transformer.sources`. |
|
55
|
|
|
|
|
56
|
|
|
""" |
|
57
|
|
|
def __init__(self, transformer, child_iterator, request_iterator=None, |
|
58
|
|
|
as_dict=False): |
|
59
|
|
|
self.transformer = transformer |
|
60
|
|
|
self.child_iterator = child_iterator |
|
61
|
|
|
self.request_iterator = request_iterator |
|
62
|
|
|
self.as_dict = as_dict |
|
63
|
|
|
|
|
64
|
|
|
def __iter__(self): |
|
65
|
|
|
return self |
|
66
|
|
|
|
|
67
|
|
|
def __next__(self): |
|
68
|
|
|
if self.request_iterator is not None: |
|
69
|
|
|
data = self.transformer.get_data( |
|
70
|
|
|
self.child_iterator, next(self.request_iterator)) |
|
71
|
|
|
else: |
|
72
|
|
|
data = self.transformer.get_data(self.child_iterator) |
|
73
|
|
|
if self.as_dict: |
|
74
|
|
|
return dict(zip(self.transformer.sources, data)) |
|
75
|
|
|
else: |
|
76
|
|
|
return data |
|
77
|
|
|
|
mila-udem /
fuel
