1
|
|
|
#! /usr/bin/env python |
2
|
|
|
# |
3
|
|
|
# Copyright (C) 2015-2016 Rich Lewis <[email protected]> |
4
|
|
|
# License: 3-clause BSD |
5
|
|
|
|
6
|
1 |
|
""" |
7
|
|
|
# skchem.pipeline.pipeline |
8
|
|
|
|
9
|
|
|
Module implementing pipelines. |
10
|
|
|
""" |
11
|
|
|
|
12
|
1 |
|
from ..utils import yaml_dump, json_dump |
13
|
1 |
|
from ..io import read_config, read_json, read_yaml |
|
|
|
|
14
|
|
|
|
15
|
|
|
|
16
|
1 |
|
def is_transform_filter(obj): |
17
|
|
|
""" Whether an object is a TransformFilter (by duck typing). """ |
18
|
|
|
return hasattr(obj, 'transform_filter') |
19
|
|
|
|
20
|
|
|
|
21
|
1 |
|
def is_filter(obj): |
22
|
|
|
""" Whether an object is a Filter (by duck typing). """ |
23
|
|
|
return hasattr(obj, 'filter') |
24
|
|
|
|
25
|
|
|
|
26
|
1 |
|
def is_transformer(obj): |
27
|
|
|
""" Whether an object is a Transformer (by duck typing). """ |
28
|
|
|
return hasattr(obj, 'transform') |
29
|
|
|
|
30
|
|
|
|
31
|
1 |
|
class Pipeline(object): |
32
|
|
|
|
33
|
|
|
""" Pipeline object. Applies filters and transformers in sequence. """ |
34
|
|
|
|
35
|
1 |
|
def __init__(self, objects): |
36
|
|
|
self.objects = objects |
37
|
|
|
|
38
|
1 |
|
def get_params(self): |
|
|
|
|
39
|
|
|
return {'objects': [obj.to_dict() for obj in self.objects]} |
40
|
|
|
|
41
|
1 |
|
def to_dict(self): |
42
|
|
|
""" Return a dictionary representation of the object.""" |
43
|
|
|
|
44
|
|
|
return {'skchem.pipeline.pipeline.Pipeline': self.get_params()} |
45
|
|
|
|
46
|
1 |
|
@classmethod |
47
|
|
|
def from_params(cls, params): |
48
|
|
|
""" Create a instance from a params dictionary. """ |
49
|
|
|
return cls([read_config(conf) for conf in params['objects']]) |
50
|
|
|
|
51
|
1 |
|
def to_json(self, target=None): |
52
|
|
|
|
53
|
|
|
""" Serialize the object as JSON. |
54
|
|
|
|
55
|
|
|
Args: |
56
|
|
|
target (str or file-like): |
57
|
|
|
A file or filepath to serialize the object to. If `None`, |
58
|
|
|
return the JSON as a string. |
59
|
|
|
|
60
|
|
|
Returns: |
61
|
|
|
None or str |
62
|
|
|
""" |
63
|
|
|
|
64
|
|
|
return json_dump(self.to_dict(), target) |
65
|
|
|
|
66
|
1 |
|
def to_yaml(self, target=None): |
67
|
|
|
|
68
|
|
|
""" Serialize the object as YAML. |
69
|
|
|
|
70
|
|
|
Args: |
71
|
|
|
target (str or file-like): |
72
|
|
|
A file or filepath to serialize the object to. If `None`, |
73
|
|
|
return the YAML as a string. |
74
|
|
|
|
75
|
|
|
Returns: |
76
|
|
|
None or str |
77
|
|
|
""" |
78
|
|
|
|
79
|
|
|
return yaml_dump(self.to_dict(), target) |
80
|
|
|
|
81
|
1 |
|
def copy(self): |
82
|
|
|
""" Return a copy of this object. """ |
83
|
|
|
return self.__class__(self.get_params()) |
84
|
|
|
|
85
|
1 |
|
def transform_filter(self, mols, y=None): |
|
|
|
|
86
|
|
|
for obj in self.objects: |
87
|
|
|
if is_transform_filter(obj): |
88
|
|
|
mols = obj.transform_filter(mols) |
89
|
|
|
elif is_filter(obj): |
90
|
|
|
mols = obj.filter(mols) |
91
|
|
|
elif is_transformer(obj): |
92
|
|
|
mols = obj.transform(mols) |
93
|
|
|
else: |
94
|
|
|
raise NotImplementedError('Cannot apply {}.'.format(obj)) |
95
|
|
|
return mols if y is None else (mols, y.reindex(mols.index)) |
96
|
|
|
|
97
|
|
|
|