1
|
|
|
"""Common functionality for input transforming.""" |
2
|
|
|
|
3
|
|
|
from annif.corpus import TransformingDocumentCorpus |
4
|
|
|
from annif.exception import ConfigurationException |
5
|
|
|
|
6
|
|
|
|
7
|
|
|
class IdentityTransform(): |
8
|
|
|
"""""""" # TODO |
9
|
|
|
|
10
|
|
|
name = 'pass' |
11
|
|
|
|
12
|
|
|
def __init__(self, project): |
13
|
|
|
self.project = project |
14
|
|
|
|
15
|
|
|
def transform_text(self, text): |
16
|
|
|
"""""" # TODO |
17
|
|
|
return text |
18
|
|
|
|
19
|
|
|
def transform_corpus(self, corpus): |
20
|
|
|
"""""" # TODO |
21
|
|
|
return TransformingDocumentCorpus(corpus, self.transform_text) |
22
|
|
|
|
23
|
|
|
|
24
|
|
|
class TransformChain(): |
25
|
|
|
"""""" # TODO |
26
|
|
|
|
27
|
|
|
def __init__(self, transform_classes, args, project): |
28
|
|
|
self.project = project |
29
|
|
|
self.transforms = self._init_transforms(transform_classes, args) |
30
|
|
|
|
31
|
|
|
def _init_transforms(self, transform_classes, args): |
32
|
|
|
transforms = [] |
33
|
|
|
for trans, (posargs, kwargs) in zip(transform_classes, args): |
34
|
|
|
try: |
35
|
|
|
transforms.append( |
36
|
|
|
trans(self.project, *posargs, **kwargs)) |
37
|
|
|
except (ValueError, TypeError): |
38
|
|
|
raise ConfigurationException( |
39
|
|
|
f"Invalid arguments to {trans.name} transform: " |
40
|
|
|
f"{posargs}, {kwargs})", |
41
|
|
|
project_id=self.project.project_id) |
42
|
|
|
return transforms |
43
|
|
|
|
44
|
|
|
def transform_text(self, text): |
45
|
|
|
for trans in self.transforms: |
46
|
|
|
text = trans.transform_text(text) |
47
|
|
|
return text |
48
|
|
|
|
49
|
|
|
def transform_corpus(self, corpus): |
50
|
|
|
for trans in self.transforms: |
51
|
|
|
corpus = trans.transform_corpus(corpus) |
52
|
|
|
return corpus |
53
|
|
|
|