|
1
|
|
|
"""Common functionality for transforming text of input documents.""" |
|
2
|
|
|
|
|
3
|
|
|
import abc |
|
4
|
|
|
from annif.corpus import TransformingDocumentCorpus |
|
5
|
|
|
from annif.exception import ConfigurationException |
|
6
|
|
|
|
|
7
|
|
|
|
|
8
|
|
|
class BaseTransform(metaclass=abc.ABCMeta): |
|
9
|
|
|
"""Base class for text transformations, which need to implement the |
|
10
|
|
|
transform function.""" |
|
11
|
|
|
|
|
12
|
|
|
name = None |
|
13
|
|
|
|
|
14
|
|
|
def __init__(self, project): |
|
15
|
|
|
self.project = project |
|
16
|
|
|
|
|
17
|
|
|
@abc.abstractmethod |
|
18
|
|
|
def transform_fn(self, text): |
|
19
|
|
|
"""Perform the text transformation.""" |
|
20
|
|
|
pass # pragma: no cover |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
class IdentityTransform(BaseTransform): |
|
24
|
|
|
"""Transform that does not modify text but simply passes it through.""" |
|
25
|
|
|
|
|
26
|
|
|
name = 'pass' |
|
27
|
|
|
|
|
28
|
|
|
def transform_fn(self, text): |
|
29
|
|
|
return text |
|
30
|
|
|
|
|
31
|
|
|
|
|
32
|
|
|
class TransformChain(): |
|
33
|
|
|
"""Class instantiating and holding the transformation objects performing |
|
34
|
|
|
the actual text transformation.""" |
|
35
|
|
|
|
|
36
|
|
|
def __init__(self, transform_classes, args, project): |
|
37
|
|
|
self.project = project |
|
38
|
|
|
self.transforms = self._init_transforms(transform_classes, args) |
|
39
|
|
|
|
|
40
|
|
|
def _init_transforms(self, transform_classes, args): |
|
41
|
|
|
transforms = [] |
|
42
|
|
|
for trans, (posargs, kwargs) in zip(transform_classes, args): |
|
43
|
|
|
try: |
|
44
|
|
|
transforms.append( |
|
45
|
|
|
trans(self.project, *posargs, **kwargs)) |
|
46
|
|
|
except (ValueError, TypeError): |
|
47
|
|
|
raise ConfigurationException( |
|
48
|
|
|
f"Invalid arguments to {trans.name} transform: " |
|
49
|
|
|
f"{posargs}, {kwargs})", |
|
50
|
|
|
project_id=self.project.project_id) |
|
51
|
|
|
return transforms |
|
52
|
|
|
|
|
53
|
|
|
def transform_text(self, text): |
|
54
|
|
|
for trans in self.transforms: |
|
55
|
|
|
text = trans.transform_fn(text) |
|
56
|
|
|
return text |
|
57
|
|
|
|
|
58
|
|
|
def transform_corpus(self, corpus): |
|
59
|
|
|
return TransformingDocumentCorpus(corpus, self.transform_text) |
|
60
|
|
|
|
NatLibFi /
Annif
