4032787010

Build Type

push

github

Committed by Artur Barseghyan

Commit Message

Bring back MD files

Run Details

1228 of 1230 relevant lines covered (99.84%)

6.27 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

100.0

/src/faker_file/providers/augment_file_from_dir/augmenters/nlpaug_augmenter.py

import nlpaug.augmenter.word as naw

from .base import BaseTextAugmenter

__author__ = "Artur Barseghyan <artur.barseghyan@gmail.com>"
__copyright__ = "2022-2023 Artur Barseghyan"
__license__ = "MIT"
__all__ = (
    "ContextualWordEmbeddingsAugmenter",
    "DEFAULT_ACTION",
    "DEFAULT_MODEL_PATH",
)

DEFAULT_MODEL_PATH = "bert-base-multilingual-cased"
DEFAULT_ACTION = "substitute"


class ContextualWordEmbeddingsAugmenter(BaseTextAugmenter):
    """Text extractor based on `ContextualWordEmbsAug` of `nlpaug`.

    Usage example:

        from faker import Faker
        from faker_file.providers.augment_file_from_dir import (
            AugmentFileFromDirProvider,
        )
        from faker_file.providers.augment_file_from_dir.augmenters import (
            nlpaug_augmenter,
        )

        FAKER = Faker()

        file = AugmentFileFromDirProvider(FAKER).augment_file_from_dir(
            text_augmenter_cls=(
                nlpaug_augmenter.ContextualWordEmbeddingsAugmenter
            ),
            text_augmenter_kwargs={
                "model_path": "bert-base-uncased",
                "action": "substitute",
            }
        )

    Refer to `nlpaug` official documentation and check examples
    for `Textual augmenters`:

        https://nlpaug.readthedocs.io/en/latest/example/example.html

    Some well working options for `model_path` are:

        - bert-base-multilingual-cased
        - bert-base-multilingual-uncased
        - bert-base-cased
        - bert-base-uncased
        - bert-base-german-cased
        - GroNLP/bert-base-dutch-cased

    Options for `action` are:

        - insert
        - substitute
    """

    model_path: str = DEFAULT_MODEL_PATH
    action: str = DEFAULT_ACTION

    def handle_kwargs(
        self: "ContextualWordEmbeddingsAugmenter", **kwargs
    ) -> None:
        """Handle kwargs."""
        if "model_path" in kwargs:
            self.model_path = kwargs["model_path"]
        if "action" in kwargs:
            self.action = kwargs["action"]

    def augment(
        self: "ContextualWordEmbeddingsAugmenter",
        text: str,
    ) -> str:
        """Augment text."""
        aug = naw.ContextualWordEmbsAug(
            model_path=self.model_path,
            action=self.action,
        )
        return aug.augment(text)[0]

1	import nlpaug.augmenter.word as naw	4✔
2
3	from .base import BaseTextAugmenter	4✔
4
5	__author__ = "Artur Barseghyan <artur.barseghyan@gmail.com>"	4✔
6	__copyright__ = "2022-2023 Artur Barseghyan"	4✔
7	__license__ = "MIT"	4✔
8	__all__ = (	4✔
9	"ContextualWordEmbeddingsAugmenter",
10	"DEFAULT_ACTION",
11	"DEFAULT_MODEL_PATH",
12	)
13
14	DEFAULT_MODEL_PATH = "bert-base-multilingual-cased"	4✔
15	DEFAULT_ACTION = "substitute"	4✔
16
17
18	class ContextualWordEmbeddingsAugmenter(BaseTextAugmenter):	4✔
19	"""Text extractor based on `ContextualWordEmbsAug` of `nlpaug`.
20
21	Usage example:
22
23	from faker import Faker
24	from faker_file.providers.augment_file_from_dir import (
25	AugmentFileFromDirProvider,
26	)
27	from faker_file.providers.augment_file_from_dir.augmenters import (
28	nlpaug_augmenter,
29	)
30
31	FAKER = Faker()
32
33	file = AugmentFileFromDirProvider(FAKER).augment_file_from_dir(
34	text_augmenter_cls=(
35	nlpaug_augmenter.ContextualWordEmbeddingsAugmenter
36	),
37	text_augmenter_kwargs={
38	"model_path": "bert-base-uncased",
39	"action": "substitute",
40	}
41	)
42
43	Refer to `nlpaug` official documentation and check examples
44	for `Textual augmenters`:
45
46	https://nlpaug.readthedocs.io/en/latest/example/example.html
47
48	Some well working options for `model_path` are:
49
50	- bert-base-multilingual-cased
51	- bert-base-multilingual-uncased
52	- bert-base-cased
53	- bert-base-uncased
54	- bert-base-german-cased
55	- GroNLP/bert-base-dutch-cased
56
57	Options for `action` are:
58
59	- insert
60	- substitute
61	"""
62
63	model_path: str = DEFAULT_MODEL_PATH	4✔
64	action: str = DEFAULT_ACTION	4✔
65
66	def handle_kwargs(	4✔
67	self: "ContextualWordEmbeddingsAugmenter", **kwargs
68	) -> None:
69	"""Handle kwargs."""
70	if "model_path" in kwargs:	4✔
71	self.model_path = kwargs["model_path"]	4✔
72	if "action" in kwargs:	4✔
73	self.action = kwargs["action"]	4✔
74
75	def augment(	4✔
76	self: "ContextualWordEmbeddingsAugmenter",
77	text: str,
78	) -> str:
79	"""Augment text."""
80	aug = naw.ContextualWordEmbsAug(	4✔
81	model_path=self.model_path,
82	action=self.action,
83	)
84	return aug.augment(text)[0]	4✔

barseghyanartur / faker-file / 4032787010

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous