Total Complexity | 4 |
Total Lines | 32 |
Duplicated Lines | 0 % |
Changes | 0 |
1 | """A simple transformation that truncates the text of input documents to a |
||
2 | given character length.""" |
||
3 | |||
4 | from __future__ import annotations |
||
5 | |||
6 | from typing import TYPE_CHECKING |
||
7 | |||
8 | from annif.exception import ConfigurationException |
||
9 | |||
10 | from . import transform |
||
11 | |||
12 | if TYPE_CHECKING: |
||
13 | from annif.project import AnnifProject |
||
14 | |||
15 | |||
16 | class InputLimiter(transform.BaseTransform): |
||
17 | name = "limit" |
||
18 | |||
19 | def __init__(self, project: AnnifProject | None, input_limit: str) -> None: |
||
20 | super().__init__(project) |
||
21 | self.input_limit = int(input_limit) |
||
22 | self._validate_value(self.input_limit) |
||
23 | |||
24 | def transform_text(self, text: str) -> str: |
||
25 | return text[: self.input_limit] |
||
26 | |||
27 | def _validate_value(self, input_limit: int) -> None: |
||
28 | if input_limit < 0: |
||
29 | raise ConfigurationException( |
||
30 | "input_limit in limit_input transform cannot be negative", |
||
31 | project_id=self.project.project_id, |
||
32 | ) |
||
33 |