|
1
|
|
|
"""Classes for supporting vocabulary files in CSV or TSV format""" |
|
2
|
|
|
|
|
3
|
|
|
from __future__ import annotations |
|
4
|
|
|
|
|
5
|
|
|
import csv |
|
6
|
|
|
import os.path |
|
7
|
|
|
from typing import TYPE_CHECKING |
|
8
|
|
|
|
|
9
|
|
|
import annif |
|
10
|
|
|
import annif.util |
|
11
|
|
|
|
|
12
|
|
|
from .skos import serialize_subjects_to_skos |
|
13
|
|
|
from .types import Subject, VocabSource |
|
14
|
|
|
|
|
15
|
|
|
if TYPE_CHECKING: |
|
16
|
|
|
from collections.abc import Generator, Iterator |
|
17
|
|
|
|
|
18
|
|
|
|
|
19
|
|
|
class VocabFileTSV(VocabSource): |
|
20
|
|
|
"""A monolingual subject vocabulary stored in a TSV file.""" |
|
21
|
|
|
|
|
22
|
|
|
def __init__(self, path: str, language: str) -> None: |
|
23
|
|
|
"""initialize the VocabFileTSV given a path to a TSV file and the |
|
24
|
|
|
language of the vocabulary""" |
|
25
|
|
|
|
|
26
|
|
|
self.path = path |
|
27
|
|
|
self.language = language |
|
28
|
|
|
|
|
29
|
|
|
def _parse_line(self, line: str) -> Iterator[Subject]: |
|
30
|
|
|
vals = line.strip().split("\t", 2) |
|
31
|
|
|
clean_uri = annif.util.cleanup_uri(vals[0]) |
|
32
|
|
|
label = vals[1] if len(vals) >= 2 else None |
|
33
|
|
|
labels = {self.language: label} if label else None |
|
34
|
|
|
notation = vals[2] if len(vals) >= 3 else None |
|
35
|
|
|
yield Subject(uri=clean_uri, labels=labels, notation=notation) |
|
36
|
|
|
|
|
37
|
|
|
@property |
|
38
|
|
|
def languages(self) -> list[str]: |
|
39
|
|
|
return [self.language] |
|
40
|
|
|
|
|
41
|
|
|
@property |
|
42
|
|
|
def subjects(self) -> Generator: |
|
43
|
|
|
with open(self.path, encoding="utf-8-sig") as subjfile: |
|
44
|
|
|
for line in subjfile: |
|
45
|
|
|
yield from self._parse_line(line) |
|
46
|
|
|
|
|
47
|
|
|
def save_skos(self, path: str) -> None: |
|
48
|
|
|
"""Save the contents of the subject vocabulary into a SKOS/Turtle |
|
49
|
|
|
file with the given path name.""" |
|
50
|
|
|
serialize_subjects_to_skos(self.subjects, path) |
|
51
|
|
|
|
|
52
|
|
|
|
|
53
|
|
|
class VocabFileCSV(VocabSource): |
|
54
|
|
|
"""A multilingual subject vocabulary stored in a CSV file.""" |
|
55
|
|
|
|
|
56
|
|
|
def __init__(self, path: str) -> None: |
|
57
|
|
|
"""initialize the VocabFileCSV given a path to a CSV file""" |
|
58
|
|
|
self.path = path |
|
59
|
|
|
|
|
60
|
|
|
def _parse_row(self, row: dict[str, str]) -> Iterator[Subject]: |
|
61
|
|
|
labels = { |
|
62
|
|
|
fname.replace("label_", ""): value or None |
|
63
|
|
|
for fname, value in row.items() |
|
64
|
|
|
if fname.startswith("label_") |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
|
|
# if there are no labels in any language, set labels to None |
|
68
|
|
|
# indicating a deprecated subject |
|
69
|
|
|
if set(labels.values()) == {None}: |
|
70
|
|
|
labels = None |
|
71
|
|
|
|
|
72
|
|
|
yield Subject( |
|
73
|
|
|
uri=annif.util.cleanup_uri(row["uri"]), |
|
74
|
|
|
labels=labels, |
|
75
|
|
|
notation=row.get("notation", None) or None, |
|
76
|
|
|
) |
|
77
|
|
|
|
|
78
|
|
|
@property |
|
79
|
|
|
def languages(self) -> list[str]: |
|
80
|
|
|
# infer the supported languages from the CSV column names |
|
81
|
|
|
with open(self.path, encoding="utf-8-sig") as csvfile: |
|
82
|
|
|
reader = csv.reader(csvfile) |
|
83
|
|
|
fieldnames = next(reader, None) |
|
84
|
|
|
|
|
85
|
|
|
return [ |
|
86
|
|
|
fname.replace("label_", "") |
|
87
|
|
|
for fname in fieldnames |
|
88
|
|
|
if fname.startswith("label_") |
|
89
|
|
|
] |
|
90
|
|
|
|
|
91
|
|
|
@property |
|
92
|
|
|
def subjects(self) -> Generator: |
|
93
|
|
|
with open(self.path, encoding="utf-8-sig") as csvfile: |
|
94
|
|
|
reader = csv.DictReader(csvfile) |
|
95
|
|
|
for row in reader: |
|
96
|
|
|
yield from self._parse_row(row) |
|
97
|
|
|
|
|
98
|
|
|
def save_skos(self, path: str) -> None: |
|
99
|
|
|
"""Save the contents of the subject vocabulary into a SKOS/Turtle |
|
100
|
|
|
file with the given path name.""" |
|
101
|
|
|
serialize_subjects_to_skos(self.subjects, path) |
|
102
|
|
|
|
|
103
|
|
|
@staticmethod |
|
104
|
|
|
def is_csv_file(path: str) -> bool: |
|
105
|
|
|
"""return True if the path looks like a CSV file""" |
|
106
|
|
|
|
|
107
|
|
|
return os.path.splitext(path)[1].lower() == ".csv" |
|
108
|
|
|
|
NatLibFi /
Annif
