|
1
|
|
|
"""Vocabulary management functionality for Annif""" |
|
2
|
|
|
|
|
3
|
|
|
from __future__ import annotations |
|
4
|
|
|
|
|
5
|
|
|
import os.path |
|
6
|
|
|
from typing import TYPE_CHECKING |
|
7
|
|
|
|
|
8
|
|
|
import annif |
|
9
|
|
|
import annif.corpus |
|
10
|
|
|
import annif.util |
|
11
|
|
|
from annif.datadir import DatadirMixin |
|
12
|
|
|
from annif.exception import NotInitializedException |
|
13
|
|
|
|
|
14
|
|
|
from .subject_index import SubjectIndexFile |
|
15
|
|
|
from .types import SubjectIndex |
|
16
|
|
|
|
|
17
|
|
|
if TYPE_CHECKING: |
|
18
|
|
|
from rdflib.graph import Graph |
|
19
|
|
|
|
|
20
|
|
|
from annif.corpus.skos import SubjectFileSKOS |
|
21
|
|
|
from annif.corpus.subject import SubjectCorpus |
|
22
|
|
|
|
|
23
|
|
|
|
|
24
|
|
|
logger = annif.logger |
|
25
|
|
|
logger.addFilter(annif.util.DuplicateFilter()) |
|
26
|
|
|
|
|
27
|
|
|
|
|
28
|
|
|
class AnnifVocabulary(DatadirMixin): |
|
29
|
|
|
"""Class representing a subject vocabulary which can be used by multiple |
|
30
|
|
|
Annif projects.""" |
|
31
|
|
|
|
|
32
|
|
|
# defaults for uninitialized instances |
|
33
|
|
|
_subjects = None |
|
34
|
|
|
|
|
35
|
|
|
# constants |
|
36
|
|
|
INDEX_FILENAME_DUMP = "subjects.dump.gz" |
|
37
|
|
|
INDEX_FILENAME_TTL = "subjects.ttl" |
|
38
|
|
|
INDEX_FILENAME_CSV = "subjects.csv" |
|
39
|
|
|
|
|
40
|
|
|
def __init__(self, vocab_id: str, datadir: str) -> None: |
|
41
|
|
|
DatadirMixin.__init__(self, datadir, "vocabs", vocab_id) |
|
42
|
|
|
self.vocab_id = vocab_id |
|
43
|
|
|
self._skos_vocab = None |
|
44
|
|
|
|
|
45
|
|
|
def _create_subject_index(self, subject_corpus: SubjectCorpus) -> SubjectIndex: |
|
46
|
|
|
subjects = SubjectIndexFile() |
|
47
|
|
|
subjects.load_subjects(subject_corpus) |
|
48
|
|
|
annif.util.atomic_save(subjects, self.datadir, self.INDEX_FILENAME_CSV) |
|
49
|
|
|
return subjects |
|
50
|
|
|
|
|
51
|
|
|
def _update_subject_index(self, subject_corpus: SubjectCorpus) -> SubjectIndex: |
|
52
|
|
|
old_subjects = self.subjects |
|
53
|
|
|
new_subjects = SubjectIndexFile() |
|
54
|
|
|
new_subjects.load_subjects(subject_corpus) |
|
55
|
|
|
updated_subjects = SubjectIndexFile() |
|
56
|
|
|
|
|
57
|
|
|
for old_subject in old_subjects: |
|
58
|
|
|
if new_subjects.contains_uri(old_subject.uri): |
|
59
|
|
|
new_subject = new_subjects[new_subjects.by_uri(old_subject.uri)] |
|
60
|
|
|
else: # subject removed from new corpus |
|
61
|
|
|
new_subject = annif.corpus.Subject( |
|
62
|
|
|
uri=old_subject.uri, labels=None, notation=None |
|
63
|
|
|
) |
|
64
|
|
|
updated_subjects.append(new_subject) |
|
65
|
|
|
for new_subject in new_subjects: |
|
66
|
|
|
if not old_subjects.contains_uri(new_subject.uri): |
|
67
|
|
|
updated_subjects.append(new_subject) |
|
68
|
|
|
annif.util.atomic_save(updated_subjects, self.datadir, self.INDEX_FILENAME_CSV) |
|
69
|
|
|
return updated_subjects |
|
70
|
|
|
|
|
71
|
|
|
@property |
|
72
|
|
|
def subjects(self) -> SubjectIndex: |
|
73
|
|
|
if self._subjects is None: |
|
74
|
|
|
path = os.path.join(self.datadir, self.INDEX_FILENAME_CSV) |
|
75
|
|
|
if os.path.exists(path): |
|
76
|
|
|
logger.debug("loading subjects from %s", path) |
|
77
|
|
|
self._subjects = SubjectIndexFile.load(path) |
|
78
|
|
|
else: |
|
79
|
|
|
raise NotInitializedException("subject file {} not found".format(path)) |
|
80
|
|
|
return self._subjects |
|
81
|
|
|
|
|
82
|
|
|
@property |
|
83
|
|
|
def skos(self) -> SubjectFileSKOS: |
|
84
|
|
|
"""return the subject vocabulary from SKOS file""" |
|
85
|
|
|
if self._skos_vocab is not None: |
|
86
|
|
|
return self._skos_vocab |
|
87
|
|
|
|
|
88
|
|
|
# attempt to load graph from dump file |
|
89
|
|
|
dumppath = os.path.join(self.datadir, self.INDEX_FILENAME_DUMP) |
|
90
|
|
|
if os.path.exists(dumppath): |
|
91
|
|
|
logger.debug(f"loading graph dump from {dumppath}") |
|
92
|
|
|
try: |
|
93
|
|
|
self._skos_vocab = annif.corpus.SubjectFileSKOS(dumppath) |
|
94
|
|
|
except ModuleNotFoundError: |
|
95
|
|
|
# Probably dump has been saved using a different rdflib version |
|
96
|
|
|
logger.debug("could not load graph dump, using turtle file") |
|
97
|
|
|
else: |
|
98
|
|
|
return self._skos_vocab |
|
99
|
|
|
|
|
100
|
|
|
# graph dump file not found - parse ttl file instead |
|
101
|
|
|
path = os.path.join(self.datadir, self.INDEX_FILENAME_TTL) |
|
102
|
|
|
if os.path.exists(path): |
|
103
|
|
|
logger.debug(f"loading graph from {path}") |
|
104
|
|
|
self._skos_vocab = annif.corpus.SubjectFileSKOS(path) |
|
105
|
|
|
# store the dump file so we can use it next time |
|
106
|
|
|
self._skos_vocab.save_skos(path) |
|
107
|
|
|
return self._skos_vocab |
|
108
|
|
|
|
|
109
|
|
|
raise NotInitializedException(f"graph file {path} not found") |
|
110
|
|
|
|
|
111
|
|
|
def __len__(self) -> int: |
|
112
|
|
|
return len(self.subjects) |
|
113
|
|
|
|
|
114
|
|
|
@property |
|
115
|
|
|
def languages(self) -> list[str]: |
|
116
|
|
|
try: |
|
117
|
|
|
return self.subjects.languages |
|
118
|
|
|
except NotInitializedException: |
|
119
|
|
|
return [] |
|
120
|
|
|
|
|
121
|
|
|
def load_vocabulary( |
|
122
|
|
|
self, |
|
123
|
|
|
subject_corpus: SubjectCorpus, |
|
124
|
|
|
force: bool = False, |
|
125
|
|
|
) -> None: |
|
126
|
|
|
"""Load subjects from a subject corpus and save them into one |
|
127
|
|
|
or more subject index files as well as a SKOS/Turtle file for later |
|
128
|
|
|
use. If force=True, replace the existing subject index completely.""" |
|
129
|
|
|
|
|
130
|
|
|
if not force and os.path.exists( |
|
131
|
|
|
os.path.join(self.datadir, self.INDEX_FILENAME_CSV) |
|
132
|
|
|
): |
|
133
|
|
|
logger.info("updating existing subject index") |
|
134
|
|
|
self._subjects = self._update_subject_index(subject_corpus) |
|
135
|
|
|
else: |
|
136
|
|
|
logger.info("creating subject index") |
|
137
|
|
|
self._subjects = self._create_subject_index(subject_corpus) |
|
138
|
|
|
|
|
139
|
|
|
skosfile = os.path.join(self.datadir, self.INDEX_FILENAME_TTL) |
|
140
|
|
|
logger.info(f"saving vocabulary into SKOS file {skosfile}") |
|
141
|
|
|
subject_corpus.save_skos(skosfile) |
|
142
|
|
|
|
|
143
|
|
|
def as_graph(self) -> Graph: |
|
144
|
|
|
"""return the vocabulary as an rdflib graph""" |
|
145
|
|
|
return self.skos.graph |
|
146
|
|
|
|
|
147
|
|
|
def dump(self) -> dict[str, str | list | int | bool]: |
|
148
|
|
|
"""return this vocabulary as a dict""" |
|
149
|
|
|
|
|
150
|
|
|
try: |
|
151
|
|
|
languages = list(sorted(self.languages)) |
|
152
|
|
|
size = len(self) |
|
153
|
|
|
loaded = True |
|
154
|
|
|
except NotInitializedException: |
|
155
|
|
|
languages = [] |
|
156
|
|
|
size = None |
|
157
|
|
|
loaded = False |
|
158
|
|
|
|
|
159
|
|
|
return { |
|
160
|
|
|
"vocab_id": self.vocab_id, |
|
161
|
|
|
"languages": languages, |
|
162
|
|
|
"size": size, |
|
163
|
|
|
"loaded": loaded, |
|
164
|
|
|
} |
|
165
|
|
|
|
NatLibFi /
Annif
