|
1
|
|
|
"""Vocabulary management functionality for Annif""" |
|
2
|
|
|
|
|
3
|
|
|
import os.path |
|
4
|
|
|
import annif |
|
5
|
|
|
import annif.corpus |
|
6
|
|
|
import annif.util |
|
7
|
|
|
from annif.datadir import DatadirMixin |
|
8
|
|
|
from annif.exception import NotInitializedException |
|
9
|
|
|
|
|
10
|
|
|
logger = annif.logger |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
class AnnifVocabulary(DatadirMixin): |
|
14
|
|
|
"""Class representing a subject vocabulary which can be used by multiple |
|
15
|
|
|
Annif projects.""" |
|
16
|
|
|
|
|
17
|
|
|
# defaults for uninitialized instances |
|
18
|
|
|
_subjects = None |
|
19
|
|
|
|
|
20
|
|
|
def __init__(self, vocab_id, datadir, language): |
|
21
|
|
|
DatadirMixin.__init__(self, datadir, 'vocabs', vocab_id) |
|
22
|
|
|
self.vocab_id = vocab_id |
|
23
|
|
|
self.language = language |
|
24
|
|
|
self._skos_vocab = None |
|
25
|
|
|
|
|
26
|
|
|
def _create_subject_index(self, subject_corpus): |
|
27
|
|
|
self._subjects = annif.corpus.SubjectIndex(subject_corpus) |
|
28
|
|
|
annif.util.atomic_save(self._subjects, self.datadir, 'subjects') |
|
29
|
|
|
|
|
30
|
|
|
def _update_subject_index(self, subject_corpus): |
|
31
|
|
|
old_subjects = self.subjects |
|
32
|
|
|
new_subjects = annif.corpus.SubjectIndex(subject_corpus) |
|
33
|
|
|
updated_subjects = annif.corpus.SubjectIndex() |
|
34
|
|
|
|
|
35
|
|
|
for uri, label, notation in old_subjects: |
|
36
|
|
|
if new_subjects.contains_uri(uri): |
|
37
|
|
|
label, notation = new_subjects[new_subjects.by_uri(uri)][1:3] |
|
38
|
|
|
else: # subject removed from new corpus |
|
39
|
|
|
label, notation = None, None |
|
40
|
|
|
updated_subjects.append(uri, label, notation) |
|
41
|
|
|
for uri, label, notation in new_subjects: |
|
42
|
|
|
if not old_subjects.contains_uri(uri): |
|
43
|
|
|
updated_subjects.append(uri, label, notation) |
|
44
|
|
|
self._subjects = updated_subjects |
|
45
|
|
|
annif.util.atomic_save(self._subjects, self.datadir, 'subjects') |
|
46
|
|
|
|
|
47
|
|
|
@property |
|
48
|
|
|
def subjects(self): |
|
49
|
|
|
if self._subjects is None: |
|
50
|
|
|
path = os.path.join(self.datadir, 'subjects') |
|
51
|
|
|
if os.path.exists(path): |
|
52
|
|
|
logger.debug('loading subjects from %s', path) |
|
53
|
|
|
self._subjects = annif.corpus.SubjectIndex.load(path) |
|
54
|
|
|
else: |
|
55
|
|
|
raise NotInitializedException( |
|
56
|
|
|
"subject file {} not found".format(path)) |
|
57
|
|
|
return self._subjects |
|
58
|
|
|
|
|
59
|
|
|
@property |
|
60
|
|
|
def skos(self): |
|
61
|
|
|
"""return the subject vocabulary from SKOS file""" |
|
62
|
|
|
if self._skos_vocab is not None: |
|
63
|
|
|
return self._skos_vocab |
|
64
|
|
|
|
|
65
|
|
|
# attempt to load graph from dump file |
|
66
|
|
|
dumppath = os.path.join(self.datadir, 'subjects.dump.gz') |
|
67
|
|
|
if os.path.exists(dumppath): |
|
68
|
|
|
logger.debug(f'loading graph dump from {dumppath}') |
|
69
|
|
|
try: |
|
70
|
|
|
self._skos_vocab = annif.corpus.SubjectFileSKOS(dumppath, |
|
71
|
|
|
self.language) |
|
72
|
|
|
except ModuleNotFoundError: |
|
73
|
|
|
# Probably dump has been saved using a different rdflib version |
|
74
|
|
|
logger.debug('could not load graph dump, using turtle file') |
|
75
|
|
|
else: |
|
76
|
|
|
return self._skos_vocab |
|
77
|
|
|
|
|
78
|
|
|
# graph dump file not found - parse ttl file instead |
|
79
|
|
|
path = os.path.join(self.datadir, 'subjects.ttl') |
|
80
|
|
|
if os.path.exists(path): |
|
81
|
|
|
logger.debug(f'loading graph from {path}') |
|
82
|
|
|
self._skos_vocab = annif.corpus.SubjectFileSKOS(path, |
|
83
|
|
|
self.language) |
|
84
|
|
|
# store the dump file so we can use it next time |
|
85
|
|
|
self._skos_vocab.save_skos(path, self.language) |
|
86
|
|
|
return self._skos_vocab |
|
87
|
|
|
|
|
88
|
|
|
raise NotInitializedException(f'graph file {path} not found') |
|
89
|
|
|
|
|
90
|
|
|
def load_vocabulary(self, subject_corpus, language): |
|
91
|
|
|
"""load subjects from a subject corpus and save them into a |
|
92
|
|
|
SKOS/Turtle file for later use""" |
|
93
|
|
|
|
|
94
|
|
|
if os.path.exists(os.path.join(self.datadir, 'subjects')): |
|
95
|
|
|
logger.info('updating existing vocabulary') |
|
96
|
|
|
self._update_subject_index(subject_corpus) |
|
97
|
|
|
else: |
|
98
|
|
|
self._create_subject_index(subject_corpus) |
|
99
|
|
|
subject_corpus.save_skos(os.path.join(self.datadir, 'subjects.ttl'), |
|
100
|
|
|
language) |
|
101
|
|
|
|
|
102
|
|
|
def as_skos_file(self): |
|
103
|
|
|
"""return the vocabulary as a file object, in SKOS/Turtle syntax""" |
|
104
|
|
|
return open(os.path.join(self.datadir, 'subjects.ttl'), 'rb') |
|
105
|
|
|
|
|
106
|
|
|
def as_graph(self): |
|
107
|
|
|
"""return the vocabulary as an rdflib graph""" |
|
108
|
|
|
return self.skos.graph |
|
109
|
|
|
|
NatLibFi /
Annif
