|
1
|
|
|
"""Support for exclude/include rules for subject vocabularies""" |
|
2
|
|
|
|
|
3
|
|
|
from rdflib import RDF, Graph, URIRef |
|
4
|
|
|
from rdflib.namespace import SKOS |
|
5
|
|
|
|
|
6
|
|
|
import annif |
|
7
|
|
|
from annif.exception import ConfigurationException |
|
8
|
|
|
|
|
9
|
|
|
from .vocab import AnnifVocabulary |
|
10
|
|
|
|
|
11
|
|
|
logger = annif.logger |
|
12
|
|
|
|
|
13
|
|
|
|
|
14
|
|
|
def resolve_uri_or_curie(graph: Graph, value: str) -> URIRef: |
|
15
|
|
|
try: |
|
16
|
|
|
# Try to expand as CURIE using the graph's namespace manager |
|
17
|
|
|
return graph.namespace_manager.expand_curie(value) |
|
18
|
|
|
except ValueError: |
|
19
|
|
|
# Not a CURIE or prefix not defined; treat as full URI |
|
20
|
|
|
return URIRef(value) |
|
21
|
|
|
|
|
22
|
|
|
|
|
23
|
|
|
def uris_by_type(graph: Graph, type_: str, action: str) -> list[str]: |
|
24
|
|
|
type_uri = resolve_uri_or_curie(graph, type_) |
|
25
|
|
|
uris = [str(uri) for uri in graph.subjects(RDF.type, type_uri)] |
|
26
|
|
|
if not uris: |
|
27
|
|
|
logger.warning(f"{action}: no concepts found with type {type_uri}") |
|
28
|
|
|
return uris |
|
29
|
|
|
|
|
30
|
|
|
|
|
31
|
|
|
def uris_by_scheme(graph: Graph, scheme: str, action: str) -> list[str]: |
|
32
|
|
|
scheme_uri = resolve_uri_or_curie(graph, scheme) |
|
33
|
|
|
uris = [str(uri) for uri in graph.subjects(SKOS.inScheme, scheme_uri)] |
|
34
|
|
|
if not uris: |
|
35
|
|
|
logger.warning(f"{action}: no concepts found in scheme {scheme_uri}") |
|
36
|
|
|
return uris |
|
37
|
|
|
|
|
38
|
|
|
|
|
39
|
|
|
def uris_by_collection(graph: Graph, collection: str, action: str) -> list[str]: |
|
40
|
|
|
collection_uri = resolve_uri_or_curie(graph, collection) |
|
41
|
|
|
uris = [str(uri) for uri in graph.objects(collection_uri, SKOS.member)] |
|
42
|
|
|
if not uris: |
|
43
|
|
|
logger.warning(f"{action}: no concepts found in collection {collection_uri}") |
|
44
|
|
|
return uris |
|
45
|
|
|
|
|
46
|
|
|
|
|
47
|
|
|
def add_uris( |
|
48
|
|
|
graph: Graph, uris_func: callable, uris_set: set[str], vals: list[str], action: str |
|
49
|
|
|
) -> None: |
|
50
|
|
|
for val in vals: |
|
51
|
|
|
uris_set.update(uris_func(graph, val, action)) |
|
52
|
|
|
|
|
53
|
|
|
|
|
54
|
|
|
def remove_uris( |
|
55
|
|
|
graph: Graph, uris_func: callable, uris_set: set[str], vals: list[str], action: str |
|
56
|
|
|
) -> None: |
|
57
|
|
|
for val in vals: |
|
58
|
|
|
for uri in uris_func(graph, val, action): |
|
59
|
|
|
uris_set.discard(uri) |
|
60
|
|
|
|
|
61
|
|
|
|
|
62
|
|
|
def kwargs_to_exclude_uris(vocab: AnnifVocabulary, kwargs: dict[str, str]) -> set[str]: |
|
63
|
|
|
exclude_uris = set() |
|
64
|
|
|
actions = { |
|
65
|
|
|
"exclude": lambda vals: exclude_uris.update( |
|
66
|
|
|
vals |
|
67
|
|
|
if "*" not in vals |
|
68
|
|
|
else uris_by_type(vocab.as_graph(), "skos:Concept", "exclude") |
|
69
|
|
|
), |
|
70
|
|
|
"exclude_type": lambda vals: add_uris( |
|
71
|
|
|
vocab.as_graph(), uris_by_type, exclude_uris, vals, "exclude_type" |
|
72
|
|
|
), |
|
73
|
|
|
"exclude_scheme": lambda vals: add_uris( |
|
74
|
|
|
vocab.as_graph(), uris_by_scheme, exclude_uris, vals, "exclude_scheme" |
|
75
|
|
|
), |
|
76
|
|
|
"exclude_collection": lambda vals: add_uris( |
|
77
|
|
|
vocab.as_graph(), |
|
78
|
|
|
uris_by_collection, |
|
79
|
|
|
exclude_uris, |
|
80
|
|
|
vals, |
|
81
|
|
|
"exclude_collection", |
|
82
|
|
|
), |
|
83
|
|
|
"include": lambda vals: exclude_uris.difference_update(vals), |
|
84
|
|
|
"include_type": lambda vals: remove_uris( |
|
85
|
|
|
vocab.as_graph(), uris_by_type, exclude_uris, vals, "include_type" |
|
86
|
|
|
), |
|
87
|
|
|
"include_scheme": lambda vals: remove_uris( |
|
88
|
|
|
vocab.as_graph(), uris_by_scheme, exclude_uris, vals, "include_scheme" |
|
89
|
|
|
), |
|
90
|
|
|
"include_collection": lambda vals: remove_uris( |
|
91
|
|
|
vocab.as_graph(), |
|
92
|
|
|
uris_by_collection, |
|
93
|
|
|
exclude_uris, |
|
94
|
|
|
vals, |
|
95
|
|
|
"include_collection", |
|
96
|
|
|
), |
|
97
|
|
|
} |
|
98
|
|
|
|
|
99
|
|
|
for key, value in kwargs.items(): |
|
100
|
|
|
vals = value.split("|") |
|
101
|
|
|
if key in actions: |
|
102
|
|
|
actions[key](vals) |
|
103
|
|
|
else: |
|
104
|
|
|
raise ConfigurationException(f"unknown vocab keyword argument {key}") |
|
105
|
|
|
|
|
106
|
|
|
return exclude_uris |
|
107
|
|
|
|
NatLibFi /
Annif
