Skip to content
Snippets Groups Projects
Commit d8ebb2ee authored by Frank Lange's avatar Frank Lange
Browse files

generalize the way 1:N metadata of learning resources is retrieved 

apply this pattern to: keywords, languages, learning resource types and media types
parent 385a1f8d
Branches
No related tags found
No related merge requests found
Pipeline #1545898 passed
Stage: container
Hide whitespace changes
Inline Side-by-side
Showing
with 132 additions and 250 deletions
project/dalia/query/items/metadata/keywords.py
+ 11
− 38
View file @ d8ebb2ee
from collections import defaultdict
from typing import Dict, List
from rdflib import RDF, URIRef, Variable
from rdflib import URIRef
from rdflib.term import Node
from project.dalia.query.utils import query_dalia_dataset
from project.dalia.query_builder.query_builder import QueryBuilder, VALUES
from project.dalia.rdf.namespace import SCHEMA, educor
from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
from project.dalia.rdf.namespace import SCHEMA
_VARIABLES = {
"lr": Variable("lr"),
"keyword": Variable("keyword")
}
def prepare_query_for_keywords_metadata_for_resources(resource_uri_refs: List[URIRef]) -> str:
var_lr = _VARIABLES["lr"]
resource_uri_ref_blocks = [[uri_ref] for uri_ref in resource_uri_refs]
return QueryBuilder().SELECT(
*_VARIABLES.values()
).WHERE(
VALUES(
[var_lr],
resource_uri_ref_blocks
),
(var_lr, RDF.type, educor.EducationalResource),
(var_lr, SCHEMA.keywords, _VARIABLES["keyword"])
).build()
def keywords_from_results(results) -> Dict[URIRef, List[str]]:
lr_keywords = defaultdict(list)
for result in results:
lr_keywords[result.lr].append(str(result.keyword))
return lr_keywords
def _keyword_from_result_item(item: Node) -> str:
return str(item)
def get_keywords_metadata_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRef, List[str]]:
......@@ -46,7 +18,8 @@ def get_keywords_metadata_for_resources(resource_uri_refs: List[URIRef]) -> Dict
:param resource_uri_refs: List of learning resource URIRefs
:return: Associations between the learning resource URIRefs and their respective list of keywords.
"""
query = prepare_query_for_keywords_metadata_for_resources(resource_uri_refs)
results = query_dalia_dataset(query)
return keywords_from_results(results)
return get_one_to_many_metadata_for_resources(
resource_uri_refs=resource_uri_refs,
relation=SCHEMA.keywords,
item_crosswalk_fn=_keyword_from_result_item
)
project/dalia/query/items/metadata/languages.py
+ 10
− 39
View file @ d8ebb2ee
from collections import defaultdict
from typing import Dict, List
from rdflib import DCTERMS, Literal, RDF, URIRef, Variable
from rdflib import DCTERMS, URIRef
from rdflib.term import Node
from project.dalia.query.items.facets.facet_objects import LANGUAGE_FACET
from project.dalia.query.utils import query_dalia_dataset
from project.dalia.query_builder.query_builder import QueryBuilder, VALUES
from project.dalia.rdf.namespace import educor
_VARIABLES = {
"lr": Variable("lr"),
"language": Variable("language")
}
def prepare_query_for_languages_for_resources(resource_uri_refs: List[URIRef]) -> str:
var_lr = _VARIABLES["lr"]
resource_uri_ref_blocks = [[uri_ref] for uri_ref in resource_uri_refs]
return QueryBuilder().SELECT(
*_VARIABLES.values()
).WHERE(
VALUES(
[var_lr],
resource_uri_ref_blocks
),
(var_lr, RDF.type, educor.EducationalResource),
(var_lr, DCTERMS.language, _VARIABLES["language"])
).build()
from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
LANGUAGE_LABEL_MAPPING = LANGUAGE_FACET.items
def languages_from_results(results) -> Dict[URIRef, List[str]]:
lr_languages = defaultdict(list)
for result in results:
lr_languages[result.lr].append(LANGUAGE_LABEL_MAPPING[result.language])
return lr_languages
def _language_from_result_item(item: Node) -> str:
return LANGUAGE_LABEL_MAPPING[item]
def get_languages_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRef, List[str]]:
......@@ -50,7 +20,8 @@ def get_languages_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRef,
:param resource_uri_refs: List of learning resource URIRefs
:return: Associations between the learning resource URIRefs and their respective list of languages.
"""
query = prepare_query_for_languages_for_resources(resource_uri_refs)
results = query_dalia_dataset(query)
return languages_from_results(results)
return get_one_to_many_metadata_for_resources(
resource_uri_refs=resource_uri_refs,
relation=DCTERMS.language,
item_crosswalk_fn=_language_from_result_item
)
project/dalia/query/items/metadata/learning_resource_types.py
+ 14
− 44
View file @ d8ebb2ee
from collections import defaultdict
from typing import Dict, List
from rdflib import RDF, URIRef, Variable
from rdflib import URIRef
from rdflib.term import Node
from project.dalia.api_models.api_models import LabelValueItem
from project.dalia.query.items.facets.facet_objects import LEARNING_RESOURCE_TYPE_FACET
from project.dalia.query.utils import query_dalia_dataset
from project.dalia.query_builder.query_builder import QueryBuilder, VALUES
from project.dalia.rdf.namespace import MoDalia, educor
_VARIABLES = {
"lr": Variable("lr"),
"learning_resource_type": Variable("learning_resource_type")
}
def prepare_query_for_learning_resource_types_for_resources(resource_uri_refs: List[URIRef]) -> str:
var_lr = _VARIABLES["lr"]
resource_uri_ref_blocks = [[uri_ref] for uri_ref in resource_uri_refs]
return QueryBuilder().SELECT(
*_VARIABLES.values()
).WHERE(
VALUES(
[var_lr],
resource_uri_ref_blocks
),
(var_lr, RDF.type, educor.EducationalResource),
(var_lr, MoDalia.hasLearningType, _VARIABLES["learning_resource_type"])
).build()
from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
from project.dalia.rdf.namespace import MoDalia
LEARNING_RESOURCE_TYPES_LABEL_MAPPING = LEARNING_RESOURCE_TYPE_FACET.items
def learning_resource_types_from_results(results) -> Dict[URIRef, List[LabelValueItem]]:
lr_learning_resource_types = defaultdict(list)
for result in results:
lr_learning_resource_types[result.lr].append(
LabelValueItem(
label=LEARNING_RESOURCE_TYPES_LABEL_MAPPING[result.learning_resource_type],
value=str(result.learning_resource_type)
)
)
return lr_learning_resource_types
def _learning_resource_type_from_result_item(item: Node) -> LabelValueItem:
return LabelValueItem(
label=LEARNING_RESOURCE_TYPES_LABEL_MAPPING[item],
value=str(item)
)
def get_learning_resource_types_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRef, List[LabelValueItem]]:
......@@ -56,7 +25,8 @@ def get_learning_resource_types_for_resources(resource_uri_refs: List[URIRef]) -
:param resource_uri_refs: List of learning resource URIRefs
:return: Associations between the learning resource URIRefs and their respective list of learning resource types.
"""
query = prepare_query_for_learning_resource_types_for_resources(resource_uri_refs)
results = query_dalia_dataset(query)
return learning_resource_types_from_results(results)
return get_one_to_many_metadata_for_resources(
resource_uri_refs=resource_uri_refs,
relation=MoDalia.hasLearningType,
item_crosswalk_fn=_learning_resource_type_from_result_item
)
project/dalia/query/items/metadata/media_types.py
+ 14
− 44
View file @ d8ebb2ee
from collections import defaultdict
from typing import Dict, List
from rdflib import RDF, URIRef, Variable
from rdflib import URIRef
from rdflib.term import Node
from project.dalia.api_models.api_models import LabelValueItem
from project.dalia.query.items.facets.facet_objects import MEDIA_TYPE_FACET
from project.dalia.query.utils import query_dalia_dataset
from project.dalia.query_builder.query_builder import QueryBuilder, VALUES
from project.dalia.rdf.namespace import MoDalia, educor
_VARIABLES = {
"lr": Variable("lr"),
"media_type": Variable("media_type")
}
def prepare_query_for_media_types_for_resources(resource_uri_refs: List[URIRef]) -> str:
var_lr = _VARIABLES["lr"]
resource_uri_ref_blocks = [[uri_ref] for uri_ref in resource_uri_refs]
return QueryBuilder().SELECT(
*_VARIABLES.values()
).WHERE(
VALUES(
[var_lr],
resource_uri_ref_blocks
),
(var_lr, RDF.type, educor.EducationalResource),
(var_lr, MoDalia.hasMediaType, _VARIABLES["media_type"])
).build()
from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
from project.dalia.rdf.namespace import MoDalia
MEDIA_TYPES_LABEL_MAPPING = MEDIA_TYPE_FACET.items
def media_types_from_results(results) -> Dict[URIRef, List[LabelValueItem]]:
lr_media_types = defaultdict(list)
for result in results:
lr_media_types[result.lr].append(
LabelValueItem(
label=MEDIA_TYPES_LABEL_MAPPING[result.media_type],
value=str(result.media_type)
)
)
return lr_media_types
def _media_type_from_result_item(item: Node) -> LabelValueItem:
return LabelValueItem(
label=MEDIA_TYPES_LABEL_MAPPING[item],
value=str(item)
)
def get_media_types_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRef, List[LabelValueItem]]:
......@@ -56,7 +25,8 @@ def get_media_types_for_resources(resource_uri_refs: List[URIRef]) -> Dict[URIRe
:param resource_uri_refs: List of learning resource URIRefs
:return: Associations between the learning resource URIRefs and their respective list of media types.
"""
query = prepare_query_for_media_types_for_resources(resource_uri_refs)
results = query_dalia_dataset(query)
return media_types_from_results(results)
return get_one_to_many_metadata_for_resources(
resource_uri_refs=resource_uri_refs,
relation=MoDalia.hasMediaType,
item_crosswalk_fn=_media_type_from_result_item
)
project/dalia/query/items/metadata/one_to_many_metadata.py 0 → 100644
+ 58
− 0
View file @ d8ebb2ee
from collections import defaultdict
from typing import Any, Callable, Dict, List
from rdflib import RDF, URIRef, Variable
from rdflib.term import Node
from project.dalia.query.utils import query_dalia_dataset
from project.dalia.query_builder.query_builder import QueryBuilder, VALUES
from project.dalia.rdf.namespace import educor
_VARIABLES = {
"lr": Variable("lr"),
"item": Variable("item")
}
def prepare_query_for_one_to_many_metadata_for_resources(resource_uri_refs: List[URIRef], relation: URIRef) -> str:
var_lr = _VARIABLES["lr"]
resource_uri_ref_blocks = [[uri_ref] for uri_ref in resource_uri_refs]
return QueryBuilder().SELECT(
*_VARIABLES.values()
).WHERE(
VALUES(
[var_lr],
resource_uri_ref_blocks
),
(var_lr, RDF.type, educor.EducationalResource),
(var_lr, relation, _VARIABLES["item"])
).build()
def _metadata_from_results(results, item_crosswalk_fn: Callable[[Node], Any]) -> Dict[URIRef, List[str]]:
lr_metadata_items = defaultdict(list)
for result in results:
lr_metadata_items[result.lr].append(item_crosswalk_fn(result.item))
return lr_metadata_items
def get_one_to_many_metadata_for_resources(
resource_uri_refs: List[URIRef],
relation: URIRef,
item_crosswalk_fn: Callable[[Node], Any],
) -> Dict[URIRef, List[Any]]:
"""
Retrieve the one-to-many metadata for each of the given learning resource URIRefs.
:param resource_uri_refs: List of learning resource URIRefs
:param relation: relation between a learning resource and the metadata items (RDF property)
:param item_crosswalk_fn: function to generate a Python object from an item received by the SPARQL call
:return: Associations between the learning resource URIRefs and their respective list of metadata items.
"""
query = prepare_query_for_one_to_many_metadata_for_resources(resource_uri_refs, relation)
results = query_dalia_dataset(query)
return _metadata_from_results(results, item_crosswalk_fn)
tests/project/dalia/query/items/metadata/test_keywords.py
+ 2
− 22
View file @ d8ebb2ee
from rdflib import URIRef
from project.dalia.query.items.metadata.keywords import (
get_keywords_metadata_for_resources,
prepare_query_for_keywords_metadata_for_resources,
)
from tests.project.dalia.utils import dedent_and_normalize, normalize
def test_prepare_query_for_keywords_metadata_for_resources():
query = prepare_query_for_keywords_metadata_for_resources([URIRef("abc"), URIRef("def")])
assert normalize(query) == dedent_and_normalize("""
SELECT ?lr ?keyword
WHERE {
VALUES ( ?lr ) {
( <abc> )
( <def> )
}
?lr <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://github.com/tibonto/educor#EducationalResource> .
?lr <https://schema.org/keywords> ?keyword .
}
""")
from project.dalia.query.items.metadata.keywords import get_keywords_metadata_for_resources
def test_get_keywords_metadata_for_resources(triplestore):
......@@ -33,10 +13,10 @@ def test_get_keywords_metadata_for_resources(triplestore):
assert lr_keywords == {
URIRef('https://id.dalia.education/learning-resource/20f0b586-8019-40d1-8f92-baa6ac168bca'): [
'Research Data Management',
'FAIR',
'I3D:bio',
'Microscopy',
'Research Data Management'
],
URIRef('https://id.dalia.education/learning-resource/1f794f99-c131-4fb6-b238-5750afa197c8'): [
'NFDI',
......
tests/project/dalia/query/items/metadata/test_languages.py
+ 1
− 21
View file @ d8ebb2ee
from rdflib import URIRef
from project.dalia.query.items.metadata.languages import (
get_languages_for_resources,
prepare_query_for_languages_for_resources,
)
from tests.project.dalia.utils import dedent_and_normalize, normalize
def test_prepare_query_for_languages_for_resources():
query = prepare_query_for_languages_for_resources([URIRef("abc"), URIRef("def")])
assert normalize(query) == dedent_and_normalize("""
SELECT ?lr ?language
WHERE {
VALUES ( ?lr ) {
( <abc> )
( <def> )
}
?lr <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://github.com/tibonto/educor#EducationalResource> .
?lr <http://purl.org/dc/terms/language> ?language .
}
""")
from project.dalia.query.items.metadata.languages import get_languages_for_resources
def test_get_languages_for_resources(triplestore):
......
tests/project/dalia/query/items/metadata/test_learning_resource_types.py
+ 1
− 21
View file @ d8ebb2ee
from rdflib import URIRef
from project.dalia.api_models.api_models import LabelValueItem
from project.dalia.query.items.metadata.learning_resource_types import (
get_learning_resource_types_for_resources,
prepare_query_for_learning_resource_types_for_resources,
)
from project.dalia.query.items.metadata.learning_resource_types import get_learning_resource_types_for_resources
from project.dalia.rdf.namespace import MoDalia
from tests.project.dalia.utils import dedent_and_normalize, normalize
def test_prepare_query_for_learning_resource_types_for_resources():
query = prepare_query_for_learning_resource_types_for_resources([URIRef("abc"), URIRef("def")])
assert normalize(query) == dedent_and_normalize("""
SELECT ?lr ?learning_resource_type
WHERE {
VALUES ( ?lr ) {
( <abc> )
( <def> )
}
?lr <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://github.com/tibonto/educor#EducationalResource> .
?lr <https://purl.org/ontology/modalia#hasLearningType> ?learning_resource_type .
}
""")
def test_get_learning_resource_types_for_resources(triplestore):
......
tests/project/dalia/query/items/metadata/test_media_types.py
+ 1
− 21
View file @ d8ebb2ee
from rdflib import URIRef
from project.dalia.api_models.api_models import LabelValueItem
from project.dalia.query.items.metadata.media_types import (
get_media_types_for_resources,
prepare_query_for_media_types_for_resources,
)
from project.dalia.query.items.metadata.media_types import get_media_types_for_resources
from project.dalia.rdf.namespace import SCHEMA
from tests.project.dalia.utils import dedent_and_normalize, normalize
def test_prepare_query_for_media_types_for_resources():
query = prepare_query_for_media_types_for_resources([URIRef("abc"), URIRef("def")])
assert normalize(query) == dedent_and_normalize("""
SELECT ?lr ?media_type
WHERE {
VALUES ( ?lr ) {
( <abc> )
( <def> )
}
?lr <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://github.com/tibonto/educor#EducationalResource> .
?lr <https://purl.org/ontology/modalia#hasMediaType> ?media_type .
}
""")
def test_get_media_types_for_resources(triplestore):
......
tests/project/dalia/query/items/metadata/test_one_to_many_metadata.py 0 → 100644
+ 20
− 0
View file @ d8ebb2ee
from rdflib import URIRef
from project.dalia.query.items.metadata.one_to_many_metadata import prepare_query_for_one_to_many_metadata_for_resources
from tests.project.dalia.utils import dedent_and_normalize, normalize
def test_prepare_query_for_one_to_many_metadata_for_resources():
query = prepare_query_for_one_to_many_metadata_for_resources([URIRef("abc"), URIRef("def")], URIRef("relation"))
assert normalize(query) == dedent_and_normalize("""
SELECT ?lr ?item
WHERE {
VALUES ( ?lr ) {
( <abc> )
( <def> )
}
?lr <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://github.com/tibonto/educor#EducationalResource> .
?lr <relation> ?item .
}
""")
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment