refactor language filtering to a single function

project/dalia/curation/suggest/disciplines.py

@@ -3,14 +3,9 @@ from typing import List, Tuple
 from rdflib import Literal, OWL, RDF, SKOS, URIRef, Variable
 
 from project.dalia.api_models.api_models import CurationSuggestDisciplinesResultItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import (
-    FILTER,
-    FILTER_EXISTS,
-    FunctionExpressions,
-    Operators,
-    QueryBuilder,
-)
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import FILTER_EXISTS, QueryBuilder
+
 
 # Note: This algorithm is quite inefficient and results in a lot of SPARQL queries.
 # Another idea would be to pull all disciplines and their narrower/broader nodes at once, e.g. via
@@ -71,12 +66,7 @@ def _prepare_query_to_get_disciplines_metadata(discipline_selection_bgp: Tuple)
             state=False
         ),
         (var_discipline, SKOS.prefLabel, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
         (var_discipline, SKOS.notation, var_notation)
     ).ORDER_BY(
         var_notation

project/dalia/curation/suggest/languages.py

@@ -8,16 +8,8 @@ from project.dalia.api_models.api_models import (
     CurationSuggestSearchRequest,
     LabelValueItem,
 )
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import (
-    Aggregates,
-    BIND,
-    FILTER,
-    FunctionExpressions,
-    OPTIONAL,
-    Operators,
-    QueryBuilder,
-)
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import Aggregates, BIND, FunctionExpressions, OPTIONAL, QueryBuilder
 from project.dalia.rdf.namespace import Jena_text, SKOS_last_call, lvont
 
 
@@ -47,12 +39,7 @@ def _where_for_text_search(query: str, var_lang: Variable, var_score: Variable,
 
     where = [
         (subject_list_for_text_search, Jena_text.query, object_list_for_text_search),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ]
 
     return tuple(where)

project/dalia/curation/suggest/learning_resource_types.py

 from typing import List
 
-from rdflib import Literal, RDFS, SKOS, Variable
+from rdflib import RDFS, SKOS, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, GROUP, Operators, QueryBuilder, UNION
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import FILTER, GROUP, Operators, QueryBuilder, UNION
 from project.dalia.rdf.namespace import MoDalia, hcrt
 
 
@@ -54,12 +54,7 @@ def prepare_query_to_get_learning_resource_types() -> str:
             ),
         ).build(),
         (var_lrt, SKOS.prefLabel, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ).ORDER_BY(
         var_label
     ).build()

project/dalia/curation/suggest/media_types.py

 from typing import List
 
-from rdflib import Literal, RDFS, SKOS, Variable
+from rdflib import RDFS, SKOS, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators, QueryBuilder
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import QueryBuilder
 from project.dalia.rdf.namespace import MoDalia
 
 
@@ -30,12 +30,7 @@ def prepare_query_to_get_media_types() -> str:
     ).WHERE(
         (var_type, RDFS.subClassOf, MoDalia.MediaType),
         (var_type, SKOS.prefLabel, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ).ORDER_BY(
         var_label
     ).build()

project/dalia/curation/suggest/proficiency_levels.py

 from typing import List
 
-from rdflib import Literal, RDF, RDFS, Variable
+from rdflib import RDF, RDFS, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators, QueryBuilder
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import QueryBuilder
 from project.dalia.rdf.namespace import MoDalia
 
 
@@ -32,12 +32,7 @@ def prepare_query_to_get_proficiency_levels() -> str:
         (var_level, RDF.type, MoDalia.Proficiency),
         (var_level, MoDalia.hasOrder, var_level_order),
         (var_level, RDFS.label, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ).ORDER_BY(
         var_level_order
     ).build()

project/dalia/curation/suggest/relation_types.py

 from typing import List
 
-from rdflib import Literal, RDFS, Variable
+from rdflib import RDFS, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators, QueryBuilder
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import QueryBuilder
 from project.dalia.rdf.namespace import MoDalia
 
 
@@ -30,12 +30,7 @@ def prepare_query_to_get_relation_types() -> str:
     ).WHERE(
         (var_type, RDFS.subPropertyOf, MoDalia.isRelatedTo),
         (var_type, RDFS.label, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ).ORDER_BY(
         var_label
     ).build()

project/dalia/curation/suggest/target_groups.py

 from typing import List
 
-from rdflib import Literal, RDFS, SKOS, Variable
+from rdflib import RDFS, SKOS, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
-from project.dalia.query.utils import query_ontologies_dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators, QueryBuilder
+from project.dalia.query.utils import filter_by_lang, query_ontologies_dataset
+from project.dalia.query_builder.query_builder import QueryBuilder
 from project.dalia.rdf.namespace import MoDalia
 
 
@@ -30,12 +30,7 @@ def prepare_query_to_get_target_groups() -> str:
     ).WHERE(
         (var_tg, RDFS.subClassOf, MoDalia.TargetGroup),
         (var_tg, SKOS.prefLabel, var_label),
-        FILTER(
-            Operators.EQ(
-                FunctionExpressions.LANG(var_label),
-                Literal("en")
-            )
-        ),
+        filter_by_lang(var_label),
     ).ORDER_BY(
         var_label
     ).build()

project/dalia/query/communities/one_to_one_metadata.py

 from typing import Dict, Iterable
 
 from django.utils.text import slugify
-from rdflib import DCTERMS, Literal, RDF, URIRef, Variable
+from rdflib import DCTERMS, RDF, URIRef, Variable
 
 from project.dalia.api_models.api_models import Community, SocialMedia
-from project.dalia.query.utils import query_dalia_dataset
-from project.dalia.query_builder.query_builder import (
-    FILTER,
-    FunctionExpressions,
-    OPTIONAL,
-    Operators,
-    QueryBuilder,
-    VALUES
-)
+from project.dalia.query.utils import filter_by_lang, query_dalia_dataset
+from project.dalia.query_builder.query_builder import OPTIONAL, QueryBuilder, VALUES
 from project.dalia.rdf.namespace import MoDalia, SCHEMA, wdt
 
 
@@ -109,12 +102,7 @@ def prepare_query_for_one_to_one_metadata_for_communities(community_uri_refs: It
         (var_community, RDF.type, MoDalia.Community),
         OPTIONAL(
             (var_community, DCTERMS.description, var_description),
-            FILTER(
-                Operators.EQ(
-                    FunctionExpressions.LANG(var_description),
-                    Literal("en")
-                )
-            )
+            filter_by_lang(var_description),
         ),
         OPTIONAL((var_community, DCTERMS.title, _VARIABLES["title"])),
         OPTIONAL((var_community, wdt.Bluesky_handle, _VARIABLES["bluesky_handle"])),

project/dalia/query/items/metadata/disciplines.py

 from typing import Collection, Dict, List
 
 from rdflib import SKOS, URIRef
-from rdflib.term import Literal, Node
+from rdflib.term import Node
 
 from project.dalia.api_models.api_models import LabelValueItem
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
 from project.dalia.query.labels.utils import remap_to_label_value_item
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import fabio
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/items/metadata/languages.py

 from typing import Collection, Dict, List
 
 from rdflib import DCTERMS, URIRef
-from rdflib.term import Literal, Node
+from rdflib.term import Node
 
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import SKOS_last_call
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS_last_call.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/items/metadata/learning_resource_types.py

 from typing import Collection, Dict, List
 
 from rdflib import SKOS, URIRef
-from rdflib.term import Literal, Node
+from rdflib.term import Node
 
 from project.dalia.api_models.api_models import LabelValueItem
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
 from project.dalia.query.labels.utils import remap_to_label_value_item
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import MoDalia
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/items/metadata/media_types.py

 from typing import Collection, Dict, List
 
 from rdflib import SKOS, URIRef
-from rdflib.term import Literal, Node
+from rdflib.term import Node
 
 from project.dalia.api_models.api_models import LabelValueItem
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
 from project.dalia.query.labels.utils import remap_to_label_value_item
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import MoDalia
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/items/metadata/proficiency_levels.py

 from typing import Collection, Dict, List
 
 from rdflib import RDFS, URIRef
-from rdflib.term import Literal, Node, Variable
+from rdflib.term import Node, Variable
 
 from project.dalia.api_models.api_models import LabelValueItem
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
 from project.dalia.query.labels.utils import remap_to_label_value_item
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import MoDalia
 
 _VAR_LEVEL_ORDER = Variable("levelOrder")
@@ -16,12 +15,7 @@ _VAR_LEVEL_ORDER = Variable("levelOrder")
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, RDFS.label, VAR_LABEL),
     (VAR_ITEM, MoDalia.hasOrder, _VAR_LEVEL_ORDER),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/items/metadata/target_groups.py

 from typing import Collection, Dict, List
 
 from rdflib import SKOS, URIRef
-from rdflib.term import Literal, Node
+from rdflib.term import Node
 
 from project.dalia.api_models.api_models import LabelValueItem
 from project.dalia.query.items.metadata.one_to_many_metadata import get_one_to_many_metadata_for_resources
 from project.dalia.query.labels.label_service import VAR_ITEM, VAR_LABEL, get_labels_for_item_uris
 from project.dalia.query.labels.utils import remap_to_label_value_item
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
 from project.dalia.rdf.namespace import MoDalia
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )
 
 

project/dalia/query/utils.py

 from enum import Enum
 
 from django.conf import settings
-from rdflib import URIRef
+from rdflib import Literal, Variable
 from rdflib.plugins.stores.sparqlstore import SPARQLStore
 from rdflib.query import Result
 
+from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+
 
 class Dataset(Enum):
     """
@@ -36,3 +38,12 @@ def _get_triplestore_endpoint_from_settings() -> str:
     if not endpoint.endswith("/"):
         endpoint += "/"
     return endpoint
+
+
+def filter_by_lang(var: Variable, lang: str = "en") -> tuple[FILTER]:
+    return FILTER(
+        Operators.EQ(
+            FunctionExpressions.LANG(var),
+            Literal(lang)
+        )
+    )

tests/project/dalia/query/labels/test_label_service.py

 from collections import OrderedDict
 
-from rdflib import Literal, SKOS, URIRef
+from rdflib import SKOS, URIRef
 
 from project.dalia.query.labels.label_service import (
     VAR_ITEM,
@@ -8,19 +8,14 @@ from project.dalia.query.labels.label_service import (
     get_labels_for_item_uris,
     prepare_query_for_labels_from_items,
 )
-from project.dalia.query.utils import Dataset
-from project.dalia.query_builder.query_builder import FILTER, FunctionExpressions, Operators
+from project.dalia.query.utils import Dataset, filter_by_lang
+from project.dalia.query_builder.query_builder import FunctionExpressions
 from project.dalia.rdf.namespace import SCHEMA
 from tests.project.dalia.utils import dedent_and_normalize, normalize
 
 _LABEL_GRAPH_PATTERN = (
     (VAR_ITEM, SKOS.prefLabel, VAR_LABEL),
-    FILTER(
-        Operators.EQ(
-            FunctionExpressions.LANG(VAR_LABEL),
-            Literal("en")
-        )
-    )
+    filter_by_lang(VAR_LABEL),
 )