feat(utils): implement new type rdf importer

apis_core/apis_entities/abc.py

@@ -1,3 +1,5 @@
+from pathlib import Path
+
 from django.db import models
 
 #########################
@@ -23,6 +25,13 @@ class Meta:
     def __str__(self):
         return f"{self.forename} {self.surname}"
 
+    @classmethod
+    def rdf_configs(cls):
+        return [
+            Path(__file__).parent / "triple_configs/E21_PersonFromDNB.toml",
+            Path(__file__).parent / "triple_configs/E21_PersonFromWikidata.toml",
+        ]
+
 
 class E53_Place(models.Model):
     label = models.CharField(blank=True, default="", max_length=4096)
@@ -35,6 +44,14 @@ class Meta:
     def __str__(self):
         return self.label
 
+    @classmethod
+    def rdf_configs(cls):
+        return [
+            Path(__file__).parent / "triple_configs/E53_PlaceFromDNB.toml",
+            Path(__file__).parent / "triple_configs/E53_PlaceFromGeonames.toml",
+            Path(__file__).parent / "triple_configs/E53_PlaceFromWikidata.toml",
+        ]
+
 
 class E74_Group(models.Model):
     label = models.CharField(blank=True, default="", max_length=4096)
@@ -44,3 +61,10 @@ class Meta:
 
     def __str__(self):
         return self.label
+
+    @classmethod
+    def rdf_configs(cls):
+        return [
+            Path(__file__).parent / "triple_configs/E74_GroupFromDNB.toml",
+            Path(__file__).parent / "triple_configs/E74_GroupFromWikidata.toml",
+        ]

apis_core/apis_entities/triple_configs/E21_PersonFromDNB.toml

@@ -0,0 +1,13 @@
+[[filters]]
+"rdf:type" = "gndo:DifferentiatedPerson"
+
+[attributes]
+forename = ["gndo:forename", "gndo:preferredNameEntityForThePerson/gndo:forename"]
+alternative_names = "gndo:variantNameForThePerson"
+surname = ["gndo:surname", "gndo:preferredNameEntityForThePerson/gndo:surname"]
+start_date_written = "gndo:dateOfBirth"
+end_date_written = "gndo:dateOfDeath"
+same_as = "owl:sameAs"
+profession = "gndo:professionOrOccupation"
+
+relations = ["gndo:placeOfDeath", "gndo:placeOfBirth"]

apis_core/apis_entities/triple_configs/E21_PersonFromWikidata.toml

@@ -0,0 +1,12 @@
+[[filters]]
+"wdt:P31" = "wd:Q5"
+
+[attributes]
+forename = "wdt:P735/rdfs:label"
+surname = "wdt:P734/rdfs:label"
+date_of_birth = "wdt:P569"
+date_of_death = "wdt:P570"
+same_as = ["owl:sameAs", "wdtn:P227", "wdtn:P1566", "wdtn:P214", "wdtn:P244"]
+
+#TODO: 
+#relations = ["wd:P20", "wd:P19"]

apis_core/apis_entities/triple_configs/E53_PlaceFromDNB.toml

@@ -0,0 +1,21 @@
+[[filters]]
+"rdf:type" = "gndo:TerritorialCorporateBodyOrAdministrativeUnit"
+
+
+[attributes]
+label = "gndo:preferredNameForThePlaceOrGeographicName"
+longitude = '''
+SELECT ?longitude
+WHERE {
+  ?subject geo:hasGeometry/geo:asWKT ?point .
+  BIND(REPLACE(str(?point), "Point \\( \\+?(-?\\d+.\\d+).*", "$1") as ?longitude)
+}
+'''
+latitude = '''
+SELECT ?latitude
+WHERE {
+  ?subject geo:hasGeometry/geo:asWKT ?point .
+  BIND(REPLACE(str(?point), "^Point\\s*\\(\\s*[+-]?\\d+\\.\\d+\\s+([+-]?\\d+\\.\\d+)\\s*\\)$", "$1") as ?latitude)
+}
+'''
+same_as = "owl:sameAs"

apis_core/apis_entities/triple_configs/E53_PlaceFromGeonames.toml

@@ -0,0 +1,9 @@
+[[filters]]
+"rdf:type" = "gn:Feature"
+
+
+[attributes]
+label = ["gn:name", "gn:officialName", "gn:alternateName"]
+latitude = "wgs84_pos:lat"
+longitude = "wgs84_pos:long"
+same_as = ["rdfs:seeAlso", "gn:wikipediaArticle"]

apis_core/apis_entities/triple_configs/E53_PlaceFromWikidata.toml

@@ -0,0 +1,22 @@
+[[filters]]
+# metropolis (large and significant city or urban area usually with millions of inhabitants)
+"wdt:P31" = "wd:Q200250"
+
+
+[attributes]
+label = ["rdfs:label", "wdt:P1448/rdfs:label"]
+longitude = '''
+SELECT ?longitude
+WHERE {
+  ?subject wdt:P625 ?geo1 .
+  BIND(REPLACE(str(?geo1), "Point\\((\\d+\\.\\d+).*$", "$1") as ?longitude)
+  }
+'''
+latitude = '''
+SELECT ?latitude
+WHERE {
+  ?subject wdt:P625 ?geo1 .
+  BIND(REPLACE(str(?geo1), "Point\\((\\d+\\.\\d+) (\\d+\\.\\d+).*$", "$2") as ?latitude)
+  }
+'''
+same_as = ["owl:sameAs", "wdtn:P227", "wdtn:P1566", "wdtn:P214", "wdtn:P244"]

apis_core/apis_entities/triple_configs/E74_GroupFromDNB.toml

@@ -0,0 +1,8 @@
+[[filters]]
+"rdf:type" = "gndo:CorporateBody"
+
+[attributes]
+label = ["gndo:preferredNameForTheCorporateBody", "gndo:variantNameForTheCorporateBody"]
+start_date = "gndo:dateOfEstablishment"
+end_date = "gndo:dateOfTermination"
+sameas = "owl:sameAs"

apis_core/apis_entities/triple_configs/E74_GroupFromWikidata.toml

@@ -0,0 +1,6 @@
+[[filters]]
+"wdt:P31" = "wd:Q414147"
+
+[attributes]
+label = ["schema:about/rdfs:label", "rdfs:label"]
+same_as = ["owl:sameAs", "wdtn:P227", "wdtn:P1566", "wdtn:P214", "wdtn:P244"]

apis_core/utils/rdf2.py

@@ -0,0 +1,82 @@
+# SPDX-FileCopyrightText: 2025 Birger Schacht
+# SPDX-License-Identifier: MIT
+
+import logging
+from collections import defaultdict
+from pathlib import Path
+
+import tomllib
+from AcdhArcheAssets.uri_norm_rules import get_normalized_uri
+from django.apps import apps
+from rdflib import RDF, BNode, Graph
+
+logger = logging.getLogger(__name__)
+
+
+def find_matching_config(graph: Graph) -> dict | None:
+    models_with_config = [
+        model for model in apps.get_models() if hasattr(model, "rdf_configs")
+    ]
+    for model in models_with_config:
+        for path in model.rdf_configs():
+            config = tomllib.loads(Path(path).read_text())
+            for _filter in config.get("filters", []):
+                try:
+                    triples = [
+                        (
+                            None,
+                            graph.namespace_manager.expand_curie(predicate),
+                            graph.namespace_manager.expand_curie(obj),
+                        )
+                        for predicate, obj in _filter.items()
+                    ]
+                    triples = [triple in graph for triple in triples]
+                    if all(triples):
+                        logger.debug("Using %s for parsing graph", path)
+                        config["model"] = model
+                        return config
+                except ValueError:
+                    pass
+    return None
+
+
+def get_something_from_uri(uri: str) -> dict | None:
+    uri = get_normalized_uri(uri)
+    graph = Graph()
+    graph.parse(uri)
+
+    if config := find_matching_config(graph):
+        result = defaultdict(list)
+        result["model"] = config["model"]
+        result["relations"] = defaultdict(list)
+
+        for attribute, curies in config.get("attributes", {}).items():
+            if isinstance(curies, str):
+                curies = [curies]
+            for curie in curies:
+                values = []
+                if curie.startswith("SELECT "):
+                    results = graph.query(curie)
+                else:
+                    results = graph.query(
+                        "SELECT ?object WHERE { ?subject " + curie + " ?object }"
+                    )
+                objects = [result[0] for result in results]
+                for obj in objects:
+                    if isinstance(obj, BNode):
+                        values.extend(
+                            [
+                                value.toPython()
+                                for value in graph.objects(subject=obj)
+                                if value != RDF.Seq
+                            ]
+                        )
+                    else:
+                        values.append(obj.toPython())
+
+                if attribute == "relations":
+                    result["relations"][curie].extend(values)
+                else:
+                    result[attribute].extend(values)
+        return dict(result)
+    return None