add entity_types db

kgdata/wikidata/__main__.py

@@ -151,6 +151,15 @@ def db_entities_attr(attr: EntAttr, output: str, compact: bool, lang: str):
         },
     )
 )
+wikidata.add_command(
+    dataset2db(
+        "entity_types",
+        format={
+            "record_type": {"type": "tuple2", "key": None, "value": None},
+            "is_sorted": False,
+        },
+    )
+)
 wikidata.add_command(
     dataset2db(
         "entity_labels",

kgdata/wikidata/db.py

@@ -209,6 +209,11 @@ def __call__(
     ser_value=str.encode,
     dbopts=small_dbopts,
 )
+get_entity_type_db: make_get_rocksdb[list[str]] = make_get_rocksdb(
+    deser_value=orjson.loads,
+    ser_value=orjson.dumps,
+    dbopts=small_dbopts,
+)
 get_entity_pagerank_db = make_get_rocksdb(
     deser_value=unpack_float,
     ser_value=pack_float,
@@ -273,6 +278,12 @@ def entity_labels(self):
             self.database_dir / "entity_labels.db", read_only=self.read_only
         )
 
+    @cached_property
+    def entity_types(self):
+        return get_entity_type_db(
+            self.database_dir / "entity_types.db", read_only=self.read_only
+        )
+
     @cached_property
     def classes(self):
         wdclasses = get_class_db(

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "kgdata"
-version = "5.0.0"
+version = "5.1.0"
 description = "Library to process dumps of knowledge graphs (Wikipedia, DBpedia, Wikidata)"
 readme = "README.md"
 authors = [{ name = "Binh Vu", email = "binh@toan2.com" }]

scripts/build.sh

@@ -96,6 +96,7 @@ wikidata_db properties
 wikidata_db entities
 wikidata_db entity_labels
 wikidata_db entity_metadata
+wikidata_db entity_types
 wikidata_db entity_outlinks
 wikidata_db entity_redirections
 wikidata_db wp2wd