Add acorn option and nested objects from dataset.

This commit adds support for enabling acorn on collection creation,
as well as during the collection update.

Also, it includes couple of new properties of type object by adding
the property type and improving the import_json function to handle
properties with camelCase naming

weaviate_cli/commands/create.py

@@ -29,7 +29,16 @@ def create() -> None:
     "--vector_index",
     default="hnsw",
     type=click.Choice(
-        ["hnsw", "flat", "dynamic", "hnsw_pq", "hnsw_bq", "hnsw_sq", "flat_bq"]
+        [
+            "hnsw",
+            "flat",
+            "dynamic",
+            "hnsw_pq",
+            "hnsw_bq",
+            "hnsw_sq",
+            "hnsw_acorn",
+            "flat_bq",
+        ]
     ),
     help="Vector index type (default: 'hnsw').",
 )

weaviate_cli/commands/update.py

@@ -26,7 +26,9 @@ def update() -> None:
 @click.option(
     "--vector_index",
     default=None,
-    type=click.Choice(["hnsw", "flat", "hnsw_pq", "hnsw_bq", "hnsw_sq", "flat_bq"]),
+    type=click.Choice(
+        ["hnsw", "flat", "hnsw_pq", "hnsw_bq", "hnsw_sq", "flat_bq", "hnsw_acorn"]
+    ),
     help='Vector index type (default: "None").',
 )
 @click.option(

weaviate_cli/managers/collection_manager.py

@@ -4,6 +4,7 @@
 from weaviate.client import WeaviateClient
 from weaviate.collections import Collection
 from weaviate.collections.classes.tenants import TenantActivityStatus
+from weaviate.classes.config import VectorFilterStrategy
 import weaviate.classes.config as wvc
 
 
@@ -86,6 +87,9 @@ def create_collection(
                     training_limit=training_limit
                 )
             ),
+            "hnsw_acorn": wvc.Configure.VectorIndex.hnsw(
+                filter_strategy=VectorFilterStrategy.ACORN
+            ),
             # Should fail at the moment as Flat and PQ are not compatible
             "flat_pq": wvc.Configure.VectorIndex.flat(
                 quantizer=wvc.Configure.VectorIndex.Quantizer.pq()
@@ -117,6 +121,7 @@ def create_collection(
             "length": wvc.Configure.inverted_index(index_property_length=True),
         }
 
+        # Collection schema
         properties: List[wvc.Property] = [
             wvc.Property(name="title", data_type=wvc.DataType.TEXT),
             wvc.Property(name="genres", data_type=wvc.DataType.TEXT),
@@ -126,6 +131,22 @@ def create_collection(
             wvc.Property(name="runtime", data_type=wvc.DataType.TEXT),
             wvc.Property(name="cast", data_type=wvc.DataType.TEXT),
             wvc.Property(name="originalLanguage", data_type=wvc.DataType.TEXT),
+            wvc.Property(
+                name="productionCountries",
+                data_type=wvc.DataType.OBJECT_ARRAY,
+                nested_properties=[
+                    wvc.Property(name="iso_3166_1", data_type=wvc.DataType.TEXT),
+                    wvc.Property(name="name", data_type=wvc.DataType.TEXT),
+                ],
+            ),
+            wvc.Property(
+                name="spokenLanguages",
+                data_type=wvc.DataType.OBJECT_ARRAY,
+                nested_properties=[
+                    wvc.Property(name="iso_639_1", data_type=wvc.DataType.TEXT),
+                    wvc.Property(name="name", data_type=wvc.DataType.TEXT),
+                ],
+            ),
             wvc.Property(name="tagline", data_type=wvc.DataType.TEXT),
             wvc.Property(name="budget", data_type=wvc.DataType.NUMBER),
             wvc.Property(name="releaseDate", data_type=wvc.DataType.DATE),
@@ -195,6 +216,9 @@ def update_collection(
             "hnsw_bq": wvc.Reconfigure.VectorIndex.hnsw(
                 quantizer=wvc.Reconfigure.VectorIndex.Quantizer.bq()
             ),
+            "hnsw_acorn": wvc.Reconfigure.VectorIndex.hnsw(
+                filter_strategy=VectorFilterStrategy.ACORN
+            ),
             "flat_bq": wvc.Reconfigure.VectorIndex.flat(
                 quantizer=wvc.Reconfigure.VectorIndex.Quantizer.bq()
             ),

weaviate_cli/managers/data_manager.py

@@ -16,7 +16,12 @@
 import importlib.resources as resources
 from pathlib import Path
 
-PROPERTY_NAME_MAPPING = {"release_date": "releaseDate"}
+PROPERTY_NAME_MAPPING = {
+    "releaseDate": "release_date",
+    "originalLanguage": "original_language",
+    "productionCountries": "production_countries",
+    "spokenLanguages": "spoken_languages",
+}
 
 
 class DataManager:
@@ -48,7 +53,7 @@ def __import_json(
                         for prop in properties:
                             prop_name = PROPERTY_NAME_MAPPING.get(prop.name, prop.name)
                             if prop_name in obj:
-                                added_obj[prop_name] = self.__convert_property_value(
+                                added_obj[prop.name] = self.__convert_property_value(
                                     obj[prop_name], prop.data_type
                                 )
                         batch.add_object(properties=added_obj)
@@ -95,6 +100,14 @@ def create_single_object() -> Dict:
             date = datetime.strptime("1980-01-01", "%Y-%m-%d")
             random_date = date + timedelta(days=random.randint(1, 15_000))
             release_date = random_date.strftime("%Y-%m-%dT%H:%M:%SZ")
+            spoken_languages = [
+                {"iso_639_1": get_random_string(3), "name": get_random_string(3)}
+                for _ in range(random.randint(1, 3))
+            ]
+            production_countries = [
+                {"iso_3166_1": get_random_string(3), "name": get_random_string(3)}
+                for _ in range(random.randint(1, 3))
+            ]
 
             prefix = "update-" if is_update else ""
             return {
@@ -111,6 +124,8 @@ def create_single_object() -> Dict:
                 "releaseDate": release_date,
                 "revenue": random.randint(1_000_000, 10_000_0000_000),
                 "status": f"{prefix}status" + get_random_string(3),
+                "spokenLanguages": spoken_languages,
+                "productionCountries": production_countries,
             }
 
         if is_update: