weaviate-cli 3.1.2__tar.gz → 3.1.3__tar.gz

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: weaviate-cli
-Version: 3.1.2
+Version: 3.1.3
 Summary: Command line interface to interact with weaviate
 Home-page: https://github.com/weaviate/weaviate-cli
 Download-URL: https://github.com/weaviate/weaviate-cli

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/commands/create.py

@@ -183,20 +183,28 @@ def create_collection_cli(
 @click.option(
     "--tenant_suffix",
     default=CreateTenantsDefaults.tenant_suffix,
-    help="The suffix to add to the tenant name (default: 'Tenant--').",
+    help="The suffix to add to the tenant name (default: 'Tenant-').",
 )
 @click.option(
     "--number_tenants",
     default=CreateTenantsDefaults.number_tenants,
     help="Number of tenants to create (default: 100).",
 )
+@click.option(
+    "--tenant_batch_size",
+    default=CreateTenantsDefaults.tenant_batch_size,
+    type=int,
+    help="Number of tenants to create in each batch (default: None, all tenants will be created in one batch).",
+)
 @click.option(
     "--state",
     default=CreateTenantsDefaults.state,
     type=click.Choice(["hot", "active", "cold", "inactive", "frozen", "offloaded"]),
 )
 @click.pass_context
-def create_tenants_cli(ctx, collection, tenant_suffix, number_tenants, state):
+def create_tenants_cli(
+    ctx, collection, tenant_suffix, number_tenants, tenant_batch_size, state
+):
     """Create tenants in Weaviate."""
 
     client = None
@@ -208,6 +216,7 @@ def create_tenants_cli(ctx, collection, tenant_suffix, number_tenants, state):
             collection=collection,
             tenant_suffix=tenant_suffix,
             number_tenants=number_tenants,
+            tenant_batch_size=tenant_batch_size,
             state=state,
         )
     except Exception as e:
@@ -301,6 +310,16 @@ def create_backup_cli(ctx, backend, backup_id, include, exclude, wait, cpu_for_b
     default=CreateDataDefaults.auto_tenants,
     help="Number of tenants for which we will send data. NOTE: Requires class with --auto_tenant_creation (default: 0).",
 )
+@click.option(
+    "--tenants",
+    default=None,
+    help="Comma separated list of tenants to send data to.",
+)
+@click.option(
+    "--tenant_suffix",
+    default=CreateTenantsDefaults.tenant_suffix,
+    help="The suffix to add to the tenant name (default: 'Tenant-'). Only used if --auto_tenants is provided.",
+)
 @click.option(
     "--vector_dimensions",
     default=CreateDataDefaults.vector_dimensions,
@@ -319,6 +338,8 @@ def create_data_cli(
     consistency_level,
     randomize,
     auto_tenants,
+    tenants,
+    tenant_suffix,
     vector_dimensions,
     uuid,
 ):
@@ -349,8 +370,10 @@ def create_data_cli(
             consistency_level=consistency_level,
             randomize=randomize,
             auto_tenants=auto_tenants,
+            tenants_list=tenants.split(",") if tenants else None,
             vector_dimensions=vector_dimensions,
             uuid=uuid,
+            tenant_suffix=tenant_suffix,
         )
     except Exception as e:
         click.echo(f"Error: {e}")

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/commands/delete.py

@@ -60,7 +60,7 @@ def delete_collection_cli(ctx: click.Context, collection: str, all: bool) -> Non
 @click.option(
     "--tenant_suffix",
     default=DeleteTenantsDefaults.tenant_suffix,
-    help="The suffix to add to the tenant name (default: 'Tenant--').",
+    help="The suffix to add to the tenant name (default: 'Tenant-').",
 )
 @click.option(
     "--number_tenants",
@@ -110,13 +110,18 @@ def delete_tenants_cli(
     type=click.Choice(["quorum", "all", "one"]),
     help="Consistency level (default: 'quorum').",
 )
+@click.option(
+    "--tenants",
+    default=None,
+    help="Comma separated list of tenants to delete data from.",
+)
 @click.option(
     "--uuid",
     default=DeleteDataDefaults.uuid,
     help="UUID of the oject to be deleted. If provided, --limit will be ignored.",
 )
 @click.pass_context
-def delete_data_cli(ctx, collection, limit, consistency_level, uuid):
+def delete_data_cli(ctx, collection, limit, consistency_level, tenants, uuid):
     """Delete data from a collection in Weaviate."""
 
     client = None
@@ -128,6 +133,7 @@ def delete_data_cli(ctx, collection, limit, consistency_level, uuid):
             collection=collection,
             limit=limit,
             consistency_level=consistency_level,
+            tenants_list=tenants.split(",") if tenants else None,
             uuid=uuid,
         )
     except Exception as e:

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/commands/update.py

@@ -121,7 +121,7 @@ def update_collection_cli(
 @click.option(
     "--tenant_suffix",
     default=UpdateTenantsDefaults.tenant_suffix,
-    help="The suffix to add to the tenant name (default: 'Tenant--').",
+    help="The suffix to add to the tenant name (default: 'Tenant-').",
 )
 @click.option(
     "--number_tenants",

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/defaults.py

@@ -30,6 +30,8 @@ PERMISSION_HELP_STRING = (
     "  --permission read_cluster"
 )
 
+MAX_OBJECTS_PER_BATCH = 5000
+
 
 @dataclass
 class CreateCollectionDefaults:
@@ -43,7 +45,7 @@ class CreateCollectionDefaults:
     auto_tenant_creation: bool = False
     auto_tenant_activation: bool = False
     force_auto_schema: bool = False
-    shards: int = 1
+    shards: int = 0
     vectorizer: Optional[str] = None
     replication_deletion_strategy: str = "no_automated_resolution"
 
@@ -51,8 +53,9 @@ class CreateCollectionDefaults:
 @dataclass
 class CreateTenantsDefaults:
     collection: str = "Movies"
-    tenant_suffix: str = "Tenant--"
+    tenant_suffix: str = "Tenant-"
     number_tenants: int = 100
+    tenant_batch_size: Optional[int] = None
     state: str = "active"
 
 
@@ -97,7 +100,7 @@ class DeleteCollectionDefaults:
 @dataclass
 class DeleteTenantsDefaults:
     collection: str = "Movies"
-    tenant_suffix: str = "Tenant--"
+    tenant_suffix: str = "Tenant-"
     number_tenants: int = 100
 
 
@@ -183,7 +186,7 @@ class UpdateCollectionDefaults:
 @dataclass
 class UpdateTenantsDefaults:
     collection: str = "Movies"
-    tenant_suffix: str = "Tenant--"
+    tenant_suffix: str = "Tenant-"
     number_tenants: int = 100
     state: str = "active"
 

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/managers/collection_manager.py

@@ -234,7 +234,7 @@ class CollectionManager:
                     ),
                 ),
                 sharding_config=(
-                    wvc.Configure.sharding(desired_count=shards) if shards > 1 else None
+                    wvc.Configure.sharding(desired_count=shards) if shards > 0 else None
                 ),
                 multi_tenancy_config=wvc.Configure.multi_tenancy(
                     enabled=multitenant,

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/managers/data_manager.py

@@ -10,16 +10,20 @@ from weaviate.classes.query import MetadataQuery
 from weaviate.collections.classes.tenants import TenantActivityStatus
 from typing import Dict, List, Optional, Union, Any
 import weaviate.classes.config as wvc
+from weaviate.classes.query import Filter
 from weaviate.collections import Collection
 from datetime import datetime, timedelta
 from weaviate_cli.defaults import (
+    MAX_OBJECTS_PER_BATCH,
     CreateDataDefaults,
+    CreateTenantsDefaults,
     QueryDataDefaults,
     UpdateDataDefaults,
     DeleteDataDefaults,
 )
 import importlib.resources as resources
 from pathlib import Path
+import math
 
 PROPERTY_NAME_MAPPING = {
     "releaseDate": "release_date",
@@ -203,7 +207,9 @@ class DataManager:
         limit: int = CreateDataDefaults.limit,
         consistency_level: str = CreateDataDefaults.consistency_level,
         randomize: bool = CreateDataDefaults.randomize,
+        tenant_suffix: str = CreateTenantsDefaults.tenant_suffix,
         auto_tenants: int = CreateDataDefaults.auto_tenants,
+        tenants_list: Optional[List[str]] = None,
         vector_dimensions: Optional[int] = CreateDataDefaults.vector_dimensions,
         uuid: Optional[str] = None,
         named_vectors: Optional[List[str]] = None,
@@ -217,14 +223,19 @@ class DataManager:
 
         col: Collection = self.client.collections.get(collection)
         try:
-            tenants = [key for key in col.tenants.get().keys()]
+            existing_tenants = [key for key in col.tenants.get().keys()]
         except Exception as e:
             # Check if the error is due to multi-tenancy being disabled
             if "multi-tenancy is not enabled" in str(e):
-                click.echo(
-                    f"Collection '{col.name}' does not have multi-tenancy enabled. Skipping tenant information collection."
-                )
-                tenants = ["None"]
+                if (
+                    tenants_list is not None
+                    or auto_tenants != CreateDataDefaults.auto_tenants
+                ):
+                    raise Exception(
+                        f"Collection '{col.name}' does not have multi-tenancy enabled. Adding data to tenants is not possible."
+                    )
+
+                existing_tenants = ["None"]
             else:
                 raise e
         if (
@@ -241,16 +252,23 @@ class DataManager:
             "all": wvc.ConsistencyLevel.ALL,
             "one": wvc.ConsistencyLevel.ONE,
         }
+        tenants = existing_tenants
         if auto_tenants > 0:
-            if tenants == "None":
-                tenants = [f"Tenant--{i}" for i in range(1, auto_tenants + 1)]
+            if tenants_list is not None:
+                raise Exception(
+                    f"Either --tenants or --auto_tenants must be provided, not both."
+                )
+            if existing_tenants == "None":
+                tenants = [f"{tenant_suffix}{i}" for i in range(1, auto_tenants + 1)]
             else:
-                if len(tenants) < auto_tenants:
+                if len(existing_tenants) < auto_tenants:
                     tenants += [
-                        f"Tenant--{i}"
-                        for i in range(len(tenants) + 1, auto_tenants + 1)
+                        f"{tenant_suffix}{i}"
+                        for i in range(len(existing_tenants) + 1, auto_tenants + 1)
                     ]
-
+        else:
+            if tenants_list is not None:
+                tenants = tenants_list
         for tenant in tenants:
             if tenant == "None":
                 collection = self.__ingest_data(
@@ -273,7 +291,7 @@ class DataManager:
                     uuid,
                     named_vectors,
                 )
-
+            collection.batch.wait_for_vector_indexing()
             if len(collection) != limit:
                 click.echo(
                     f"Error occurred while ingesting data for tenant '{tenant}'. Check number of objects inserted."
@@ -402,25 +420,43 @@ class DataManager:
             print(f"Object deleted: {uuid} into class '{collection.name}'")
             return 1
 
-        res = collection.query.fetch_objects(limit=num_objects)
-        if len(res.objects) == 0:
-            print(
-                f"No objects found in class '{collection.name}'. Insert objects first using <ingest data> command"
+        # Calculate the number of full batches and handle any remaining objects
+        # Use math.ceil to ensure we process all objects, even if num_objects < MAX_OBJECTS_PER_BATCH
+        iterations = math.ceil(num_objects / MAX_OBJECTS_PER_BATCH)
+        deleted_objects = 0
+
+        for _ in range(iterations):
+            # Determine how many objects to fetch in this batch
+            batch_size = min(MAX_OBJECTS_PER_BATCH, num_objects - deleted_objects)
+            if batch_size <= 0:
+                break
+
+            res = collection.query.fetch_objects(limit=batch_size)
+            if len(res.objects) == 0:
+                click.echo(
+                    f"No objects found in class '{collection.name}'. Insert objects first using <ingest data> command"
+                )
+                return deleted_objects
+
+            ids = [o.uuid for o in res.objects]
+            collection.with_consistency_level(cl).data.delete_many(
+                where=Filter.by_id().contains_any(ids)
             )
-            return -1
-        data_objects = res.objects
+            deleted_objects += len(ids)
 
-        for obj in data_objects:
-            collection.with_consistency_level(cl).data.delete_by_id(uuid=obj.uuid)
+            # If we've deleted fewer objects than expected, there might not be any more objects
+            if len(ids) < batch_size:
+                break
 
-        print(f"Deleted {num_objects} objects into class '{collection.name}'")
-        return num_objects
+        print(f"Deleted {deleted_objects} objects from class '{collection.name}'")
+        return deleted_objects
 
     def delete_data(
         self,
         collection: str = DeleteDataDefaults.collection,
         limit: int = DeleteDataDefaults.limit,
         consistency_level: str = DeleteDataDefaults.consistency_level,
+        tenants_list: Optional[List[str]] = None,
         uuid: Optional[str] = DeleteDataDefaults.uuid,
     ) -> None:
 
@@ -433,14 +469,14 @@ class DataManager:
 
         col: Collection = self.client.collections.get(collection)
         try:
-            tenants = [key for key in col.tenants.get().keys()]
+            existing_tenants = [key for key in col.tenants.get().keys()]
         except Exception as e:
             # Check if the error is due to multi-tenancy being disabled
             if "multi-tenancy is not enabled" in str(e):
                 click.echo(
                     f"Collection '{col.name}' does not have multi-tenancy enabled. Skipping tenant information collection."
                 )
-                tenants = ["None"]
+                existing_tenants = ["None"]
 
         cl_map = {
             "quorum": wvc.ConsistencyLevel.QUORUM,
@@ -448,6 +484,11 @@ class DataManager:
             "one": wvc.ConsistencyLevel.ONE,
         }
 
+        if tenants_list is not None:
+            tenants = tenants_list
+        else:
+            tenants = existing_tenants
+
         for tenant in tenants:
             if tenant == "None":
                 ret = self.__delete_data(col, limit, cl_map[consistency_level], uuid)

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli/managers/tenant_manager.py

@@ -1,3 +1,4 @@
+from typing import Optional
 import click
 import semver
 
@@ -20,6 +21,7 @@ class TenantManager:
         collection: str = CreateTenantsDefaults.collection,
         tenant_suffix: str = CreateTenantsDefaults.tenant_suffix,
         number_tenants: int = CreateTenantsDefaults.number_tenants,
+        tenant_batch_size: Optional[int] = CreateTenantsDefaults.tenant_batch_size,
         state: str = CreateTenantsDefaults.state,
     ) -> None:
         """
@@ -33,11 +35,10 @@ class TenantManager:
 
         Raises:
             Exception: If the collection does not exist, multi-tenancy is not enabled,
-                       tenants already exist, or if there is a mismatch in tenant activity status.
+                       or if there is a mismatch in tenant activity status.
         """
 
         if not self.client.collections.exists(collection):
-
             raise Exception(
                 f"Class '{collection}' does not exist in Weaviate. Create first using <create class>"
             )
@@ -46,7 +47,6 @@ class TenantManager:
         collection = self.client.collections.get(collection)
 
         if not collection.config.get().multi_tenancy_config.enabled:
-
             raise Exception(
                 f"Collection '{collection.name}' does not have multi-tenancy enabled. Recreate or modify the class with: <create class>"
             )
@@ -61,45 +61,94 @@ class TenantManager:
         }
 
         existing_tenants = collection.tenants.get()
-        if existing_tenants:
+        new_tenant_names = []
 
-            raise Exception(
-                f"Tenants already exist in class '{collection.name}'. Update their status using ./update_tenants.py or delete them using <delete tenants> command"
-            )
-        else:
-            collection.tenants.create(
-                [
-                    Tenant(
-                        name=f"{tenant_suffix}{i}",
-                        activity_status=tenant_state_map[state],
+        if existing_tenants:
+            # Check if existing tenants follow the same suffix pattern
+            existing_tenant_names = list(existing_tenants.keys())
+            for tenant_name in existing_tenant_names:
+                if tenant_name.startswith(tenant_suffix):
+                    try:
+                        # Try to extract the index part after the suffix
+                        int(tenant_name[len(tenant_suffix) :])
+                    except ValueError:
+                        raise Exception(
+                            f"Existing tenant '{tenant_name}' does not follow the expected pattern '{tenant_suffix}N' where N is a number. "
+                            f"Please use a different tenant_suffix or delete existing tenants."
+                        )
+                else:
+                    raise Exception(
+                        f"Existing tenant '{tenant_name}' does not use the provided tenant_suffix '{tenant_suffix}'. "
+                        f"Please use the same tenant_suffix as existing tenants or delete existing tenants."
                     )
-                    for i in range(number_tenants)
-                ]
-            )
 
-        # get_by_names is only available after 1.25.0
-        if version.compare(semver.Version.parse("1.25.0")) < 0:
-            tenants_list = {
-                name: tenant
-                for name, tenant in collection.tenants.get().items()
-                if name.startswith(tenant_suffix)
-            }
+            # Find the highest index among existing tenants
+            highest_index = -1
+            for tenant_name in existing_tenant_names:
+                try:
+                    index = int(tenant_name[len(tenant_suffix) :])
+                    highest_index = max(highest_index, index)
+                except ValueError:
+                    continue
+
+            # Generate new tenant names starting from the next index
+            start_index = highest_index + 1
+            new_tenant_names = [
+                f"{tenant_suffix}{i}"
+                for i in range(start_index, start_index + number_tenants)
+            ]
         else:
-            tenants_list = collection.tenants.get_by_names(
-                [f"{tenant_suffix}{i}" for i in range(number_tenants)]
+            # No existing tenants, create tenants starting from index 0
+            new_tenant_names = [f"{tenant_suffix}{i}" for i in range(number_tenants)]
+
+        # Create the new tenants
+        tenants = [
+            Tenant(
+                name=name,
+                activity_status=tenant_state_map[state],
             )
-        assert (
-            len(tenants_list) == number_tenants
-        ), f"Expected {number_tenants} tenants, but found {len(tenants_list)}"
-        for tenant in tenants_list.values():
-            if tenant.activity_status != tenant_state_map[state]:
+            for name in new_tenant_names
+        ]
+
+        if tenants:  # Only call create if there are tenants to create
+            if tenant_batch_size:
+                for i in range(0, len(tenants), tenant_batch_size):
+                    batch = tenants[i : i + tenant_batch_size]
+                    collection.tenants.create(batch)
+            else:
+                collection.tenants.create(tenants)
+
+            # Verify the newly created tenants
+            if version.compare(semver.Version.parse("1.25.0")) < 0:
+                created_tenants = {
+                    name: tenant
+                    for name, tenant in collection.tenants.get().items()
+                    if name in new_tenant_names
+                }
+            else:
+                created_tenants = collection.tenants.get_by_names(new_tenant_names)
 
+            # Verify all requested tenants were created
+            if len(created_tenants) != len(new_tenant_names):
+                missing_tenants = set(new_tenant_names) - set(created_tenants.keys())
                 raise Exception(
-                    f"Tenant '{tenant.name}' has activity status '{tenant.activity_status}', but expected '{tenant_state_map[state]}'"
+                    f"Not all requested tenants were created. Missing: {', '.join(missing_tenants)}"
                 )
-        click.echo(
-            f"{len(tenants_list)} tenants added with tenant status '{tenant.activity_status}' for collection '{collection.name}'"
-        )
+
+            # Verify tenant activity status
+            for tenant_name, tenant in created_tenants.items():
+                if tenant.activity_status != tenant_state_map[state]:
+                    raise Exception(
+                        f"Tenant '{tenant_name}' has activity status '{tenant.activity_status}', but expected '{tenant_state_map[state]}'"
+                    )
+
+            click.echo(
+                f"{len(new_tenant_names)} tenants added with tenant status '{tenant_state_map[state]}' for collection '{collection.name}'"
+            )
+        else:
+            click.echo(
+                f"No new tenants were created for collection '{collection.name}'"
+            )
 
     def delete_tenants(
         self,

{weaviate_cli-3.1.2 → weaviate_cli-3.1.3}/weaviate_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: weaviate-cli
-Version: 3.1.2
+Version: 3.1.3
 Summary: Command line interface to interact with weaviate
 Home-page: https://github.com/weaviate/weaviate-cli
 Download-URL: https://github.com/weaviate/weaviate-cli