Skip to content

Dynamic array field filters fail in Milvus Lite 3.1.0 #348

Description

@zc277584121

Summary

Milvus Lite 3.1.0 can store and return dynamic array fields, and dynamic scalar filters work, but filtering on a dynamic array field using array indexing fails with unknown field.

The same array filter works on Milvus Standalone and Zilliz Cloud.

Environment

  • pymilvus==3.0.0
  • milvus-lite==3.1.0
  • Milvus Lite server version reported by get_server_version(): milvus_lite-3.1.0

Comparison backends tested:

Backend Version reported rank == 1 topic == "a" array_field[0] < 1
Milvus Lite milvus_lite-3.1.0 works works fails with unknown field 'array_field'
Milvus Standalone 2.5.13 works works works
Milvus Standalone 3.0-beta works works works
Zilliz Cloud Compatible with Milvus 2.6 works works works

Reproduction

import os
import tempfile
from pymilvus import DataType, MilvusClient

uri = os.path.join(tempfile.mkdtemp(), "milvus.db")
client = MilvusClient(uri=uri)
print(client.get_server_version())

collection_name = "dynamic_array_filter_repro"

schema = client.create_schema(auto_id=True, enable_dynamic_field=True)
schema.add_field(field_name="pk", datatype=DataType.INT64, is_primary=True, auto_id=True)
schema.add_field(field_name="text", datatype=DataType.VARCHAR, max_length=1024)
schema.add_field(field_name="vector", datatype=DataType.FLOAT_VECTOR, dim=2)

index_params = client.prepare_index_params()
index_params.add_index(field_name="vector", index_type="AUTOINDEX", metric_type="L2")

client.create_collection(
    collection_name=collection_name,
    schema=schema,
    index_params=index_params,
    consistency_level="Strong",
)
client.insert(
    collection_name=collection_name,
    data=[
        {"text": "foo", "vector": [0.0, 0.0], "rank": 1, "topic": "a", "array_field": [0, 1, 2]},
        {"text": "bar", "vector": [1.0, 1.0], "rank": 2, "topic": "b", "array_field": [1, 2, 3]},
    ],
)
client.flush(collection_name)
client.load_collection(collection_name)

for expr in ["rank == 1", 'topic == "a"', "array_field[0] < 1"]:
    result = client.search(
        collection_name=collection_name,
        data=[[0.0, 0.0]],
        anns_field="vector",
        search_params={"metric_type": "L2", "params": {}},
        limit=2,
        filter=expr,
        output_fields=["pk", "text", "rank", "topic", "array_field"],
    )
    print(expr, [hit["entity"] for hit in result[0]])

Actual output on Milvus Lite 3.1.0

milvus_lite-3.1.0
rank == 1 [{'pk': 1, 'text': 'foo', 'rank': 1, 'topic': 'a', 'array_field': [0, 1, 2]}]
topic == "a" [{'pk': 1, 'text': 'foo', 'rank': 1, 'topic': 'a', 'array_field': [0, 1, 2]}]
MilvusException: unknown field 'array_field' at column 1
  array_field[0] < 1
  ^^^^^^^^^^^
available fields: ['pk', 'text', 'vector']

Expected behavior

Since the dynamic array field is stored and returned correctly, and the same filter works on Milvus Standalone and Zilliz Cloud, Milvus Lite should support array_field[0] < 1 for dynamic array fields as well.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions