[Bug]: When using text_match as a field name, the text match expression will throw an error: mismatched input 'text_match' expecting Identifier: invalid parameter.

[zhuwenxing]

Is there an existing issue for this?

• I have searched the existing issues

Environment

- Milvus version:master/2.5
- Deployment mode(standalone or cluster):
- MQ type(rocksmq, pulsar or kafka):    
- SDK version(e.g. pymilvus v2.0.0rc2):
- OS(Ubuntu or CentOS): 
- CPU/Memory: 
- GPU: 
- Others:

Current Behavior

=== start connecting to Milvus     ===

Does collection hello_match_text exist in Milvus: False

=== Create collection `hello_match_text` ===


=== Start inserting entities       ===

Number of insert entities in Milvus: 6

=== Start querying with `text_match(text_match, 'milvus')` ===

2025-03-03 11:54:58,769 [ERROR][handler]: RPC error: [query], <MilvusException: (code=1100, message=failed to create query plan: cannot parse expression: text_match(text_match, 'milvus'), error: line 1:11 mismatched input 'text_match' expecting Identifier: invalid parameter)>, <Time:{'RPC start': '2025-03-03 11:54:58.753119', 'RPC error': '2025-03-03 11:54:58.768924'}> (decorators.py:140)
2025-03-03 11:54:58,769 [ERROR][query]: Failed to query collection: match_text (milvus_client.py:486)

Expected Behavior

If text_match cannot be used as a field name, an error should be thrown when creating the collection, rather than waiting until the text match operation is executed to report the error.

Steps To Reproduce

Milvus Log

reporduce code

import numpy as np

from pymilvus import (
    MilvusClient,
    Function,
    FunctionType,
    DataType,
)

fmt = "\n=== {:30} ===\n"
collection_name = "match_text"
dim = 8

#################################################################################
# 1. connect to Milvus
# Add a new connection alias `default` for Milvus server in `localhost:19530`
print(fmt.format("start connecting to Milvus"))
milvus_client = MilvusClient("http://10.104.32.135:19530")

has_collection = milvus_client.has_collection(collection_name, timeout=5)
print(f"Does collection hello_match_text exist in Milvus: {has_collection}")
if has_collection:
    milvus_client.drop_collection(collection_name)


schema = milvus_client.create_schema()
schema.add_field("id", DataType.INT64, is_primary=True, auto_id=False)
# set analyzer params in match_text field for more situations
# default as analyzer_params = {"type": "standard"}
schema.add_field("text_match", DataType.VARCHAR, max_length=1000, enable_analyzer=True, enable_match=True),
schema.add_field("embeddings", DataType.FLOAT_VECTOR, dim=dim)

print(fmt.format("Create collection `hello_match_text`"))

index_params = milvus_client.prepare_index_params()
index_params.add_index(
    "embeddings",
    index_type= "AUTOINDEX",
    metric_type= "IP"
)

milvus_client.create_collection(collection_name, schema=schema, index_params=index_params, consistency_level="Strong")

################################################################################
# 3. insert data
# We are going to insert 6 rows of data into `hello_match_text`
# Data to be inserted must be organized in fields.
#
# The insert() method returns:
# - either automatically generated primary keys by Milvus if auto_id=True in the schema;
# - or the existing primary key field from the entities if auto_id=False in the schema.

print(fmt.format("Start inserting entities"))

rng = np.random.default_rng(seed=19530)
num_entities = 6
keywords = ["milvus", "match", "search", "query", "analyzer", "tokenizer"]
embeddings = rng.random((num_entities, dim), np.float32)

entities = [{
        "id": i,
        "text_match":f"This is a test match_text {i} with keywords: {keywords[i]}",
        "embeddings": embeddings[i]
    } for i in range(num_entities)
]

insert_result = milvus_client.insert(collection_name, entities)
print(f"Number of insert entities in Milvus: {insert_result['insert_count']}")  # check the num_entities
milvus_client.flush(collection_name)

# ###############################################################################
# 4. query and scalar filtering search with text match
# After data were inserted into Milvus and indexed, you can perform:
# - query with text match expression
# - search data with text match filter

# -----------------------------------------------------------------------------
# query based text match with single keyword filter
filter = f"text_match(text_match, '{keywords[0]}')"
print(fmt.format(f"Start querying with `{filter}`"))

result = milvus_client.query(collection_name, filter, output_fields=["text_match"])
print(f"query result:\n-{result}")

Anything else?

No response

[dosubot]

The error you're encountering is due to the use of text_match as a field name, which conflicts with a reserved keyword in Milvus. Milvus has a mechanism to validate field names during collection creation to ensure they do not use reserved keywords. However, it seems that text_match is not being flagged during this validation process, leading to the error only appearing during query execution.

To resolve this issue, you should avoid using text_match as a field name. Instead, choose a different name that does not conflict with reserved keywords. This will prevent the error from occurring during query execution.

If you believe that text_match should be flagged as a reserved keyword during collection creation, it might be worth checking if the keyword list in the Milvus codebase needs updating or if there's a bug in the validation logic. The validateFieldName function is responsible for checking if a field name is a reserved keyword ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[zhuwenxing]

related to: #35873

#35873 (comment) should be updated

[yanliang567]

/assign @xiaocai2333
/unassign

[xiaocai2333]

yeah, we need to update the keywords.