DynamoDB Table Resource is_in ConditionExpression overwrites data if more than 1 item is in the value list

[KidAmnesiac1]

Describe the bug

When using a ConditionalExpression is_in filter with more than one item in the filter value, the values used to filter overwrite data in the DynamoDB item

Expected Behavior

ConditionalExpressions should not change DynamoDB item

Current Behavior

Values in ConditionalExpression appear in DynamoDB item after update

Reproduction Steps

• Create table via console called test_table with hash key test_key
• Run the code below

import boto3
from boto3.dynamodb.conditions import Attr

ddb = boto3.resource("dynamodb").Table("test_table")

ddb.put_item(
    Item={
        "test_key": "foo-key",
        "filter_val": "foo",
        "val1": 123,
    },
)


resp = ddb.update_item(
    Key={"test_key": "foo-key"},
    UpdateExpression="SET #f1=:v1",
    ExpressionAttributeNames={"#f1": "val1"},
    ExpressionAttributeValues={":v1": 321},
    ConditionExpression=Attr("filter_val").is_in(["foo", "bar"]),
    ReturnValues="ALL_NEW",
)


print(resp["Attributes"])

# {'filter_val': 'foo', 'val1': 'bar', 'test_key': 'foo-key'}

see above the the value for val1 was originally set to 123, and was meant to be update to 321 but instead was updated to bar.

Possible Solution

No response

Additional Information/Context

No response

SDK version used

1.34.49

Environment details (OS name and version, etc.)

Amazon Linux 2

[tim-finnigan]

Thanks for reaching out and your patience here. This update to your function appears to work for me. Can you confirm that this accomplishes your use case?

resp = ddb.update_item(
    Key={"test_key": "foo-key"},
    UpdateExpression="SET #f1=:v1",
    ExpressionAttributeNames={"#f1": "val1"},
    ExpressionAttributeValues={":v1": 321, ":valFoo": "foo", ":valBar": "bar"},
    ConditionExpression="filter_val IN (:valFoo, :valBar)",
    ReturnValues="ALL_NEW",
)

Linking is_in documentation for reference: https://boto3.amazonaws.com/v1/documentation/api/latest/reference/customizations/dynamodb.html#boto3.dynamodb.conditions.Attr.is_in

[KidAmnesiac1]

@tim-finnigan Thanks for getting back to me.

Yes, this modification works.

Am I misinterpreting the docs is seems using boto3.dynamodb.conditions.Attr is preferred to strings.

It would be good if is_in behaved like your example as it would allow for cleaner/easier composability of filters vs. string manipulation.

[tim-finnigan]

Thanks for following up, it looks like this updating adding filter_val to ExpressionAttributeNames does what you're looking for:

resp = ddb.update_item(
    Key={"id": "foo-key"},
    UpdateExpression="SET #f1=:v1",
    ExpressionAttributeNames={"#f1": "filter_val"},
    ExpressionAttributeValues={":v1": 321},
    ConditionExpression=Attr("filter_val").is_in(["foo", "bar"]),
    ReturnValues="ALL_NEW",
)