Filter Values raises AttributeError for ordinary text inputs #15729

Bug Description

The legacy Filter Values (FilterDataValues / DataFilterComponent) component raises AttributeError: 'str' object has no attribute 'text' when its Filter Key and Filter Value are ordinary text, including the default values. Data/JSON/Message values passed through normal input validation also become text and hit the same error.

The component has no replacement (replacement = []), so saved flows still need this operation. filter_data() reads .text even though MessageTextInput supplies strings. The exception occurs before the empty-input and missing-filter guards as well.

Reproduction

from lfx.components.processing.filter_data_values import DataFilterComponent
from lfx.schema import Data

component = DataFilterComponent().set(
    input_data=[Data(data={"route": "CMIP"}), Data(data={"route": "OTHER"})],
    filter_key="route",
    filter_value="CMIP",
    operator="equals",
)
component.filter_data()

Observed on unchanged release-1.13.0 at bb6e7cf93b6eab2fc68c0c6dbd7914770e3ed2c6:

filter_key: str = self.filter_key.text
                  ^^^^^^^^^^^^^^^^^^^^
AttributeError: 'str' object has no attribute 'text'

Expected behavior

Return the CMIP record. Accept text directly while retaining direct SDK Data/Message inputs through get_text(). Existing comparison operators and empty-input behavior should stay the same.

Operating System

Windows 11

Langflow Version

lfx/langflow-base 1.13.0 from the active release candidate branch above.

Python Version

3.12.14

Additional context

The proposed two-line production change is covered by focused component tests: 11 regressions fail on the original code and two direct SDK object controls pass. With the change, 14 tests pass; four inherited historical-version checks skip because this new test class has no version fixtures. No external services or model calls are needed. Searches for FilterDataValues, filter_data_values, and Filter Values found no existing repair for this failure. AI-assisted investigation and local validation by OpenAI Codex; no human review is claimed.