feat: expose Data Designer run config

[lipikaramaswamy]

Summary

Adds an optional data_designer_run_config parameter to Anonymizer(...) so callers can configure Data Designer runtime controls such as buffer_size and max_in_flight_tasks while still using Anonymizer's normal model config/provider loading path.

The name is intentionally explicit: it sits next to the existing advanced data_designer constructor parameter and avoids overloading run_config, which could be confused with AnonymizerConfig or anonymizer.run(config=...).

This keeps execution/backpressure tuning on the runtime constructor instead of putting it on AnonymizerConfig, which remains focused on anonymization behavior. The PR also re-exports RunConfig from anonymizer so examples can use the public Anonymizer import path.

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation update
• Refactoring

Testing

• make test passes locally
• make check passes locally (format + lint + typecheck + lock-check)
• Added/updated tests for changes

Targeted validation run:

• uv run pytest tests/interface/test_anonymizer_interface.py -q
• uv run ruff check src/anonymizer/interface/anonymizer.py tests/interface/test_anonymizer_interface.py

Documentation

• If docs changed: make docs-build passes locally

No docs changed. This small API hook is intended to support self-hosted Data Designer runtime/backpressure configuration without bypassing Anonymizer's provider setup.

Related Issues

[greptile-apps]

Greptile Summary

This PR adds an optional data_designer_run_config: RunConfig | None parameter to Anonymizer.__init__ so callers can configure Data Designer runtime knobs (e.g. buffer_size, max_in_flight_tasks) without bypassing Anonymizer's provider-loading path. RunConfig is also re-exported from the top-level anonymizer package for discoverability.

• RunConfig is correctly placed under TYPE_CHECKING in anonymizer.py (guarded from runtime import) while being eagerly imported in __init__.py for public API exposure, consistent with the ModelProvider pattern.
• set_run_config is applied unconditionally to whichever DataDesigner instance ends up in self._data_designer — both the Anonymizer-managed and caller-supplied paths — and both are covered by new unit tests.

Confidence Score: 5/5

Safe to merge — the change is a small, additive parameter with no breaking surface and no mutations to existing paths when the new parameter is omitted.

The new data_designer_run_config parameter is fully opt-in and defaults to None, so all existing call sites are unaffected. The RunConfig import is correctly annotation-only in anonymizer.py and correctly eager in __init__.py for public API re-export. Both DataDesigner paths (managed and caller-supplied) are exercised by dedicated unit tests. No logic is removed or restructured.

No files require special attention.

Important Files Changed

Filename	Overview
src/anonymizer/init.py	Exports RunConfig as a top-level public symbol with an eager import, consistent with the existing ModelProvider pattern.
src/anonymizer/interface/anonymizer.py	Adds data_designer_run_config parameter; RunConfig is correctly placed under TYPE_CHECKING so it is annotation-only; set_run_config is applied after the managed/supplied DataDesigner branch and before NddAdapter construction.
tests/interface/test_anonymizer_interface.py	Two new tests cover both code paths (Anonymizer-managed DataDesigner and caller-supplied DataDesigner), asserting set_run_config is called with the supplied RunConfig.

Sequence Diagram

%%{init: {'theme': 'neutral'}}%%
sequenceDiagram
    participant Caller
    participant Anonymizer
    participant DataDesigner

    Caller->>Anonymizer: "__init__(data_designer_run_config=RunConfig(...))"
    alt data_designer is None (managed)
        Anonymizer->>DataDesigner: DataDesigner(artifact_path, model_providers)
        DataDesigner-->>Anonymizer: instance
    else data_designer supplied by caller
        Caller->>Anonymizer: "__init__(data_designer=existing_dd, data_designer_run_config=RunConfig(...))"
        Anonymizer->>Anonymizer: "self._data_designer = existing_dd"
    end
    Anonymizer->>DataDesigner: set_run_config(data_designer_run_config)
    Anonymizer->>Anonymizer: "NddAdapter(data_designer=self._data_designer)"

Loading

%%{init: {'theme': 'base', 'themeVariables': {"darkMode": true, "background": "#0d1117", "primaryColor": "#21262d", "primaryTextColor": "#e6edf3", "primaryBorderColor": "#8b949e", "lineColor": "#8b949e", "textColor": "#e6edf3", "edgeLabelBackground": "#161b22", "actorBkg": "#21262d", "actorBorder": "#8b949e", "actorTextColor": "#e6edf3", "actorLineColor": "#8b949e", "signalColor": "#8b949e", "signalTextColor": "#e6edf3", "noteBkgColor": "#373320", "noteBorderColor": "#d4a72c", "noteTextColor": "#f0e6c0", "labelBoxBkgColor": "#21262d", "labelBoxBorderColor": "#8b949e", "labelTextColor": "#e6edf3", "loopTextColor": "#e6edf3", "activationBkgColor": "#30363d", "activationBorderColor": "#8b949e"}}}%%
sequenceDiagram
    participant Caller
    participant Anonymizer
    participant DataDesigner

    Caller->>Anonymizer: "__init__(data_designer_run_config=RunConfig(...))"
    alt data_designer is None (managed)
        Anonymizer->>DataDesigner: DataDesigner(artifact_path, model_providers)
        DataDesigner-->>Anonymizer: instance
    else data_designer supplied by caller
        Caller->>Anonymizer: "__init__(data_designer=existing_dd, data_designer_run_config=RunConfig(...))"
        Anonymizer->>Anonymizer: "self._data_designer = existing_dd"
    end
    Anonymizer->>DataDesigner: set_run_config(data_designer_run_config)
    Anonymizer->>Anonymizer: "NddAdapter(data_designer=self._data_designer)"

Loading

_{Reviews (3): Last reviewed commit: "refactor: defer run config type import" | Re-trigger Greptile}

[greptile-apps]

RunConfig is only used in the type annotation for data_designer_run_config — it is never instantiated or called at runtime. The file already uses from __future__ import annotations (PEP 563), which stringifies all annotations so the class doesn't need to be present at import time. The existing convention in this file is to put annotation-only data_designer imports under TYPE_CHECKING (see DataDesignerConfigBuilder on line 89). Moving RunConfig there avoids an unnecessary eager import of data_designer.config.run_config on every module load.

Suggested change

	from data_designer.config.models import ModelProvider
	from data_designer.config.run_config import RunConfig
	from data_designer.config.utils.io_helpers import load_config_file
	from data_designer.config.models import ModelProvider
	from data_designer.config.utils.io_helpers import load_config_file

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

[asteier2026]

Looks good.