SynthesizeData Command

Command that generates synthetic data using the DataSynthesizer registry and stores it in DataManager

Overview

SynthesizeData is a non-undoable command that generates synthetic data using a registered generator from the DataSynthesizer module and stores the result in DataManager. This enables data synthesis to participate in command sequences, pipelines, and scripted workflows.

Source files:

  • src/Commands/SynthesizeData.hpp / .cpp
  • src/Commands/SynthesizeData.test.cpp

All code lives in the commands namespace with no Qt dependency.

Parameters

struct SynthesizeDataParams {
    std::string output_key;      // DataManager key for the result
    std::string generator_name;  // Registry lookup (e.g., "SineWave")
    std::string output_type;     // Data type hint (e.g., "AnalogTimeSeries")
    rfl::Generic parameters;     // Generator-specific params (forwarded as JSON)
    std::string time_key = "time"; // TimeFrame association
};
Field Description
output_key Key under which the generated data will be stored in DataManager
generator_name Name of a registered generator (e.g., "SineWave", "GaussianNoise")
output_type Data type string for metadata/filtering (e.g., "AnalogTimeSeries")
parameters Generator-specific parameters as a JSON object, forwarded to the generator
time_key TimeFrame key to associate with the generated data (defaults to "time")

Supported Generators

Any generator registered in the GeneratorRegistry. As of Milestone 1b:

  • SineWave — Periodic sine wave
  • SquareWave — Periodic square wave with configurable duty cycle
  • TriangleWave — Periodic triangle wave
  • GaussianNoise — i.i.d. Gaussian noise (deterministic with seed)
  • UniformNoise — i.i.d. uniform noise (deterministic with seed)

New generators are automatically available without modifying this command.

Undo

Not undoable (isUndoable() returns false).

Execution Flow

  1. Look up the generator by generator_name in GeneratorRegistry::instance().
  2. If the generator is not found, return CommandResult::error().
  3. Serialize the parameters field to a JSON string.
  4. Call GeneratorRegistry::generate(name, params_json).
  5. If generation fails, return CommandResult::error().
  6. Store the result via DataManager::setData(output_key, variant, TimeKey(time_key)).
  7. Return CommandResult::ok({output_key}).

Example JSON

{
    "command_name": "SynthesizeData",
    "parameters": {
        "output_key": "test_sine",
        "generator_name": "SineWave",
        "output_type": "AnalogTimeSeries",
        "parameters": {
            "num_samples": 1000,
            "amplitude": 2.0,
            "frequency": 0.01,
            "phase": 0.0,
            "dc_offset": 0.0
        },
        "time_key": "time"
    }
}

Linker Requirements

Because DataSynthesizer generators use static registration (RAII objects in anonymous namespaces), any binary that executes SynthesizeData must link the DataSynthesizer library with --whole-archive (Linux), -force_load (macOS), or /WHOLEARCHIVE (MSVC). Without this, the linker discards “unused” translation units and generators will not be registered at runtime.

Testing

Tests are in src/Commands/SynthesizeData.test.cpp and exercise:

  • Factory creation from JSON and rfl::Generic
  • Execution with a SineWave generator
  • Determinism: same seed produces identical output
  • Error handling for unknown generators
  • Serialization round-trip (toJson → createCommandFromJson)
  • Custom time_key propagation
  • Introspection (getAvailableCommands, isKnownCommandName)