core.pipeline_runtime¶

Shared runtime abstraction for executing pipelines on audio files.

triton.core.pipeline_runtime provides the core execution engine for applying pipelines. It handles:

Run ID generation and directory organization
Step execution and output isolation
Pipeline file writing and patching
Run metadata and logging

This module is used by both the CLI and GUI to ensure consistent pipeline behavior across interfaces.

Key Functions¶

`new_pipeline_run_id() -> str`¶

Generate a new unique run ID based on current timestamp.

Returns a string like run_20260331_143022_abc123.

`pipeline_run_dir(project: Project, pipeline_name: str, run_id: str) -> Path`¶

Get the output directory for a pipeline run.

Returns project.path / data/derived/pipelines/<pipeline_name>/<run_id>.

`run_pipeline_on_file(project: Project, pipeline: Pipeline, input_file: Path, run_id: str, row_id: str | None = None, overrides: dict[str, dict[str, object]] | None = None) -> dict`¶

Execute a pipeline on a single audio file with optional parameter overrides.

Parameters: - project: Loaded project config - pipeline: Pipeline definition - input_file: Path to input audio file - run_id: Run identifier - row_id: Optional row label (e.g., row_0, row_1) for matrix runs - overrides: Step-level parameter overrides ({step_name: {option: value}})

Returns: Dict with execution metadata including outputs and step logs.

Usage Example¶

from pathlib import Path
from triton.core import new_pipeline_run_id, run_pipeline_on_file
from triton.core.project import load_project_config, load_project_pipelines

project = load_project_config("my-project")
pipelines = load_project_pipelines("my-project")
pipeline = pipelines[0]

run_id = new_pipeline_run_id()
result = run_pipeline_on_file(
    project, 
    pipeline, 
    input_file=Path("my-project/data/raw/audio.wav"),
    run_id=run_id,
    overrides={"0": {"target_peak": 0.8}},  # Override step 0's target_peak
)

print(f"Run completed: {result}")

Design Notes¶

Each run is organized into isolated subdirectories to prevent output collisions.
Parameter overrides are step-level (by index or name) for cleaner CLI/UI design.
Run metadata is logged in run.log.jsonl for reproducibility and debugging.