| |
| |
| |
| |
| |
| |
| |
|
|
| """Adapt benchmark CLI arguments to reusable application/report services.""" |
|
|
| from __future__ import annotations |
|
|
| import argparse |
| import json |
| from pathlib import Path |
|
|
| from speech_text_data_aligners.application.benchmark import ( |
| BenchmarkOptions, |
| benchmark_backends, |
| ) |
| from speech_text_data_aligners.backends.registry import ( |
| BackendRegistry, |
| default_backend_registry, |
| ) |
| from speech_text_data_aligners.benchmarking.reports import ( |
| preflight_report_bundle, |
| render_report, |
| write_report_bundle, |
| ) |
| from speech_text_data_aligners.core.errors import ConfigurationError |
|
|
|
|
| def configure_parser(parser: argparse.ArgumentParser) -> None: |
| """Add benchmark composition/presentation arguments to a subcommand parser. |
| |
| Parameters |
| ---------- |
| parser: Parser owned by the CLI composition root. |
| |
| Side Effects: |
| Registers arguments and the command handler on ``parser``. |
| |
| """ |
| parser.add_argument("metadata_csv", type=Path) |
| parser.add_argument("reference_csv", type=Path) |
| parser.add_argument("audio_dir", type=Path) |
| parser.add_argument("--backend", action="append", required=True) |
| parser.add_argument("--backend-config-json", default="{}") |
| parser.add_argument("--strict", action="store_true") |
| parser.add_argument("--output-base", type=Path) |
| parser.add_argument("--overwrite", action="store_true") |
| parser.add_argument( |
| "--format", choices=("json", "csv", "markdown"), default="markdown" |
| ) |
| parser.set_defaults(handler=run) |
|
|
|
|
| def run( |
| args: argparse.Namespace, |
| *, |
| registry: BackendRegistry | None = None, |
| ) -> int: |
| """Construct selected backends and delegate benchmark work to library services. |
| |
| Parameters |
| ---------- |
| args: Parsed benchmark arguments. |
| registry: Optional injected lazy registry for tests/custom composition. |
| |
| Returns |
| ------- |
| Zero after writing a bundle or printing the requested report format. |
| |
| Raises |
| ------ |
| ConfigurationError: If backend names repeat or configuration JSON is invalid. |
| SpeechTextAlignerError: Propagated typed construction/report/service failures. |
| |
| Side Effects: |
| Calls backends and either writes a report bundle or prints one report to stdout. |
| |
| """ |
| if len(args.backend) != len(set(args.backend)): |
| raise ConfigurationError("benchmark backend names must be unique") |
| try: |
| config = json.loads(args.backend_config_json) |
| except json.JSONDecodeError as error: |
| raise ConfigurationError("backend configuration is not valid JSON") from error |
| if not isinstance(config, dict): |
| raise ConfigurationError("backend configuration JSON must be an object") |
| if args.output_base is not None: |
| |
| preflight_report_bundle( |
| args.output_base, |
| metadata_csv=args.metadata_csv, |
| reference_csv=args.reference_csv, |
| overwrite=args.overwrite, |
| ) |
| selected_registry = registry or default_backend_registry() |
| backends = { |
| name: selected_registry.create(name, config=config.get(name, {})) |
| for name in args.backend |
| } |
| report = benchmark_backends( |
| backends=backends, |
| metadata_csv=args.metadata_csv, |
| reference_csv=args.reference_csv, |
| audio_dir=args.audio_dir, |
| options=BenchmarkOptions(strict=args.strict), |
| ) |
| if args.output_base is not None: |
| write_report_bundle( |
| report, |
| args.output_base, |
| metadata_csv=args.metadata_csv, |
| reference_csv=args.reference_csv, |
| overwrite=args.overwrite, |
| ) |
| else: |
| print(render_report(report, args.format), end="") |
| return 0 |
|
|
