Quick Start¶

Install¶

After the package is released on PyPI:

python -m pip install legopolitics

For a local source checkout:

python -m pip install -e ".[dev]"

Check the environment¶

legopolitics doctor

This reports Python, FFmpeg, PyTorch, CUDA, optional adapters, output-directory access, and release-review status without exposing secret tokens.

Run the lightweight pipeline¶

legopolitics analyze-video \
  --video input.mp4 \
  --config configs/minimal.yaml \
  --output outputs/input

The minimal configuration avoids large model downloads. It performs probing, hybrid frame sampling, quality and composition measurements, manifests, and research exports.

Inspect the result¶

legopolitics validate --output outputs/input
legopolitics inspect-run --manifest outputs/input/run_manifest.sqlite

Open outputs/input/video_summary.xlsx for the researcher-facing workbook and outputs/input/report.html for the local report.

Python API¶

from pathlib import Path

from legopolitics import AnalysisConfig, LegoPoliticsAnalyzer

config = AnalysisConfig.from_yaml(Path("configs/minimal.yaml"))
result = LegoPoliticsAnalyzer(config).analyze_video(
    video_path=Path("input.mp4"),
    output_dir=Path("outputs/input"),
)

print(result.output_paths)

Continue with the complete tutorial to add a custom detector, tracking, vision-language models, OCR, audio, research context, narrative coding, and validation.