init_project.py

import os

PROJECT_ROOT = "."

DIRS = [
    f"{PROJECT_ROOT}/core",
    f"{PROJECT_ROOT}/config",
    f"{PROJECT_ROOT}/data/uploads",
    f"{PROJECT_ROOT}/data/vector_stores",
    f"{PROJECT_ROOT}/utils",
    f"{PROJECT_ROOT}/tests",
]

TEST_PIPELINE = """\
import pytest
from config.pipeline_configs import ALL_PIPELINES

def test_pipeline_registry():
    assert len(ALL_PIPELINES) >= 4
    for key, cfg in ALL_PIPELINES.items():
        assert cfg.chunk_size > 0
        assert cfg.top_k > 0
"""

if __name__ == "__main__":
    for d in DIRS:
        os.makedirs(d, exist_ok=True)
        # __init__.py for packages
        if "data" not in d:
            init_path = os.path.join(d, "__init__.py")
            if not os.path.exists(init_path):
                open(init_path, "w").close()

    with open(f"{PROJECT_ROOT}/tests/test_pipeline.py", "w") as f:
        f.write(TEST_PIPELINE)

    print("✅ Phase 1 project skeleton created.")
    print("Next steps:")
    print("1. cd rag_optimizer")
    print("2. python -m venv venv")
    print("3. source venv/bin/activate  # or venv\\\\Scripts\\\\activate on Windows")
    print("4. pip install -r requirements.txt")
    print("5. cp .env.example .env && fill your keys")