Results database with MLFlow

[stevenstetzler]

Address #1

[Copilot]

Pull Request Overview

This PR integrates MLflow for experiment tracking and result management, addressing issue #1. It refactors the main search function to be callable programmatically and adds database storage capabilities for search results using SQLAlchemy.

Key changes:

• Extracted core search logic into run_search() function for programmatic access
• Added MLflow integration via run_search_mlflow() for experiment tracking
• Implemented database models and result compilation utilities for persistent storage

Reviewed Changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tests/test_search.py	Added tests for the new run_search and run_search_mlflow functions
src/find_asteroids/search.py	Refactored main() to extract run_search() and added run_search_mlflow() wrapper with MLflow tracking
src/find_asteroids/results.py	New module for reading and compiling results from local directories or MLflow, with database storage support
src/find_asteroids/models.py	New SQLAlchemy models for storing experiments, runs, and search results in a database
pyproject.toml	Added sqlalchemy dependency and mlflow as optional pipeline dependency

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The loop iterates over tags without validation. If tags is an empty list (the default), this will work, but if a tag is not a two-element tuple, this will fail. While the CLI parsing at line 353 ensures tuples, when called directly via run_search_mlflow(), invalid tag formats could cause an unpacking error.

Suggested change

	for k, v in tags:
	for tag in tags:
	if not (isinstance(tag, (tuple, list)) and len(tag) == 2):
	raise ValueError(f"Each tag must be a tuple or list of length 2, got: {tag!r}")
	k, v = tag

[Copilot]

Use the logging module instead of print statements for consistency with the rest of the codebase. Import and use log.info() or log.debug() instead.

Suggested change

	def read_results(results_dir : Path, name : str, output_format='ecsv'):
	import astropy.table
	if type(results_dir) is str:
	results_dir = Path(results_dir)
	for p in sorted(results_dir.glob("*/"+f"{name}.{output_format}"), key=lambda x: int(x.parent.name)):
	print("reading", p)
	import logging
	log = logging.getLogger(__name__)
	def read_results(results_dir : Path, name : str, output_format='ecsv'):
	import astropy.table
	if type(results_dir) is str:
	results_dir = Path(results_dir)
	for p in sorted(results_dir.glob("*/"+f"{name}.{output_format}"), key=lambda x: int(x.parent.name)):
	log.info(f"reading {p}")