Catty Steve
84b3b7d8e8
Adds a complete pytest system test suite for the HoneyBiscuitWorkshop CI/CD system: - runner.sh: orchestrates build, Docker image creation, and test execution - Dockerfile: Arch Linux-based test image with Python venv - conftest.py: shared fixtures (baker_bin, run_baker, work_dir, llm_env) - probes/: environment probes for checking users, files, packages, processes - llm/: LLM-based log judgment system with caching for semantic test evaluation - fixtures/: test configs and scripts for prebake, bake, and finalize stages - test_prebake.py, test_bake.py, test_finalize.py: test suites for each pipeline stage Also refactors finalize.yml.tmpl notification config to use a templates system.
13 lines
353 B
Python
13 lines
353 B
Python
"""
|
|
LLM-based log judgment system for system tests.
|
|
|
|
Uses LLM APIs to evaluate non-deterministic test outputs (logs, stderr, etc.)
|
|
where simple string matching is insufficient.
|
|
"""
|
|
|
|
from .client import LLMClient
|
|
from .cache import LLMCache
|
|
from .judge import judge_log, JudgmentResult
|
|
|
|
__all__ = ["LLMClient", "LLMCache", "judge_log", "JudgmentResult"]
|