mirror of
https://github.com/NousResearch/hermes-agent.git
synced 2026-07-30 19:09:28 +00:00
6b81590c55
Systematic prune per AGENTS.md test policy, one pass over every major test tree (gateway, hermes_cli, tools, agent, run_agent, plugins, cli, cron, tui_gateway, honcho/openviking, root-level): - DELETE: source-reading tests (read_text/getsource on prod files), change-detector tests (exact catalog counts, model-name snapshots, config version literals), mock-echo tests (assert a mock returns what it was told), assertion-free/trivial tests, near-duplicate parametrizations (boundaries + one representative kept), async/sync twin duplicates, cosmetic within-file variations. - KEEP (mandatory): security/redaction/approval guards, message-role alternation invariants, prompt-caching/deterministic-call-id invariants, issue-number regression tests (deduped), E2E tests. - 6 test files deleted outright (script-style/no-assert or fully redundant); conftest.py, fakes/, fixtures/ untouched. - tests/acp/conftest.py added: autouse fixture stubs the live models.dev/GitHub/Copilot/Anthropic inventory fetches that ACP server tests performed on every session create — test_server.py 147s → 3.4s, and the tests are now genuinely hermetic. - Sleep-based slowness shrunk where safe (codex_ttfb_watchdog, compression_concurrent_fork, etc.); no wall-clock assertion tightened. Verification: full hermetic suite via scripts/run_tests.sh — 2439 files, 31,130 tests passed, 0 failed, 0 flaky retries, 315s wall (baseline: 583s wall, 13,564s subprocess CPU).
84 lines
4 KiB
Python
84 lines
4 KiB
Python
"""Tests for MiniMax model validation via static catalog (issues #12611, #12460, #12399, #12547).
|
|
|
|
MiniMax and MiniMax-CN providers don't expose /v1/models, so validate_requested_model()
|
|
must validate against the static catalog instead of probing the live API.
|
|
"""
|
|
|
|
from unittest.mock import patch
|
|
|
|
import pytest
|
|
|
|
from hermes_cli.models import validate_requested_model
|
|
|
|
|
|
class TestMiniMaxModelValidation:
|
|
"""Test that validate_requested_model handles MiniMax providers correctly."""
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def _isolate_minimax(self):
|
|
"""Ensure MiniMax catalog is used even if a live /v1/models endpoint exists."""
|
|
# Simulate fetch_api_models returning None (i.e., /v1/models is unreachable),
|
|
# proving that the catalog path is taken.
|
|
probe_payload = {
|
|
"models": None,
|
|
"probed_url": "https://api.minimax.io/v1/models",
|
|
"resolved_base_url": "https://api.minimax.io/v1",
|
|
"suggested_base_url": None,
|
|
"used_fallback": False,
|
|
}
|
|
with patch("hermes_cli.models.fetch_api_models", return_value=None), \
|
|
patch("hermes_cli.models.probe_api_models", return_value=probe_payload):
|
|
yield
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Test 1: A known MiniMax model is accepted with recognized=True
|
|
# -------------------------------------------------------------------------
|
|
def test_valid_minimax_model_accepted(self):
|
|
result = validate_requested_model("MiniMax-M2.7", "minimax")
|
|
assert result["accepted"] is True
|
|
assert result["persist"] is True
|
|
assert result["recognized"] is True
|
|
assert result["message"] is None
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Test 1b: Case-insensitive lookup matches catalog entries
|
|
# -------------------------------------------------------------------------
|
|
|
|
def test_valid_minimax_model_uppercase(self):
|
|
result = validate_requested_model("MINIMAX-M2.7", "minimax")
|
|
assert result["accepted"] is True
|
|
assert result["recognized"] is True
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Test 2: A near-match model on minimax-cn triggers a suggestion (not auto-correct)
|
|
# -------------------------------------------------------------------------
|
|
def test_near_match_minimax_cn_suggests_similar(self):
|
|
# "MiniMax-M2.7-highspeed" is somewhat similar to "MiniMax-M2.7" (ratio ~0.71)
|
|
# but below the 0.9 auto-correct cutoff. It should be accepted with a
|
|
# recognized=False and a similar-models suggestion (ratio > 0.5).
|
|
result = validate_requested_model("MiniMax-M2.7-highspeed", "minimax-cn")
|
|
assert result["accepted"] is True
|
|
assert result["persist"] is True
|
|
assert result["recognized"] is False
|
|
# Should NOT auto-correct (ratio 0.71 < 0.9)
|
|
assert "corrected_model" not in result
|
|
# But should suggest similar models (ratio 0.71 > 0.5)
|
|
assert "MiniMax-M2.7" in result["message"]
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Test 3: A completely unknown model is accepted (not rejected) with a warning
|
|
# -------------------------------------------------------------------------
|
|
|
|
# -------------------------------------------------------------------------
|
|
# Test 4: Verify catalog path is used (probe_api_models returns None)
|
|
# -------------------------------------------------------------------------
|
|
def test_minimax_uses_catalog_not_api_probe(self):
|
|
"""Ensure that when fetch_api_models returns None, the catalog is still checked."""
|
|
# The _isolate_minimax fixture already patches fetch_api_models to return None.
|
|
# If we reach the catalog path, MiniMax-M2.5 should be found and recognized.
|
|
result = validate_requested_model("MiniMax-M2.5", "minimax")
|
|
assert result["accepted"] is True
|
|
assert result["recognized"] is True
|
|
assert result["message"] is None
|
|
|
|
|