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).
48 lines
1.8 KiB
Python
48 lines
1.8 KiB
Python
"""Verify compression trigger excludes reasoning/completion tokens (#12026).
|
|
|
|
Thinking models (GLM-5.1, QwQ, DeepSeek R1) inflate completion_tokens with
|
|
reasoning tokens that don't consume context window space. The compression
|
|
trigger must use only prompt_tokens so sessions aren't prematurely split.
|
|
"""
|
|
|
|
import types
|
|
|
|
|
|
def _make_agent_stub(prompt_tokens, completion_tokens, threshold_tokens):
|
|
"""Create a minimal stub that exercises the compression check path."""
|
|
compressor = types.SimpleNamespace(
|
|
last_prompt_tokens=prompt_tokens,
|
|
last_completion_tokens=completion_tokens,
|
|
threshold_tokens=threshold_tokens,
|
|
)
|
|
# Replicate the fixed logic from run_agent.py ~line 11273
|
|
if compressor.last_prompt_tokens > 0:
|
|
real_tokens = compressor.last_prompt_tokens # Fixed: no completion
|
|
else:
|
|
real_tokens = 0
|
|
return real_tokens, compressor
|
|
|
|
|
|
class TestCompressionTriggerExcludesReasoning:
|
|
def test_high_reasoning_tokens_should_not_trigger_compression(self):
|
|
"""With the old bug, 40k prompt + 80k reasoning = 120k > 100k threshold.
|
|
After the fix, only 40k prompt is compared — no compression."""
|
|
real_tokens, comp = _make_agent_stub(
|
|
prompt_tokens=40_000,
|
|
completion_tokens=80_000, # reasoning-heavy model
|
|
threshold_tokens=100_000,
|
|
)
|
|
assert real_tokens == 40_000
|
|
assert real_tokens < comp.threshold_tokens, (
|
|
"Should NOT trigger compression — only prompt tokens matter"
|
|
)
|
|
|
|
|
|
def test_zero_prompt_tokens_falls_back(self):
|
|
"""When provider returns 0 prompt tokens, real_tokens is 0 (fallback path)."""
|
|
real_tokens, _ = _make_agent_stub(
|
|
prompt_tokens=0,
|
|
completion_tokens=50_000,
|
|
threshold_tokens=100_000,
|
|
)
|
|
assert real_tokens == 0
|