mirror of
https://github.com/turnstonelabs/turnstone.git
synced 2026-08-12 23:12:23 -06:00
Patrick Buckley
63e9205e83
reconstruct_messages(repair=True) did two things: strip a trailing incomplete tool-call turn AND synthesize cancellation results for mid-conversation orphans. The mid-orphan synth was a near-duplicate of lowering.repair_wire_messages — same detector, same contiguous insert past interspersed system turns, same cancellation string — running at the wrong layer (storage, on every load). Drop it: load is now trailing-strip only (boot-crash recovery), and the mid-orphan synth happens once, at send, in lowering.repair_wire_messages — the single place the wire path fills orphans. The session send path gets it via _prepare_wire_messages; export, which bypasses that path, now runs repair_wire_messages itself (otherwise a mid-conversation orphan would serialize as an unanswered tool_call). The duplicated cancellation string goes with the synth — CANCELLED_TOOL_RESULT lives only in lowering now. Safe: a bare mid-orphan is harmless between load and send (token count is additive, /history reads repair=False, compaction summarizes to text), and every wire path repairs it. Reconstruct tests updated to the new load contract; an export mid-orphan test added.
215 lines
8.9 KiB
Python
215 lines
8.9 KiB
Python
"""Unit tests for the workstream export serializer (issue #613).
|
|
|
|
Drives through a REAL storage backend (the ``backend`` fixture is a
|
|
SQLite ``StorageBackend``): seed workstreams / messages / attachments,
|
|
call :func:`export_workstream`, parse the returned bytes, and assert
|
|
structural facts. No hand-built message dicts are injected straight
|
|
into the serializer as the sole gate — the pipeline order (attach
|
|
reasoning → sanitize) is what these tests guard.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import io
|
|
import json
|
|
import zipfile
|
|
|
|
from turnstone.core.export import (
|
|
WorkstreamNotFoundError,
|
|
_attach_reasoning_content,
|
|
_build_openai_json,
|
|
export_workstream,
|
|
)
|
|
|
|
USER = "u1"
|
|
|
|
|
|
def _assistants(messages: list[dict]) -> list[dict]:
|
|
return [m for m in messages if m.get("role") == "assistant"]
|
|
|
|
|
|
def _parse_messages(data: bytes) -> list[dict]:
|
|
return json.loads(data)["messages"]
|
|
|
|
|
|
def _seed_interactive_turn(backend, ws_id: str) -> None:
|
|
"""user + assistant(tool_call) + tool + assistant."""
|
|
tc = [
|
|
{
|
|
"id": "call_a1",
|
|
"type": "function",
|
|
"function": {"name": "run", "arguments": "{}"},
|
|
}
|
|
]
|
|
backend.register_workstream(ws_id, user_id=USER, title="T", kind="interactive")
|
|
backend.save_message(ws_id, "user", "go")
|
|
backend.save_message(ws_id, "assistant", "working", tool_calls=json.dumps(tc))
|
|
backend.save_message(ws_id, "tool", "ran ok", tool_name="run", tool_call_id="call_a1")
|
|
backend.save_message(ws_id, "assistant", "done")
|
|
|
|
|
|
def test_openai_json_envelope_shape(backend):
|
|
_seed_interactive_turn(backend, "ws1")
|
|
result = export_workstream(backend, "ws1")
|
|
|
|
assert result.content_type == "application/json"
|
|
assert result.filename == "ws1.json"
|
|
top_keys = sorted(json.loads(result.data).keys())
|
|
assert top_keys == ["messages"]
|
|
|
|
|
|
def test_reasoning_content_present_thinking(backend):
|
|
pc = [{"type": "thinking", "thinking": "R1", "signature": "sig"}]
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "ok", provider_data=json.dumps(pc))
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
reasoning = [m.get("reasoning_content") for m in _assistants(messages)]
|
|
assert reasoning == ["R1"]
|
|
|
|
|
|
def test_reasoning_content_present_reasoning_text(backend):
|
|
pc = [{"type": "reasoning_text", "text": "R2", "source": "synth"}]
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "ok", provider_data=json.dumps(pc))
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
reasoning = [m.get("reasoning_content") for m in _assistants(messages)]
|
|
assert reasoning == ["R2"]
|
|
|
|
|
|
def test_reasoning_content_present_responses(backend):
|
|
pc = [{"type": "reasoning", "summary": [{"type": "summary_text", "text": "R3"}]}]
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "ok", provider_data=json.dumps(pc))
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
reasoning_text = _assistants(messages)[0].get("reasoning_content")
|
|
assert reasoning_text is not None
|
|
assert "R3" in reasoning_text
|
|
|
|
|
|
def test_no_underscore_keys_leak(backend):
|
|
pc = [{"type": "thinking", "thinking": "R1", "signature": "sig"}]
|
|
_seed_interactive_turn(backend, "ws1")
|
|
backend.save_message("ws1", "assistant", "more", provider_data=json.dumps(pc))
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
leaked = sorted({k for m in messages for k in m if isinstance(k, str) and k.startswith("_")})
|
|
assert leaked == []
|
|
|
|
|
|
def test_image_url_kept_document_inlined(backend):
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
msg_id = backend.save_message("ws1", "user", "see attached")
|
|
backend.save_attachment("att_img", "ws1", USER, "pic.png", "image/png", 4, "image", b"\x89PNG")
|
|
backend.save_attachment("att_doc", "ws1", USER, "notes.txt", "text/plain", 5, "text", b"hello")
|
|
backend.set_message_attachments("ws1", msg_id, ["att_img", "att_doc"])
|
|
backend.save_message("ws1", "assistant", "got it")
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
user_msg = next(m for m in messages if m.get("role") == "user")
|
|
parts = user_msg["content"]
|
|
part_types = [p.get("type") for p in parts]
|
|
document_texts = [
|
|
p.get("text", "")
|
|
for p in parts
|
|
if p.get("type") == "text" and "<document name=" in p.get("text", "")
|
|
]
|
|
|
|
assert "image_url" in part_types
|
|
assert document_texts != []
|
|
|
|
|
|
def test_assistant_without_reasoning_has_no_reasoning_content(backend):
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "ok")
|
|
|
|
messages = _parse_messages(export_workstream(backend, "ws1").data)
|
|
assistant = _assistants(messages)[0]
|
|
assert "reasoning_content" not in assistant
|
|
|
|
|
|
def test_coordinator_zip_parent_plus_children(backend):
|
|
backend.register_workstream("coord", user_id=USER, title="C", kind="coordinator")
|
|
backend.save_message("coord", "user", "coordinate")
|
|
backend.save_message("coord", "assistant", "spawning")
|
|
backend.register_workstream("c1", user_id=USER, kind="interactive", parent_ws_id="coord")
|
|
backend.register_workstream("c2", user_id=USER, kind="interactive", parent_ws_id="coord")
|
|
for child in ("c1", "c2"):
|
|
backend.save_message(child, "user", "do x")
|
|
backend.save_message(child, "assistant", "x done")
|
|
|
|
result = export_workstream(backend, "coord", children=True)
|
|
assert result.content_type == "application/zip"
|
|
assert result.filename == "coord.zip"
|
|
|
|
zf = zipfile.ZipFile(io.BytesIO(result.data))
|
|
names = sorted(zf.namelist())
|
|
expected = sorted(["coord.json", "children/c1.json", "children/c2.json"])
|
|
assert names == expected
|
|
|
|
top_keys = [sorted(json.loads(zf.read(name)).keys()) for name in names]
|
|
assert top_keys == [["messages"], ["messages"], ["messages"]]
|
|
|
|
|
|
def test_coordinator_default_parent_only(backend):
|
|
backend.register_workstream("coord", user_id=USER, title="C", kind="coordinator")
|
|
backend.save_message("coord", "user", "coordinate")
|
|
backend.save_message("coord", "assistant", "done")
|
|
backend.register_workstream("c1", user_id=USER, kind="interactive", parent_ws_id="coord")
|
|
|
|
result = export_workstream(backend, "coord", children=False)
|
|
assert result.content_type == "application/json"
|
|
assert result.filename == "coord.json"
|
|
|
|
|
|
def test_export_unknown_ws_raises(backend):
|
|
try:
|
|
export_workstream(backend, "does-not-exist")
|
|
except WorkstreamNotFoundError as exc:
|
|
assert "does-not-exist" in str(exc)
|
|
else:
|
|
raise AssertionError("expected WorkstreamNotFoundError")
|
|
|
|
|
|
def test_attach_reasoning_runs_before_sanitize(backend):
|
|
pc = [{"type": "thinking", "thinking": "R1", "signature": "sig"}]
|
|
backend.register_workstream("ws1", user_id=USER, kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "ok", provider_data=json.dumps(pc))
|
|
|
|
attached = _attach_reasoning_content(backend.load_messages("ws1", repair=True))
|
|
assistant = _assistants(attached)[0]
|
|
# Pre-sanitize: reasoning stamped AND the raw provider lane still present.
|
|
assert assistant.get("reasoning_content") == "R1"
|
|
assert "_provider_content" in assistant
|
|
|
|
# Full pipeline output: provider lane is gone, reasoning survives.
|
|
messages = _parse_messages(_build_openai_json(backend, "ws1"))
|
|
leaked = [k for m in messages for k in m if isinstance(k, str) and k.startswith("_")]
|
|
assert leaked == []
|
|
assert _assistants(messages)[0].get("reasoning_content") == "R1"
|
|
|
|
|
|
def test_mid_orphan_tool_call_exports_with_cancellation(backend):
|
|
"""A mid-conversation orphaned tool_call (no result) exports with a
|
|
synthesized cancellation: export bypasses the session send path, so it runs
|
|
the send-time orphan repair itself (load is trailing-strip only)."""
|
|
tc = [{"id": "call_x", "type": "function", "function": {"name": "run", "arguments": "{}"}}]
|
|
backend.register_workstream("ws1", user_id=USER, title="T", kind="interactive")
|
|
backend.save_message("ws1", "user", "go")
|
|
backend.save_message("ws1", "assistant", "working", tool_calls=json.dumps(tc))
|
|
# A user turn after the orphan keeps it mid-conversation (not stripped).
|
|
backend.save_message("ws1", "user", "never mind")
|
|
|
|
messages = _parse_messages(_build_openai_json(backend, "ws1"))
|
|
tool_msgs = [m for m in messages if m.get("role") == "tool"]
|
|
assert len(tool_msgs) == 1
|
|
assert tool_msgs[0]["tool_call_id"] == "call_x"
|
|
assert "cancelled" in tool_msgs[0]["content"].lower()
|