Raven-1/tests/test_rpc_server_socket.py
Handsome-wzw 7b45527b5b fix(tools): an array sent as its own JSON is still an array
## Summary

Reported as "deliver_files is broken". It is not: **the tool never ran.**

Every `deliver_files` call on one install was refused before dispatch -- five in a row in a single session -- and nothing was ever delivered:

```
Error: Invalid parameters for tool 'deliver_files': files should be array
```

What the model sent was the array, spelled as a string:

```json
{"files": "[{\"path\": \"...poster.png\", \"title\": \"...\"}]", "message": "..."}
```

`cast_params` already forgives that spelling for every scalar type -- `"3"` for an integer, `"true"` for a boolean -- and did not for arrays or objects, which are exactly the types a model is most likely to stringify.

## Type

- [x] Fix
- [ ] Feature

## The edge of the forgiving is the design

When the value is a string, try `json.loads`, and take the result **only if it is the type the schema asked for**. A string that is not JSON, or is JSON of the wrong shape, is returned unchanged so the validator still names the real problem -- the point is to widen what is accepted, not to swallow a type error.

**Nothing here can raise.** Turning "your argument has the wrong type" into a traceback would be worse than the refusal it replaces, so the parse is wrapped and failure falls through.

**A `type: "string"` field is never parsed.** JSON is ordinary text in a message body or a file's content, and replacing it would hand the tool something it never asked for. Two independent things prevent it today (the string branch returns before the parse is reached, and the parse only wins on a type match), which is why the case pinning it is a tripwire rather than a load-bearing assertion -- said so in its own docstring rather than left to look like coverage.

## Not just deliver_files

Six built-in tools declare an array or object parameter and were all exposed:

| tool | parameter |
|---|---|
| `deliver_files` | `files` |
| `ask_user` | `questions` |
| `image_generate` | `images` |
| `message` | `media` |
| `create_playbook` | `skills` |
| `video_generate` | `params` |

The same failure was already seen from the other side: `acp_client/acp_agent.py` records `update_task_state` and `operations should be array` from a real dispatch.

## Verification

```
make lint-python    ruff check + ruff format --check, 1850 files
make lint-imports   9 contracts kept, 0 broken
pytest tests/       19,699 passed, 7 failed (all pre-existing, see below)
```

- [x] Relevant tests pass locally
- [x] Relevant lint / type checks pass locally
- [ ] User-facing docs or screenshots -- no user-facing string

**End to end, through the real tool**, with the exact payload that failed five times in production:

```
files sent as:            str -> [{"path": ".../poster.png", "title": "..."}]
returned:                 Delivered 1 file(s): poster.png (72 B).
still "should be array":  False
registered in the store:  True
```

Twelve cases, five mutations, each killing at least one:

| mutation | red |
|---|---|
| drop the whole branch (the old behaviour) | 3 |
| raise instead of falling through on a parse failure | 5 |
| accept the parse whatever type it yielded | 1 |
| do not re-cast the parsed value, so items are not cast | 1 |
| widen the tuple to include `string` | **survived -- unreachable** |

The last is worth stating rather than hiding: that mutation cannot change behaviour, because the `type == "string"` branch returns earlier and the type check would refuse the result anyway. I also tried moving the new block above that branch, alone and combined with widening the tuple, and neither reddens anything either. The invariant holds by construction; the case that documents it says so.

**Pre-existing failures, checked against the base.** 7 failed in the full run, all in `tests/test_security_network.py` (`test_a_redirect_to_a_dangerous_address_is_refused` and siblings). Running that file alone on this same tree gives 4 -- the count is order-dependent -- and the change touches only `raven/agent/tools/params.py` plus a new test file, neither of which `raven/security/network.py` imports. The failures are the box's resolver: it returns poisoned answers roughly half the time, which I measured separately (`en.wikipedia.org` came back as `2001::1` on 3 of 6 lookups). Not introduced here, and worth someone's attention on its own.

## Risk

- [x] Security impact considered -- widens what reaches a tool's own validation, not what bypasses it. A parsed value is still validated against the same schema, and a parse that yields the wrong type is refused exactly as before
- [x] Backward compatibility considered -- a call that worked before takes the identical path; the new branch is reached only by a `str` where an array or object is declared, which previously always failed
- [x] Rollback path is clear -- single-commit revert

## Related Issues

N/A

## Swept

`mr-review-patterns` pre-submit sweep, over `git diff origin/main...HEAD`.

- **1 (a write location moved)** -- nothing moved; a branch was added to an existing dispatch on `target_type`. Grepped every caller of `cast_params`: the registry is the only one, and it runs `validate_params` right after, which is the reader this affects.
- **2 (a value with several entry points)** -- the cast is the single funnel every tool argument passes through, which is why the fix belongs here and not in `deliver_files`.
- **3 (a guard)** -- checked both directions: what it wrongly lets through (JSON of the wrong type, non-JSON text -- both refused, with cases) and what it wrongly refuses (a real list, unaffected, with a case).
- **4 (tests)** -- five mutations above, one honestly reported as unreachable. Failing-ID sets diffed against the base rather than counts.
- **6 (stale text)** -- the module docstring enumerated what is coerced ("a string that should be an integer"); it now names arrays and objects too. Grepped the tracked tree for other statements of the rule: none.
- **9 (error handling)** -- the new path adds one way to end and it is caught: `json.loads` raises `ValueError` on bad text and `TypeError` on a non-string, and both fall through to the untouched value.
- **11** -- `git ls-files | grep -c node_modules` is 0; the commit-message gate was run, not eyeballed.

Not applicable, with reasons: **3b, 3c, 3d, 3e** (no tool schema produced, no state moved, one lane, no assertion lost its symbol); **4b, 4c, 4d** (no platform branch, no shell, no rendered number); **5, 5b** (no dependency; `json` is stdlib and already imported across this package); **6b, 6c** (no backup, no generated artifact); **7, 7b** (nothing on disk); **8** (no asset, no secret); **10** (no prompt text).

## Deliberately not fixed

The error text still ends with `[Analyze the error above and try a different approach.]`, and the model plainly did not -- it sent the same shape five times. A message naming the expected spelling (`files must be a JSON array, not a string containing one`) would give it something to act on. That is a change to the registry's error surface, affects every tool, and deserves its own argument.

## Moved from !594

Opened on a compliant source branch. AGENTS.md 2.1 asks for a 3-5 word snake_case description and the previous branch had seven; GitLab cannot change an existing merge request's source branch, so the change moves rather than the name.

Same commits, rebased onto `c2a9768f`. Both code findings from the review on !594 are already in them -- `RecursionError` caught with its regression case, and the implementation comment reduced to the two non-obvious reasons. Nothing else changed, so that review carries.
2026-09-09 15:42:36 +00:00

274 lines
10 KiB
Python

"""Regression: RpcServer must speak proper JSON-RPC over a unix domain socket.
P0 dogfood blocker (2026-05-15) — typing ``/cha`` in ``raven tui`` produced
repeating ``pipe closed by peer or os.write(pipe, data) raised exception.``
warnings (asyncio.unix_events) that bled into the Ink render. Root cause:
``connect_write_pipe`` builds a ``_UnixWritePipeTransport`` that registers a
reader callback on the WRITE fd to detect peer EOF; on a bidirectional
SOCK_STREAM socket the very first inbound byte trips that callback and the
write side silently closes. Every subsequent RPC response is dropped, and
after 5 dropped writes asyncio starts logging the "pipe closed" warning.
The production transport in ``tui_commands.run_subprocess_with_rpc`` is a
unix socket whose fd is dup'd into ``request_fd`` and ``notify_fd``, so this
bug was triggered on every interactive session. The fix is to detect the
socket fds and use ``connect_accepted_socket`` (full-duplex selector
transport) instead of the pipe pair.
These tests exercise the bug directly without spawning Node:
1. ``test_socket_roundtrip_after_inbound_byte`` — sends a request, reads the
response. Pre-fix this hangs because the write transport closes on the
first inbound byte and the response never reaches the client.
2. ``test_socket_no_asyncio_pipe_warnings`` — sends three requests in
sequence and asserts the asyncio.unix_events logger emits no
``"pipe closed"`` warning.
"""
from __future__ import annotations
import asyncio
import json
import logging
import socket
import tempfile
from collections.abc import Iterator
from pathlib import Path
import pytest
from raven.rpc.dispatcher import Dispatcher
from raven.rpc.methods import register_aligned_methods
from raven.rpc.server import RpcServer
# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
async def _wire_paired_socket() -> tuple[socket.socket, socket.socket, socket.socket, Path]:
"""Build the same fd topology that ``run_subprocess_with_rpc`` builds.
Returns ``(listening_sock, client_sock, server_conn, tmp_dir)``; the caller
hands ``server_conn`` to ``RpcServer`` the way the production path does.
"""
tmp = Path(tempfile.mkdtemp(prefix="eve-test-rpc-"))
spath = tmp / "sock"
server_sock = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
server_sock.bind(str(spath))
server_sock.listen(1)
server_sock.setblocking(False)
client = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
client.setblocking(False)
loop = asyncio.get_running_loop()
await loop.sock_connect(client, str(spath))
conn, _ = await loop.sock_accept(server_sock)
conn.setblocking(False)
return server_sock, client, conn, tmp
async def _read_one_frame(client: socket.socket, *, timeout: float = 2.0) -> bytes:
"""Read a single newline-terminated frame from ``client``."""
loop = asyncio.get_running_loop()
buf = bytearray()
deadline = loop.time() + timeout
while not buf.endswith(b"\n"):
remaining = deadline - loop.time()
if remaining <= 0:
break
try:
chunk = await asyncio.wait_for(loop.sock_recv(client, 65536), timeout=remaining)
except asyncio.TimeoutError:
break
if not chunk:
break
buf.extend(chunk)
return bytes(buf)
async def _send(client: socket.socket, method: str, params: dict, rid: int) -> None:
loop = asyncio.get_running_loop()
frame = json.dumps({"jsonrpc": "2.0", "id": rid, "method": method, "params": params}) + "\n"
await loop.sock_sendall(client, frame.encode())
@pytest.fixture()
def _capture_asyncio_warnings() -> Iterator[list[str]]:
"""Capture WARNING records emitted by the ``asyncio`` logger."""
captured: list[str] = []
class _H(logging.Handler):
def emit(self, record: logging.LogRecord) -> None:
captured.append(record.getMessage())
handler = _H(level=logging.WARNING)
log = logging.getLogger("asyncio")
log.addHandler(handler)
prev_level = log.level
log.setLevel(logging.WARNING)
try:
yield captured
finally:
log.removeHandler(handler)
log.setLevel(prev_level)
# ---------------------------------------------------------------------------
# Tests
# ---------------------------------------------------------------------------
@pytest.mark.asyncio
async def test_send_frame_waits_for_a_peer_that_stopped_reading() -> None:
"""A client that stops reading must stall the emitter, not fill a buffer.
The TUI stops reading whenever its event loop is busy. Without flow control
``send_frame`` returns as fast as the runtime can produce frames, the
backlog accumulates in this process, and the whole batch lands on the
client at once when it recovers -- which is how a slow client becomes a
stuck one. Here the peer never reads, so sends must stop; once it drains,
they must resume.
"""
server_sock, client, conn, tmp = await _wire_paired_socket()
try:
disp = Dispatcher()
register_aligned_methods(disp)
server = RpcServer(disp, sock=conn)
serve_task = asyncio.create_task(server.serve_forever())
await server.started.wait()
sent = 0
async def pump() -> None:
nonlocal sent
payload = "x" * 4_000
# Bounded so a regression fails on the assertions below instead of
# buffering until the run is killed -- which is what losing the
# wait actually does.
while sent < 5_000:
await server.send_frame({"jsonrpc": "2.0", "method": "event", "params": {"t": payload}})
sent += 1
pump_task = asyncio.create_task(pump())
try:
# Asserted as "stops moving", not as a frame count: how much the
# socket and the transport absorb before the high-water mark is
# platform-dependent, but with no peer reading it must settle.
await asyncio.sleep(0.3)
stalled_at = sent
assert stalled_at > 0, "nothing was written at all"
await asyncio.sleep(0.2)
assert sent == stalled_at, f"emitter kept writing to an unread peer ({stalled_at} -> {sent})"
# Drain the peer; the emitter must pick up again.
loop = asyncio.get_running_loop()
for _ in range(200):
if sent > stalled_at:
break
try:
await asyncio.wait_for(loop.sock_recv(client, 1 << 20), timeout=0.1)
except asyncio.TimeoutError:
pass
assert sent > stalled_at, "emitter did not resume once the peer drained"
finally:
pump_task.cancel()
try:
await pump_task
except asyncio.CancelledError:
pass
serve_task.cancel()
try:
await serve_task
except asyncio.CancelledError:
pass
finally:
client.close()
conn.close()
server_sock.close()
@pytest.mark.asyncio
async def test_socket_roundtrip_after_inbound_byte() -> None:
"""Hello + slash-style requests round-trip with responses delivered.
Pre-fix this hangs: the write transport closed on first inbound byte and
no response ever reached the client. The 2 s per-frame timeout in
``_read_one_frame`` would expire and the assertion would fail.
"""
server_sock, client, conn, tmp = await _wire_paired_socket()
try:
disp = Dispatcher()
register_aligned_methods(disp)
server = RpcServer(disp, sock=conn)
serve_task = asyncio.create_task(server.serve_forever())
await server.started.wait()
# 1) handshake
await _send(client, "system.hello", {"client_version": "0.0.2"}, 1)
hello = await _read_one_frame(client)
assert hello, "no handshake response — write transport closed early"
hello_obj = json.loads(hello.decode().strip())
assert hello_obj["id"] == 1
assert "result" in hello_obj
assert hello_obj["result"]["server_version"]
# 2) three follow-up requests (simulating /cha autocomplete keystrokes)
for rid in (2, 3, 4):
await _send(client, "system.ping", {}, rid)
resp = await _read_one_frame(client)
assert resp, f"no response for ping #{rid} — transport closed"
obj = json.loads(resp.decode().strip())
assert obj["id"] == rid
assert obj["result"]["pong"] is True
serve_task.cancel()
try:
await serve_task
except asyncio.CancelledError:
pass
finally:
client.close()
conn.close()
server_sock.close()
@pytest.mark.asyncio
async def test_socket_no_asyncio_pipe_warnings(_capture_asyncio_warnings: list[str]) -> None:
"""Three sequential RPC requests must NOT trigger the asyncio pipe warning.
The pre-fix warning text is:
``"pipe closed by peer or os.write(pipe, data) raised exception."``
"""
server_sock, client, conn, tmp = await _wire_paired_socket()
try:
disp = Dispatcher()
register_aligned_methods(disp)
server = RpcServer(disp, sock=conn)
serve_task = asyncio.create_task(server.serve_forever())
await server.started.wait()
await _send(client, "system.hello", {"client_version": "0.0.2"}, 1)
await _read_one_frame(client)
# Fire 10 pings; pre-fix the asyncio warning starts after 5 dropped writes.
for rid in range(2, 12):
await _send(client, "system.ping", {}, rid)
await _read_one_frame(client)
serve_task.cancel()
try:
await serve_task
except asyncio.CancelledError:
pass
pipe_warnings = [m for m in _capture_asyncio_warnings if "pipe closed by peer" in m]
assert not pipe_warnings, f"asyncio.unix_events emitted pipe-closed warnings: {pipe_warnings!r}"
finally:
client.close()
conn.close()
server_sock.close()