Files
AutoGPT/autogpt_platform/backend/backend/copilot/sdk/compaction.py
Zamil Majdy cfe22e5a8f fix(backend/copilot): sync TranscriptBuilder with CLI on mid-stream compaction (#12401)
## Summary
- **Root cause**: `TranscriptBuilder` accumulates all raw SDK stream
messages including pre-compaction content. When the CLI compacts
mid-stream, the uploaded transcript was still uncompacted, causing
"Prompt is too long" errors on the next `--resume` turn.
- **Fix**: Detect mid-stream compaction via the `PreCompact` hook, read
the CLI's session file to get the compacted entries (summary +
post-compaction messages), and call
`TranscriptBuilder.replace_entries()` to sync it with the CLI's active
context. This ensures the uploaded transcript always matches what the
CLI sees.
- **Key changes**:
- `CompactionTracker`: stores `transcript_path` from `PreCompact` hook,
one-shot `compaction_just_ended` flag that correctly resets for multiple
compactions
- `read_compacted_entries()`: reads CLI session JSONL, finds
`isCompactSummary: true` entry, returns it + all entries after. Includes
path validation against the CLI projects directory.
- `TranscriptBuilder.replace_entries()`: clears and replaces all entries
with compacted ones, preserving `isCompactSummary` entries (which have
`type: "summary"` that would normally be stripped)
- `load_previous()`: also preserves `isCompactSummary` entries when
loading a previously compacted transcript
- Service stream loop: after compaction ends, reads compacted entries
and syncs TranscriptBuilder

## Test plan
- [x] 69 tests pass across `compaction_test.py` and `transcript_test.py`
- [x] Tests cover: one-shot flag behavior, multiple compactions within a
query, transcript path storage, path traversal rejection,
`read_compacted_entries` (7 tests), `replace_entries` (4 tests),
`load_previous` with compacted content (2 tests)
- [x] Pre-commit hooks pass (lint, format, typecheck)
- [ ] Manual test: trigger compaction in a multi-turn session and verify
the uploaded transcript reflects compaction
2026-03-13 22:17:46 +00:00

274 lines
9.4 KiB
Python

"""Compaction tracking for SDK-based chat sessions.
Encapsulates the state machine and event emission for context compaction,
both pre-query (history compressed before SDK query) and SDK-internal
(PreCompact hook fires mid-stream).
All compaction-related helpers live here: event builders, message filtering,
persistence, and the ``CompactionTracker`` state machine.
"""
import asyncio
import logging
import uuid
from dataclasses import dataclass, field
from ..constants import COMPACTION_DONE_MSG, COMPACTION_TOOL_NAME
from ..model import ChatMessage, ChatSession
from ..response_model import (
StreamBaseResponse,
StreamFinishStep,
StreamStartStep,
StreamToolInputAvailable,
StreamToolInputStart,
StreamToolOutputAvailable,
)
logger = logging.getLogger(__name__)
@dataclass
class CompactionResult:
"""Result of emit_end_if_ready — bundles events with compaction metadata.
Eliminates the need for separate ``compaction_just_ended`` checks,
preventing TOCTOU races between the emit call and the flag read.
"""
events: list[StreamBaseResponse] = field(default_factory=list)
just_ended: bool = False
transcript_path: str = ""
# ---------------------------------------------------------------------------
# Event builders (private — use CompactionTracker or compaction_events)
# ---------------------------------------------------------------------------
def _start_events(tool_call_id: str) -> list[StreamBaseResponse]:
"""Build the opening events for a compaction tool call."""
return [
StreamStartStep(),
StreamToolInputStart(toolCallId=tool_call_id, toolName=COMPACTION_TOOL_NAME),
StreamToolInputAvailable(
toolCallId=tool_call_id, toolName=COMPACTION_TOOL_NAME, input={}
),
]
def _end_events(tool_call_id: str, message: str) -> list[StreamBaseResponse]:
"""Build the closing events for a compaction tool call."""
return [
StreamToolOutputAvailable(
toolCallId=tool_call_id,
toolName=COMPACTION_TOOL_NAME,
output=message,
),
StreamFinishStep(),
]
def _new_tool_call_id() -> str:
return f"compaction-{uuid.uuid4().hex[:12]}"
# ---------------------------------------------------------------------------
# Public event builder
# ---------------------------------------------------------------------------
def emit_compaction(session: ChatSession) -> list[StreamBaseResponse]:
"""Create, persist, and return a self-contained compaction tool call.
Convenience for callers that don't use ``CompactionTracker`` (e.g. the
legacy non-SDK streaming path in ``service.py``).
"""
tc_id = _new_tool_call_id()
evts = compaction_events(COMPACTION_DONE_MSG, tool_call_id=tc_id)
_persist(session, tc_id, COMPACTION_DONE_MSG)
return evts
def compaction_events(
message: str, tool_call_id: str | None = None
) -> list[StreamBaseResponse]:
"""Emit a self-contained compaction tool call (already completed).
When *tool_call_id* is provided it is reused (e.g. for persistence that
must match an already-streamed start event). Otherwise a new ID is
generated.
"""
tc_id = tool_call_id or _new_tool_call_id()
return _start_events(tc_id) + _end_events(tc_id, message)
# ---------------------------------------------------------------------------
# Message filtering
# ---------------------------------------------------------------------------
def filter_compaction_messages(
messages: list[ChatMessage],
) -> list[ChatMessage]:
"""Remove synthetic compaction tool-call messages (UI-only artifacts).
Strips assistant messages whose only tool calls are compaction calls,
and their corresponding tool-result messages.
"""
compaction_ids: set[str] = set()
filtered: list[ChatMessage] = []
for msg in messages:
if msg.role == "assistant" and msg.tool_calls:
for tc in msg.tool_calls:
if tc.get("function", {}).get("name") == COMPACTION_TOOL_NAME:
compaction_ids.add(tc.get("id", ""))
real_calls = [
tc
for tc in msg.tool_calls
if tc.get("function", {}).get("name") != COMPACTION_TOOL_NAME
]
if not real_calls and not msg.content:
continue
if msg.role == "tool" and msg.tool_call_id in compaction_ids:
continue
filtered.append(msg)
return filtered
# ---------------------------------------------------------------------------
# Persistence
# ---------------------------------------------------------------------------
def _persist(session: ChatSession, tool_call_id: str, message: str) -> None:
"""Append compaction tool-call + result to session messages.
Compaction events are synthetic so they bypass the normal adapter
accumulation. This explicitly records them so they survive a page refresh.
"""
session.messages.append(
ChatMessage(
role="assistant",
content="",
tool_calls=[
{
"id": tool_call_id,
"type": "function",
"function": {
"name": COMPACTION_TOOL_NAME,
"arguments": "{}",
},
}
],
)
)
session.messages.append(
ChatMessage(role="tool", content=message, tool_call_id=tool_call_id)
)
# ---------------------------------------------------------------------------
# CompactionTracker — state machine for streaming sessions
# ---------------------------------------------------------------------------
class CompactionTracker:
"""Tracks compaction state and yields UI events.
Two compaction paths:
1. **Pre-query** — history compressed before the SDK query starts.
Call :meth:`emit_pre_query` to yield a self-contained tool call.
2. **SDK-internal** — ``PreCompact`` hook fires mid-stream.
Call :meth:`emit_start_if_ready` on heartbeat ticks and
:meth:`emit_end_if_ready` when a message arrives.
"""
def __init__(self) -> None:
self._compact_start = asyncio.Event()
self._start_emitted = False
self._done = False
self._tool_call_id = ""
self._transcript_path: str = ""
def on_compact(self, transcript_path: str = "") -> None:
"""Callback for the PreCompact hook. Stores transcript_path."""
if (
self._transcript_path
and transcript_path
and self._transcript_path != transcript_path
):
logger.warning(
"[Compaction] Overwriting transcript_path %s -> %s",
self._transcript_path,
transcript_path,
)
self._transcript_path = transcript_path
self._compact_start.set()
# ------------------------------------------------------------------
# Pre-query compaction
# ------------------------------------------------------------------
def emit_pre_query(self, session: ChatSession) -> list[StreamBaseResponse]:
"""Emit + persist a self-contained compaction tool call."""
self._done = True
return emit_compaction(session)
# ------------------------------------------------------------------
# SDK-internal compaction
# ------------------------------------------------------------------
def reset_for_query(self) -> None:
"""Reset per-query state before a new SDK query."""
self._done = False
self._start_emitted = False
self._tool_call_id = ""
self._transcript_path = ""
def emit_start_if_ready(self) -> list[StreamBaseResponse]:
"""If the PreCompact hook fired, emit start events (spinning tool)."""
if self._compact_start.is_set() and not self._start_emitted and not self._done:
self._compact_start.clear()
self._start_emitted = True
self._tool_call_id = _new_tool_call_id()
return _start_events(self._tool_call_id)
return []
async def emit_end_if_ready(self, session: ChatSession) -> CompactionResult:
"""If compaction is in progress, emit end events and persist.
Returns a ``CompactionResult`` with ``just_ended=True`` and the
captured ``transcript_path`` when a compaction cycle completes.
This avoids a separate flag check (TOCTOU-safe).
"""
# Yield so pending hook tasks can set compact_start
await asyncio.sleep(0)
if self._done:
return CompactionResult()
if not self._start_emitted and not self._compact_start.is_set():
return CompactionResult()
if self._start_emitted:
# Close the open spinner
done_events = _end_events(self._tool_call_id, COMPACTION_DONE_MSG)
persist_id = self._tool_call_id
else:
# PreCompact fired but start never emitted — self-contained
persist_id = _new_tool_call_id()
done_events = compaction_events(
COMPACTION_DONE_MSG, tool_call_id=persist_id
)
transcript_path = self._transcript_path
self._compact_start.clear()
self._start_emitted = False
self._done = True
self._transcript_path = ""
_persist(session, persist_id, COMPACTION_DONE_MSG)
return CompactionResult(
events=done_events, just_ended=True, transcript_path=transcript_path
)