mirror of
https://github.com/headroomlabs-ai/headroom.git
synced 2026-11-22 13:29:19 +08:00
feat(proxy): let /transformations/feed omit message bodies for number-only pollers (#3672)
## Description `/transformations/feed` always returns `request_messages`, `compressed_messages` and `response_content` for every entry, and `RequestLogger.get_recent_with_messages` builds them with `asdict()`, which deep-copies each transcript first. A poller that only reads the per-request numbers (tokens, model, transforms) still pays for a full copy and a multi-MB JSON serialization on the event loop for each pull. Measured on a desktop client polling every 20 s: a `limit=100` pull is ~44 MB, of which ~400 bytes per entry are read, and three concurrent pulls take `/stats?cached=1` from 45 ms to 1.3 s. This adds `?include_messages=0` (default `1`, byte-identical output to today). When off, the three body fields are omitted from the response and never walked server-side. Closes # ## Type of Change - [x] Performance improvement ## Changes Made - `RequestLogger.get_recent_with_messages(n, include_messages=True)`: when `False`, builds each entry from the dataclass fields minus the three payload fields, deep-copying only what it returns (same idiom as #3613 for `get_recent`). - `/transformations/feed` accepts `include_messages: bool = True` and omits the three body keys when it is off. Everything else in the item, including `log_full_messages`, is unchanged. - Each item also carries `uncached_input_tokens`, `cache_write_tokens` and `cache_read_tokens` (already stored on `RequestLog`, never exposed by the feed). A number-only poller can then rate `tokens_saved` against new input (uncached + cache_write) the way `/stats` `new_input_savings_percent` does; the item's full-transcript `savings_percent` is unchanged. Present in both the full and the slim response. - Tests: logger test proves the payloads are not traversed (a leaf whose `__deepcopy__` raises); endpoint test proves the slim item equals the full item minus the three keys and carries the cache split. ## Testing - [x] Unit tests pass (`pytest`) - [x] Linting passes (`ruff check .`) - [x] Type checking passes (`mypy headroom`) - [x] New tests added for new functionality - [x] Manual testing performed ### Test Output ```text $ uv run --frozen --extra dev pytest tests/test_proxy/test_transformations_feed.py tests/test_proxy/test_request_logger.py -q 9 passed in 4.53s $ uvx ruff check headroom/proxy/request_logger.py headroom/proxy/server.py tests/test_proxy/test_transformations_feed.py tests/test_proxy/test_request_logger.py All checks passed! $ uv run --frozen --extra dev mypy headroom/proxy/request_logger.py headroom/proxy/server.py Success: no issues found in 2 source files ``` ## Real Behavior Proof - Environment: macOS 15, Python 3.12, this branch, in-process `RequestLogger` seeded with 100 entries carrying ~160 KB of messages each (the size of a mid-session Claude Code turn). - Exact command / steps: `get_recent_with_messages(100)` and `get_recent_with_messages(100, include_messages=False)`, timed, then `json.dumps` of each result, timed and sized. - Observed result: include_messages=True: copy 29.4 ms, json 56.5 ms, 24.45 MB. include_messages=False: copy 1.3 ms, json 1.9 ms, 0.05 MB. Same 100 entries, same numeric fields. - Not tested: the dashboard UI (it does not pass the parameter, so it gets the default, unchanged response). ## Runtime Rollout Safety - Rollout-managed feature(s): none - Minimum rollout channel: n/a (opt-in query parameter) - Stable/default behavior changed: additive only; without the parameter every existing key is byte-identical, and three integer keys are added to each item - Kill switch / disable path: callers stop passing `include_messages=0` - Unsafe override required: no - Qualification impact: none - Rollback path: revert the commit; no persisted state or schema involved ## Review Readiness - [x] I have performed a self-review - [x] This PR is ready for human review ## Checklist - [x] My code follows the project's style guidelines - [x] I have performed a self-review of my code - [x] I have commented my code, particularly in hard-to-understand areas - [ ] I have made corresponding changes to the documentation - [x] My changes generate no new warnings - [x] I have added tests that prove my fix is effective or that my feature works - [x] New and existing unit tests pass locally with my changes - [x] I did **not** edit `CHANGELOG.md` — it is generated by release-please from my Conventional Commit PR title (a CI guard enforces this) ## Screenshots (if applicable) N/A ## Additional Notes - Documentation: the endpoint has no reference page; the parameter is documented in the handler docstring. - Overlaps #3613 on the `_HEAVY_FIELDS` constant and the import lines; both hunks are identical so either merge order applies cleanly.
This commit is contained in:
@@ -165,10 +165,23 @@ class RequestLogger:
|
||||
for e in entries
|
||||
]
|
||||
|
||||
def get_recent_with_messages(self, n: int = 20) -> list[dict]:
|
||||
"""Get recent log entries including full request/response messages."""
|
||||
def get_recent_with_messages(self, n: int = 20, include_messages: bool = True) -> list[dict]:
|
||||
"""Get recent log entries including full request/response messages.
|
||||
|
||||
``include_messages=False`` returns the same entries without
|
||||
``request_messages`` / ``compressed_messages`` / ``response_content``,
|
||||
and without walking them: ``asdict`` deep-copies every field, so a
|
||||
caller that only reads the per-request numbers otherwise pays for a
|
||||
full copy of each transcript (~160 KB per Claude Code turn, ~44 MB per
|
||||
``limit=100`` feed pull) on the event loop.
|
||||
"""
|
||||
entries = list(self._logs)[-n:]
|
||||
return [asdict(e) for e in entries]
|
||||
if include_messages:
|
||||
return [asdict(e) for e in entries]
|
||||
return [
|
||||
{f.name: deepcopy(getattr(e, f.name)) for f in fields(e) if f.name not in _HEAVY_FIELDS}
|
||||
for e in entries
|
||||
]
|
||||
|
||||
def stats(self) -> dict:
|
||||
"""Get logging statistics."""
|
||||
|
||||
+36
-23
@@ -4967,9 +4967,15 @@ def create_app(config: ProxyConfig | None = None) -> FastAPI:
|
||||
return proxy.metrics.savings_tracker.history_response(history_mode=history_mode)
|
||||
|
||||
@app.get("/transformations/feed", dependencies=[Depends(_require_loopback)])
|
||||
async def transformations_feed(limit: int = 20):
|
||||
async def transformations_feed(limit: int = 20, include_messages: bool = True):
|
||||
"""Get recent message transformations for the live feed.
|
||||
|
||||
``?include_messages=0`` omits the three message-body fields, and skips
|
||||
copying them server-side, for pollers that only read the per-request
|
||||
numbers. A ``limit=100`` pull with bodies is ~44 MB and its
|
||||
serialization blocks the event loop; the dashboard's live feed keeps
|
||||
the default.
|
||||
|
||||
Loopback-only: when ``log_full_messages`` is enabled this returns the
|
||||
full request/response message bodies (prompt content and completions)
|
||||
via ``request_messages`` / ``compressed_messages`` / ``response_content``.
|
||||
@@ -4987,29 +4993,36 @@ def create_app(config: ProxyConfig | None = None) -> FastAPI:
|
||||
log_full_messages = proxy.config.log_full_messages if proxy else False
|
||||
|
||||
if proxy and proxy.logger:
|
||||
logs = proxy.logger.get_recent_with_messages(limit)
|
||||
logs = proxy.logger.get_recent_with_messages(limit, include_messages=include_messages)
|
||||
for log in logs:
|
||||
transformations.append(
|
||||
{
|
||||
"request_id": log.get("request_id"),
|
||||
"timestamp": log.get("timestamp"),
|
||||
"provider": resolve_display_provider(
|
||||
log.get("provider"),
|
||||
openai_api_url=proxy.config.openai_api_url,
|
||||
provider_name=proxy.config.provider_name,
|
||||
),
|
||||
"model": log.get("model"),
|
||||
"input_tokens_original": log.get("input_tokens_original"),
|
||||
"input_tokens_optimized": log.get("input_tokens_optimized"),
|
||||
"tokens_saved": log.get("tokens_saved"),
|
||||
"savings_percent": log.get("savings_percent"),
|
||||
"transforms_applied": log.get("transforms_applied", []),
|
||||
"request_messages": log.get("request_messages"),
|
||||
"compressed_messages": log.get("compressed_messages"),
|
||||
"response_content": log.get("response_content"),
|
||||
"turn_id": log.get("turn_id"),
|
||||
}
|
||||
)
|
||||
item = {
|
||||
"request_id": log.get("request_id"),
|
||||
"timestamp": log.get("timestamp"),
|
||||
"provider": resolve_display_provider(
|
||||
log.get("provider"),
|
||||
openai_api_url=proxy.config.openai_api_url,
|
||||
provider_name=proxy.config.provider_name,
|
||||
),
|
||||
"model": log.get("model"),
|
||||
"input_tokens_original": log.get("input_tokens_original"),
|
||||
"input_tokens_optimized": log.get("input_tokens_optimized"),
|
||||
"tokens_saved": log.get("tokens_saved"),
|
||||
"savings_percent": log.get("savings_percent"),
|
||||
"transforms_applied": log.get("transforms_applied", []),
|
||||
"turn_id": log.get("turn_id"),
|
||||
# Per-request prefix-cache split, so a number-only poller can put
|
||||
# tokens_saved on the new-input basis /stats reports as
|
||||
# new_input_savings_percent (saved / (saved + uncached + cache_write))
|
||||
# instead of the full-transcript basis of savings_percent.
|
||||
"uncached_input_tokens": log.get("uncached_input_tokens", 0),
|
||||
"cache_write_tokens": log.get("cache_write_tokens", 0),
|
||||
"cache_read_tokens": log.get("cache_read_tokens", 0),
|
||||
}
|
||||
if include_messages:
|
||||
item["request_messages"] = log.get("request_messages")
|
||||
item["compressed_messages"] = log.get("compressed_messages")
|
||||
item["response_content"] = log.get("response_content")
|
||||
transformations.append(item)
|
||||
|
||||
return {"transformations": transformations, "log_full_messages": log_full_messages}
|
||||
|
||||
|
||||
@@ -132,3 +132,27 @@ def test_get_recent_never_walks_message_payloads():
|
||||
again = logger.get_recent(10)[0]
|
||||
assert again["tags"] == {"agent": "codex", "meta": {"depth": 1}}
|
||||
assert again["savings_breakdown"] == [{"tokens": 60}]
|
||||
|
||||
|
||||
def test_get_recent_with_messages_can_skip_payloads_without_walking_them():
|
||||
"""A feed poller that only reads the numbers passes include_messages=False;
|
||||
the three body fields are absent and never traversed (asdict deep-copied
|
||||
them). A leaf that refuses to be deep-copied proves the walk is gone."""
|
||||
|
||||
class _NoCopy:
|
||||
def __deepcopy__(self, memo):
|
||||
raise AssertionError("get_recent_with_messages walked a message payload")
|
||||
|
||||
logger = RequestLogger(log_file=None, log_full_messages=True)
|
||||
logger.log(
|
||||
_entry(
|
||||
request_messages=[{"role": "user", "content": _NoCopy()}],
|
||||
compressed_messages=[{"role": "user", "content": _NoCopy()}],
|
||||
response_content="ok",
|
||||
)
|
||||
)
|
||||
|
||||
slim = logger.get_recent_with_messages(10, include_messages=False)
|
||||
assert slim[0]["tokens_saved"] == 60
|
||||
assert slim[0]["transforms_applied"] == ["kompress:user:0.4"]
|
||||
assert not {"request_messages", "compressed_messages", "response_content"} & slim[0].keys()
|
||||
|
||||
@@ -7,6 +7,7 @@ pytest.importorskip("fastapi")
|
||||
|
||||
from httpx import ASGITransport, AsyncClient
|
||||
|
||||
from headroom.proxy.models import RequestLog
|
||||
from headroom.proxy.server import create_app
|
||||
|
||||
|
||||
@@ -67,3 +68,55 @@ async def test_transformations_feed_respects_limit(app):
|
||||
|
||||
data = response.json()
|
||||
assert len(data["transformations"]) <= 5
|
||||
|
||||
|
||||
@pytest.mark.asyncio
|
||||
async def test_transformations_feed_can_omit_message_bodies(app):
|
||||
"""``?include_messages=0`` is for pollers that only read the numbers: the
|
||||
three body fields are absent (not null) and every other field is the same
|
||||
as the full response."""
|
||||
app.state.proxy.logger.log(
|
||||
RequestLog(
|
||||
request_id="r1",
|
||||
timestamp="2026-04-24T10:00:00Z",
|
||||
provider="anthropic",
|
||||
model="claude-sonnet-4-6",
|
||||
input_tokens_original=100,
|
||||
input_tokens_optimized=40,
|
||||
output_tokens=10,
|
||||
tokens_saved=60,
|
||||
savings_percent=60.0,
|
||||
optimization_latency_ms=1.0,
|
||||
total_latency_ms=20.0,
|
||||
tags={},
|
||||
cache_hit=False,
|
||||
transforms_applied=["kompress:user:0.4"],
|
||||
cache_read_tokens=1000,
|
||||
cache_write_tokens=5,
|
||||
uncached_input_tokens=30,
|
||||
request_messages=[{"role": "user", "content": "hi"}],
|
||||
compressed_messages=[{"role": "user", "content": "hi"}],
|
||||
response_content="ok",
|
||||
)
|
||||
)
|
||||
async with AsyncClient(
|
||||
transport=ASGITransport(app=app, client=("127.0.0.1", 12345)),
|
||||
base_url="http://127.0.0.1",
|
||||
) as client:
|
||||
full = (await client.get("/transformations/feed?limit=1")).json()
|
||||
slim = (await client.get("/transformations/feed?limit=1&include_messages=0")).json()
|
||||
|
||||
full_item = full["transformations"][0]
|
||||
slim_item = slim["transformations"][0]
|
||||
assert full_item["request_messages"] == [{"role": "user", "content": "hi"}]
|
||||
bodies = {"request_messages", "compressed_messages", "response_content"}
|
||||
assert not bodies & slim_item.keys()
|
||||
# The prefix-cache split rides along, so a poller can rate tokens_saved
|
||||
# against new input (uncached + cache_write) like /stats does.
|
||||
assert (
|
||||
slim_item["uncached_input_tokens"],
|
||||
slim_item["cache_write_tokens"],
|
||||
slim_item["cache_read_tokens"],
|
||||
) == (30, 5, 1000)
|
||||
assert {k: v for k, v in full_item.items() if k not in bodies} == slim_item
|
||||
assert slim["log_full_messages"] == full["log_full_messages"]
|
||||
|
||||
Reference in New Issue
Block a user