fix(history): defer full transcript hydration to model sends (#5929)

* fix(history): defer full hydration to model sends

* fix(session): key hydration on real rows, fork through get_session

Two regressions from the display/model-context split, both reproducible
against dev.

The hydration gate compared the cached transcript against the
denormalized sessions.message_count column. That column drifts in normal
operation — _persist_message swallows a failed insert while add_message
has already appended in memory, so the next successful persist writes
rows+1 — and _db_to_session re-read the same column after each reload, so
the shortfall never closed. Every send, edit, delete and truncate on a
warm session re-selected the whole message table: the cost this change
set out to remove, relocated onto the hot path. The other direction was
just as bad — a persist for an uncached session writes message_count = 0,
and a stale-low counter with a partly filled cache meant no hydration at
all and a silently truncated transcript for the model.

sync_session_metadata now reconciles message_count against COUNT(*) on
chat_messages (one indexed count inside the connection it already opens),
and _db_to_session trusts the rows it just loaded. A hydrate always
closes the gap, so the next read is a cache hit.

fork_session read session_manager.sessions directly and never hydrated.
keep_count indexes into source.history, and display pagination no longer
fills that cache, so forking after a restart returned HTTP 200 with an
empty conversation and no error surfaced. It goes through get_session
now.

_hydrate_session_history_from_db is gone with its helper: get_session is
the hydration seam, and rebuilding session.history from raw rows in the
display fallback overwrote the parsed multimodal content and the _db_id
edit/delete keys that had just been set.

Tests drive a real SessionManager over a temp DB instead of a stub that
only proved the stub hydrates — both drift directions, the send path
warm and cold, and a fork taken after a restart. All five fail without
this change. The brittle SQL-text assertions are dropped; the page
bounds are already proven by the response body.

* fix(history): route pagination through canonical handler

---------

Co-authored-by: Léo <leograndcontact@gmail.com>
This commit is contained in:
RaresKeY
2026-08-10 19:39:21 +01:00
committed by GitHub
co-authored by Léo
parent dbeed4b63f
commit d449a9d431
6 changed files with 614 additions and 82 deletions
+41 -12
View File
@@ -194,7 +194,12 @@ class SessionManager:
is_important=getattr(db_session, 'is_important', False) or False,
)
session.message_count = getattr(db_session, 'message_count', len(history))
# The rows just loaded are the whole transcript, so they — not the
# denormalized sessions.message_count column — are the truth for this
# cached object. get_session's hydration gate compares against this
# number; seeding it from a drifted column would ask for a reload that
# can never close the gap.
session.message_count = len(history)
return session
# ------------------------------------------------------------------
@@ -398,30 +403,50 @@ class SessionManager:
# ------------------------------------------------------------------
def get_session(self, session_id: str) -> Session:
"""Get a session by ID, loading from DB if needed.
"""Get a session by ID, loading complete DB history when needed.
Sessions seeded by `load_sessions` start with empty history. The
first read here hydrates them with the message rows.
Sessions seeded by ``load_sessions`` start with empty history, and a
cached session can also become partially stale. Refresh metadata first,
then hydrate whenever the cached transcript is short of the stored rows.
Model-send routes enter through this method before building context,
while paginated display history reads SQLite directly.
The gate compares against ``sync_session_metadata``'s reconciled count
(the real ``chat_messages`` total), never the denormalized column, so a
hydrate always closes the gap and the next read is a cache hit.
"""
if session_id not in self.sessions:
self._load_session_from_db(session_id)
else:
cached = self.sessions[session_id]
# Lazy hydrate: metadata-only entries get their messages on first read.
if not cached.history and getattr(cached, "message_count", 0) > 0:
self._load_session_from_db(session_id)
# Keep model/endpoint metadata fresh. Endpoint deletion can clear the
# DB row while a session object is still cached in RAM.
# DB row while a session object is still cached in RAM. Refreshing first
# also exposes the authoritative message count before completeness is
# checked.
self.sync_session_metadata(session_id)
cached = self.sessions[session_id]
cached_count = len(cached.history or [])
stored_count = int(getattr(cached, "message_count", 0) or 0)
if cached_count < stored_count:
self._load_session_from_db(session_id)
# Update last_accessed
self._touch_session(session_id)
return self.sessions[session_id]
def sync_session_metadata(self, session_id: str) -> bool:
"""Refresh non-message session fields from the DB into the cached object."""
"""Refresh non-message session fields from the DB into the cached object.
``message_count`` is reconciled against the real ``chat_messages`` rows
rather than copied from the denormalized ``sessions.message_count``
column. That column drifts in normal operation — ``_persist_message``
swallows a failed insert but ``add_message`` has already appended in
memory, so the next successful persist writes rows+1, and a persist for
an uncached session writes 0. Hydration keys off this number: a
drifted-high column would reload the whole transcript on every warm
read, and a drifted-low one would leave the model a truncated one.
"""
session = self.sessions.get(session_id)
if session is None:
return False
@@ -444,7 +469,11 @@ class SessionManager:
session.archived = db_session.archived
session.owner = getattr(db_session, "owner", None)
session.is_important = getattr(db_session, "is_important", False) or False
session.message_count = getattr(db_session, "message_count", session.message_count) or 0
session.message_count = (
db.query(DbChatMessage)
.filter(DbChatMessage.session_id == session_id)
.count()
)
return True
except Exception as e:
logger.error(f"Error syncing session metadata {session_id}: {e}")