fix(chat): keep in-flight answers alive across in-app view switches

Root cause (owner repro, verified in a real browser 2026-09-06): the
five navbar views (Chat, RAG, Sources, Tuning, History) were separate
HTML documents, so a navbar click was a REAL cross-document navigation
— the chat page unloaded, the in-flight SSE fetch was aborted, and the
phase-48 teardown (app/api/chat.py `finally`, "chat: turn cancelled")
stopped the model. Observed: send question -> click RAG mid-stream ->
click Chat -> the answer never finished: no `query_log` row, and on
return a dangling question with no brain record (the pre-token pagehide
partial persist skips because `acc` is empty).

Phase-48 LOCKED-DECISION REFINEMENT (owner-confirmed 2026-09-06,
flagged per AGENTS.md rule 3, not silently deviated): "real navigation
cancels the fetch" now means LEAVING THE APP — tab close,
external/other-document navigation, the Stop button. In-app navbar
switches are client-side view switches and no longer cancel.

Fix — Option A (SPA shell), chosen over B (Service Worker owns the
stream) and C (server-side turn registry + resume):
- frontend/index.html is the shell: ONE `<main id="main">` holds the
  five `<section class="view">` blocks; hidden views carry BOTH
  `hidden` and `inert` (WCAG — no focus/keyboard traversal). The
  shared header, the single `doc-modal-*` skeleton, and the
  `#app-version` footer each exist exactly once; the per-view copies
  from the four folded pages are dropped.
- New frontend/assets/router.js (vanilla module — no framework, no
  bundler, No-CDN rule intact): lazy-imports a view module on FIRST
  show only (mount-once, hide-forever — the chat view's in-flight SSE
  reader persists across switches; that persistence IS the fix);
  intercepts same-shell navbar links with preventDefault +
  history.pushState (never a document load); handles popstate; single
  writer of `.nav-link` active state (is-active + aria-current),
  document.title, and the per-view meta description (values carried
  over from the old pages' heads, brand-resolved at write time).
- Each folded page's JS becomes `export async function mount(root)` —
  root-scoped queries; `initSharedHeader()` dropped (the header boots
  once in the shell via the chat module; the admin flag comes from the
  same cached `fetchIsAdmin()` promise — zero extra requests).
- app/main.py: a small list-driven route factory serves the shell for
  /tuning.html, /sources.html, /git-sources.html, /history.html —
  registered AFTER the API routers and BEFORE the static catch-all
  (routes-first). The phase-33 caching middleware applies no-cache +
  `?v=` rewriting unchanged; app/core/caching.py needed NO change
  (the view paths did not change — pinned by the integration tests).
- The four old view .html files are DELETED (one source of truth);
  deep links to the old URLs keep working (the router picks the view
  from the pathname); `/?chat=<id>` is unaffected; the Containerfile
  bundles router.js (inlining the lazy view modules) and drops the
  folded page files.
- app/schemas.py: HistoryTurn.text cap 4000 -> 32000 — the shell
  keeps long saved answers in the chat, and the old cap (stricter than
  the 24_000-char total history budget) 422-rejected any second turn
  in such a chat (found by the phase-42 E2E suite on the shell).

Boundaries: login.html, shared.html, doc-edit.html, document.html
REMAIN separate documents (flow pages, not navbar tabs); a mid-stream
navigation to doc-edit/document.html still cancels per phase 48
(follow-up candidate, out of scope). The SSE API is unchanged. Real
departures still cancel the turn — phase 48 intact (pinned by
tests/e2e/test_stop_generation.py, unchanged, and by the new suite's
real-departure control).

Tests:
- Phase-20 suite REWRITTEN to the new semantics
  (tests/e2e/test_sources_midstream_bug.py): a navbar switch no longer
  cancels — the stream survives the switch and the FULL answer
  settles; the pagehide partial persist REMAINS for real departures
  (the partial's exact shape — first streamed chunk prefix, no done
  metadata — is still pinned there).
- NEW story suite tests/e2e/test_nav_switch_keeps_stream.py (mock
  LLM): the owner repro (send -> RAG mid-stream -> Chat: window
  sentinel survives = same document, FULL answer, exactly one brain
  turn in bor.chat.v1, exactly one settled query_log row, auto-saved
  row matches) + the same mid-stream switch against the other three
  views + the real-departure-still-cancels control + the no-switch
  baseline.
- tests/unit/test_frontend_router.py: source-level pins of the router
  invariants (click interceptor targets ONLY same-shell view paths,
  pushState-only switches, mount-once guard, hidden+inert pair,
  single-writer active state/title); shell-route integration tests
  (each folded path serves the shell with no-cache + `?v=` body; a
  non-view path still 404s); the file-reading unit pins re-pointed at
  the shell (the four view files are gone — the shell is the source
  of truth).

Verification (this commit): full suite green — 1565 unit+integration
tests, app/ coverage 99% (>90% floor); ruff + pyright clean; the
phase's E2E suites green in isolation (house protocol, AGENTS.md rule
9). Owner repro verified in a real browser against the real LLM
(dev server :8010, headful Chromium): "tell me about everquest" ->
RAG mid-stream -> Chat — the answer completed with one brain bubble
and no error banner, `query_log` gained exactly one settled row
(deflected=True: the dev KB holds no EverQuest docs — the settle, not
the topic, is the proof), zero "chat: turn cancelled" lines for that
turn; the control (real navigation to /shared.html mid-stream) still
cancelled (no settled row, the cancel line logged, the partial
persisted on return). Screenshots: .agents/screenshots/76_manual_*.

Phase 76 (76_spa_nav_shell) complete — moved to
.agents/phases/complete/.
This commit is contained in:
2026-09-06 06:31:31 -04:00
parent 7e567bddf3
commit ffa919b8bf
78 changed files with 5548 additions and 3244 deletions
+61 -5
View File
@@ -150,12 +150,23 @@ def test_suggestions_honors_bor_suggestions_env_override(monkeypatch) -> None:
("path", "marker"),
[
("/", "Brain of Reese"),
("/sources.html", "Knowledge base"),
# Phase 76 (task 02): /sources.html + /git-sources.html are SHELL
# routes — the body is the shell (index.html) whose static
# <title> is "Brain of Reese" (the per-view title is set
# CLIENT-side by the router, invisible to httpx). The marker
# asserts the shell body (the view section is inside it) instead
# of the old page's title.
("/sources.html", 'id="view-rag"'), # phase 76: shell route (was "Knowledge base")
("/document.html", "Brain of Reese"), # phase 10: viewer page
("/login.html", "Sign in"), # phase 16: admin sign-in page
("/tuning.html", "Global Tuning"), # phase 27: global tuning page
("/git-sources.html", "Git sources"), # phase 35: admin git sources page
("/history.html", "Saved chats"), # phase 50: admin saved-chats page
# Phase 76 (task 01): /tuning.html is a SHELL route — same
# shell-body marker pattern as the two task-02 paths above.
("/tuning.html", 'id="view-tuning"'), # phase 76: shell route (was "Global Tuning")
("/git-sources.html", 'id="view-git-sources"'), # phase 76: shell route (was "Git sources")
# Phase 76 (task 03): /history.html is a SHELL route too — the
# shell-body marker (the History view section is inside the
# shell; the old standalone page's title is client-side now).
("/history.html", 'id="view-history"'), # phase 76: shell route (was "Saved chats")
("/shared.html", "Shared conversation"), # phase 51: anonymous shared page
],
)
@@ -194,7 +205,7 @@ def test_index_page_no_cache_with_versioned_asset_refs(client) -> None:
@pytest.mark.parametrize(
"path",
["/sources.html", "/document.html", "/login.html", "/tuning.html",
"/git-sources.html", "/history.html", # phase 50: + the History page
"/git-sources.html", "/history.html", # phase 50: + History (shell route, task 03)
"/shared.html"], # phase 51: + the anonymous shared page
)
def test_html_pages_no_cache_with_versioned_refs(client, path: str) -> None:
@@ -208,6 +219,51 @@ def test_html_pages_no_cache_with_versioned_refs(client, path: str) -> None:
assert f"?v={asset_version()}" in r.text
@pytest.mark.parametrize(
("path", "view_id", "old_title"),
[
("/tuning.html", 'id="view-tuning"', "Global Tuning · Brain of Reese"),
("/sources.html", 'id="view-rag"', "Sources · Brain of Reese"),
("/git-sources.html", 'id="view-git-sources"', "Git sources · Brain of Reese"),
("/history.html", 'id="view-history"', "Saved chats · Brain of Reese"), # phase 76 task 03
],
)
def test_shell_routes_serve_the_shell_no_cache_versioned(
client, path: str, view_id: str, old_title: str
) -> None:
"""Phase 76 (task 01: /tuning.html; task 02: /sources.html +
/git-sources.html; task 03: /history.html — all four non-chat
navbar views): every shell route serves the shell
(``frontend/index.html``), so the phase-33 page contract applies to
it exactly as to a static page: 200, text/html, ``Cache-Control:
no-cache``, ``?v=<token>`` asset refs, no validators (the
middleware wraps the whole app and lists the path in ``HTML_PAGES``
— unchanged). The body IS the shell: the view section is inside it,
the old standalone page's title is gone. The static catch-all stays
intact: an unknown path still 404s. (The conditional-GET
revalidation contract for these paths is pinned by
``tests/integration/test_caching_revalidation.py``.)"""
from app.core.caching import asset_version
r = client.get(path)
assert r.status_code == 200
assert r.headers["content-type"].split(";", 1)[0] == "text/html"
assert r.headers["cache-control"] == "no-cache"
assert "etag" not in r.headers
assert "last-modified" not in r.headers
assert f"?v={asset_version()}" in r.text
# The body is the SHELL: the chat view AND the route's view section
# are inside it…
assert 'id="view-chat"' in r.text
assert view_id in r.text
# …with the shell's static title (the per-view title is the
# router's client-side job), and the old page's own <title> is gone.
assert "<title>Brain of Reese</title>" in r.text
assert f"{old_title}</title>" not in r.text
# The catch-all is intact: an unknown path still 404s.
assert client.get("/nonexistent.html").status_code == 404
def test_index_html_variant_no_cache_versioned(client) -> None:
"""/index.html is the same page as / — same caching treatment."""
from app.core.caching import asset_version
+23 -2
View File
@@ -68,9 +68,30 @@ def file_validators(page_file: Path) -> tuple[str, str]:
return headers["etag"], headers["last-modified"]
#: Phase 76 (task 01): the shell-served view paths — the URL is a
#: navbar view, the file on disk is the SHELL (the shell route in
#: app/main.py serves frontend/index.html for it). The etag
#: computation below must use the file that actually backs the
#: response, or the conditional-GET probe would carry a validator no
#: browser ever saw. Tasks 02/03 extended this as the remaining views
#: folded in (task 03 — History — completes the set: all four
#: non-chat navbar views). The page CONTRACT itself is unchanged: the
#: phase-33 middleware wraps the whole app and lists the path in
#: HTML_PAGES, so the shell-route response is normalized exactly like
#: a static page (200, no-cache, ?v=, no validators — asserted by
#: _assert_page_contract below).
SHELL_BACKED_PAGES = {
"/tuning.html": "index.html", # phase 76 task 01
"/sources.html": "index.html", # phase 76 task 02
"/git-sources.html": "index.html", # phase 76 task 02
"/history.html": "index.html", # phase 76 task 03
}
def _page_file(path: str) -> Path:
"""The static file backing a page path (``/`` → ``index.html``)."""
name = path.lstrip("/") or "index.html"
"""The static file backing a page path (``/`` → ``index.html``;
the shell-served view paths → the shell, ``SHELL_BACKED_PAGES``)."""
name = SHELL_BACKED_PAGES.get(path, path.lstrip("/") or "index.html")
file = FRONTEND / name
assert file.is_file(), f"missing page file for {path}: {file}"
return file
@@ -242,6 +242,18 @@ def test_header_module_is_imported_not_directly_loaded() -> None:
js_path = ASSETS / name
assert js_path.is_file(), f"page script missing: {js_path}"
body = js_path.read_text(encoding="utf-8")
# Phase 76: the shell's router (router.js) is NOT a view module
# — the shell's shared header boots via the chat module (app.js,
# which imports header.js), and the view modules the router
# lazy-imports (tuning.js, …) import header.js themselves, so
# the single-evaluation contract holds without the router
# importing it (it owns the view switch, not the header).
if name == "router.js":
assert 'from "./header.js"' not in body, (
"router.js must not import the header module — the chat "
"module boots the shell's ONE header"
)
continue
assert re.search(r"""from\s+["']\./header\.js["']""", body), (
f"{name}: must import the shared header module relatively "
f'("from \\"./header.js\\"")'