brain-of-reese/tests/e2e/test_edit_summaries.py

"""Phase 57 E2E (Playwright): edit the AI-generated summary in the
viewer — and watch it get re-embedded.

TODO.md L4: "Be able to edit the summaries for documents in the RAG.
Click an edit button in summary box and change the summary that the AI
created. re-embed that document after changing the summary."
Run in isolation (DB must be up: ``podman compose up -d db``):

    uv run pytest tests/e2e/test_edit_summaries.py -v --no-cov

The fixture KB is a story-dedicated directory
(``tests/fixtures/summary_edit_kb/`` — the shared ``tests/fixtures/docs/``
and the phase-30 ``summary_kb/`` stay pinned at their files) with ONE
non-markdown A9 document:

* ``quadlet/llamacpp.container`` — a podman quadlet unit (the
  ``container`` extension is in the A9 default family, so the DEFAULT
  import scope walks it — no ``BOR_IMPORT_EXTENSIONS`` override). At
  import the mock ``lite`` model (``SUMMARY_MODE`` marker,
  ``tests/e2e/mock_llm.py``) reduces it to the deterministic 24-token
  digest — the first 24 tokens of the file, the header comment line —
  stored on ``documents.summary`` and indexed as one ``is_summary``
  chunk. The rest of the file is deliberately token-diluted
  (config keys the digest never contains), and the sentinel
  ``RESE-EDIT-SUMMARY-SENTINEL-b41d`` sits on the document's LAST line —
  outside the 24-token digest — so the raw ``<pre>`` content is
  distinguishable from the stored summary (the digest/sentinel mechanic
  of ``tests/e2e/test_document_summaries.py``).

DB isolation: the fixture's source name (``summary_edit_kb``) is
distinctive — the suite never asserts on absolute row counts and
deletes the rows it creates in a ``finally`` (other suites' documents
stay untouched in the shared E2E database).

The admin edits through the REAL browser flow (form login → the
viewer's Edit button → the inline editor → Save); the re-embed itself
is then verified against the live database (the chunk's NEW content, a
FRESH non-NULL vector, the content chunks byte-for-byte untouched — the
D4 re-embed scope) and against the content endpoint (phase 79 superseded the
phase-16 soft rule: the endpoint is require_user, so the pin runs
under the admin session — the shape is unchanged).
"""
from __future__ import annotations

import asyncio
from collections.abc import Iterator
from pathlib import Path
from threading import Thread
from typing import Any

import httpx
import pytest
from playwright.sync_api import Page, expect
from sqlalchemy import select

from app.config import Settings
from app.db import SessionLocal
from app.models import Chunk, Document
from app.rag.importer import ImportSummary, import_sources
from app.rag.llm import LLMClient
from e2e.auth_helpers import ADMIN_PASSWORD, login
from e2e.mock_llm import TOKEN_RE

REPO = Path(__file__).resolve().parents[2]
FIXTURES = REPO / "tests" / "fixtures" / "summary_edit_kb"
SOURCE = FIXTURES.name  # "summary_edit_kb" — distinctive, never asserted by count
DOC_PATH = "quadlet/llamacpp.container"
#: Encoded viewer URL query value (slash → %2F, same as the chips /
#: Sources table links build it).
DOC_URL_PATH = "quadlet%2Fllamacpp.container"
SENTINEL = "RESE-EDIT-SUMMARY-SENTINEL-b41d"

#: The hand-edited summary (test 1) — a distinctive sentence no part of
#: the fixture or its digest contains, so the round-trip assertion can
#: never pass against the old text.
NEW_SUMMARY = (
    "Hand-edited: the quadlet unit serves the homelab's local model on "
    "port 8081. (RESE-HANDEDIT-77ab)"
)
#: The modal-surface test's own edit (test 4) — likewise distinctive.
MODAL_SUMMARY = (
    "Edited from the modal: quadlet unit for the local inference server. "
    "(RESE-MODAL-EDIT-3cd9)"
)


# --- Importer + thread helpers (test_document_summaries.py pattern) ---


async def _import_fixtures(mock_port: int) -> ImportSummary:
    kwargs: dict[str, Any] = {
        "_env_file": None,
        "llm_base_url": f"http://127.0.0.1:{mock_port}/v1",
    }
    settings = Settings(**kwargs)  # pyright: ignore[reportCallIssue]
    return await import_sources([FIXTURES], LLMClient(settings))


def _run_in_thread(coro: Any) -> Any:
    """Run a coroutine on a worker thread.

    Playwright's sync API keeps an asyncio loop running on the test thread,
    so ``asyncio.run`` cannot be called directly from a test body.
    """
    box: dict[str, Any] = {}

    def runner() -> None:
        try:
            box["value"] = asyncio.run(coro)
        except BaseException as e:  # noqa: BLE001 — re-raised on the test thread
            box["error"] = e

    t = Thread(target=runner)
    t.start()
    t.join()
    if "error" in box:
        raise box["error"]
    return box["value"]


def _delete_source_rows() -> None:
    """Delete every row of this suite's distinctive source (chunks
    cascade with the document rows)."""
    with SessionLocal() as db:
        for doc in db.scalars(select(Document).where(Document.source == SOURCE)).all():
            db.delete(doc)
        db.commit()


@pytest.fixture(autouse=True)
def seeded_kb(mock_llm: int, db_ready: None) -> Iterator[ImportSummary]:
    """Seed the story-dedicated fixture for one test (the DEFAULT import
    scope — ``container`` is A9) and delete every row it creates
    afterwards (DB isolation — see the module docstring)."""
    _delete_source_rows()  # idempotent: leftovers from a crashed run
    summary = _run_in_thread(_import_fixtures(mock_llm))
    assert summary.formats == {"container": 1}
    assert summary.added == 1
    assert summary.summaries == 1 and summary.summary_errors == 0
    assert summary.errors == 0
    try:
        yield summary
    finally:
        _delete_source_rows()


def _expected_summary(content: str, source: str, path: str) -> str:
    """The mock lite model's byte-stable digest + the code pointer line.

    Mirrors ``mock_llm.compose_answer``'s ``SUMMARY_MODE`` branch (first
    24 tokens of the document content) plus the summarizer's
    deterministic ``Source:`` line — no model output is ever trusted.
    """
    digest = " ".join(TOKEN_RE.findall(content.lower())[:24])
    return f"This document covers {digest}.\nSource: {source}/{path}"


def _chunk_state() -> dict[str, Any]:
    """The fixture document's DB state, read back through a fresh session.

    ``total`` — the document's chunk count (never a table-wide count —
    DB isolation); ``summary`` — ``documents.summary``; ``summary_*`` —
    the single ``is_summary`` chunk (content + vector read-back);
    ``raw_contents`` — the content chunks' text, sorted (the D4 pin:
    the content chunks are byte-for-byte untouched by a summary edit).
    """
    with SessionLocal() as db:
        doc = db.scalar(
            select(Document).where(Document.source == SOURCE, Document.path == DOC_PATH)
        )
        assert doc is not None, "fixture doc was not imported"
        chunks = db.scalars(select(Chunk).where(Chunk.document_id == doc.id)).all()
    summary_chunks = [c for c in chunks if c.is_summary]
    assert len(summary_chunks) <= 1, f"more than one is_summary chunk: {len(summary_chunks)}"
    sc = summary_chunks[0] if summary_chunks else None
    return {
        "total": len(chunks),
        "summary": doc.summary,
        "summary_count": len(summary_chunks),
        "summary_content": sc.content if sc else None,
        "summary_vec": list(sc.embedding) if sc and sc.embedding is not None else None,
        "raw_contents": sorted(c.content for c in chunks if not c.is_summary),
    }


def _api_summary(app_url: str) -> str | None:
    """The content endpoint's ``summary``. Phase 79: the endpoint is
    require_user (the phase-16 soft rule is superseded) — the fresh
    httpx client signs in as the admin first (every caller is an
    admin-flow test)."""
    client = httpx.Client(timeout=10)
    r = client.post(f"{app_url}/api/login", json={"password": ADMIN_PASSWORD})
    assert r.status_code == 204
    r = client.get(
        f"{app_url}/api/documents/content",
        params={"source": SOURCE, "path": DOC_PATH},
    )
    assert r.status_code == 200, r.text
    return r.json()["summary"]


def _doc_url(app_url: str) -> str:
    return f"{app_url}/document.html?source={SOURCE}&path={DOC_URL_PATH}"


# ---------------------------------------------------------------------------
# 1. Admin: edit → Save → the panel, the API, and the DB all agree —
#    and the is_summary chunk carries a FRESH embedding (D4 re-embed)
# ---------------------------------------------------------------------------


def test_admin_edits_summary(page: Page, app_url: str) -> None:
    page.set_default_timeout(30_000)
    content = (FIXTURES / DOC_PATH).read_text(encoding="utf-8")
    assert SENTINEL in content.splitlines()[-1]  # last line, by design
    expected = _expected_summary(content, SOURCE, DOC_PATH)
    digest_line, pointer_line = expected.split("\n", 1)

    before = _chunk_state()
    assert before["summary"] == expected  # the mock digest is the stored summary
    assert before["summary_count"] == 1
    assert before["summary_vec"] is not None

    login(page, app_url)
    page.goto(_doc_url(app_url))

    # The .doc-summary panel shows the digest (digest + pointer lines) —
    # and nothing from the diluted raw body ("PublishPort" is deeper in
    # the file, outside the 24-token digest).
    panel = page.locator(".doc-summary")
    expect(panel).to_have_count(1)
    expect(panel).to_be_visible()
    expect(panel.locator(".doc-summary-title")).to_have_text("Summary")
    expect(panel).to_contain_text(digest_line)
    expect(panel).to_contain_text(pointer_line)
    expect(panel).not_to_contain_text("PublishPort")
    # The original still renders below, sentinel and all.
    expect(page.locator("#doc-content pre.doc-raw")).to_contain_text(SENTINEL)

    # The admin-only Edit button (phase 57, D4 — the viewer itself is
    # public; only the affordance is gated).
    edit = panel.locator(".doc-summary-edit")
    expect(edit).to_be_visible()
    expect(edit).to_have_text("Edit")

    # Edit → inline editor: textarea PREFILLED with the current summary,
    # Save / Cancel, and the role=status live region.
    edit.click()
    editor = page.locator(".doc-summary-editor")
    expect(editor).to_be_visible()
    expect(editor).to_have_value(expected)
    expect(page.locator(".doc-summary-save")).to_be_visible()
    expect(page.locator(".doc-summary-cancel")).to_be_visible()
    status = page.locator(".doc-summary-status")
    expect(status).to_have_attribute("role", "status")
    expect(status).to_have_attribute("aria-live", "polite")

    # Replace the text with the distinctive hand-edit and Save.
    page.fill(".doc-summary-editor", NEW_SUMMARY)
    page.click(".doc-summary-save")

    # Live-region confirmation, and the panel text is the NEW summary
    # (re-rendered through the textContent contract — the digest is gone).
    expect(status).to_have_text("Summary updated.")
    expect(panel.locator(".doc-summary-text")).to_have_text(NEW_SUMMARY)
    expect(panel).not_to_contain_text(digest_line)

    # Public read (no cookie): the content endpoint serves the new text.
    assert _api_summary(app_url) == NEW_SUMMARY

    # The re-embed, verified in the DB (D4): the is_summary chunk's
    # content is the new text with a FRESH non-NULL vector; the total
    # chunk count is unchanged and the content chunks are byte-for-byte
    # untouched — only the summary changed, so only it was re-embedded.
    after = _chunk_state()
    assert after["total"] == before["total"]  # count unchanged by an update
    assert after["summary_count"] == 1
    assert after["summary"] == NEW_SUMMARY
    assert after["summary_content"] == NEW_SUMMARY
    assert after["summary_vec"] is not None  # re-embedded, non-NULL
    assert after["summary_vec"] != before["summary_vec"]  # a FRESH vector
    assert after["raw_contents"] == before["raw_contents"]  # content chunks untouched


# ---------------------------------------------------------------------------
# 2. Admin: an empty save CLEARS — the panel disappears, summary NULL,
#    the is_summary chunk row is gone (count −1)
# ---------------------------------------------------------------------------


def test_admin_clears_summary(page: Page, app_url: str) -> None:
    page.set_default_timeout(30_000)
    content = (FIXTURES / DOC_PATH).read_text(encoding="utf-8")
    expected = _expected_summary(content, SOURCE, DOC_PATH)

    before = _chunk_state()
    assert before["summary"] == expected
    assert before["summary_count"] == 1

    login(page, app_url)
    page.goto(_doc_url(app_url))
    panel = page.locator(".doc-summary")
    expect(panel).to_have_count(1)
    expect(panel.locator(".doc-summary-edit")).to_be_visible()

    # Select-all + delete — clear the prefilled editor — then Save.
    panel.locator(".doc-summary-edit").click()
    page.fill(".doc-summary-editor", "")
    page.click(".doc-summary-save")

    # "Summary cleared." in the live region, then the panel leaves the
    # DOM (the renderer only draws it for non-empty summaries — the
    # removal lands a short beat after the confirmation). The original
    # content below is untouched.
    expect(page.locator(".doc-summary-status")).to_have_text("Summary cleared.")
    expect(page.locator(".doc-summary")).to_have_count(0, timeout=8_000)
    expect(page.locator("#doc-content pre.doc-raw")).to_contain_text(SENTINEL)

    # The public endpoint now reports no summary…
    assert _api_summary(app_url) is None

    # …and the DB agrees: summary NULL, the is_summary row deleted
    # (count −1), the content chunks byte-for-byte untouched.
    after = _chunk_state()
    assert after["total"] == before["total"] - 1  # the is_summary chunk is gone
    assert after["summary"] is None
    assert after["summary_count"] == 0
    assert after["summary_vec"] is None
    assert after["raw_contents"] == before["raw_contents"]


# ---------------------------------------------------------------------------
# 3. Anonymous: the digest renders, but no Edit button — and the
#    endpoint is 403 (the public viewer is byte-for-byte phase 36)
# ---------------------------------------------------------------------------


def test_anonymous_cannot(page: Page, app_url: str) -> None:
    page.set_default_timeout(30_000)
    content = (FIXTURES / DOC_PATH).read_text(encoding="utf-8")
    expected = _expected_summary(content, SOURCE, DOC_PATH)

    # Fresh context (the function-scoped page fixture — no login).
    # Phase 79 (task 05): the content endpoint is require_user — the
    # anonymous viewer meets the inline token gate (the content fetch
    # never runs — no summary panel, no not-found card either), so the
    # edit affordance is absent by construction.
    page.goto(_doc_url(app_url))
    expect(page.locator("#doc-auth-gate")).to_be_visible(timeout=15_000)
    expect(page.locator("#doc-title")).to_have_text("Loading…")
    expect(page.locator(".doc-summary")).to_have_count(0)
    expect(page.locator(".doc-summary-edit")).to_have_count(0)
    expect(page.locator(".doc-summary-head")).to_have_count(0)
    expect(page.locator(".doc-summary-editor")).to_have_count(0)

    # The endpoint is admin-gated (D4): an anonymous PATCH → 403
    # "admin only" (a fresh httpx client carries no session), and the
    # stored summary is untouched.
    r = httpx.patch(
        f"{app_url}/api/documents/summary",
        json={"source": SOURCE, "path": DOC_PATH, "summary": "not allowed"},
        timeout=10,
    )
    assert r.status_code == 403
    assert r.json() == {"detail": "admin only"}
    assert _chunk_state()["summary"] == expected  # untouched


# ---------------------------------------------------------------------------
# 4. Modal surface: the SAME shared renderer — the Sources-page modal
#    carries the Edit button too, and a save from it round-trips
# ---------------------------------------------------------------------------


def test_admin_modal_surface_edit(page: Page, app_url: str) -> None:
    """One core, two surfaces (phase 26/36): the document modal from the
    Sources table renders through the SAME ``renderDocument`` — so the
    admin gets the Edit button in the modal as well, and a save from
    there hits the same endpoint + DB row (the page test is unchanged
    in shape; this pins the second surface)."""
    page.set_default_timeout(30_000)
    content = (FIXTURES / DOC_PATH).read_text(encoding="utf-8")
    expected = _expected_summary(content, SOURCE, DOC_PATH)
    digest_line, _ = expected.split("\n", 1)

    before = _chunk_state()
    assert before["summary"] == expected

    login(page, app_url)  # lands on /sources.html (the catalog is admin-only)
    # Phase 97: the catalog is the drill-down tree — the row lives at
    # the quadlet level (the drill is the only change; the phase-57
    # summary-edit flow on the viewer itself is untouched).
    for name in (SOURCE, "quadlet"):
        page.click(f'#folders-tbody a.folder-link:text-is("{name}")')
    row = page.locator("#docs-tbody tr", has_text=DOC_PATH)
    expect(row).to_have_count(1)
    before_tabs = len(page.context.pages)
    row.locator("td:nth-child(2) a.doc-link").click()
    assert len(page.context.pages) == before_tabs, "row link must not open a new tab"

    # The modal shows the panel with the digest + the admin Edit button.
    expect(page.locator(".doc-modal")).to_be_visible()
    expect(page.locator("#doc-modal-title")).to_have_text("llamacpp")
    modal_panel = page.locator("#doc-modal .doc-summary")
    expect(modal_panel).to_have_count(1)
    expect(modal_panel).to_be_visible()
    expect(modal_panel).to_contain_text(digest_line)
    edit = modal_panel.locator(".doc-summary-edit")
    expect(edit).to_be_visible()

    # Edit → prefilled editor → replace → Save → the modal's panel
    # reflects the new text and the live region confirms.
    edit.click()
    editor = page.locator("#doc-modal .doc-summary-editor")
    expect(editor).to_be_visible()
    expect(editor).to_have_value(expected)
    page.fill("#doc-modal .doc-summary-editor", MODAL_SUMMARY)
    page.click("#doc-modal .doc-summary-save")
    expect(page.locator("#doc-modal .doc-summary-status")).to_have_text("Summary updated.")
    expect(modal_panel.locator(".doc-summary-text")).to_have_text(MODAL_SUMMARY)

    # Same endpoint, same DB row: the public read and the chunk agree —
    # count unchanged, fresh vector, content chunks untouched.
    assert _api_summary(app_url) == MODAL_SUMMARY
    after = _chunk_state()
    assert after["total"] == before["total"]
    assert after["summary_count"] == 1
    assert after["summary"] == MODAL_SUMMARY
    assert after["summary_content"] == MODAL_SUMMARY
    assert after["summary_vec"] is not None
    assert after["summary_vec"] != before["summary_vec"]
    assert after["raw_contents"] == before["raw_contents"]