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

"""Phase 102 E2E (Playwright): extensionless files (``Dockerfile``,
``Containerfile``) sync when their name is a ``BOR_IMPORT_EXTENSIONS``
token — import → chunks → mock ``lite`` summary → drill-down tree →
viewer badge, plus the negative control and the anonymous gate.

The owner's defect: files without extensions never got synced, so
``Dockerfile`` / ``Containerfile`` were skipped even with
``dockerfile,containerfile`` in the env. Phase 102's rule: a suffix-less
file imports iff its lowercased FULL filename equals a token (exact
name, case-insensitive), the import ``formats=`` counter keys it by the
matched token, and the viewer badge is truthful (``dockerfile``, not
the generic ``text``).

The subject is the extensionless scope (``import_extensions=
"md,dockerfile,containerfile"``); the story-dedicated fixture
(``tests/fixtures/extensionless_kb/``) is seeded in-process against the
deterministic mock LLM — the phase-02/56 seeding-thread pattern — with
``Dockerfile`` / ``Containerfile`` carrying their own unique sentinels
and ``Makefile`` as the negative control (no token names it).

Run in isolation (DB must be up: ``podman compose up -d db``):

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

DB isolation: the fixture's source name (``extensionless_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).

App-under-test scope pin (the phase-61 leak-guard pattern): the viewer
badge is rendered by the APP from its own ``get_settings().
import_extension_set`` (``GET /api/documents/content``), so the app
process must carry exactly the scope under test — a module-level
``BOR_IMPORT_EXTENSIONS`` override (process env ranks above the
operator's local gitignored ``.env`` when the ``app_server`` fixture
snapshots ``os.environ``). In isolation (AGENTS.md rule 9) the session
app is spawned by this file's first test, so the pin lands determinis-
tically.

Phase 97 adaptation: the catalog is the DRILL-DOWN TREE — the fixture's
files live at the SOURCE level (no subfolders), so a single drill
reaches every asserted row; the drill is the only change, the asserted
rows/links/modal are unchanged.
"""
from __future__ import annotations

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

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 Document
from app.rag.importer import ImportSummary, import_sources
from app.rag.llm import LLMClient
from e2e.auth_helpers import login
from tests.e2e.mock_llm import TOKEN_RE

#: The app-under-test scope pin (see the module docstring) — exactly the
#: scope this suite proves, no matter what the operator's local ``.env``
#: carries. Must run before the session ``app_server`` fixture snapshots
#: ``os.environ`` (module import precedes fixture setup).
os.environ["BOR_IMPORT_EXTENSIONS"] = "md,dockerfile,containerfile"

REPO = Path(__file__).resolve().parents[2]
FIXTURES = REPO / "tests" / "fixtures" / "extensionless_kb"
SOURCE = FIXTURES.name  # "extensionless_kb" — distinctive, never asserted by count
DOCKER_REL = "Dockerfile"
CONTAINER_REL = "Containerfile"
NOTES_REL = "notes.md"
MAKE_REL = "Makefile"  # negative control — no token in the scope under test
DOCKER_SENTINEL = "DOCKERFILE-PROBE-SENTINEL-7a3e"


async def _import_fixtures(mock_port: int, extensions: str) -> ImportSummary:
    kwargs: dict[str, Any] = {
        "_env_file": None,
        "llm_base_url": f"http://127.0.0.1:{mock_port}/v1",
        "import_extensions": extensions,
    }
    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()


def _drill(page: Page, *names: str) -> None:
    """Drill one level at a time (phase 97; phase 167: an UNCACHED level
    costs exactly one /api/docs/tree/node fetch, a held level costs
    nothing, and no click ever changes the URL): each name is the EXACT
    text of the source/folder link at the current level."""
    for name in names:
        page.click(f'#folders-tbody a.folder-link:text-is("{name}")')


def _go_top(page: Page) -> None:
    """Back to the top level: the breadcrumb's top-level link (call
    between drills only — the breadcrumb is hidden at the top)."""
    page.locator("#kb-crumb a.kb-crumb-link").first.click()


def _summary_lines(path: str) -> tuple[str, str]:
    """(digest line, pointer line) of the stored phase-30 summary for a
    fixture file — mirrors the mock ``SUMMARY_MODE`` branch (first 24
    tokens of the document content) plus the code pointer line (the
    ``test_summary_in_viewer.py`` house helper)."""
    content = (FIXTURES / path).read_text(encoding="utf-8")
    digest = " ".join(TOKEN_RE.findall(content.lower())[:24])
    expected = f"This document covers {digest}.\nSource: {SOURCE}/{path}"
    digest_line, pointer_line = expected.split("\n", 1)
    return digest_line, pointer_line


@pytest.fixture(autouse=True)
def extensionless_kb(mock_llm: int, db_ready: None) -> Iterator[ImportSummary]:
    """Seed the fixture with the extensionless scope
    (``md,dockerfile,containerfile``) for one test 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, "md,dockerfile,containerfile"))
    try:
        yield summary
    finally:
        _delete_source_rows()


def test_admin_tree_lists_name_token_files_and_viewer_badge(
    page: Page, app_url: str, extensionless_kb: ImportSummary
) -> None:
    # The seed saw exactly the three in-scope files in their formats —
    # the two extensionless build files walked by name token, chunked,
    # and summarized; the Makefile negative control stayed out.
    assert extensionless_kb.formats == {"dockerfile": 1, "containerfile": 1, "md": 1}
    assert "unknown" not in extensionless_kb.formats
    assert (extensionless_kb.added, extensionless_kb.errors) == (3, 0)

    login(page, app_url)  # phase 16: the catalog is admin-only
    # All three fixture files live at the SOURCE level (no subfolders —
    # one drill reaches them, phase 97).
    _drill(page, SOURCE)
    for name in (DOCKER_REL, CONTAINER_REL, NOTES_REL):
        expect(page.locator("#docs-tbody tr", has_text=name)).to_have_count(1)
    # The negative control: no token names Makefile exactly — absent.
    expect(page.locator("#docs-tbody tr", has_text=MAKE_REL)).to_have_count(0)

    # The Dockerfile row's path link opens the SAME-PAGE modal and its
    # meta row shows the D3 badge (``dockerfile`` — not the generic
    # ``text`` fallback) + the source badge + the stem title.
    row = page.locator("#docs-tbody tr", has_text=DOCKER_REL)
    link = row.locator("td:nth-child(2) a.doc-link")
    expect(link).to_have_count(1)
    expect(link).to_have_attribute("title", DOCKER_REL)
    before = len(page.context.pages)
    link.click()
    assert len(page.context.pages) == before, "clicking a row link must not open a new tab"
    expect(page.locator("#doc-modal-meta .doc-source-badge")).to_have_text(SOURCE)
    expect(page.locator("#doc-modal-meta .format-badge")).to_have_text("dockerfile")
    expect(page.locator("#doc-modal-title")).to_have_text(DOCKER_REL)

    # Non-markdown content renders as escaped monospace text in a pre —
    # the sentinel proves it is THIS document's content.
    pre = page.locator("#doc-modal-content pre.doc-raw")
    expect(pre).to_have_count(1)
    expect(pre).to_contain_text(DOCKER_SENTINEL)

    # The phase-30 summary line renders above the content (house
    # assertion style — test_summary_in_viewer.py): the deterministic
    # mock digest + the code pointer line, in the labeled panel.
    digest_line, pointer_line = _summary_lines(DOCKER_REL)
    panel = page.locator("#doc-modal .doc-summary")
    expect(panel).to_have_count(1)
    expect(panel).to_be_visible()
    expect(panel).to_have_attribute("aria-label", "Summary")
    expect(panel).to_contain_text(digest_line)
    expect(panel).to_contain_text(pointer_line)

    # Still on the Sources page: no navigation happened.
    assert page.url == app_url + "/sources.html", f"navigated away: {page.url}"


def test_anonymous_sources_gate_and_no_api_docs(
    page: Page, app_url: str, extensionless_kb: ImportSummary
) -> None:
    """A fresh anonymous context (function-scoped ``page`` = new
    browser context, no cookies): the sign-in gate renders and the page
    never calls ``/api/docs`` — the phase-16 pin, regression-checked
    with the extensionless KB seeded; the API itself 403s anonymous
    callers."""
    api_docs_calls: list[str] = []
    page.on(
        "request",
        lambda r: api_docs_calls.append(r.url) if "/api/docs" in r.url else None,
    )

    page.goto(f"{app_url}/sources.html")
    # The gate, with its sign-in link — not a redirect.
    gate = page.locator("#sources-gate")
    expect(gate).to_be_visible()
    expect(gate).to_contain_text("Sign in to view the full catalog")
    expect(gate.locator("a[href='/login.html?next=/sources.html']")).to_have_count(1)
    # Stat cards + table hidden…
    expect(page.locator("#stat-cards")).to_be_hidden()
    expect(page.locator("#docs-table")).to_be_hidden()
    expect(page.locator("#sources-empty")).to_be_hidden()
    # …and NO /api/docs call was ever made.
    assert api_docs_calls == [], f"anonymous sources page called /api/docs: {api_docs_calls}"
    # And the API gate itself: an anonymous GET /api/docs 403s (phase 16
    # — the router sits behind require_admin).
    assert page.request.get(f"{app_url}/api/docs").status == 403