Files
thoughtsync/tests/test_integration.py
T
bvandeusen 761c3b5e82
CI & Build / Build now, or wait for Android? (push) Successful in 2s
CI & Build / Python lint (push) Successful in 2s
CI & Build / TypeScript typecheck (push) Successful in 6s
CI & Build / Python tests (push) Failing after 8s
CI & Build / integration (push) Successful in 20s
CI & Build / Build & push image (push) Skipped
server: the body is the checklist here too, and note_items is dropped
M304 steps 3 and the server half of 4. The client half landed in 668f7fa; these
belong in one deploy, and the protocol floor below is what enforces that.

notes/checklist.py is the Python half of a grammar that now exists three times —
here, core/src/local/derive.rs, and (next) frontend/src/notes/markdown.ts. That
triplication is the deliberate cost: the alternative is a round trip to the server
before a phone can draw a checkbox. Each copy names the other two, and each is
tested against the same table of cases, including the near-misses that must stay
prose: `-[ ] x`, `- []`, `- [ ]x`, a `[ ]` mid-sentence.

Routes: add/update/delete items stop touching rows and rewrite note.body, all
through one _rewrite_body that runs the same sequence the PATCH route runs for a
body change — because it IS a body change. Revisions, #tag reconciliation, the
name, and link unfurls therefore happen in one place rather than three routes each
remembering to.

The reorder route is gone (rule 22). Reordering a checklist is moving a line, and
no client ever called it — the only reference in the tree was a test asserting the
route existed.

The API still returns `items`, DERIVED from the body on the way out. That is not a
second source of truth and it cannot disagree with the body it came from; it keeps
the web client working across the rest of this milestone and saves any consumer
that only wants to draw checkboxes from carrying a parser.

Export drops its separate items block, in both formats. The body already ends with
those exact lines, so writing them again would double every checklist in an export
and then double it again on re-import. Import still ACCEPTS items, because a Keep
takeout has a list and not a blob; it folds them in before the Note is built, so
display_title and _reconcile_tags both see the finished text.

Protocol 3 on both sides now. A v2 client is refused rather than half-served —
which matters more than I first said: _apply_note_items returned early on an absent
`items` key, so an un-bumped v3 client against a v2 server would not have LOST the
rows, it would have kept them and then had the migration fold them a second time.
Duplicated lists rather than missing ones. The floor prevents both.

Migration 0027 folds every existing row into its note's body and drops the table.
It inlines its own copy of the fold on purpose — a migration has to keep producing
what it produced the day it ran — and a test pins that copy against the app's until
they are allowed to diverge. updated_at is deliberately untouched: a client holding
an unpushed edit keeps the newer timestamp, so last-write-wins keeps its work
instead of the migration silently winning.

The downgrade is honest rather than faithful. It recreates an empty note_items and
leaves the bodies alone, because once items are lines nothing distinguishes one this
migration wrote from one somebody typed, and a downgrade that guessed would eat
hand-written lists. Recreating the table is still necessary: 0015's downgrade drops
a trigger ON note_items, and IF EXISTS covers the trigger, not the table.
2026-08-24 08:03:37 -04:00

476 lines
19 KiB
Python

"""The real-Postgres lane (family rule 6).
Everything else in this suite is deliberately DB-free, which means the schema the
migrations build has never been checked against the models that read it. That gap is
what this file closes, and it is not theoretical: M13 dropped three columns and
rebuilt a generated column, and until now `alembic upgrade head` ran for the first
time when the operator's container started.
Marked `integration` and excluded from the unit lane by `-m "not integration"`, so a
workstation without Postgres runs the rest of the suite unchanged.
The schema comes from real migrations, never `metadata.create_all` (rule 82) — the
point is to test what actually ships, and `create_all` would build a schema no
deployment has ever seen.
"""
from __future__ import annotations
import uuid
from datetime import datetime, timedelta, timezone
import pytest
import pytest_asyncio
from sqlalchemy import select, text
from thoughtsync import ratelimit
from thoughtsync.app import create_app
from thoughtsync.db import dispose_engine, session_scope
from thoughtsync.models.note import Note
from thoughtsync.models.user import User
from thoughtsync.settings import get_setting, live, refresh_live, reset_live, set_settings
from thoughtsync.notes.checklist import parse_items, set_item_checked
from thoughtsync.notes.helpers import derive_display_title
from thoughtsync.models.note_link_preview import NoteLinkPreview
from thoughtsync.models.note_revision import NoteRevision
from thoughtsync.revisions import REVISION_WINDOW_MINUTES, should_snapshot
from thoughtsync.unfurl_queue import _unfurl_new_urls, detect_urls
pytestmark = pytest.mark.integration
# Every table the tests touch, child-first so FKs never block the truncate.
# RESTART IDENTITY + CASCADE keeps this honest if a table gains children later.
_TABLES = "notes, note_revisions, note_labels, note_link_previews, labels, users"
@pytest_asyncio.fixture
async def db():
"""A session against the migrated database, wiped before each test.
Wiped BEFORE rather than after so a failed test leaves its rows behind to look at.
"""
async with session_scope() as session:
await session.execute(text(f"TRUNCATE {_TABLES} RESTART IDENTITY CASCADE"))
await session.commit()
yield session
await dispose_engine()
@pytest_asyncio.fixture
async def app_client(db):
"""A test client against the real app, over the migrated database.
The credential throttle is process-global and its counters outlive a single
test, so they are cleared here — otherwise a suite that registers a few times
starts handing out 429s for reasons that have nothing to do with the test.
"""
ratelimit.reset_all()
yield create_app().test_client()
ratelimit.reset_all()
@pytest_asyncio.fixture
async def owner(db):
"""A user to hang notes off — `notes.owner_id` is a real foreign key."""
user = User(email=f"{uuid.uuid4().hex}@example.test", display_name="Integration")
db.add(user)
await db.commit()
await db.refresh(user)
return user
async def test_the_migrated_schema_matches_the_models(db, owner):
"""The check that has never run: insert through the ORM, read it back.
A column the models expect and the migrations never created — or the reverse —
fails right here, instead of when a container starts.
"""
note = Note(owner_id=owner.id, body="a thought", display_title="a thought")
db.add(note)
await db.commit()
await db.refresh(note)
found = await db.scalar(select(Note).where(Note.id == note.id))
assert found is not None
assert found.body == "a thought"
assert found.display_title == "a thought"
async def test_the_dropped_columns_are_actually_gone(db):
"""M13 dropped three. If a migration silently no-opped, this is where it shows."""
cols = set(
(
await db.execute(
text("SELECT column_name FROM information_schema.columns WHERE table_name = 'notes'")
)
)
.scalars()
.all()
)
assert "title" not in cols, "notes.title should have gone in 0026"
assert "kind" not in cols, "notes.kind should have gone in 0025"
assert "display_title" in cols and "body" in cols
rev_cols = set(
(
await db.execute(
text("SELECT column_name FROM information_schema.columns WHERE table_name = 'note_revisions'")
)
)
.scalars()
.all()
)
assert "title" not in rev_cols, "note_revisions.title should have gone in 0026"
tables = set(
(await db.execute(text("SELECT table_name FROM information_schema.tables WHERE table_schema = 'public'")))
.scalars()
.all()
)
assert "note_links" not in tables, "note_links should have gone in 0024"
async def test_the_search_vector_was_rebuilt_over_the_name(db, owner):
"""0026 had to drop and recreate a STORED GENERATED column.
Postgres refuses to drop a column another generated column depends on, so getting
this wrong doesn't produce a subtly wrong ranking — it produces a migration that
won't run at all. Worth proving the replacement actually indexes something.
"""
note = Note(owner_id=owner.id, body="ferry tickets\nbook before friday", display_title="ferry tickets")
db.add(note)
await db.commit()
hit = await db.scalar(
text(
"SELECT count(*) FROM notes "
"WHERE search_vector @@ websearch_to_tsquery('english', :q)"
).bindparams(q="ferry")
)
assert hit == 1
# The NAME is weight A and the body weight B, which is what makes a name match
# rank above a body-only one. Both must be in the vector at all.
body_only = await db.scalar(
text(
"SELECT count(*) FROM notes "
"WHERE search_vector @@ websearch_to_tsquery('english', :q)"
).bindparams(q="friday")
)
assert body_only == 1
async def test_a_note_keeps_its_prose_on_both_sides_of_its_list(db, owner):
"""The shape M304 made expressible at all.
The old model could not hold this: a row had a position in a table and none in the
text, so a checklist could only ever render AFTER the body. Prose, list, prose is
the case that proves the storage changed, not just the styling.
"""
body = "weekend shop\n\n- [ ] milk\n- [x] eggs\n\nback before six"
note = Note(owner_id=owner.id, body=body, display_title=derive_display_title(body))
db.add(note)
await db.commit()
stored = await db.scalar(select(Note.body).where(Note.id == note.id))
assert [(i.text, i.checked) for i in parse_items(stored)] == [("milk", False), ("eggs", True)]
assert stored.splitlines()[0] == "weekend shop"
assert stored.splitlines()[-1] == "back before six"
async def test_ticking_an_item_is_a_body_edit(db, owner):
"""What replaced `_apply_note_items`: there is no separate thing left to apply.
The regression that function guarded against — a sync silently eating a checklist
off a note that also had a body — cannot recur, because there is nothing to delete.
A pushed body either has the lines or it does not.
"""
body = "packing\n\n- [ ] socks"
note = Note(owner_id=owner.id, body=body, display_title="packing")
db.add(note)
await db.commit()
note.body = set_item_checked(note.body, 0, True)
await db.commit()
stored = await db.scalar(select(Note.body).where(Note.id == note.id))
assert stored == "packing\n\n- [x] socks"
assert parse_items(stored)[0].checked
# The prose is untouched — a tick rewrites one line, not the note.
assert stored.splitlines()[0] == "packing"
async def test_a_note_with_only_a_list_still_has_a_name(db, owner):
"""The hole that made removing the title unsafe, still closed — by a different
mechanism. There is no item table to fall back to any more; the name comes from
the first line with its marker stripped, because calling the note "- [ ] milk"
would show someone the storage instead of the note.
"""
body = "- [ ] milk\n- [ ] eggs"
note = Note(owner_id=owner.id, body=body, display_title=derive_display_title(body))
db.add(note)
await db.commit()
assert (await db.scalar(select(Note.display_title).where(Note.id == note.id))) == "milk"
async def test_auto_unfurl_stores_a_preview_and_skips_what_is_cached(db, owner, monkeypatch):
"""The background pass, run inline so the assertions are deterministic.
The network is stubbed — this is about what reaches the DATABASE, not about
parsing someone's OpenGraph tags (unfurl.py's own tests cover that). What matters
here is the part only a real database can show: the unique constraint holding, the
upsert going to the right row, and a second pass not re-fetching.
"""
note = Note(
owner_id=owner.id,
body="read https://example.com/a and https://example.com/b",
display_title="read https://example.com/a and https://example.com/b",
)
db.add(note)
await db.commit()
calls: list[str] = []
async def fake_unfurl(url):
calls.append(url)
return {"url": url, "title": f"T {url}", "description": None, "image_url": None, "site_name": "example.com"}
monkeypatch.setattr("thoughtsync.unfurl_queue.unfurl", fake_unfurl)
await _unfurl_new_urls(note.id, note.body)
assert sorted(calls) == ["https://example.com/a", "https://example.com/b"]
rows = (await db.scalars(select(NoteLinkPreview).where(NoteLinkPreview.note_id == note.id))).all()
assert {r.url for r in rows} == {"https://example.com/a", "https://example.com/b"}
assert all(r.title.startswith("T ") for r in rows)
# A second pass over an unchanged body fetches nothing — the whole reason
# `schedule` is safe to call on every save.
calls.clear()
await _unfurl_new_urls(note.id, note.body)
assert calls == []
async def test_auto_unfurl_drops_a_preview_whose_url_left_the_body(db, owner, monkeypatch):
"""A slow fetch must not resurrect a link the person deleted mid-flight."""
note = Note(owner_id=owner.id, body="https://example.com/gone", display_title="x")
db.add(note)
await db.commit()
async def fake_unfurl(url):
# Simulate the body changing while the request was in the air.
return {"url": url, "title": "T", "description": None, "image_url": None, "site_name": None}
monkeypatch.setattr("thoughtsync.unfurl_queue.unfurl", fake_unfurl)
note.body = "changed my mind"
await db.commit()
await _unfurl_new_urls(note.id, "https://example.com/gone")
rows = (await db.scalars(select(NoteLinkPreview).where(NoteLinkPreview.note_id == note.id))).all()
assert rows == [], "a preview was stored for a URL the note no longer contains"
async def test_detection_agrees_with_what_gets_stored(db, owner, monkeypatch):
"""The detector and the storage path read the same body the same way."""
body = "one https://example.com/x. two (https://example.com/y) three"
assert detect_urls(body) == ["https://example.com/x", "https://example.com/y"]
note = Note(owner_id=owner.id, body=body, display_title="one")
db.add(note)
await db.commit()
async def fake_unfurl(url):
return {"url": url, "title": "T", "description": None, "image_url": None, "site_name": None}
monkeypatch.setattr("thoughtsync.unfurl_queue.unfurl", fake_unfurl)
await _unfurl_new_urls(note.id, body)
stored = {
r for r in (await db.scalars(select(NoteLinkPreview.url).where(NoteLinkPreview.note_id == note.id))).all()
}
assert stored == set(detect_urls(body))
async def test_registration_closes_itself_once_an_admin_exists(app_client, db):
"""The gap this removes: registration was open between "my account exists" and
"I remembered to turn it off", and on a public host that gap starts at DNS.
Runs against a real database because it is the interaction between two writes —
the user row and the settings row — inside one transaction.
"""
# The instance is empty (the fixture truncated it), so this is the first account:
# allowed unconditionally, and it becomes the admin.
first = await app_client.post(
"/api/auth/register",
json={"email": "owner@example.test", "password": "a-long-enough-password"},
)
assert first.status_code == 201
assert (await first.get_json())["is_admin"] is True
# …and the door shut behind it.
async with session_scope() as fresh:
assert await get_setting(fresh, "allow_registration") is False
second = await app_client.post(
"/api/auth/register",
json={"email": "stranger@example.test", "password": "a-long-enough-password"},
)
assert second.status_code == 403
# Re-opening it deliberately still works — that is how a second person gets in
# until invites exist.
async with session_scope() as fresh:
await set_settings(fresh, {"allow_registration": True})
await fresh.commit()
third = await app_client.post(
"/api/auth/register",
json={"email": "invited@example.test", "password": "a-long-enough-password"},
)
assert third.status_code == 201
assert (await third.get_json())["is_admin"] is False
async def test_security_settings_are_live_and_bounded(app_client, db):
"""The security values are settings now, not constants — so saving one has to take
effect without a restart, and a dangerous value has to be refused.
Real database because the whole point is the round trip: write through the admin
API, re-read into the cache the throttle consults, observe the new number.
"""
# An admin to authenticate as. First account, so it is allowed and becomes admin.
reset_live()
created = await app_client.post(
"/api/auth/register",
json={"email": "admin@example.test", "password": "a-long-enough-password"},
)
assert created.status_code == 201
# Defaults are what the registry says.
async with session_scope() as fresh:
await refresh_live(fresh)
assert live("trusted_proxy_hops") == 1
assert live("signin_limit_per_account") == 10
# A value that would disable the protection is REFUSED, not clamped — storing a
# different number than the one typed is how somebody ends up believing a limit
# is set to something it is not.
bad = await app_client.patch("/api/settings", json={"signin_limit_per_account": 0})
assert bad.status_code == 400
assert "at least" in (await bad.get_json())["error"]
# …and so is a hop count that would trust anything a caller sent.
bad_hops = await app_client.patch("/api/settings", json={"trusted_proxy_hops": 99})
assert bad_hops.status_code == 400
# A legitimate change applies to the cache the throttle reads, immediately.
ok = await app_client.patch(
"/api/settings", json={"signin_limit_per_account": 3, "trusted_proxy_hops": 2}
)
assert ok.status_code == 200
assert live("signin_limit_per_account") == 3
assert live("trusted_proxy_hops") == 2
# And it is persisted, not just cached.
async with session_scope() as fresh:
assert await get_setting(fresh, "trusted_proxy_hops") == 2
reset_live()
async def test_the_security_group_reaches_the_admin_ui(app_client, db):
"""Every security value has to be visible and editable, which is the whole reason
they moved out of the environment."""
created = await app_client.post(
"/api/auth/register",
json={"email": "admin2@example.test", "password": "a-long-enough-password"},
)
assert created.status_code == 201
resp = await app_client.get("/api/settings")
assert resp.status_code == 200
rows = (await resp.get_json())["settings"]
security = {r["key"]: r for r in rows if r["group"] == "Security"}
assert set(security) == {
"trusted_proxy_hops",
"signin_limit_per_account",
"signin_limit_per_address",
"signin_window_minutes",
"register_limit_per_address",
"register_window_minutes",
}
# The UI renders a number input from these, and it cannot offer a safe range it
# was never told about.
for row in security.values():
assert row["type"] == "int"
assert row["minimum"] is not None and row["maximum"] is not None
assert row["description"], f"{row['key']} has no description to explain itself"
async def _revision_count(db, note_id) -> int:
rows = (await db.scalars(select(NoteRevision.id).where(NoteRevision.note_id == note_id))).all()
return len(rows)
async def test_a_session_of_edits_costs_one_revision(db, owner):
"""The change that makes autosave affordable.
Version history used to snapshot on EVERY body write, so the clients saved as
rarely as they could — only when an editor closed — and a crash mid-session lost
everything typed. Durability was paying for history. Now a sitting earns one
revision no matter how many times it is written, so a client can write whenever
it likes.
"""
note = Note(owner_id=owner.id, body="one", display_title="one")
db.add(note)
await db.commit()
# A session's worth of autosaves.
for text_ in ("one two", "one two three", "one two three four"):
if await should_snapshot(db, note.id, note.body, text_):
db.add(NoteRevision(note_id=note.id, body=note.body))
note.body = text_
await db.commit()
assert await _revision_count(db, note.id) == 1
# And it is the body as it was BEFORE the sitting, not some midpoint — which is
# what makes one-per-session the useful granularity rather than an arbitrary one.
kept = (await db.scalars(select(NoteRevision.body).where(NoteRevision.note_id == note.id))).all()
assert kept == ["one"]
async def test_rewriting_the_same_text_is_not_a_version(db, owner):
note = Note(owner_id=owner.id, body="unchanged", display_title="unchanged")
db.add(note)
await db.commit()
assert await should_snapshot(db, note.id, note.body, "unchanged") is False
assert await _revision_count(db, note.id) == 0
async def test_a_later_sitting_earns_its_own_revision(db, owner):
"""The window has to REOPEN, or a note edited daily would keep only its first
version forever — which would be a worse history than the one we replaced."""
note = Note(owner_id=owner.id, body="today", display_title="today")
db.add(note)
await db.flush()
# A revision from longer ago than one session: the clock is not mocked, the row
# is simply written with an older timestamp, which is what the query reads.
stale = datetime.now(timezone.utc) - timedelta(minutes=REVISION_WINDOW_MINUTES + 1)
db.add(NoteRevision(note_id=note.id, body="yesterday", created_at=stale))
await db.commit()
assert await should_snapshot(db, note.id, note.body, "tomorrow") is True
async def test_a_revision_inside_the_window_blocks_another(db, owner):
note = Note(owner_id=owner.id, body="draft", display_title="draft")
db.add(note)
await db.flush()
db.add(NoteRevision(note_id=note.id, body="earlier", created_at=datetime.now(timezone.utc)))
await db.commit()
assert await should_snapshot(db, note.id, note.body, "draft revised") is False