feat(mail): read and index Nextcloud Mail via the Mail OCS API

Add read-only support for the Nextcloud Mail app, plus semantic indexing
of mail messages. The MCP server never speaks IMAP/POP3 itself: it calls
the Mail app's CSRF-free OCS API (/ocs/v2.php/apps/mail/api/...) with the
existing Basic-Auth app-password flow and an OCS-APIRequest header, and the
Mail app handles IMAP server-side.

- client/mail.py: MailClient (accounts, mailboxes, messages, message,
  attachment), OCS-envelope aware.
- models/mail.py: Pydantic models with the API's camelCase aliases.
- server/mail.py: 5 read-only MCP tools (mail.read scope), registered in
  AVAILABLE_APPS.
- Vector pipeline: new "mail_message" doc_type wired into scanner
  (scan_mail_messages, newest-N per mailbox), processor (body -> markdown
  embedding), per-id verifier, and context expansion.
- Tests: client API, model round-trips, verifier behavior; consent-backstop
  test now derives its allowed set from INDEXED_DOC_TYPES.
- README + semantic-search docstrings updated.

Requires Mail 5.x / Nextcloud 32+ and a mail account configured in the
Mail app. Follow-up: astrolabe must advertise "mail_message" in its
enabled_doc_types capability for search under admin doc_type restriction.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Chris Coutinho
2026-06-20 11:53:47 +02:00
co-authored by Claude Opus 4.8
parent a9d36a8aee
commit 3074622455
18 changed files with 1413 additions and 7 deletions
View File
+206
View File
@@ -0,0 +1,206 @@
"""Unit tests for MailClient API methods."""
import logging
from typing import Any
import httpx
import pytest
from nextcloud_mcp_server.client.mail import MailClient
from tests.client.conftest import create_mock_response
logger = logging.getLogger(__name__)
# Mark all tests in this module as unit tests
pytestmark = pytest.mark.unit
def _ocs_response(data: Any, status_code: int = 200) -> httpx.Response:
"""Wrap a payload in the standard OCS envelope."""
return create_mock_response(
status_code=status_code,
json_data={
"ocs": {
"meta": {"status": "ok", "statuscode": status_code, "message": "OK"},
"data": data,
}
},
)
async def test_list_accounts_unwraps_ocs_envelope(mocker):
"""list_accounts returns the ocs.data payload."""
mock_response = _ocs_response(
[
{"id": 1, "email": "alice@example.com", "isDelegated": False},
{"id": 2, "email": "bob@example.com", "isDelegated": False},
]
)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
accounts = await client.list_accounts()
assert len(accounts) == 2
assert accounts[0]["id"] == 1
assert accounts[0]["email"] == "alice@example.com"
# Correct URL, OCS header, and format=json param.
args, kwargs = mock_make_request.call_args
assert args == ("GET", "/ocs/v2.php/apps/mail/api/account/list")
assert kwargs["headers"]["OCS-APIRequest"] == "true"
assert kwargs["params"]["format"] == "json"
async def test_get_mailboxes_passes_account_id(mocker):
"""get_mailboxes sends accountId and unwraps the list."""
mock_response = _ocs_response(
[
{
"databaseId": 10,
"id": "SU5CT1g=",
"name": "INBOX",
"displayName": "INBOX",
"accountId": 1,
"specialUse": ["inbox"],
"unread": 3,
}
]
)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
mailboxes = await client.get_mailboxes(account_id=1)
assert len(mailboxes) == 1
assert mailboxes[0]["databaseId"] == 10
assert mailboxes[0]["specialUse"] == ["inbox"]
args, kwargs = mock_make_request.call_args
assert args == ("GET", "/ocs/v2.php/apps/mail/api/mailboxes")
assert kwargs["params"]["accountId"] == 1
async def test_list_messages_builds_params(mocker):
"""list_messages forwards limit/cursor/filter/view query params."""
mock_response = _ocs_response(
[
{
"databaseId": 100,
"subject": "Hello",
"dateInt": 1700000000,
"from": [{"label": "Alice", "email": "alice@example.com"}],
"to": [{"label": "Bob", "email": "bob@example.com"}],
"mailboxId": 10,
}
]
)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
messages = await client.list_messages(
10, cursor=42, filter="hello", limit=50, view="threaded"
)
assert len(messages) == 1
assert messages[0]["databaseId"] == 100
args, kwargs = mock_make_request.call_args
assert args == ("GET", "/ocs/v2.php/apps/mail/api/mailboxes/10/messages")
assert kwargs["params"]["limit"] == 50
assert kwargs["params"]["cursor"] == 42
assert kwargs["params"]["filter"] == "hello"
assert kwargs["params"]["view"] == "threaded"
async def test_list_messages_omits_optional_params(mocker):
"""Optional params are omitted when not supplied; limit always present."""
mock_response = _ocs_response([])
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
await client.list_messages(10)
_, kwargs = mock_make_request.call_args
params = kwargs["params"]
assert params["limit"] == 20 # default
assert "cursor" not in params
assert "filter" not in params
assert "view" not in params
async def test_get_message_unwraps_full_message(mocker):
"""get_message returns the full message dict."""
mock_response = _ocs_response(
{
"id": 100,
"subject": "Hello",
"hasHtmlBody": True,
"body": "<p>Hi there</p>",
"from": [{"label": "Alice", "email": "alice@example.com"}],
"attachments": [
{
"id": "1.2",
"fileName": "doc.pdf",
"mime": "application/pdf",
"size": 1024,
}
],
}
)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
message = await client.get_message(100)
assert message["id"] == 100
assert message["hasHtmlBody"] is True
assert message["attachments"][0]["fileName"] == "doc.pdf"
args, _ = mock_make_request.call_args
assert args == ("GET", "/ocs/v2.php/apps/mail/api/message/100")
async def test_get_attachment_unwraps_json(mocker):
"""get_attachment returns the JSON attachment object (not a binary download)."""
mock_response = _ocs_response(
{"name": "doc.pdf", "mime": "application/pdf", "size": 1024, "content": "abc"}
)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mock_make_request = mocker.patch.object(
MailClient, "_make_request", return_value=mock_response
)
client = MailClient(mock_client, "testuser")
attachment = await client.get_attachment(100, "1.2")
assert attachment["name"] == "doc.pdf"
assert attachment["content"] == "abc"
args, _ = mock_make_request.call_args
assert args == ("GET", "/ocs/v2.php/apps/mail/api/message/100/attachment/1.2")
async def test_empty_data_returns_empty_list(mocker):
"""A null ocs.data payload degrades to an empty list for list endpoints."""
mock_response = _ocs_response(None)
mock_client = mocker.AsyncMock(spec=httpx.AsyncClient)
mocker.patch.object(MailClient, "_make_request", return_value=mock_response)
client = MailClient(mock_client, "testuser")
assert await client.list_accounts() == []
+92
View File
@@ -12,6 +12,7 @@ from nextcloud_mcp_server.search.algorithms import SearchResult
from nextcloud_mcp_server.search.verification import (
_verify_deck_cards,
_verify_files,
_verify_mail_messages,
_verify_news_items,
_verify_notes,
get_supported_doc_types,
@@ -222,6 +223,97 @@ async def test_verify_notes_string_doc_id_matches_production(mocker):
notes_client.get_note.assert_awaited_once_with(42)
# ---------------------------------------------------------------------------
# Mail verifier (per-id, mirrors the note verifier)
# ---------------------------------------------------------------------------
@pytest.mark.unit
async def test_verify_mail_200_keeps_all(mocker):
mail_client = SimpleNamespace(get_message=mocker.AsyncMock(return_value={"id": 42}))
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client, [_make_result(42, doc_type="mail_message")], _sem()
)
assert result == {"42"}
mail_client.get_message.assert_awaited_once_with(42)
@pytest.mark.unit
async def test_verify_mail_404_drops(mocker):
mail_client = SimpleNamespace(
get_message=mocker.AsyncMock(side_effect=_http_error(404))
)
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client, [_make_result(42, doc_type="mail_message")], _sem()
)
assert result == set()
@pytest.mark.unit
async def test_verify_mail_403_drops(mocker):
mail_client = SimpleNamespace(
get_message=mocker.AsyncMock(side_effect=_http_error(403))
)
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client, [_make_result(42, doc_type="mail_message")], _sem()
)
assert result == set()
@pytest.mark.unit
async def test_verify_mail_transient_5xx_keeps(mocker):
mail_client = SimpleNamespace(
get_message=mocker.AsyncMock(side_effect=_http_error(503))
)
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client, [_make_result(42, doc_type="mail_message")], _sem()
)
assert result == {"42"}
@pytest.mark.unit
async def test_verify_mail_non_numeric_id_keeps(mocker):
mail_client = SimpleNamespace(get_message=mocker.AsyncMock())
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client, [_make_result("not-a-number", doc_type="mail_message")], _sem()
)
assert result == {"not-a-number"}
# Malformed id is kept without any network call.
mail_client.get_message.assert_not_awaited()
@pytest.mark.unit
async def test_verify_mail_mixed_outcomes(mocker):
async def fake_get(message_id: int):
if message_id == 20:
raise _http_error(404) # deleted
return {"id": message_id}
mail_client = SimpleNamespace(get_message=mocker.AsyncMock(side_effect=fake_get))
client = SimpleNamespace(mail=mail_client, username="alice")
result = await _verify_mail_messages(
client,
[
_make_result(10, doc_type="mail_message"),
_make_result(20, doc_type="mail_message"),
_make_result(30, doc_type="mail_message"),
],
_sem(),
)
assert result == {"10", "30"}
# ---------------------------------------------------------------------------
# News batch verifier
# ---------------------------------------------------------------------------
+109
View File
@@ -0,0 +1,109 @@
"""Unit tests for Mail Pydantic models (alias mapping from the OCS API)."""
import pytest
from nextcloud_mcp_server.models.mail import (
GetMessageResponse,
ListAccountsResponse,
MailAccount,
MailMailbox,
MailMessage,
MailMessageSummary,
)
pytestmark = pytest.mark.unit
def test_account_maps_is_delegated_alias():
account = MailAccount(**{"id": 1, "email": "a@example.com", "isDelegated": True})
assert account.id == 1
assert account.is_delegated is True
def test_mailbox_maps_camelcase_aliases():
mailbox = MailMailbox(
**{
"databaseId": 10,
"id": "SU5CT1g=",
"name": "INBOX",
"displayName": "Inbox",
"accountId": 1,
"specialUse": ["inbox"],
"unread": 5,
}
)
assert mailbox.database_id == 10
assert mailbox.account_id == 1
assert mailbox.display_name == "Inbox"
assert mailbox.special_use == ["inbox"]
assert mailbox.unread == 5
def test_message_summary_maps_from_and_dateint():
summary = MailMessageSummary(
**{
"databaseId": 100,
"subject": "Hello",
"dateInt": 1700000000,
"from": [{"label": "Alice", "email": "alice@example.com"}],
"to": [{"email": "bob@example.com"}],
"mailboxId": 10,
"previewText": "snippet",
"flags": {"seen": True, "hasAttachments": True},
}
)
assert summary.database_id == 100
assert summary.date_int == 1700000000
assert summary.from_[0].label == "Alice"
assert summary.to[0].email == "bob@example.com"
assert summary.mailbox_id == 10
assert summary.preview_text == "snippet"
assert summary.flags is not None
assert summary.flags.seen is True
assert summary.flags.has_attachments is True
def test_full_message_maps_body_and_attachments():
message = MailMessage(
**{
"id": 100,
"subject": "Hello",
"hasHtmlBody": True,
"body": "<p>Hi</p>",
"from": [{"label": "Alice", "email": "alice@example.com"}],
"attachments": [
{
"id": "1.2",
"fileName": "doc.pdf",
"mime": "application/pdf",
"size": 1024,
}
],
}
)
assert message.id == 100
assert message.has_html_body is True
assert message.body == "<p>Hi</p>"
assert message.attachments[0].file_name == "doc.pdf"
assert message.attachments[0].id == "1.2"
def test_message_tolerates_missing_optional_fields():
"""A 206 partial response may omit the body."""
message = MailMessage(**{"id": 100})
assert message.id == 100
assert message.body is None
assert message.has_html_body is False
assert message.attachments == []
def test_response_models_wrap_results():
resp = ListAccountsResponse(
results=[MailAccount(id=1, email="a@example.com")], total_count=1
)
assert resp.success is True
assert resp.total_count == 1
assert resp.results[0].email == "a@example.com"
msg_resp = GetMessageResponse(message=MailMessage(id=5, subject="Hi"))
assert msg_resp.message.id == 5
@@ -104,7 +104,9 @@ async def test_noop_when_allowed_is_none(monkeypatch):
async def test_noop_when_all_text_types_allowed(monkeypatch):
send = AsyncMock()
allowed = frozenset({"note", "news_item", "deck_card", "file"})
# Derive from INDEXED_DOC_TYPES so a newly-indexed text type doesn't make
# this "all allowed" set silently incomplete (and trip the backstop).
allowed = frozenset(scanner_module.INDEXED_DOC_TYPES)
queued = await _enqueue_deletes_for_disabled_types(
"alice", _producer(send), allowed, 1
)