Chris CoutinhoandClaude Opus 4.7
2e2a098bee
The /webhooks/nextcloud endpoint was a no-op stub that logged the payload and returned 200 OK; webhook deletions never reached Qdrant. Compounding that, _get_webhook_uri() registered the docker-compose internal hostname (http://mcp:8000) with Nextcloud whenever /.dockerenv existed — including ECS Fargate — so cloud deployments were registering a URL NC could not resolve. - New vector/webhook_parser.py extracts a DocumentTask from NodeCreatedEvent / NodeWrittenEvent / BeforeNodeDeletedEvent payloads scoped to */files/Notes/*.md (matching the registered preset filters). - New vector/webhook_receiver.py pushes that task onto the same send-stream the scanner uses (app.state.document_send_stream), with 503 when sync is not running so NC retries delivery. - _get_webhook_uri() now prefers NEXTCLOUD_MCP_SERVER_URL over the /.dockerenv branch, so the explicit public URL set on cloud tasks wins; docker-compose dev still falls back to the internal name when no public URL is configured. Calendar / Tables event parsing is intentionally out of scope here. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
87 lines
2.9 KiB
Python
87 lines
2.9 KiB
Python
"""Parse Nextcloud webhook payloads into DocumentTask objects.
|
|
|
|
Maps Nextcloud webhook events to vector-sync DocumentTasks. The handler at
|
|
``/webhooks/nextcloud`` calls :func:`extract_document_task` and forwards any
|
|
non-None result to the same processor send-stream the scanner uses.
|
|
|
|
Currently scoped to file (note) events. Calendar / Tables events fall through
|
|
to ``None`` for now; those parsers can be added in follow-up changes.
|
|
|
|
See ADR-010 for the design and ``webhook-testing-findings.md`` for real
|
|
captured payloads.
|
|
"""
|
|
|
|
import logging
|
|
import re
|
|
|
|
from nextcloud_mcp_server.vector.scanner import DocumentTask
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
_FILE_EVENT_CREATED = "OCP\\Files\\Events\\Node\\NodeCreatedEvent"
|
|
_FILE_EVENT_WRITTEN = "OCP\\Files\\Events\\Node\\NodeWrittenEvent"
|
|
_FILE_EVENT_BEFORE_DELETED = "OCP\\Files\\Events\\Node\\BeforeNodeDeletedEvent"
|
|
|
|
# Matches paths inside any user's Notes folder ending in .md, e.g.
|
|
# "/admin/files/Notes/Sub/Note.md" or "/alice/files/Notes/foo.md".
|
|
_NOTES_PATH_RE = re.compile(r"^/[^/]+/files/Notes/.+\.md$")
|
|
|
|
|
|
def extract_document_task(payload: dict) -> DocumentTask | None:
|
|
"""Convert a Nextcloud webhook payload into a DocumentTask.
|
|
|
|
Returns None for any event we don't (yet) handle, or any event whose
|
|
target isn't a markdown file under a user's Notes folder. Callers should
|
|
treat None as "ignored" — not an error.
|
|
"""
|
|
try:
|
|
event = payload["event"]
|
|
event_class = event["class"]
|
|
user_id = payload["user"]["uid"]
|
|
except (KeyError, TypeError):
|
|
logger.debug("Webhook payload missing user/event/class fields")
|
|
return None
|
|
|
|
if event_class in (
|
|
_FILE_EVENT_CREATED,
|
|
_FILE_EVENT_WRITTEN,
|
|
_FILE_EVENT_BEFORE_DELETED,
|
|
):
|
|
return _parse_file_event(event_class, event, user_id, payload.get("time", 0))
|
|
|
|
logger.debug("Ignoring webhook for unsupported event: %s", event_class)
|
|
return None
|
|
|
|
|
|
def _parse_file_event(
|
|
event_class: str, event: dict, user_id: str, time: int
|
|
) -> DocumentTask | None:
|
|
node = event.get("node") or {}
|
|
path = node.get("path", "")
|
|
node_id = node.get("id")
|
|
|
|
if not _NOTES_PATH_RE.match(path):
|
|
# Not a note file — could be a parent folder, an unrelated file, etc.
|
|
return None
|
|
|
|
if node_id is None:
|
|
# BeforeNodeDeletedEvent should still carry node.id; if it doesn't
|
|
# we can't address the Qdrant points to delete. Skip rather than
|
|
# guess — the polling scanner will catch up via its grace period.
|
|
logger.warning(
|
|
"Webhook %s for note %s missing node.id; falling back to scanner",
|
|
event_class,
|
|
path,
|
|
)
|
|
return None
|
|
|
|
operation = "delete" if event_class == _FILE_EVENT_BEFORE_DELETED else "index"
|
|
|
|
return DocumentTask(
|
|
user_id=user_id,
|
|
doc_id=str(node_id),
|
|
doc_type="note",
|
|
operation=operation,
|
|
modified_at=int(time),
|
|
)
|