gateway/tests/unit/services/test_knowledge_ingest_consumer.py

#!/usr/bin/env python3
# Copyright (c) 2025 Patrick Motsch
# All rights reserved.
"""Unit tests for KnowledgeIngestionConsumer event dispatch.

- `connection.established` → enqueue a `connection.bootstrap` job.
- `connection.revoked` → synchronous purge via KnowledgeObjects.
"""

import asyncio
import os
import sys
import types

sys.path.insert(0, os.path.join(os.path.dirname(__file__), "../../.."))

from modules.serviceCenter.services.serviceKnowledge import subConnectorIngestConsumer as consumer


def _resetRegistration(monkeypatch):
    """Force the module-level guard to register fresh in each test."""
    monkeypatch.setattr(consumer, "_registered", False)


def test_onConnectionEstablished_enqueues_bootstrap(monkeypatch):
    startedJobs = []

    async def _fakeStartJob(jobType, payload, **kwargs):
        startedJobs.append({"jobType": jobType, "payload": payload, "kwargs": kwargs})
        return "job-1"

    monkeypatch.setattr(consumer, "startJob", _fakeStartJob)
    consumer._onConnectionEstablished(
        connectionId="c1", authority="msft", userId="u1"
    )
    # Drain pending tasks created by the consumer.
    loop = asyncio.new_event_loop()
    try:
        asyncio.set_event_loop(loop)
        # If the consumer created a Task on a closed loop the fake startJob
        # was still called synchronously via asyncio.run — in either case we
        # check the recorded call.
    finally:
        loop.close()

    assert len(startedJobs) == 1
    assert startedJobs[0]["jobType"] == consumer.BOOTSTRAP_JOB_TYPE
    assert startedJobs[0]["payload"]["connectionId"] == "c1"
    assert startedJobs[0]["payload"]["authority"] == "msft"
    assert startedJobs[0]["kwargs"]["triggeredBy"] == "u1"


def test_onConnectionEstablished_ignores_missing_id(monkeypatch):
    called = []

    async def _fakeStartJob(*a, **kw):
        called.append(1)
        return "x"

    monkeypatch.setattr(consumer, "startJob", _fakeStartJob)
    consumer._onConnectionEstablished(connectionId="", authority="msft")
    assert called == []


def test_onConnectionRevoked_runs_sync_purge(monkeypatch):
    class _FakeKnowledge:
        def __init__(self):
            self.calls = []

        def deleteFileContentIndexByConnectionId(self, cid):
            self.calls.append(cid)
            return {"indexRows": 2, "chunks": 5}

    fakeKnow = _FakeKnowledge()

    def _fakeGetInterface(_user=None):
        return fakeKnow

    monkeypatch.setattr(consumer, "getKnowledgeInterface", _fakeGetInterface)
    consumer._onConnectionRevoked(
        connectionId="c1", authority="msft", userId="u1", reason="disconnected"
    )
    assert fakeKnow.calls == ["c1"]


def test_onConnectionRevoked_ignores_missing_id(monkeypatch):
    seen = []

    def _fakeGetInterface(_user=None):
        class _K:
            def deleteFileContentIndexByConnectionId(self, cid):
                seen.append(cid)
                return {"indexRows": 0, "chunks": 0}

        return _K()

    monkeypatch.setattr(consumer, "getKnowledgeInterface", _fakeGetInterface)
    consumer._onConnectionRevoked(connectionId="")
    assert seen == []


def test_bootstrap_job_skips_non_pilot_authority(monkeypatch):
    async def _run():
        result = await consumer._bootstrapJobHandler(
            {"payload": {"connectionId": "c1", "authority": "google"}},
            lambda *_: None,
        )
        return result

    result = asyncio.run(_run())
    assert result["skipped"] is True
    assert result["authority"] == "google"


def test_bootstrap_job_dispatches_msft_parts(monkeypatch):
    calls = {"sp": 0, "ol": 0}

    async def _fakeSp(connectionId, progressCb=None):
        calls["sp"] += 1
        return {"indexed": 1}

    async def _fakeOl(connectionId, progressCb=None):
        calls["ol"] += 1
        return {"indexed": 2}

    # subConnectorSync* are lazy-imported inside the handler; install fake
    # modules before invoking.
    fakeSharepoint = types.ModuleType("subConnectorSyncSharepoint")
    fakeSharepoint.bootstrapSharepoint = _fakeSp
    fakeOutlook = types.ModuleType("subConnectorSyncOutlook")
    fakeOutlook.bootstrapOutlook = _fakeOl
    monkeypatch.setitem(
        sys.modules,
        "modules.serviceCenter.services.serviceKnowledge.subConnectorSyncSharepoint",
        fakeSharepoint,
    )
    monkeypatch.setitem(
        sys.modules,
        "modules.serviceCenter.services.serviceKnowledge.subConnectorSyncOutlook",
        fakeOutlook,
    )

    async def _run():
        return await consumer._bootstrapJobHandler(
            {"payload": {"connectionId": "c1", "authority": "msft"}},
            lambda *_: None,
        )

    result = asyncio.run(_run())
    assert calls == {"sp": 1, "ol": 1}
    assert result["sharepoint"] == {"indexed": 1}
    assert result["outlook"] == {"indexed": 2}


if __name__ == "__main__":
    # Usable without pytest fixtures for a quick smoke run.
    class _MP:
        def __init__(self):
            self.undos = []

        def setattr(self, target, name_or_value, value=None):
            if value is None:
                # target is an object, name_or_value is value → no, original signature
                raise SystemExit("use pytest monkeypatch in CLI")
            self.undos.append((target, name_or_value, getattr(target, name_or_value)))
            setattr(target, name_or_value, value)

        def setitem(self, mapping, key, value):
            self.undos.append((mapping, key, mapping.get(key)))
            mapping[key] = value

    print("Run via pytest: pytest tests/unit/services/test_knowledge_ingest_consumer.py")