gateway/tests/unit/services/test_bootstrap_outlook.py

#!/usr/bin/env python3
# Copyright (c) 2025 Patrick Motsch
# All rights reserved.
"""Bootstrap Outlook tests with a fake adapter + knowledge service.

Verifies:
- Well-known folders (inbox, sentitems) are discovered via Graph.
- Each message produces a `requestIngestion` call with sourceKind=outlook_message
  and structured contentObjects (header / snippet / body).
- Pagination via `@odata.nextLink` is followed.
- changeKey is forwarded as contentVersion → idempotency.
"""

import asyncio
import os
import sys
from types import SimpleNamespace

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

from modules.serviceCenter.services.serviceKnowledge.subConnectorSyncOutlook import (
    bootstrapOutlook,
    OutlookBootstrapLimits,
    _syntheticMessageId,
    _buildContentObjects,
)


class _FakeOutlookAdapter:
    def __init__(self, messages_by_folder, paginated_folder=None, page2=None):
        self._folders = {"inbox": "INBOX-ID", "sentitems": "SENT-ID"}
        self._messages = messages_by_folder
        self._paginated_folder = paginated_folder
        self._page2 = page2 or []
        self.requested_endpoints = []

    async def _graphGet(self, endpoint: str):
        self.requested_endpoints.append(endpoint)
        if endpoint.startswith("me/mailFolders/") and "/messages" not in endpoint:
            wellKnown = endpoint.split("/")[-1]
            fid = self._folders.get(wellKnown)
            if not fid:
                return {"error": "not found"}
            return {"id": fid, "displayName": wellKnown}
        # message page request: e.g. me/mailFolders/INBOX-ID/messages?...
        for fid, messages in self._messages.items():
            if f"me/mailFolders/{fid}/messages" in endpoint:
                page = {"value": messages}
                if fid == self._paginated_folder and "skiptoken" not in endpoint:
                    page["@odata.nextLink"] = (
                        "https://graph.microsoft.com/v1.0/"
                        f"me/mailFolders/{fid}/messages?$skiptoken=abc"
                    )
                elif fid == self._paginated_folder and "skiptoken" in endpoint:
                    page = {"value": self._page2}
                return page
        return {"value": []}

    async def browse(self, path):
        return []


class _FakeKnowledgeService:
    def __init__(self, duplicateIds=None):
        self.calls = []
        self._duplicates = duplicateIds or set()

    async def requestIngestion(self, job):
        self.calls.append(job)
        status = "duplicate" if job.sourceId in self._duplicates else "indexed"
        return SimpleNamespace(
            jobId=job.sourceId, status=status, contentHash="h",
            fileId=job.sourceId, index=None, error=None,
        )


def _msg(mid: str, subject: str = "Hi", change: str = "ck1"):
    return {
        "id": mid,
        "subject": subject,
        "from": {"emailAddress": {"name": "Alice", "address": "a@x.com"}},
        "toRecipients": [{"emailAddress": {"name": "Bob", "address": "b@x.com"}}],
        "ccRecipients": [],
        "receivedDateTime": "2025-01-01T10:00:00Z",
        "bodyPreview": "Hello world",
        "body": {"contentType": "text", "content": "Hello world\nThis is the body."},
        "internetMessageId": f"<{mid}@local>",
        "hasAttachments": False,
        "changeKey": change,
    }


def test_buildContentObjects_emits_header_snippet_body():
    parts = _buildContentObjects(_msg("m1"), maxBodyChars=8000)
    ids = [p["contentObjectId"] for p in parts]
    assert ids == ["header", "snippet", "body"]
    header = parts[0]["data"]
    assert "Subject: Hi" in header
    assert "From: Alice <a@x.com>" in header
    assert "To: Bob <b@x.com>" in header


def test_bootstrap_outlook_indexes_messages_from_inbox_and_sent():
    adapter = _FakeOutlookAdapter({
        "INBOX-ID": [_msg("m1"), _msg("m2")],
        "SENT-ID": [_msg("m3")],
    })
    knowledge = _FakeKnowledgeService()
    connection = SimpleNamespace(mandateId="m1", userId="u1")

    async def _run():
        return await bootstrapOutlook(
            connectionId="c1",
            adapter=adapter,
            connection=connection,
            knowledgeService=knowledge,
            limits=OutlookBootstrapLimits(maxAgeDays=None),
        )

    result = asyncio.run(_run())
    assert result["indexed"] == 3
    sourceIds = {c.sourceId for c in knowledge.calls}
    assert sourceIds == {
        _syntheticMessageId("c1", "m1"),
        _syntheticMessageId("c1", "m2"),
        _syntheticMessageId("c1", "m3"),
    }
    for job in knowledge.calls:
        assert job.sourceKind == "outlook_message"
        assert job.mimeType == "message/rfc822"
        assert job.provenance["connectionId"] == "c1"
        assert job.provenance["service"] == "outlook"
        assert job.contentVersion == "ck1"
        assert any(co["contentObjectId"] == "header" for co in job.contentObjects)


def test_bootstrap_outlook_follows_pagination():
    adapter = _FakeOutlookAdapter(
        messages_by_folder={"INBOX-ID": [_msg("m1")], "SENT-ID": []},
        paginated_folder="INBOX-ID",
        page2=[_msg("m2"), _msg("m3")],
    )
    knowledge = _FakeKnowledgeService()
    connection = SimpleNamespace(mandateId="m1", userId="u1")

    async def _run():
        return await bootstrapOutlook(
            connectionId="c1",
            adapter=adapter,
            connection=connection,
            knowledgeService=knowledge,
            limits=OutlookBootstrapLimits(maxAgeDays=None),
        )

    result = asyncio.run(_run())
    assert result["indexed"] == 3


def test_bootstrap_outlook_reports_duplicates():
    adapter = _FakeOutlookAdapter({
        "INBOX-ID": [_msg("m1"), _msg("m2")],
        "SENT-ID": [],
    })
    duplicates = {
        _syntheticMessageId("c1", "m1"),
        _syntheticMessageId("c1", "m2"),
    }
    knowledge = _FakeKnowledgeService(duplicateIds=duplicates)
    connection = SimpleNamespace(mandateId="m1", userId="u1")

    async def _run():
        return await bootstrapOutlook(
            connectionId="c1",
            adapter=adapter,
            connection=connection,
            knowledgeService=knowledge,
            limits=OutlookBootstrapLimits(maxAgeDays=None),
        )

    result = asyncio.run(_run())
    assert result["indexed"] == 0
    assert result["skippedDuplicate"] == 2


if __name__ == "__main__":
    test_buildContentObjects_emits_header_snippet_body()
    test_bootstrap_outlook_indexes_messages_from_inbox_and_sent()
    test_bootstrap_outlook_follows_pagination()
    test_bootstrap_outlook_reports_duplicates()
    print("OK — bootstrapOutlook tests passed")