homelab-codex-ws/jobs/gmail-bulk-import/tests/test_importer.py

"""Unit tests for Gmail bulk importer — no DB, no external services required."""
from __future__ import annotations

import mailbox
from datetime import datetime, timezone
from pathlib import Path

import pytest

from gmail_bulk_import.importer import _message_id, _parse_date, run_import


def _make_msg(headers: dict) -> mailbox.mboxMessage:
    msg = mailbox.mboxMessage()
    for k, v in headers.items():
        msg[k] = v
    return msg


def _make_mbox(path: Path, msgs: list[mailbox.mboxMessage]) -> None:
    mbox = mailbox.mbox(str(path), create=True)
    for msg in msgs:
        mbox.add(msg)
    mbox.flush()
    mbox.close()


def _msg_with_id(mid: str = "<test-unique@example.com>") -> mailbox.mboxMessage:
    return _make_msg({
        "From": "alice@example.com",
        "Message-ID": mid,
        "Date": "Tue, 10 Jun 2025 12:00:00 +0000",
        "Subject": "Hello",
    })


class TestMessageId:
    def test_uses_message_id_header(self):
        msg = _make_msg({"Message-ID": "<foo-123@example.com>"})
        assert _message_id(msg) == "foo-123@example.com"

    def test_strips_angle_brackets(self):
        msg = _make_msg({"Message-ID": "  <bar@baz.com>  "})
        assert _message_id(msg) == "bar@baz.com"

    def test_fallback_to_sha256_when_missing(self):
        msg = _make_msg({"From": "a@b.com", "Subject": "no id"})
        result = _message_id(msg)
        assert result.startswith("sha256-")
        assert len(result) == len("sha256-") + 32

    def test_sha256_is_stable(self):
        msg = _make_msg({"From": "a@b.com", "Subject": "stable"})
        assert _message_id(msg) == _message_id(msg)


class TestParseDate:
    def test_parses_rfc2822(self):
        msg = _make_msg({"Date": "Tue, 10 Jun 2025 12:00:00 +0000"})
        ts = _parse_date(msg)
        assert ts == datetime(2025, 6, 10, 12, 0, 0, tzinfo=timezone.utc)

    def test_result_is_utc_aware(self):
        msg = _make_msg({"Date": "Tue, 10 Jun 2025 12:00:00 +0000"})
        assert _parse_date(msg).tzinfo is not None

    def test_falls_back_to_epoch_on_bad_date(self):
        msg = _make_msg({"Date": "not-a-date"})
        assert _parse_date(msg) == datetime(1970, 1, 1, tzinfo=timezone.utc)

    def test_falls_back_to_epoch_when_missing(self):
        msg = _make_msg({"From": "a@b.com"})
        assert _parse_date(msg) == datetime(1970, 1, 1, tzinfo=timezone.utc)

    def test_converts_offset_to_utc(self):
        msg = _make_msg({"Date": "Tue, 10 Jun 2025 14:00:00 +0200"})
        ts = _parse_date(msg)
        assert ts == datetime(2025, 6, 10, 12, 0, 0, tzinfo=timezone.utc)


class TestRunImport:
    async def test_dry_run_counts_without_writing(self, tmp_path):
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id("<a@e.com>"), _msg_with_id("<b@e.com>")])

        stats = await run_import(mbox_path, archive, dry_run=True)

        assert stats["processed"] == 2
        assert stats["imported"] == 2
        assert not archive.exists()

    async def test_imports_eml_to_archive(self, tmp_path):
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id()])

        stats = await run_import(mbox_path, archive)

        assert stats["imported"] == 1
        assert stats["skipped"] == 0
        assert stats["errors"] == 0
        eml_files = list(archive.rglob("*.eml"))
        assert len(eml_files) == 1

    async def test_eml_stored_under_gmail_source(self, tmp_path):
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id()])

        await run_import(mbox_path, archive)

        eml_files = list(archive.rglob("*.eml"))
        assert eml_files[0].parts[len(archive.parts)] == "gmail"

    async def test_idempotent_second_run_skips(self, tmp_path):
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id()])

        await run_import(mbox_path, archive)
        stats2 = await run_import(mbox_path, archive)

        assert stats2["imported"] == 0
        assert stats2["skipped"] == 1
        assert stats2["errors"] == 0

    async def test_two_distinct_messages_both_imported(self, tmp_path):
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id("<msg1@e.com>"), _msg_with_id("<msg2@e.com>")])

        stats = await run_import(mbox_path, archive)

        assert stats["imported"] == 2
        assert stats["errors"] == 0
        assert len(list(archive.rglob("*.eml"))) == 2

    async def test_archive_only_mode_no_dsn(self, tmp_path):
        """Without a DSN, job runs archive-only (no DB connection attempted)."""
        mbox_path = tmp_path / "mail.mbox"
        archive = tmp_path / "archive"
        _make_mbox(mbox_path, [_msg_with_id()])

        stats = await run_import(mbox_path, archive, dsn=None)

        assert stats["imported"] == 1
        assert list(archive.rglob("*.eml"))