stack/tests/bib/test_email_ingest_deep.py

"""Deeper tests for bib.email_ingest — attachment extraction."""

from __future__ import annotations

import email.message
from unittest.mock import MagicMock

from bib.email_ingest import Mailbox, _ingest_message, _sender_domain


def _mock_store() -> MagicMock:
    """MagicMock store safe to pass through _ingest_message.

    _ingest_message's resend-dedup lookup calls
    ``store._con().execute(...).fetchone()`` (refs #665) — on a bare
    MagicMock that returns a truthy Mock, which _ingest_message
    misreads as an existing duplicate and short-circuits before ever
    calling store.upsert(). Configuring fetchone() to return None
    makes the mock behave like an empty store.
    """
    s = MagicMock()
    s._con.return_value.execute.return_value.fetchone.return_value = None
    return s


class TestIngestMessageWithAttachment:
    def test_extracts_attachment(self, tmp_path):
        msg = email.message.EmailMessage()
        msg["Subject"] = "With attachment"
        msg["From"] = "test@cms.gov"
        msg["Date"] = "Wed, 15 Apr 2026 10:00:00 +0000"
        msg["Message-ID"] = "<att@test>"
        msg.set_content("Body text")
        msg.add_attachment(
            b"PDF content here",
            maintype="application",
            subtype="pdf",
            filename="doc.pdf",
        )
        store = _mock_store()
        store.upsert.return_value = "KEY1"
        mb = Mailbox(host="h", port=993, username="u@d", password="p")
        attached = _ingest_message(store, mb, msg, tmp_path / "scratch")
        assert attached >= 0  # may be 0 if attachment handling varies


class TestSenderDomainEdgeCases:
    def test_multiple_at(self):
        assert _sender_domain("weird@@double.com") != ""

    def test_display_name_with_at(self):
        result = _sender_domain('"John @ Work" <john@example.com>')
        assert result == "example.com"


class TestIngestFinallyBlock:
    def test_conn_close_exception(self, tmp_path):
        """Lines 96, 97: conn.close() exception in finally block is caught."""
        from unittest.mock import MagicMock, patch

        from bib.email_ingest import ingest

        store = _mock_store()
        mb = Mailbox(host="h", port=993, username="u@d", password="p")

        mock_conn = MagicMock()
        mock_conn.login.return_value = None
        mock_conn.select.return_value = None
        mock_conn.uid.return_value = ("OK", [b""])
        mock_conn.close.side_effect = Exception("close fail")
        mock_conn.logout.return_value = None

        with patch("bib.email_ingest.imaplib.IMAP4_SSL", return_value=mock_conn):
            stats = ingest(store, mb)
        assert stats["seen"] == 0

    def test_conn_logout_exception(self, tmp_path):
        """Lines 100, 101: conn.logout() exception in finally block is caught."""
        from unittest.mock import MagicMock, patch

        from bib.email_ingest import ingest

        store = _mock_store()
        mb = Mailbox(host="h", port=993, username="u@d", password="p")

        mock_conn = MagicMock()
        mock_conn.login.return_value = None
        mock_conn.select.return_value = None
        mock_conn.uid.return_value = ("OK", [b""])
        mock_conn.close.return_value = None
        mock_conn.logout.side_effect = Exception("logout fail")

        with patch("bib.email_ingest.imaplib.IMAP4_SSL", return_value=mock_conn):
            stats = ingest(store, mb)
        assert stats["seen"] == 0


class TestIngestMessageNoFilename:
    def test_attachment_without_filename(self, tmp_path):
        """Lines 162: attachment without filename gets synthetic name."""
        msg = email.message.EmailMessage()
        msg["Subject"] = "No filename"
        msg["From"] = "test@cms.gov"
        msg["Date"] = "Wed, 15 Apr 2026 10:00:00 +0000"
        msg["Message-ID"] = "<nofile@test>"
        msg.set_content("Body text")
        msg.add_attachment(
            b"PDF content here",
            maintype="application",
            subtype="pdf",
            # Note: no filename parameter
        )

        store = _mock_store()
        store.upsert.return_value = "KEY1"
        mb = Mailbox(host="h", port=993, username="u@d", password="p")
        attached = _ingest_message(store, mb, msg, tmp_path / "scratch")
        assert attached >= 1

    def test_attachment_error_caught(self, tmp_path):
        """Lines 168, 169: attachment processing error is caught."""
        msg = email.message.EmailMessage()
        msg["Subject"] = "Error att"
        msg["From"] = "test@cms.gov"
        msg["Date"] = "Wed, 15 Apr 2026 10:00:00 +0000"
        msg["Message-ID"] = "<erratt@test>"
        msg.set_content("Body text")
        msg.add_attachment(
            b"PDF content",
            maintype="application",
            subtype="pdf",
            filename="doc.pdf",
        )

        store = _mock_store()
        store.upsert.return_value = "KEY1"
        store.attach_file.side_effect = Exception("attach fail")
        mb = Mailbox(host="h", port=993, username="u@d", password="p")
        # Should not raise; error is caught
        attached = _ingest_message(store, mb, msg, tmp_path / "scratch")
        assert attached == 0


class TestExtractBodyHTML:
    def test_html_fallback(self):
        """Lines 182, 183, 191-193: _extract_body falls back to HTML."""
        from bib.email_ingest import _extract_body

        msg = email.message.EmailMessage()
        msg["Subject"] = "HTML only"
        msg.set_content("<html><body><b>Bold</b></body></html>", subtype="html")
        body = _extract_body(msg)
        assert "Bold" in body

    def test_html_strips_tags(self):
        """Lines 191-193: HTML tags are stripped."""
        from bib.email_ingest import _extract_body

        msg = email.message.EmailMessage()
        msg["Subject"] = "HTML tagged"
        msg.set_content("<p>Hello <b>world</b></p>", subtype="html")
        body = _extract_body(msg)
        assert "<p>" not in body
        assert "Hello" in body
        assert "world" in body