PecHub/worker/tests/unit/test_sync_parsing.py

"""
Unit test: parsing EML e classificazione tipo PEC.
"""

import pytest

from app.imap.sync import _decode_header, _extract_addresses, _parse_date, _parse_eml, _classify_pec_type
import email


# ─── Test _decode_header ─────────────────────────────────────────────────────

def test_decode_header_plain():
    assert _decode_header("Hello World") == "Hello World"


def test_decode_header_none():
    assert _decode_header(None) is None


def test_decode_header_encoded():
    # Header UTF-8 base64 encoded
    encoded = "=?utf-8?b?UEVDIHRlc3Q=?="  # "PEC test"
    assert _decode_header(encoded) == "PEC test"


def test_decode_header_encoded_iso():
    # Header ISO-8859-1 quoted printable
    encoded = "=?iso-8859-1?q?Multa_n=2E_123?="  # "Multa n. 123"
    result = _decode_header(encoded)
    assert result is not None
    assert "Multa" in result


# ─── Test _extract_addresses ──────────────────────────────────────────────────

def test_extract_addresses_single():
    addrs = _extract_addresses("test@example.com")
    assert "test@example.com" in addrs


def test_extract_addresses_multiple():
    addrs = _extract_addresses("a@x.com, b@y.com, c@z.com")
    assert len(addrs) == 3
    assert "a@x.com" in addrs
    assert "b@y.com" in addrs


def test_extract_addresses_with_display_name():
    addrs = _extract_addresses('"Mario Rossi" <mario@comune.it>')
    assert "mario@comune.it" in addrs


def test_extract_addresses_empty():
    assert _extract_addresses(None) == []
    assert _extract_addresses("") == []


# ─── Test _parse_date ─────────────────────────────────────────────────────────

def test_parse_date_valid():
    date = _parse_date("Wed, 18 Mar 2026 14:00:00 +0100")
    assert date is not None
    assert date.year == 2026
    assert date.month == 3
    assert date.day == 18


def test_parse_date_none():
    assert _parse_date(None) is None


def test_parse_date_invalid():
    assert _parse_date("not-a-date") is None


# ─── Test _classify_pec_type ──────────────────────────────────────────────────

def test_classify_pec_type_avvenuta_consegna():
    msg = email.message_from_string(
        "From: server@pec.it\r\n"
        "X-TipoRicevuta: avvenuta-consegna\r\n"
        "\r\n"
    )
    assert _classify_pec_type(msg) == "avvenuta_consegna"


def test_classify_pec_type_accettazione():
    msg = email.message_from_string(
        "From: server@pec.it\r\n"
        "X-Ricevuta: accettazione\r\n"
        "\r\n"
    )
    assert _classify_pec_type(msg) == "accettazione"


def test_classify_pec_type_posta_certificata():
    msg = email.message_from_string(
        "From: mittente@pec.it\r\n"
        "Subject: Messaggio PEC\r\n"
        "\r\n"
    )
    assert _classify_pec_type(msg) == "posta_certificata"


def test_classify_pec_type_x_tipo_prevalente():
    """X-TipoRicevuta ha precedenza su X-Ricevuta."""
    msg = email.message_from_string(
        "From: server@pec.it\r\n"
        "X-Ricevuta: accettazione\r\n"
        "X-TipoRicevuta: avvenuta-consegna\r\n"
        "\r\n"
    )
    assert _classify_pec_type(msg) == "avvenuta_consegna"


# ─── Test _parse_eml completo ─────────────────────────────────────────────────

RAW_EML = b"""From: mittente@pec.it
To: destinatario@pec.it
Cc: copia@pec.it
Subject: Test PEC Fase 2
Message-ID: <test123@pec.it>
Date: Wed, 18 Mar 2026 10:00:00 +0100
Content-Type: text/plain; charset=utf-8

Corpo del messaggio di test.
"""


def test_parse_eml_subject():
    parsed = _parse_eml(RAW_EML)
    assert parsed["subject"] == "Test PEC Fase 2"


def test_parse_eml_from():
    parsed = _parse_eml(RAW_EML)
    assert parsed["from_address"] == "mittente@pec.it"


def test_parse_eml_to():
    parsed = _parse_eml(RAW_EML)
    assert "destinatario@pec.it" in parsed["to_addresses"]


def test_parse_eml_cc():
    parsed = _parse_eml(RAW_EML)
    assert "copia@pec.it" in parsed["cc_addresses"]


def test_parse_eml_message_id():
    parsed = _parse_eml(RAW_EML)
    assert parsed["message_id_header"] == "<test123@pec.it>"


def test_parse_eml_body_text():
    parsed = _parse_eml(RAW_EML)
    assert "Corpo del messaggio" in (parsed.get("body_text") or "")


def test_parse_eml_no_attachments():
    parsed = _parse_eml(RAW_EML)
    assert parsed["has_attachments"] is False


RAW_EML_WITH_ATTACHMENT = b"""From: mittente@pec.it
To: destinatario@pec.it
Subject: PEC con allegato
Date: Wed, 18 Mar 2026 10:00:00 +0100
Content-Type: multipart/mixed; boundary="----boundary123"

------boundary123
Content-Type: text/plain; charset=utf-8

Testo del messaggio.

------boundary123
Content-Type: application/pdf; name="documento.pdf"
Content-Disposition: attachment; filename="documento.pdf"
Content-Transfer-Encoding: base64

JVBERi0xLjQ=

------boundary123--
"""


def test_parse_eml_with_attachment():
    parsed = _parse_eml(RAW_EML_WITH_ATTACHMENT)
    assert parsed["has_attachments"] is True
    assert "Testo del messaggio" in (parsed.get("body_text") or "")


def test_parse_eml_empty():
    parsed = _parse_eml(b"")
    # Non deve sollevare eccezione
    assert isinstance(parsed, dict)