fase 3

2026-06-16 12:45:42 +02:00 · 2026-03-18 17:43:03 +01:00
parent d80d912fb3
commit c89c08c397
10 changed files with 2352 additions and 123 deletions
@@ -0,0 +1,509 @@
+"""
+Test unitari per app.parsers.eml_parser.
+
+Copertura:
+  - Parsing header (Subject, From, To, Cc, Message-ID, Date)
+  - Decodifica RFC 2047 (UTF-8, ISO-8859-1, base64, quoted-printable)
+  - Estrazione body text/plain e text/html
+  - Estrazione allegati (singoli e multipli)
+  - Gestione EML-in-EML (message/rfc822)
+  - Flag has_attachments (solo allegati non-PEC-system)
+  - Allegati PEC di sistema (daticert.xml, postacert.eml)
+  - EML vuoto / malformato (no crash)
+  - Messaggio non-multipart
+"""
+
+import email
+import textwrap
+
+import pytest
+
+from app.parsers.eml_parser import (
+    AttachmentInfo,
+    ParsedEmail,
+    decode_header,
+    extract_addresses,
+    parse_date,
+    parse_eml,
+)
+
+
+# ─── Fixture EML ──────────────────────────────────────────────────────────────
+
+
+SIMPLE_EML = b"""\
+From: mittente@pec.it
+To: destinatario@pec.it
+Cc: copia@pec.it
+Subject: Test PEC Fase 3
+Message-ID: <test123@pec.it>
+Date: Wed, 18 Mar 2026 14:00:00 +0100
+Content-Type: text/plain; charset=utf-8
+
+Corpo del messaggio di test.
+Seconda riga.
+"""
+
+MULTIPART_EML = b"""\
+From: mittente@pec.it
+To: dest@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====--
+"""
+
+MULTIPART_HTML_EML = b"""\
+From: mittente@pec.it
+To: dest@pec.it
+Subject: PEC multipart/alternative
+Date: Wed, 18 Mar 2026 10:00:00 +0100
+Content-Type: multipart/alternative; boundary="====alt===="
+
+--====alt====
+Content-Type: text/plain; charset=utf-8
+
+Testo piano.
+
+--====alt====
+Content-Type: text/html; charset=utf-8
+
+<html><body><p>Testo HTML.</p></body></html>
+
+--====alt====--
+"""
+
+RECEIPT_EML_WITH_NESTED = b"""\
+From: posta-certificata@pec.aruba.it
+To: mittente@pec.it
+Subject: CONSEGNA: Test PEC Fase 3
+X-Ricevuta: avvenuta-consegna
+X-Riferimento-Message-ID: <orig001@pec.it>
+Date: Wed, 18 Mar 2026 14:05:00 +0100
+Content-Type: multipart/mixed; boundary="====receipt===="
+
+--====receipt====
+Content-Type: text/plain; charset=utf-8
+
+Il messaggio e' stato consegnato al destinatario.
+
+--====receipt====
+Content-Type: application/xml; name="daticert.xml"
+Content-Disposition: attachment; filename="daticert.xml"
+
+<?xml version="1.0"?><PostaCertificata versione="2.3"></PostaCertificata>
+
+--====receipt====
+Content-Type: message/rfc822
+Content-Disposition: inline
+
+From: mittente@pec.it
+To: destinatario@pec.it
+Subject: Test PEC Fase 3
+Message-ID: <orig001@pec.it>
+Date: Wed, 18 Mar 2026 14:00:00 +0100
+Content-Type: text/plain; charset=utf-8
+
+Corpo del messaggio originale.
+
+--====receipt====--
+"""
+
+MULTIPLE_ATTACHMENTS_EML = b"""\
+From: a@pec.it
+To: b@pec.it
+Subject: PEC con allegati multipli
+Date: Wed, 18 Mar 2026 10:00:00 +0100
+Content-Type: multipart/mixed; boundary="====multi===="
+
+--====multi====
+Content-Type: text/plain; charset=utf-8
+
+Corpo.
+
+--====multi====
+Content-Type: application/pdf; name="doc1.pdf"
+Content-Disposition: attachment; filename="doc1.pdf"
+Content-Transfer-Encoding: base64
+
+AAEC
+
+--====multi====
+Content-Type: application/pdf; name="doc2.pdf"
+Content-Disposition: attachment; filename="doc2.pdf"
+Content-Transfer-Encoding: base64
+
+BAEC
+
+--====multi====--
+"""
+
+PEC_SYSTEM_EML = b"""\
+From: posta-certificata@pec.aruba.it
+To: mittente@pec.it
+Subject: ACCETTAZIONE: Test
+X-Ricevuta: accettazione
+Date: Wed, 18 Mar 2026 14:01:00 +0100
+Content-Type: multipart/mixed; boundary="====sys===="
+
+--====sys====
+Content-Type: text/plain; charset=utf-8
+
+Ricevuta di accettazione.
+
+--====sys====
+Content-Type: application/xml; name="daticert.xml"
+Content-Disposition: attachment; filename="daticert.xml"
+
+<?xml version="1.0"?><PostaCertificata versione="2.3"></PostaCertificata>
+
+--====sys====
+Content-Type: message/rfc822
+Content-Disposition: inline; filename="postacert.eml"
+
+From: mittente@pec.it
+To: dest@pec.it
+Subject: Test originale
+
+Corpo.
+
+--====sys====--
+"""
+
+
+# ─── Test decode_header ───────────────────────────────────────────────────────
+
+
+class TestDecodeHeader:
+
+    def test_stringa_semplice(self):
+        assert decode_header("Hello World") == "Hello World"
+
+    def test_none_ritorna_none(self):
+        assert decode_header(None) is None
+
+    def test_stringa_vuota_ritorna_none(self):
+        assert decode_header("") is None
+
+    def test_utf8_base64(self):
+        # "PEC test" in base64 UTF-8
+        encoded = "=?utf-8?b?UEVDIHRlc3Q=?="
+        assert decode_header(encoded) == "PEC test"
+
+    def test_iso8859_quoted_printable(self):
+        # "Multa n. 123" in QP ISO-8859-1
+        encoded = "=?iso-8859-1?q?Multa_n=2E_123?="
+        result = decode_header(encoded)
+        assert result is not None
+        assert "Multa" in result
+        assert "123" in result
+
+    def test_multipart_header(self):
+        # Header con più parti encodate
+        encoded = "=?utf-8?b?UEVD?= =?utf-8?b?IHRlc3Q=?="
+        result = decode_header(encoded)
+        assert result is not None
+        assert "PEC" in result
+
+    def test_stringa_gia_decodificata(self):
+        assert decode_header("Oggetto normale") == "Oggetto normale"
+
+
+# ─── Test extract_addresses ───────────────────────────────────────────────────
+
+
+class TestExtractAddresses:
+
+    def test_singolo_indirizzo(self):
+        addrs = extract_addresses("test@example.com")
+        assert "test@example.com" in addrs
+
+    def test_multipli_indirizzi(self):
+        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
+        assert "c@z.com" in addrs
+
+    def test_display_name(self):
+        addrs = extract_addresses('"Mario Rossi" <mario@comune.it>')
+        assert "mario@comune.it" in addrs
+
+    def test_none_ritorna_lista_vuota(self):
+        assert extract_addresses(None) == []
+
+    def test_stringa_vuota_ritorna_lista_vuota(self):
+        assert extract_addresses("") == []
+
+
+# ─── Test parse_date ──────────────────────────────────────────────────────────
+
+
+class TestParseDate:
+
+    def test_data_valida(self):
+        d = parse_date("Wed, 18 Mar 2026 14:00:00 +0100")
+        assert d is not None
+        assert d.year == 2026
+        assert d.month == 3
+        assert d.day == 18
+
+    def test_none_ritorna_none(self):
+        assert parse_date(None) is None
+
+    def test_stringa_invalida_ritorna_none(self):
+        assert parse_date("non-una-data") is None
+
+    def test_data_senza_timezone_aggiunge_utc(self):
+        d = parse_date("18 Mar 2026 14:00:00 +0000")
+        assert d is not None
+        assert d.tzinfo is not None
+
+
+# ─── Test parse_eml – messaggio semplice ──────────────────────────────────────
+
+
+class TestParseEmlSimple:
+
+    def test_subject(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.subject == "Test PEC Fase 3"
+
+    def test_from_address(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.from_address == "mittente@pec.it"
+
+    def test_to_addresses(self):
+        p = parse_eml(SIMPLE_EML)
+        assert "destinatario@pec.it" in p.to_addresses
+
+    def test_cc_addresses(self):
+        p = parse_eml(SIMPLE_EML)
+        assert "copia@pec.it" in p.cc_addresses
+
+    def test_message_id(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.message_id == "<test123@pec.it>"
+
+    def test_date(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.date is not None
+        assert p.date.year == 2026
+
+    def test_body_text(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.body_text is not None
+        assert "Corpo del messaggio" in p.body_text
+        assert "Seconda riga" in p.body_text
+
+    def test_no_html(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.body_html is None
+
+    def test_no_attachments(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.attachments == []
+        assert p.has_attachments is False
+
+    def test_raw_message_presente(self):
+        p = parse_eml(SIMPLE_EML)
+        assert p.raw_message is not None
+        assert isinstance(p.raw_message, email.message.Message)
+
+
+# ─── Test parse_eml – multipart con allegato ──────────────────────────────────
+
+
+class TestParseEmlMultipart:
+
+    def test_body_text_estratto(self):
+        p = parse_eml(MULTIPART_EML)
+        assert p.body_text is not None
+        assert "Testo del messaggio" in p.body_text
+
+    def test_allegato_trovato(self):
+        p = parse_eml(MULTIPART_EML)
+        assert len(p.attachments) == 1
+        att = p.attachments[0]
+        assert att.filename == "documento.pdf"
+        assert att.content_type == "application/pdf"
+        assert att.size_bytes > 0
+        assert att.checksum_sha256 is not None
+        assert len(att.checksum_sha256) == 64
+
+    def test_has_attachments_true(self):
+        p = parse_eml(MULTIPART_EML)
+        assert p.has_attachments is True
+
+    def test_allegati_multipli(self):
+        p = parse_eml(MULTIPLE_ATTACHMENTS_EML)
+        filenames = [a.filename for a in p.attachments]
+        assert "doc1.pdf" in filenames
+        assert "doc2.pdf" in filenames
+        assert len(p.attachments) == 2
+
+
+# ─── Test parse_eml – multipart/alternative ───────────────────────────────────
+
+
+class TestParseEmlAlternative:
+
+    def test_body_text_e_html(self):
+        p = parse_eml(MULTIPART_HTML_EML)
+        assert p.body_text is not None
+        assert "Testo piano" in p.body_text
+        assert p.body_html is not None
+        assert "<html>" in p.body_html
+        assert "Testo HTML" in p.body_html
+
+    def test_no_attachments_in_alternative(self):
+        p = parse_eml(MULTIPART_HTML_EML)
+        assert p.has_attachments is False
+
+
+# ─── Test parse_eml – ricevuta con EML-in-EML ────────────────────────────────
+
+
+class TestParseEmlReceiptWithNested:
+
+    def test_body_text_ricevuta(self):
+        p = parse_eml(RECEIPT_EML_WITH_NESTED)
+        assert p.body_text is not None
+        assert "consegnato" in p.body_text.lower()
+
+    def test_allegato_xml_daticert(self):
+        p = parse_eml(RECEIPT_EML_WITH_NESTED)
+        filenames = [a.filename for a in p.attachments]
+        assert "daticert.xml" in filenames
+
+    def test_allegato_xml_e_pec_system(self):
+        p = parse_eml(RECEIPT_EML_WITH_NESTED)
+        xml_att = next(a for a in p.attachments if a.filename == "daticert.xml")
+        assert xml_att.is_pec_system is True
+
+    def test_eml_annidato_trovato(self):
+        """Il messaggio originale annidato deve essere presente come allegato."""
+        p = parse_eml(RECEIPT_EML_WITH_NESTED)
+        eml_atts = [a for a in p.attachments if a.content_type == "message/rfc822"]
+        assert len(eml_atts) >= 1
+
+    def test_has_attachments_false_quando_solo_system(self):
+        """has_attachments deve essere False se ci sono solo allegati PEC di sistema."""
+        p = parse_eml(PEC_SYSTEM_EML)
+        # daticert.xml e postacert.eml sono entrambi system → has_attachments = False
+        assert p.has_attachments is False
+
+    def test_allegati_sistema_marcati_correttamente(self):
+        p = parse_eml(PEC_SYSTEM_EML)
+        for att in p.attachments:
+            if att.filename in ("daticert.xml", "postacert.eml"):
+                assert att.is_pec_system is True, f"{att.filename} dovrebbe essere is_pec_system=True"
+
+
+# ─── Test parse_eml – edge cases ─────────────────────────────────────────────
+
+
+class TestParseEmlEdgeCases:
+
+    def test_eml_vuoto_no_eccezione(self):
+        p = parse_eml(b"")
+        assert isinstance(p, ParsedEmail)
+        assert p.subject is None
+        assert p.body_text is None
+        assert p.attachments == []
+
+    def test_eml_malformato_no_eccezione(self):
+        p = parse_eml(b"questo non e' un EML valido\x00\xFF")
+        assert isinstance(p, ParsedEmail)
+
+    def test_headers_mancanti(self):
+        raw = b"Content-Type: text/plain\r\n\r\nSolo corpo."
+        p = parse_eml(raw)
+        assert p.subject is None
+        assert p.from_address is None
+        assert p.to_addresses == []
+
+    def test_body_con_encoding_windows1252(self):
+        raw = (
+            b"From: a@pec.it\r\nTo: b@pec.it\r\n"
+            b"Content-Type: text/plain; charset=windows-1252\r\n\r\n"
+            b"Buonagiornata\xe0 tutti"
+        )
+        p = parse_eml(raw)
+        assert p.body_text is not None
+        assert "Buonagiornata" in p.body_text
+
+    def test_attachments_senza_filename_ignorati(self):
+        """
+        Un part senza filename non deve essere aggiunto come allegato
+        se non è text/plain o text/html.
+        """
+        raw = (
+            b"From: a@pec.it\r\nTo: b@pec.it\r\n"
+            b'Content-Type: multipart/mixed; boundary="B"\r\n\r\n'
+            b"--B\r\nContent-Type: text/plain\r\n\r\nBody\r\n"
+            b"--B\r\nContent-Type: application/octet-stream\r\n"
+            b"Content-Disposition: attachment\r\n\r\nDATA\r\n"
+            b"--B--\r\n"
+        )
+        p = parse_eml(raw)
+        # L'allegato senza filename non deve comparire
+        for att in p.attachments:
+            assert att.filename is not None and att.filename != ""
+
+    def test_checksum_sha256_corretto(self):
+        """Il checksum SHA-256 dell'allegato deve essere valido."""
+        import hashlib
+        p = parse_eml(MULTIPART_EML)
+        assert len(p.attachments) == 1
+        att = p.attachments[0]
+        expected = hashlib.sha256(att.content).hexdigest()
+        assert att.checksum_sha256 == expected
+
+
+# ─── Test AttachmentInfo dataclass ────────────────────────────────────────────
+
+
+class TestAttachmentInfoDataclass:
+
+    def test_campi_base(self):
+        import hashlib
+        content = b"test content"
+        att = AttachmentInfo(
+            filename="test.pdf",
+            content_type="application/pdf",
+            content=content,
+            size_bytes=len(content),
+            checksum_sha256=hashlib.sha256(content).hexdigest(),
+        )
+        assert att.filename == "test.pdf"
+        assert att.content_type == "application/pdf"
+        assert att.size_bytes == 12
+        assert att.is_inline is False
+        assert att.is_pec_system is False
+
+    def test_inline_flag(self):
+        import hashlib
+        content = b"img"
+        att = AttachmentInfo(
+            filename="img.png",
+            content_type="image/png",
+            content=content,
+            size_bytes=len(content),
+            checksum_sha256=hashlib.sha256(content).hexdigest(),
+            is_inline=True,
+        )
+        assert att.is_inline is True