| 1 | """Email channel implementation using IMAP polling + SMTP replies.""" |
| 2 | |
| 3 | import asyncio |
| 4 | import html |
| 5 | import imaplib |
| 6 | import re |
| 7 | import smtplib |
| 8 | import ssl |
| 9 | from datetime import date |
| 10 | from email import policy |
| 11 | from email.header import decode_header, make_header |
| 12 | from email.message import EmailMessage |
| 13 | from email.parser import BytesParser |
| 14 | from email.utils import parseaddr |
| 15 | from fnmatch import fnmatch |
| 16 | from pathlib import Path |
| 17 | from typing import Any |
| 18 | |
| 19 | from loguru import logger |
| 20 | from pydantic import Field |
| 21 | |
| 22 | from nanobot.bus.events import OutboundMessage |
| 23 | from nanobot.bus.queue import MessageBus |
| 24 | from nanobot.channels.base import BaseChannel |
| 25 | from nanobot.config.paths import get_media_dir |
| 26 | from nanobot.config.schema import Base |
| 27 | from nanobot.utils.helpers import safe_filename |
| 28 | |
| 29 | |
| 30 | class EmailConfig(Base): |
| 31 | """Email channel configuration (IMAP inbound + SMTP outbound).""" |
| 32 | |
| 33 | enabled: bool = False |
| 34 | consent_granted: bool = False |
| 35 | |
| 36 | imap_host: str = "" |
| 37 | imap_port: int = 993 |
| 38 | imap_username: str = "" |
| 39 | imap_password: str = "" |
| 40 | imap_mailbox: str = "INBOX" |
| 41 | imap_use_ssl: bool = True |
| 42 | |
| 43 | smtp_host: str = "" |
| 44 | smtp_port: int = 587 |
| 45 | smtp_username: str = "" |
| 46 | smtp_password: str = "" |
| 47 | smtp_use_tls: bool = True |
| 48 | smtp_use_ssl: bool = False |
| 49 | from_address: str = "" |
| 50 | |
| 51 | auto_reply_enabled: bool = True |
| 52 | poll_interval_seconds: int = 30 |
| 53 | mark_seen: bool = True |
| 54 | max_body_chars: int = 12000 |
| 55 | subject_prefix: str = "Re: " |
| 56 | allow_from: list[str] = Field(default_factory=list) |
| 57 | |
| 58 | # Email authentication verification (anti-spoofing) |
| 59 | verify_dkim: bool = True # Require Authentication-Results with dkim=pass |
| 60 | verify_spf: bool = True # Require Authentication-Results with spf=pass |
| 61 | |
| 62 | # Attachment handling — set allowed types to enable (e.g. ["application/pdf", "image/*"], or ["*"] for all) |
| 63 | allowed_attachment_types: list[str] = Field(default_factory=list) |
| 64 | max_attachment_size: int = 2_000_000 # 2MB per attachment |
| 65 | max_attachments_per_email: int = 5 |
| 66 | |
| 67 | |
| 68 | class EmailChannel(BaseChannel): |
| 69 | """ |
| 70 | Email channel. |
| 71 | |
| 72 | Inbound: |
| 73 | - Poll IMAP mailbox for unread messages. |
| 74 | - Convert each message into an inbound event. |
| 75 | |
| 76 | Outbound: |
| 77 | - Send responses via SMTP back to the sender address. |
| 78 | """ |
| 79 | |
| 80 | name = "email" |
| 81 | display_name = "Email" |
| 82 | _IMAP_MONTHS = ( |
| 83 | "Jan", |
| 84 | "Feb", |
| 85 | "Mar", |
| 86 | "Apr", |
| 87 | "May", |
| 88 | "Jun", |
| 89 | "Jul", |
| 90 | "Aug", |
| 91 | "Sep", |
| 92 | "Oct", |
| 93 | "Nov", |
| 94 | "Dec", |
| 95 | ) |
| 96 | _IMAP_RECONNECT_MARKERS = ( |
| 97 | "disconnected for inactivity", |
| 98 | "eof occurred in violation of protocol", |
| 99 | "socket error", |
| 100 | "connection reset", |
| 101 | "broken pipe", |
| 102 | "bye", |
| 103 | ) |
| 104 | _IMAP_MISSING_MAILBOX_MARKERS = ( |
| 105 | "mailbox doesn't exist", |
| 106 | "select failed", |
| 107 | "no such mailbox", |
| 108 | "can't open mailbox", |
| 109 | "does not exist", |
| 110 | ) |
| 111 | |
| 112 | @classmethod |
| 113 | def default_config(cls) -> dict[str, Any]: |
| 114 | return EmailConfig().model_dump(by_alias=True) |
| 115 | |
| 116 | def __init__(self, config: Any, bus: MessageBus): |
| 117 | if isinstance(config, dict): |
| 118 | config = EmailConfig.model_validate(config) |
| 119 | super().__init__(config, bus) |
| 120 | self.config: EmailConfig = config |
| 121 | self._self_addresses = self._collect_self_addresses() |
| 122 | self._last_subject_by_chat: dict[str, str] = {} |
| 123 | self._last_message_id_by_chat: dict[str, str] = {} |
| 124 | self._processed_uids: set[str] = set() # Capped to prevent unbounded growth |
| 125 | self._MAX_PROCESSED_UIDS = 100000 |
| 126 | |
| 127 | async def start(self) -> None: |
| 128 | """Start polling IMAP for inbound emails.""" |
| 129 | if not self.config.consent_granted: |
| 130 | logger.warning( |
| 131 | "Email channel disabled: consent_granted is false. " |
| 132 | "Set channels.email.consentGranted=true after explicit user permission." |
| 133 | ) |
| 134 | return |
| 135 | |
| 136 | if not self._validate_config(): |
| 137 | return |
| 138 | |
| 139 | self._running = True |
| 140 | if not self.config.verify_dkim and not self.config.verify_spf: |
| 141 | logger.warning( |
| 142 | "Email channel: DKIM and SPF verification are both DISABLED. " |
| 143 | "Emails with spoofed From headers will be accepted. " |
| 144 | "Set verify_dkim=true and verify_spf=true for anti-spoofing protection." |
| 145 | ) |
| 146 | logger.info("Starting Email channel (IMAP polling mode)...") |
| 147 | |
| 148 | poll_seconds = max(5, int(self.config.poll_interval_seconds)) |
| 149 | while self._running: |
| 150 | try: |
| 151 | inbound_items = await asyncio.to_thread(self._fetch_new_messages) |
| 152 | for item in inbound_items: |
| 153 | sender = item["sender"] |
| 154 | subject = item.get("subject", "") |
| 155 | message_id = item.get("message_id", "") |
| 156 | |
| 157 | if subject: |
| 158 | self._last_subject_by_chat[sender] = subject |
| 159 | if message_id: |
| 160 | self._last_message_id_by_chat[sender] = message_id |
| 161 | |
| 162 | await self._handle_message( |
| 163 | sender_id=sender, |
| 164 | chat_id=sender, |
| 165 | content=item["content"], |
| 166 | media=item.get("media") or None, |
| 167 | metadata=item.get("metadata", {}), |
| 168 | ) |
| 169 | except Exception as e: |
| 170 | logger.error("Email polling error: {}", e) |
| 171 | |
| 172 | await asyncio.sleep(poll_seconds) |
| 173 | |
| 174 | async def stop(self) -> None: |
| 175 | """Stop polling loop.""" |
| 176 | self._running = False |
| 177 | |
| 178 | async def send(self, msg: OutboundMessage) -> None: |
| 179 | """Send email via SMTP.""" |
| 180 | if not self.config.consent_granted: |
| 181 | logger.warning("Skip email send: consent_granted is false") |
| 182 | return |
| 183 | |
| 184 | if not self.config.smtp_host: |
| 185 | logger.warning("Email channel SMTP host not configured") |
| 186 | return |
| 187 | |
| 188 | to_addr = msg.chat_id.strip() |
| 189 | if not to_addr: |
| 190 | logger.warning("Email channel missing recipient address") |
| 191 | return |
| 192 | |
| 193 | # Determine if this is a reply (recipient has sent us an email before) |
| 194 | is_reply = to_addr in self._last_subject_by_chat |
| 195 | force_send = bool((msg.metadata or {}).get("force_send")) |
| 196 | |
| 197 | # autoReplyEnabled only controls automatic replies, not proactive sends |
| 198 | if is_reply and not self.config.auto_reply_enabled and not force_send: |
| 199 | logger.info("Skip automatic email reply to {}: auto_reply_enabled is false", to_addr) |
| 200 | return |
| 201 | |
| 202 | base_subject = self._last_subject_by_chat.get(to_addr, "nanobot reply") |
| 203 | subject = self._reply_subject(base_subject) |
| 204 | if msg.metadata and isinstance(msg.metadata.get("subject"), str): |
| 205 | override = msg.metadata["subject"].strip() |
| 206 | if override: |
| 207 | subject = override |
| 208 | |
| 209 | email_msg = EmailMessage() |
| 210 | email_msg["From"] = self.config.from_address or self.config.smtp_username or self.config.imap_username |
| 211 | email_msg["To"] = to_addr |
| 212 | email_msg["Subject"] = subject |
| 213 | email_msg.set_content(msg.content or "") |
| 214 | |
| 215 | in_reply_to = self._last_message_id_by_chat.get(to_addr) |
| 216 | if in_reply_to: |
| 217 | email_msg["In-Reply-To"] = in_reply_to |
| 218 | email_msg["References"] = in_reply_to |
| 219 | |
| 220 | try: |
| 221 | await asyncio.to_thread(self._smtp_send, email_msg) |
| 222 | except Exception as e: |
| 223 | logger.error("Error sending email to {}: {}", to_addr, e) |
| 224 | raise |
| 225 | |
| 226 | def _validate_config(self) -> bool: |
| 227 | missing = [] |
| 228 | if not self.config.imap_host: |
| 229 | missing.append("imap_host") |
| 230 | if not self.config.imap_username: |
| 231 | missing.append("imap_username") |
| 232 | if not self.config.imap_password: |
| 233 | missing.append("imap_password") |
| 234 | if not self.config.smtp_host: |
| 235 | missing.append("smtp_host") |
| 236 | if not self.config.smtp_username: |
| 237 | missing.append("smtp_username") |
| 238 | if not self.config.smtp_password: |
| 239 | missing.append("smtp_password") |
| 240 | |
| 241 | if missing: |
| 242 | logger.error("Email channel not configured, missing: {}", ', '.join(missing)) |
| 243 | return False |
| 244 | return True |
| 245 | |
| 246 | def _smtp_send(self, msg: EmailMessage) -> None: |
| 247 | timeout = 30 |
| 248 | if self.config.smtp_use_ssl: |
| 249 | with smtplib.SMTP_SSL( |
| 250 | self.config.smtp_host, |
| 251 | self.config.smtp_port, |
| 252 | timeout=timeout, |
| 253 | ) as smtp: |
| 254 | smtp.login(self.config.smtp_username, self.config.smtp_password) |
| 255 | smtp.send_message(msg) |
| 256 | return |
| 257 | |
| 258 | with smtplib.SMTP(self.config.smtp_host, self.config.smtp_port, timeout=timeout) as smtp: |
| 259 | if self.config.smtp_use_tls: |
| 260 | smtp.starttls(context=ssl.create_default_context()) |
| 261 | smtp.login(self.config.smtp_username, self.config.smtp_password) |
| 262 | smtp.send_message(msg) |
| 263 | |
| 264 | def _fetch_new_messages(self) -> list[dict[str, Any]]: |
| 265 | """Poll IMAP and return parsed unread messages.""" |
| 266 | return self._fetch_messages( |
| 267 | search_criteria=("UNSEEN",), |
| 268 | mark_seen=self.config.mark_seen, |
| 269 | dedupe=True, |
| 270 | limit=0, |
| 271 | ) |
| 272 | |
| 273 | def fetch_messages_between_dates( |
| 274 | self, |
| 275 | start_date: date, |
| 276 | end_date: date, |
| 277 | limit: int = 20, |
| 278 | ) -> list[dict[str, Any]]: |
| 279 | """ |
| 280 | Fetch messages in [start_date, end_date) by IMAP date search. |
| 281 | |
| 282 | This is used for historical summarization tasks (e.g. "yesterday"). |
| 283 | """ |
| 284 | if end_date <= start_date: |
| 285 | return [] |
| 286 | |
| 287 | return self._fetch_messages( |
| 288 | search_criteria=( |
| 289 | "SINCE", |
| 290 | self._format_imap_date(start_date), |
| 291 | "BEFORE", |
| 292 | self._format_imap_date(end_date), |
| 293 | ), |
| 294 | mark_seen=False, |
| 295 | dedupe=False, |
| 296 | limit=max(1, int(limit)), |
| 297 | ) |
| 298 | |
| 299 | def _fetch_messages( |
| 300 | self, |
| 301 | search_criteria: tuple[str, ...], |
| 302 | mark_seen: bool, |
| 303 | dedupe: bool, |
| 304 | limit: int, |
| 305 | ) -> list[dict[str, Any]]: |
| 306 | messages: list[dict[str, Any]] = [] |
| 307 | cycle_uids: set[str] = set() |
| 308 | |
| 309 | for attempt in range(2): |
| 310 | try: |
| 311 | self._fetch_messages_once( |
| 312 | search_criteria, |
| 313 | mark_seen, |
| 314 | dedupe, |
| 315 | limit, |
| 316 | messages, |
| 317 | cycle_uids, |
| 318 | ) |
| 319 | return messages |
| 320 | except Exception as exc: |
| 321 | if attempt == 1 or not self._is_stale_imap_error(exc): |
| 322 | raise |
| 323 | logger.warning("Email IMAP connection went stale, retrying once: {}", exc) |
| 324 | |
| 325 | return messages |
| 326 | |
| 327 | def _fetch_messages_once( |
| 328 | self, |
| 329 | search_criteria: tuple[str, ...], |
| 330 | mark_seen: bool, |
| 331 | dedupe: bool, |
| 332 | limit: int, |
| 333 | messages: list[dict[str, Any]], |
| 334 | cycle_uids: set[str], |
| 335 | ) -> None: |
| 336 | """Fetch messages by arbitrary IMAP search criteria.""" |
| 337 | mailbox = self.config.imap_mailbox or "INBOX" |
| 338 | |
| 339 | if self.config.imap_use_ssl: |
| 340 | client = imaplib.IMAP4_SSL(self.config.imap_host, self.config.imap_port) |
| 341 | else: |
| 342 | client = imaplib.IMAP4(self.config.imap_host, self.config.imap_port) |
| 343 | |
| 344 | try: |
| 345 | client.login(self.config.imap_username, self.config.imap_password) |
| 346 | try: |
| 347 | status, _ = client.select(mailbox) |
| 348 | except Exception as exc: |
| 349 | if self._is_missing_mailbox_error(exc): |
| 350 | logger.warning("Email mailbox unavailable, skipping poll for {}: {}", mailbox, exc) |
| 351 | return messages |
| 352 | raise |
| 353 | if status != "OK": |
| 354 | logger.warning("Email mailbox select returned {}, skipping poll for {}", status, mailbox) |
| 355 | return messages |
| 356 | |
| 357 | status, data = client.search(None, *search_criteria) |
| 358 | if status != "OK" or not data: |
| 359 | return messages |
| 360 | |
| 361 | ids = data[0].split() |
| 362 | if limit > 0 and len(ids) > limit: |
| 363 | ids = ids[-limit:] |
| 364 | for imap_id in ids: |
| 365 | status, fetched = client.fetch(imap_id, "(BODY.PEEK[] UID)") |
| 366 | if status != "OK" or not fetched: |
| 367 | continue |
| 368 | |
| 369 | raw_bytes = self._extract_message_bytes(fetched) |
| 370 | if raw_bytes is None: |
| 371 | continue |
| 372 | |
| 373 | uid = self._extract_uid(fetched) |
| 374 | if uid and uid in cycle_uids: |
| 375 | continue |
| 376 | if dedupe and uid and uid in self._processed_uids: |
| 377 | continue |
| 378 | |
| 379 | parsed = BytesParser(policy=policy.default).parsebytes(raw_bytes) |
| 380 | sender = parseaddr(parsed.get("From", ""))[1].strip().lower() |
| 381 | if not sender: |
| 382 | continue |
| 383 | if self._is_self_address(sender): |
| 384 | logger.info("Email from {} ignored: matches bot-owned address", sender) |
| 385 | self._remember_processed_uid(uid, dedupe, cycle_uids) |
| 386 | if mark_seen: |
| 387 | client.store(imap_id, "+FLAGS", "\\Seen") |
| 388 | continue |
| 389 | |
| 390 | # --- Anti-spoofing: verify Authentication-Results --- |
| 391 | spf_pass, dkim_pass = self._check_authentication_results(parsed) |
| 392 | if self.config.verify_spf and not spf_pass: |
| 393 | logger.warning( |
| 394 | "Email from {} rejected: SPF verification failed " |
| 395 | "(no 'spf=pass' in Authentication-Results header)", |
| 396 | sender, |
| 397 | ) |
| 398 | self._remember_processed_uid(uid, dedupe, cycle_uids) |
| 399 | continue |
| 400 | if self.config.verify_dkim and not dkim_pass: |
| 401 | logger.warning( |
| 402 | "Email from {} rejected: DKIM verification failed " |
| 403 | "(no 'dkim=pass' in Authentication-Results header)", |
| 404 | sender, |
| 405 | ) |
| 406 | self._remember_processed_uid(uid, dedupe, cycle_uids) |
| 407 | continue |
| 408 | |
| 409 | subject = self._decode_header_value(parsed.get("Subject", "")) |
| 410 | date_value = parsed.get("Date", "") |
| 411 | message_id = parsed.get("Message-ID", "").strip() |
| 412 | body = self._extract_text_body(parsed) |
| 413 | |
| 414 | if not body: |
| 415 | body = "(empty email body)" |
| 416 | |
| 417 | body = body[: self.config.max_body_chars] |
| 418 | content = ( |
| 419 | f"[EMAIL-CONTEXT] Email received.\n" |
| 420 | f"From: {sender}\n" |
| 421 | f"Subject: {subject}\n" |
| 422 | f"Date: {date_value}\n\n" |
| 423 | f"{body}" |
| 424 | ) |
| 425 | |
| 426 | # --- Attachment extraction --- |
| 427 | attachment_paths: list[str] = [] |
| 428 | if self.config.allowed_attachment_types: |
| 429 | saved = self._extract_attachments( |
| 430 | parsed, |
| 431 | uid or "noid", |
| 432 | allowed_types=self.config.allowed_attachment_types, |
| 433 | max_size=self.config.max_attachment_size, |
| 434 | max_count=self.config.max_attachments_per_email, |
| 435 | ) |
| 436 | for p in saved: |
| 437 | attachment_paths.append(str(p)) |
| 438 | content += f"\n[attachment: {p.name} — saved to {p}]" |
| 439 | |
| 440 | metadata = { |
| 441 | "message_id": message_id, |
| 442 | "subject": subject, |
| 443 | "date": date_value, |
| 444 | "sender_email": sender, |
| 445 | "uid": uid, |
| 446 | } |
| 447 | messages.append( |
| 448 | { |
| 449 | "sender": sender, |
| 450 | "subject": subject, |
| 451 | "message_id": message_id, |
| 452 | "content": content, |
| 453 | "metadata": metadata, |
| 454 | "media": attachment_paths, |
| 455 | } |
| 456 | ) |
| 457 | |
| 458 | self._remember_processed_uid(uid, dedupe, cycle_uids) |
| 459 | |
| 460 | if mark_seen: |
| 461 | client.store(imap_id, "+FLAGS", "\\Seen") |
| 462 | finally: |
| 463 | try: |
| 464 | client.logout() |
| 465 | except Exception: |
| 466 | pass |
| 467 | |
| 468 | def _collect_self_addresses(self) -> set[str]: |
| 469 | """Return normalized email addresses owned by this channel instance.""" |
| 470 | candidates = ( |
| 471 | self.config.from_address, |
| 472 | self.config.smtp_username, |
| 473 | self.config.imap_username, |
| 474 | ) |
| 475 | normalized = { |
| 476 | addr |
| 477 | for candidate in candidates |
| 478 | if (addr := self._normalize_address(candidate)) |
| 479 | } |
| 480 | return normalized |
| 481 | |
| 482 | @staticmethod |
| 483 | def _normalize_address(value: str) -> str: |
| 484 | """Normalize an address or mailbox-like identifier for comparisons.""" |
| 485 | raw = (value or "").strip() |
| 486 | if not raw: |
| 487 | return "" |
| 488 | parsed = parseaddr(raw)[1].strip().lower() |
| 489 | if parsed: |
| 490 | return parsed |
| 491 | if "@" in raw: |
| 492 | return raw.lower() |
| 493 | return "" |
| 494 | |
| 495 | def _is_self_address(self, sender: str) -> bool: |
| 496 | """Return True when an inbound sender belongs to the bot itself.""" |
| 497 | normalized_sender = self._normalize_address(sender) |
| 498 | return bool(normalized_sender) and normalized_sender in self._self_addresses |
| 499 | |
| 500 | def _remember_processed_uid(self, uid: str, dedupe: bool, cycle_uids: set[str]) -> None: |
| 501 | """Track a fetched UID so skipped messages are not reprocessed forever.""" |
| 502 | if not uid: |
| 503 | return |
| 504 | cycle_uids.add(uid) |
| 505 | if dedupe: |
| 506 | self._processed_uids.add(uid) |
| 507 | # mark_seen is the primary dedup; this set is a safety net |
| 508 | if len(self._processed_uids) > self._MAX_PROCESSED_UIDS: |
| 509 | # Evict a random half to cap memory; mark_seen is the primary dedup |
| 510 | self._processed_uids = set(list(self._processed_uids)[len(self._processed_uids) // 2:]) |
| 511 | |
| 512 | @classmethod |
| 513 | def _is_stale_imap_error(cls, exc: Exception) -> bool: |
| 514 | message = str(exc).lower() |
| 515 | return any(marker in message for marker in cls._IMAP_RECONNECT_MARKERS) |
| 516 | |
| 517 | @classmethod |
| 518 | def _is_missing_mailbox_error(cls, exc: Exception) -> bool: |
| 519 | message = str(exc).lower() |
| 520 | return any(marker in message for marker in cls._IMAP_MISSING_MAILBOX_MARKERS) |
| 521 | |
| 522 | @classmethod |
| 523 | def _format_imap_date(cls, value: date) -> str: |
| 524 | """Format date for IMAP search (always English month abbreviations).""" |
| 525 | month = cls._IMAP_MONTHS[value.month - 1] |
| 526 | return f"{value.day:02d}-{month}-{value.year}" |
| 527 | |
| 528 | @staticmethod |
| 529 | def _extract_message_bytes(fetched: list[Any]) -> bytes | None: |
| 530 | for item in fetched: |
| 531 | if isinstance(item, tuple) and len(item) >= 2 and isinstance(item[1], (bytes, bytearray)): |
| 532 | return bytes(item[1]) |
| 533 | return None |
| 534 | |
| 535 | @staticmethod |
| 536 | def _extract_uid(fetched: list[Any]) -> str: |
| 537 | for item in fetched: |
| 538 | if isinstance(item, tuple) and item and isinstance(item[0], (bytes, bytearray)): |
| 539 | head = bytes(item[0]).decode("utf-8", errors="ignore") |
| 540 | m = re.search(r"UID\s+(\d+)", head) |
| 541 | if m: |
| 542 | return m.group(1) |
| 543 | return "" |
| 544 | |
| 545 | @staticmethod |
| 546 | def _decode_header_value(value: str) -> str: |
| 547 | if not value: |
| 548 | return "" |
| 549 | try: |
| 550 | return str(make_header(decode_header(value))) |
| 551 | except Exception: |
| 552 | return value |
| 553 | |
| 554 | @classmethod |
| 555 | def _extract_text_body(cls, msg: Any) -> str: |
| 556 | """Best-effort extraction of readable body text.""" |
| 557 | if msg.is_multipart(): |
| 558 | plain_parts: list[str] = [] |
| 559 | html_parts: list[str] = [] |
| 560 | for part in msg.walk(): |
| 561 | if part.get_content_disposition() == "attachment": |
| 562 | continue |
| 563 | content_type = part.get_content_type() |
| 564 | try: |
| 565 | payload = part.get_content() |
| 566 | except Exception: |
| 567 | payload_bytes = part.get_payload(decode=True) or b"" |
| 568 | charset = part.get_content_charset() or "utf-8" |
| 569 | payload = payload_bytes.decode(charset, errors="replace") |
| 570 | if not isinstance(payload, str): |
| 571 | continue |
| 572 | if content_type == "text/plain": |
| 573 | plain_parts.append(payload) |
| 574 | elif content_type == "text/html": |
| 575 | html_parts.append(payload) |
| 576 | if plain_parts: |
| 577 | return "\n\n".join(plain_parts).strip() |
| 578 | if html_parts: |
| 579 | return cls._html_to_text("\n\n".join(html_parts)).strip() |
| 580 | return "" |
| 581 | |
| 582 | try: |
| 583 | payload = msg.get_content() |
| 584 | except Exception: |
| 585 | payload_bytes = msg.get_payload(decode=True) or b"" |
| 586 | charset = msg.get_content_charset() or "utf-8" |
| 587 | payload = payload_bytes.decode(charset, errors="replace") |
| 588 | if not isinstance(payload, str): |
| 589 | return "" |
| 590 | if msg.get_content_type() == "text/html": |
| 591 | return cls._html_to_text(payload).strip() |
| 592 | return payload.strip() |
| 593 | |
| 594 | @staticmethod |
| 595 | def _check_authentication_results(parsed_msg: Any) -> tuple[bool, bool]: |
| 596 | """Parse Authentication-Results headers for SPF and DKIM verdicts. |
| 597 | |
| 598 | Returns: |
| 599 | A tuple of (spf_pass, dkim_pass) booleans. |
| 600 | """ |
| 601 | spf_pass = False |
| 602 | dkim_pass = False |
| 603 | for ar_header in parsed_msg.get_all("Authentication-Results") or []: |
| 604 | ar_lower = ar_header.lower() |
| 605 | if re.search(r"\bspf\s*=\s*pass\b", ar_lower): |
| 606 | spf_pass = True |
| 607 | if re.search(r"\bdkim\s*=\s*pass\b", ar_lower): |
| 608 | dkim_pass = True |
| 609 | return spf_pass, dkim_pass |
| 610 | |
| 611 | @classmethod |
| 612 | def _extract_attachments( |
| 613 | cls, |
| 614 | msg: Any, |
| 615 | uid: str, |
| 616 | *, |
| 617 | allowed_types: list[str], |
| 618 | max_size: int, |
| 619 | max_count: int, |
| 620 | ) -> list[Path]: |
| 621 | """Extract and save email attachments to the media directory. |
| 622 | |
| 623 | Returns list of saved file paths. |
| 624 | """ |
| 625 | if not msg.is_multipart(): |
| 626 | return [] |
| 627 | |
| 628 | saved: list[Path] = [] |
| 629 | media_dir = get_media_dir("email") |
| 630 | |
| 631 | for part in msg.walk(): |
| 632 | if len(saved) >= max_count: |
| 633 | break |
| 634 | if part.get_content_disposition() != "attachment": |
| 635 | continue |
| 636 | |
| 637 | content_type = part.get_content_type() |
| 638 | if not any(fnmatch(content_type, pat) for pat in allowed_types): |
| 639 | logger.debug("Email attachment skipped (type {}): not in allowed list", content_type) |
| 640 | continue |
| 641 | |
| 642 | payload = part.get_payload(decode=True) |
| 643 | if payload is None: |
| 644 | continue |
| 645 | if len(payload) > max_size: |
| 646 | logger.warning( |
| 647 | "Email attachment skipped: size {} exceeds limit {}", |
| 648 | len(payload), |
| 649 | max_size, |
| 650 | ) |
| 651 | continue |
| 652 | |
| 653 | raw_name = part.get_filename() or "attachment" |
| 654 | sanitized = safe_filename(raw_name) or "attachment" |
| 655 | dest = media_dir / f"{uid}_{sanitized}" |
| 656 | |
| 657 | try: |
| 658 | dest.write_bytes(payload) |
| 659 | saved.append(dest) |
| 660 | logger.info("Email attachment saved: {}", dest) |
| 661 | except Exception as exc: |
| 662 | logger.warning("Failed to save email attachment {}: {}", dest, exc) |
| 663 | |
| 664 | return saved |
| 665 | |
| 666 | @staticmethod |
| 667 | def _html_to_text(raw_html: str) -> str: |
| 668 | text = re.sub(r"<\s*br\s*/?>", "\n", raw_html, flags=re.IGNORECASE) |
| 669 | text = re.sub(r"<\s*/\s*p\s*>", "\n", text, flags=re.IGNORECASE) |
| 670 | text = re.sub(r"<[^>]+>", "", text) |
| 671 | return html.unescape(text) |
| 672 | |
| 673 | def _reply_subject(self, base_subject: str) -> str: |
| 674 | subject = (base_subject or "").strip() or "nanobot reply" |
| 675 | prefix = self.config.subject_prefix or "Re: " |
| 676 | if subject.lower().startswith("re:"): |
| 677 | return subject |
| 678 | return f"{prefix}{subject}" |
| 679 |