derp/plugins/emailcheck.py

"""Plugin: SMTP email verification via MX resolution and raw SMTP conversation."""

from __future__ import annotations

import asyncio
import logging
import re
import socket

from derp.dns import QTYPES, build_query, parse_response
from derp.http import create_connection
from derp.plugin import command

log = logging.getLogger(__name__)

_MAX_BATCH = 5
_SMTP_TIMEOUT = 15
_EMAIL_RE = re.compile(r"^[a-zA-Z0-9._%+\-]+@[a-zA-Z0-9.\-]+\.[a-zA-Z]{2,}$")

# Tor DNS resolver for MX lookups
_DNS_ADDR = "10.200.1.13"
_DNS_PORT = 9053


def _resolve_mx(domain: str) -> list[str]:
    """Resolve MX records for a domain via Tor DNS. Returns list of MX hosts."""
    query = build_query(domain, QTYPES["MX"])
    sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
    sock.settimeout(5)
    try:
        sock.sendto(query, (_DNS_ADDR, _DNS_PORT))
        data, _ = sock.recvfrom(4096)
    finally:
        sock.close()

    rcode, results = parse_response(data)
    if rcode != 0 or not results:
        return []

    # MX results are "priority host" format
    mx_hosts = []
    for r in results:
        parts = r.split(None, 1)
        if len(parts) == 2:
            mx_hosts.append(parts[1].rstrip("."))
        else:
            mx_hosts.append(r.rstrip("."))

    # Sort by priority (already parsed as "prio host")
    return mx_hosts


def _resolve_a(domain: str) -> str | None:
    """Resolve A record for a domain as fallback. Returns first IP or None."""
    query = build_query(domain, QTYPES["A"])
    sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
    sock.settimeout(5)
    try:
        sock.sendto(query, (_DNS_ADDR, _DNS_PORT))
        data, _ = sock.recvfrom(4096)
    finally:
        sock.close()

    rcode, results = parse_response(data)
    if rcode == 0 and results:
        return results[0]
    return None


def _smtp_check(email: str, mx_host: str) -> tuple[int, str]:
    """Raw SMTP conversation via SOCKS proxy. Returns (response_code, response_text)."""
    sock = create_connection((mx_host, 25), timeout=_SMTP_TIMEOUT)
    try:
        fp = sock.makefile("rb")

        def _read_reply() -> tuple[int, str]:
            lines = []
            while True:
                line = fp.readline()
                if not line:
                    break
                text = line.decode("utf-8", errors="replace").rstrip()
                lines.append(text)
                # Multi-line: "250-..." continues, "250 ..." ends
                if len(text) >= 4 and text[3] == " ":
                    break
            if not lines:
                return 0, "no response"
            try:
                code = int(lines[-1][:3])
            except (ValueError, IndexError):
                code = 0
            return code, lines[-1][4:] if len(lines[-1]) > 4 else lines[-1]

        # Read banner
        code, text = _read_reply()
        if code != 220:
            return code, f"banner: {text}"

        # EHLO
        sock.sendall(b"EHLO derp.bot\r\n")
        code, text = _read_reply()
        if code != 250:
            return code, f"EHLO: {text}"

        # MAIL FROM
        sock.sendall(b"MAIL FROM:<>\r\n")
        code, text = _read_reply()
        if code != 250:
            return code, f"MAIL FROM: {text}"

        # RCPT TO
        sock.sendall(f"RCPT TO:<{email}>\r\n".encode())
        code, text = _read_reply()

        # QUIT
        try:
            sock.sendall(b"QUIT\r\n")
        except OSError:
            pass

        return code, text
    finally:
        sock.close()


@command("emailcheck", help="SMTP verify: !emailcheck <email> [email2 ...]", admin=True)
async def cmd_emailcheck(bot, message):
    """Check email deliverability via SMTP RCPT TO verification.

    Usage:
        !emailcheck user@example.com                    Single check
        !emailcheck user@example.com user2@test.org     Batch (max 5)
    """
    parts = message.text.split()
    if len(parts) < 2:
        await bot.reply(message, "Usage: !emailcheck <email> [email2 ...]")
        return

    emails = parts[1:1 + _MAX_BATCH]

    # Validate format
    for email in emails:
        if not _EMAIL_RE.match(email):
            await bot.reply(message, f"Invalid email format: {email}")
            return

    loop = asyncio.get_running_loop()

    async def _check_one(email: str) -> str:
        domain = email.split("@", 1)[1]

        # Resolve MX
        try:
            mx_hosts = await loop.run_in_executor(None, _resolve_mx, domain)
        except Exception as exc:
            log.debug("emailcheck: MX lookup failed for %s: %s", domain, exc)
            mx_hosts = []

        # Fallback to A record
        if not mx_hosts:
            try:
                a_record = await loop.run_in_executor(None, _resolve_a, domain)
            except Exception:
                a_record = None
            if a_record:
                mx_hosts = [domain]
            else:
                return f"{email} -- no MX or A record for {domain}"

        # Try each MX host
        for mx in mx_hosts:
            try:
                code, text = await loop.run_in_executor(None, _smtp_check, email, mx)
                return f"{email} -- SMTP {code} {text} (mx: {mx})"
            except Exception as exc:
                log.debug("emailcheck: SMTP failed for %s via %s: %s", email, mx, exc)
                continue

        return f"{email} -- all MX hosts unreachable"

    if len(emails) > 1:
        await bot.reply(message, f"Checking {len(emails)} addresses...")

    results = await asyncio.gather(*[_check_one(e) for e in emails])
    for line in results:
        await bot.reply(message, line)