Skip to content
Snippets Groups Projects
labrea-idea.py 20.3 KiB
Newer Older
#!/usr/bin/python
# -*- coding: utf-8 -*-

import os
import sys
import re
import time
import optparse
import signal
import contextlib
import uuid
import json
import socket
import resource
import atexit
import logging
import logging.handlers
from itertools import izip
try:
    from collections import OrderedDict
except ImportError:
    from ordereddict import OrderedDict


class WindowContextMgr(object):

    def __init__(self, window=60*10, timeout=60*5, ideagen=None):
        self.contexts = {}
        self.update_timestamp = 0
        self.window = window
        self.timeout = timeout
        self.ideagen = ideagen
        self.first_update_queue = OrderedDict()
        self.last_update_queue = OrderedDict()
    def expire_queue(self, queue, window):
        ctx_to_del = []
        for ctx, timestamp in queue.iteritems():
            if timestamp >= self.update_timestamp - window:
            closed = self.ctx_close(self.contexts[ctx])
            if closed is not None:
                aggr_events.append(closed)
            ctx_to_del.append(ctx)
        for ctx in ctx_to_del:
            del self.contexts[ctx]
            del self.first_update_queue[ctx]
            del self.last_update_queue[ctx]
    def process(self, event=None, timestamp=None):
Pavel Kácha's avatar
Pavel Kácha committed
        if timestamp > self.update_timestamp:
            self.update_timestamp = timestamp
        aggr_events.extend(self.expire_queue(self.first_update_queue, self.window))
        aggr_events.extend(self.expire_queue(self.last_update_queue, self.timeout))

        if event is not None:
            ctx = self.match(event)
            if ctx is not None:
                if ctx not in self.contexts:
                    self.contexts[ctx] = self.ctx_create(event)
                    self.first_update_queue[ctx] = self.update_timestamp
                    self.last_update_queue[ctx] = self.update_timestamp
                    del self.last_update_queue[ctx]
                    self.last_update_queue[ctx] = self.update_timestamp
Pavel Kácha's avatar
Pavel Kácha committed
    def close(self, timestamp):
        if timestamp is not None and timestamp > self.update_timestamp:
            self.update_timestamp = timestamp
        aggr_events = []
        for context in self.contexts.values():
            closed = self.ctx_close(context)
            if closed is not None:
                aggr_events.append(closed)
        self.contexts = {}
        self.first_update_queue = OrderedDict()
        self.last_update_queue = OrderedDict()
        return aggr_events


class PingContextMgr(WindowContextMgr):

    def match(self, event):
        return event.src_ip if 'Ping' in event.message else None

    def ctx_create(self, event):
        ctx = dict(
            src_ip=event.src_ip,
            tgt_ips=set([event.tgt_ip]),
            count=1,
            first_update=self.update_timestamp,
            last_update=self.update_timestamp
        )
        return ctx

    def ctx_append(self, ctx, event):
        ctx["tgt_ips"].add(event.tgt_ip)
        ctx["count"] += 1
        ctx["last_update"] = self.update_timestamp

    def ctx_close(self, ctx):
        return self.ideagen.gen_idea(
            src=ctx["src_ip"],
            src_ports=None,
            targets=[(ip, []) for ip in ctx["tgt_ips"]],
            detect_time=self.update_timestamp,
            event_time=ctx["first_update"],
            cease_time=ctx["last_update"],
            count=ctx["count"],
            template="ping"
        )


class ConnectContextMgr(WindowContextMgr):

    def match(self, event):
        return event.src_ip if 'Connect' in event.message else None

    def ctx_create(self, event):
        ctx = dict(
            src_ip=event.src_ip,
            src_ports=set([event.src_port]),
            tgt_ips_ports={event.tgt_ip: set([event.tgt_port])},
            count=1,
            first_update=self.update_timestamp,
            last_update=self.update_timestamp
        )
        return ctx

    def ctx_append(self, ctx, event):
        tgt_ports = ctx["tgt_ips_ports"].setdefault(event.tgt_ip, set())
        tgt_ports.add(event.tgt_port)
        ctx["src_ports"].add(event.src_port)
        ctx["count"] += 1
        ctx["last_update"] = self.update_timestamp

    def ctx_close(self, ctx):
        return self.ideagen.gen_idea(
            src=ctx["src_ip"],
            src_ports=ctx["src_ports"],
            targets=ctx["tgt_ips_ports"].items(),
            detect_time=self.update_timestamp,
            event_time=ctx["first_update"],
            cease_time=ctx["last_update"],
            count=ctx["count"],
            template="connect"
        )


class InboundContextMgr(ConnectContextMgr):

    def match(self, event):
        return event.src_ip if 'Inbound' in event.message else None

    def ctx_close(self, ctx):
        return self.ideagen.gen_idea(
            src=ctx["src_ip"],
            src_ports=ctx["src_ports"],
            targets=ctx["tgt_ips_ports"].items(),
            detect_time=self.update_timestamp,
            event_time=ctx["first_update"],
            cease_time=ctx["last_update"],
            count=ctx["count"],
            template="synack"
class FileWatcher(object):

    def __init__(self, filename, tail=True):
        self.filename = filename
        self.open()
        self.line_buffer = ""
        if tail and self.f:
            self.f.seek(0, os.SEEK_END)

    def open(self):
        try:
            self.f = open(self.filename, "r")
            st = os.fstat(self.f.fileno())
            self.inode, self.size = st.st_ino, st.st_size
        except IOError:
            self.f = None
            self.inode = -1
            self.size = -1

    def _check_reopen(self):
        try:
            st = os.stat(self.filename)
            cur_inode, cur_size = st.st_ino, st.st_size
        except OSError as e:
            cur_inode = -1
            cur_size = -1
        if cur_inode != self.inode or cur_size < self.size:
            self.close()
            self.open()

    def readline(self):
        if not self.f:
            self.open()
            if not self.f:
                return self.line_buffer
        res = self.f.readline()
        if not res:
            self._check_reopen()
            if not self.f:
                return self.line_buffer
            res = self.f.readline()
        if not res.endswith("\n"):
            self.line_buffer += res
        else:
            res = self.line_buffer + res
            self.line_buffer = ""
            return res

    def close(self):
        try:
            if self.f:
                self.f.close()
        except IOError:
            pass
        self.inode = -1
        self.size = -1

    def __repr__(self):
        return '%s("%s")' % (type(self).__name__, self.filename)

    def __enter__(self):
        return self

    def __exit__(self, exc_type, exc_val, exc_tb):
        self.close()
        return False

    def __iter__(self):
        return self

    def next(self):
        return self.readline()


class IdeaGen(object):

    def __init__(self, name, test=False):
        self.name = name
        self.test = test
        self.template = {
            "connect": {
                "category": ["Recon.Scanning"],
                "description": "TCP connections/scan",
                "note": "Connections from remote host to never assigned IP",
                "proto": ["tcp"]
            },
            "ping": {
                "category": ["Recon.Scanning"],
                "description": "Ping scan",
                "template": "labrea-002",
                "note": "Ping requests from remote host to never assigned IP",
                "proto": ["icmp"]
                "category": ["Availability.DoS"],
                "description": "Unsolicited TCP SYN/ACK connections/scan",
                "note": "Unsolicited SYN/ACK packet received from remote host to never assigned IP",
                "source_type": ["Backscatter"],
                "source_to_target": True,
                "proto": ["tcp"]
            }
        }

    def format_timestamp(self, epoch):
        return "%04d-%02d-%02dT%02d:%02d:%02dZ" % time.gmtime(epoch)[:6]

    def gen_idea(self, src, src_ports, targets, detect_time, event_time, cease_time, count, template):
        tmpl = self.template[template]
        isource = {
            "IP6" if ":" in src else "IP4": [src],
            "Proto": tmpl["proto"]
        if "source_type" in tmpl:
            isource["Type"] = tmpl["source_type"]
        if src_ports:
            isource["Port"] = [int(port) for port in src_ports]
        # Fold multiple IPs with the same portset
        folded_tgt = {}
        for tgt, ports in targets:
            folded_tgt.setdefault(frozenset(ports), []).append(tgt)
        itargets = []
        for ports, tgt in folded_tgt.items():
            itarget = {"Proto": tmpl["proto"]}
            tgts4 = [ip for ip in tgt if ":" not in ip]
            tgts6 = [ip for ip in tgt if ":" in ip]
            if tgts4:
                itarget["IP4"] = tgts4
            if tgts6:
                itarget["IP6"] = tgts6
            if ports:
                itarget["Port"] = [int(port) for port in ports]
            itargets.append(itarget)
        inode = {
            "SW": ["LaBrea"],
            "Type": ["Connection", "Tarpit"],
            "Name": self.name
        }
        idea = {
            "Format": "IDEA0",
            "ID": str(uuid.uuid4()),
            "Category": tmpl["category"] + (["Test"] if self.test else []),
            "Description": tmpl["description"],
            "DetectTime": self.format_timestamp(detect_time),
            "EventTime": self.format_timestamp(event_time),
            "CeaseTime": self.format_timestamp(cease_time),
Pavel Kácha's avatar
Pavel Kácha committed
            "ConnCount": count,
            "Note": tmpl["note"],
            "_CESNET": {
                "EventTemplate": tmpl["template"],
            },
            "Target": itargets,
            "Node": [inode]
        }
        if tmpl.get("source_to_target", False):
            idea["Target"].append(isource)
        else:
            idea["Source"] = [isource]


class Filer(object):

    def __init__(self, directory):
        self.basedir = self._ensure_path(directory)
        self.tmp = self._ensure_path(pth.join(self.basedir, "tmp"))
        self.incoming = self._ensure_path(pth.join(self.basedir, "incoming"))
        self.hostname = socket.gethostname()
        self.pid = os.getpid()

    def _ensure_path(self, p):
        try:
            os.mkdir(p)
        except OSError:
            if not pth.isdir(p):
                raise
        return p

    def _get_new_name(self, fd=None):
        (inode, device) = os.fstat(fd)[1:3] if fd else (0, 0)
        return "%s.%d.%f.%d.%d" % (
            self.hostname, self.pid, time.time(), device, inode)

    def create_unique_file(self):
        # First find and open name unique within tmp
        tmpname = None
        while not tmpname:
            tmpname = self._get_new_name()
            try:
                fd = os.open(pth.join(self.tmp, tmpname), os.O_CREAT | os.O_RDWR | os.O_EXCL)
            except OSError as e:
                if e.errno != errno.EEXIST:
                    raise   # other errors than duplicates should get noticed
                tmpname = None
        # Now we know the device/inode, rename to make unique within system
        newname = self._get_new_name(fd)
        os.rename(pth.join(self.tmp, tmpname), pth.join(self.tmp, newname))
        nf = os.fdopen(fd, "w")
        return nf, newname

    def publish_file(self, short_name):
        os.rename(pth.join(self.tmp, short_name), pth.join(self.incoming, short_name))
        work_dir=None, chroot_dir=None,
        umask=None, uid=None, gid=None,
        pidfile=None, files_preserve=[], signals={}):
    # Dirs, limits, users
    if chroot_dir is not None:
        os.chdir(chroot_dir)
        os.chroot(chroot_dir)
    if umask is not None:
        os.umask(umask)
    if work_dir is not None:
        os.chdir(work_dir)
    if gid is not None:
        os.setgid(gid)
    if uid is not None:
        os.setuid(uid)
    # Doublefork, split session
    if os.fork() > 0:
    try:
        os.setsid()
    except OSError:
        pass
    if os.fork() > 0:
        os._exit(0)
    # Setup signal handlers
    for (signum, handler) in signals.items():
        signal.signal(signum, handler)
    # Close descriptors
    descr_preserve = set(f.fileno() for f in files_preserve)
    maxfd = resource.getrlimit(resource.RLIMIT_NOFILE)[1]
    if maxfd == resource.RLIM_INFINITY:
        maxfd = 65535
    for fd in range(maxfd, 3, -1):  # 3 means omit stdin, stdout, stderr
        if fd not in descr_preserve:
            try:
                os.close(fd)
            except Exception:
                pass
    # Redirect stdin, stdout, stderr to /dev/null
    devnull = os.open(os.devnull, os.O_RDWR)
    for fd in range(3):
        os.dup2(devnull, fd)
    # PID file
    if pidfile is not None:
        pidd = os.open(pidfile, os.O_RDWR | os.O_CREAT | os.O_EXCL | os.O_TRUNC)
        os.write(pidd, str(os.getpid())+"\n")
        os.close(pidd)
        # Define and setup atexit closure
        @atexit.register
        def unlink_pid():
            try:
                os.unlink(pidfile)
            except Exception:
                pass


def save_events(aggr, filer):
    for event in aggr:
        f, name = filer.create_unique_file()
        with f:
            f.write(json.dumps(event, ensure_ascii=True))
        filer.publish_file(name)
        logging.info(
            "Saved %s %s \"%s\" (%d) as file %s" % (
                event["ID"], "+".join(event["Category"]), event["Description"], event["ConnCount"], name))


RE_LIST = (
    # 1493035442 Initial Connect - tarpitting: 89.163.242.15 56736 -> 195.113.254.182 9898
    # 1493037991 Inbound SYN/ACK: 185.62.190.15 21001 -> 195.113.252.222 15584
    (
        re.compile(r'([0-9]+) ([^:]*:) ([^ ]+) ([0-9]+) -> ([^ ]+) ([0-9]+).*'),
        namedtuple("connect_tuple", ("timestamp", "message", "src_ip", "src_port", "tgt_ip", "tgt_port"))
    ),
    # 1493035442 Responded to a Ping: 88.86.96.25 -> 195.113.253.87 *
    (
        re.compile(r'([0-9]+) ([^:]*:) ([^ ]+) -> ([^ ]+).*'),
        namedtuple("ping_tuple", ("timestamp", "message", "src_ip", "tgt_ip"))
    )
)


def match_event(line):
    for labrea_re, event_tuple in RE_LIST:
        match = labrea_re.match(line)
        if match:
            return event_tuple(*match.groups())
    logging.info("Unmatched line: \"%s\"" % line.replace("\n", r"\n"))
    return None


def get_args():
    optp = optparse.OptionParser(
        usage="usage: %prog [options] logfile ...",
        description="Watch LaBrea logfiles and generate Idea events into directory")
    optp.add_option(
        "-w", "--window",
        default=900,
        dest="window",
        type="int",
        action="store",
        help="max detection window (default: %default)")
    optp.add_option(
        "-t", "--timeout",
        default=300,
        dest="timeout",
        type="int",
        action="store",
        help="detection timeout (default: %default)")
    optp.add_option(
        "-n", "--name",
        default=None,
        dest="name",
        type="string",
        action="store",
        help="Warden client name")
    optp.add_option(
        "--test",
        default=False,
        dest="test",
        action="store_true",
        help="Add Test category")
    optp.add_option(
        "-o", "--oneshot",
        default=False,
        dest="oneshot",
        action="store_true",
        help="process files and quit (do not daemonize)")
    optp.add_option(
        "--poll",
Pavel Kácha's avatar
Pavel Kácha committed
        default=1,
        dest="poll",
        type="int",
        action="store",
        help="log file polling interval")
    optp.add_option(
        "-d", "--dir",
        default=None,
        dest="dir",
        type="string",
        action="store",
        help="Target directory (mandatory)"),
    optp.add_option(
        "-p", "--pid",
        default=pth.join("/var/run", pth.splitext(pth.basename(sys.argv[0]))[0] + ".pid"),
        dest="pid",
        type="string",
        action="store",
        help="create PID file with this name (default: %default)")
    optp.add_option(
        "-u", "--uid",
        default=None,
        dest="uid",
        type="int",
        action="store",
        help="user id to run under")
    optp.add_option(
        "-g", "--gid",
        default=None,
        dest="gid",
        type="int",
        action="store",
        help="group id to run under")
    optp.add_option(
        "-v", "--verbose",
        dest="verbose",
        action="store_true",
        help="turn on debug logging")
    optp.add_option(
        "--log",
        default="local7",
        dest="log",
        type="string",
        action="store",
        help="syslog facility or log file name (default: %default)")
    optp.add_option(
        "--realtime",
        default=True,
        dest="realtime",
        action="store_true",
        help="use system time along with log timestamps (default)")
    optp.add_option(
        "--norealtime",
        dest="realtime",
        action="store_false",
        help="don't system time, use solely log timestamps")
    return optp


running_flag = True
reload_flag = False

def terminate_me(signum, frame):
    global running_flag
    running_flag = False

def reload_me(signum, frame):
    global reload_flag
    reload_flag = True

def main():
    global reload_flag
    optp = get_args()
    opts, args = optp.parse_args()
    if not args or opts.name is None or opts.dir is None:

    logformat = "%(filename)s[%(process)d] %(message)s"
    logger = logging.getLogger()
        handler = logging.StreamHandler()
        handler.setFormatter(logging.Formatter("%(asctime)s " + logformat))
        if "/" in opts.log:
            handler = logging.handlers.WatchedFileHandler(opts.log)
            handler.setFormatter(logging.Formatter("%(asctime)s " + logformat))
        else:
            handler = logging.handlers.SysLogHandler(address="/dev/log", facility=opts.log)
            handler.setFormatter(logging.Formatter(logformat))
    logger.addHandler(handler)
    logger.setLevel(logging.DEBUG if opts.verbose else logging.INFO)

    try:
        if opts.oneshot:
            signal.signal(signal.SIGINT, terminate_me)
            signal.signal(signal.SIGTERM, terminate_me)
            files = [open(arg) for arg in args]
        else:
            daemonize(
                pidfile=opts.pid,
                uid=opts.uid,
                gid=opts.gid,
                signals={
                    signal.SIGINT: terminate_me,
                    signal.SIGTERM: terminate_me,
                    signal.SIGHUP: reload_me
                })
            files = [FileWatcher(arg) for arg in args]

        ideagen = IdeaGen(name=opts.name, test=opts.test)
        filer = Filer(opts.dir)
        contexts = [
            context(window=opts.window, timeout=opts.timeout, ideagen=ideagen)
            for context in [PingContextMgr, ConnectContextMgr, InboundContextMgr]
        ]

        with contextlib.nested(*files):
            for line_set in izip(*files):
                for line in line_set:
                    if not line:
                        break

                    event = match_event(line)

                    if event or opts.realtime:
                        timestamp = int(event.timestamp) if event else int(time.time())
                            save_events(context.process(event, timestamp), filer)

                if not running_flag:
                    break
                if reload_flag:
                    for f in files:
                        f.close()
                        f.open()
                if not any(line_set):
                    time.sleep(opts.poll)

        for context in contexts:
Pavel Kácha's avatar
Pavel Kácha committed
            timestamp = int(time.time()) if opts.realtime else None
            save_events(context.close(timestamp), filer)

    except Exception:
        logging.exception("Exception")