Skip to content
Snippets Groups Projects
warden_server.py 35 KiB
Newer Older
#!/usr/bin/python
# -*- coding: utf-8 -*-
#
# Copyright (C) 2011-2013 Cesnet z.s.p.o
# Use of this source is governed by a 3-clause BSD-style license, see LICENSE file.

import sys
import logging
import logging.handlers
import ConfigParser
from traceback import format_tb
import M2Crypto.X509
import json
Pavel Kácha's avatar
Pavel Kácha committed
import MySQLdb as my
import MySQLdb.cursors as mycursors
from uuid import uuid4
from time import time, gmtime
from math import trunc
from io import BytesIO
from urlparse import parse_qs
from os import path

# for local version of up to date jsonschema
sys.path.append(path.join(path.dirname(__file__), "..", "lib"))

from jsonschema import Draft4Validator, FormatChecker


VERSION = "3.0-not-even-alpha"


class Error(Exception):

    def __init__(self, message, error=500, method=None,
            detail=None, exc=(None, None, None)):
        self.error = int(error)
        self.method = method
        self.message = message
        self.detail = detail
        (self.exctype, self.excval, self.exctb) = exc or sys.exc_info()
        self.cause = self.excval # compatibility with other exceptions


    def __str__(self):
        out = []
        out.append("Error(%s)" % (self.error))
        if self.method is not None:
            out.append(" in \"%s\"" % self.method)
        if self.message is not None:
            out.append(": %s" % self.message)
        if self.excval is not None:
            out.append(" - cause was %s: %s" % (type(self.excval).__name__, str(self.excval)))
        return "".join(out)


    def info_str(self):
        return ("Detail: %s" % self.detail) or ""


    def debug_str(self):
        out = []
        if self.excval is not None:
            out.append("Exception %s: %s\n" % (type(self.excval).__name__, str(self.excval)))
        if self.exctb is not None:
            out.append("Traceback:\n%s" % "".join(format_tb(self.exctb)))
        return "".join(out)


    def to_dict(self):
        d = {}
        if self.error is not None:
            d["error"] = self.error
        if self.method is not None:
            d["method"] = self.method
        if self.message is not None:
            d["message"] = self.message
        if self.detail is not None:
            d["detail"] = self.detail
        if self.excval is not None:
            d["message"] = d["message"] + ", cause was %s: %s" % (type(self.excval).__name__, str(self.excval))
        return d



def get_clean_root_logger(level=logging.INFO):
    """ Attempts to get logging module into clean slate state """

    # We want to be able to set up at least stderr logger before any
    # configuration is read, and then later get rid of it and set up
    # whatever administrator requires.
    # However, there can exist only one logger, but we want to get a clean
    # slate everytime we initialize StreamLogger or FileLogger... which
    # is not exactly supported by logging module.
    # So, we look directly inside logger class and clean up handlers/filters
    # manually.
    logger = logging.getLogger()  # no need to create new
    logger.setLevel(level)
    while logger.handlers:
        logger.removeHandler(logger.handlers[0])
    while logger.filters:
        logger.removeFilter(logger.filters[0])
    return logger



def StreamLogger(stream=sys.stderr, level=logging.INFO):
    """ Fallback handler just for setup, not meant to be used from
        configuration file because during wsgi query stdout/stderr
        is forbidden.
    """

    fhand = logging.StreamHandler(stream)
    fform = logging.Formatter('%(asctime)s %(filename)s[%(process)d]: (%(levelname)s) %(message)s')
    fhand.setFormatter(fform)
    logger = get_clean_root_logger(level)
    logger.addHandler(fhand)



def FileLogger(filename, level=logging.INFO):

    fhand = logging.FileHandler(filename)
    fform = logging.Formatter('%(asctime)s %(filename)s[%(process)d]: (%(levelname)s) %(message)s')
    fhand.setFormatter(fform)
    logger = get_clean_root_logger(level)
    logger.addHandler(fhand)
    logging.info("Initialized FileLogger(filename=\"%s\", \"%s\")" % (filename, level))



def SysLogger(socket="/dev/log", facility=logging.handlers.SysLogHandler.LOG_DAEMON, level=logging.INFO):

    fhand = logging.handlers.SysLogHandler(address=socket, facility=facility)
    fform = logging.Formatter('%(filename)s[%(process)d]: (%(levelname)s) %(message)s')
    fhand.setFormatter(fform)
    logger = get_clean_root_logger(level)
    logger.addHandler(fhand)
    logging.info("Initialized SysLogger(socket=\"%s\", facility=\"%s\", level=\"%s\")" % (socket, facility, level))



class Object(object):

    def __str__(self):
        return "%s()" % type(self).__name__



class NoAuthenticator(Object):

    def __init__(self):
        Object.__init__(self)

    def authenticate (self, env):
        return "anybody"    # or None


Michal Kostenec's avatar
Michal Kostenec committed
    def authorize(self, env, client, method, event, args):
        return (client is not None)



class X509Authenticator(NoAuthenticator):

    def __init__(self, db):
        self.db = db
        NoAuthenticator.__init__(self)


    def __str__(self):
        return "%s(db=%s)" % (type(self).__name__, type(self.db).__name__)


    def get_cert_dns_names(self, pem):

        cert = M2Crypto.X509.load_cert_string(pem)

        subj = cert.get_subject()
        commons = [n.get_data().as_text() for n in subj.get_entries_by_nid(subj.nid["CN"])]

        ext = cert.get_ext("subjectAltName")
        extstrs = [val.strip() for val in ext.get_value().split(",")]
        altnames = [val[4:] for val in extstrs if val.startswith("DNS:")]

        # bit of mangling to get rid of duplicates and leave commonname first
        firstcommon = commons[0]
        return [firstcommon] + list(set(altnames+commons) - set([firstcommon]))


    def authenticate (self, env):
        names = self.get_cert_dns_names(env["SSL_CLIENT_CERT"])
Pavel Kácha's avatar
Pavel Kácha committed
        return self.db.get_client_by_name(names)
Michal Kostenec's avatar
Michal Kostenec committed
    def authorize(self, env, client, method, event, args):
Michal Kostenec's avatar
Michal Kostenec committed
        # Authorize for debug
        if (method == 'getDebug'):
            return client if client[0]['debug'] == 1 else None

Michal Kostenec's avatar
Michal Kostenec committed
        cl = None
        service = event['Node'][0]['Name']
Michal Kostenec's avatar
Michal Kostenec committed
        test = 'Test' in event['Category']        
Michal Kostenec's avatar
Michal Kostenec committed
        
Michal Kostenec's avatar
Michal Kostenec committed
        for clx in client:
            if clx['service'] == service:
                cl = clx
Michal Kostenec's avatar
Michal Kostenec committed
                break

        if cl is None:
            return None
        
Michal Kostenec's avatar
Michal Kostenec committed
        # Authorize for sending events
Michal Kostenec's avatar
Michal Kostenec committed
        if ((method == 'sendEvents' and cl['write'] == 1) or
Michal Kostenec's avatar
Michal Kostenec committed
            (method == 'sendEvents' and cl['test'] == 1 and test)):
Michal Kostenec's avatar
Michal Kostenec committed
            return cl

        return None

class NoValidator(Object):

    def check(self, event):
        return []


class JSONSchemaValidator(NoValidator):

    def __init__(self, filename=None):
        self.path = filename or path.join(path.dirname(__file__), "idea.schema")
        with open(self.path) as f:
            self.schema = json.load(f)
        self.validator = Draft4Validator(self.schema, format_checker=FormatChecker())


    def __str__(self):
        return "%s(filename=\"%s\")" % (type(self).__name__, self.path)


    def check(self, event):

        def sortkey(k):
            """ Treat keys as lowercase, prefer keys with less path segments """
            return (len(k.path), "/".join(str(k.path)).lower())

        res = []
        for error in sorted(self.validator.iter_errors(event), key=sortkey):
            res.append(
                "Validation error: key \"%s\", value \"%s\", expected - %s, error message - %s\n" % (
                    u"/".join(str(v) for v in error.path),
                    error.instance,
                    error.schema.get('description', 'no additional info'),
                    error.message))

        return res



Pavel Kácha's avatar
Pavel Kácha committed
class MySQL(Object):

    def __init__(self, host, user, password, dbname, port):
        self.host = host
        self.user = user
        self.password = password
        self.dbname = dbname
        self.port = port

        self.con = my.connect(host=self.host, user=self.user, passwd=self.password,
            db=self.dbname, port=self.port, cursorclass=mycursors.DictCursor)
        self.crs = self.con.cursor()


    def __str__(self):
        return "%s(host='%s', user='%s', dbname='%s', port=%d)" % (
            type(self).__name__, self.host, self.user, self.dbname, self.port)


    def get_client_by_name(self, name):
Michal Kostenec's avatar
Michal Kostenec committed
        format_strings = ','.join(['%s'] * len(name))
Michal Kostenec's avatar
Michal Kostenec committed
        self.crs.execute("SELECT cl.`id`, cl.`hostname`, s.`service`, s.`service_id`, s.`identity`, cl.`read`, s.`write`, s.`test`, cl.`debug` FROM `clients` cl LEFT JOIN `services` s ON cl.`id` = s.`client_id` WHERE cl.`valid` = 1 AND s.`valid` = 1 AND `hostname` IN (%s)" % format_strings, tuple(name))
Michal Kostenec's avatar
Michal Kostenec committed
        row = self.crs.fetchall()
        
        return row if row else None
Pavel Kácha's avatar
Pavel Kácha committed


    def get_debug(self):
Michal Kostenec's avatar
Michal Kostenec committed
        self.crs.execute("SELECT VERSION() AS VER")
Pavel Kácha's avatar
Pavel Kácha committed
        row = self.crs.fetchone()
        return {
            "db": "MySQL",
            "version": row["VER"]
        }


    def get_status(self):
        return {}


    def gen_random_idea(self):

        def get_precise_timestamp():
            t = time()
            us = trunc((t-trunc(t))*1000000)
            g = gmtime(t)
            iso = '%04d-%02d-%02dT%02d:%02d:%02d.%0dZ' % (g[0:6]+(us,))
            return iso

        return {
           "Format": "IDEA0",
           "ID": str(uuid4()),
           "DetectTime": get_precise_timestamp(),
           "Category": ["Test"],
        }


    def fetch_events(self, client, id, count,
            cat=None, nocat=None,
            tag=None, notag=None,
            group=None, nogroup=None):
Michal Kostenec's avatar
Michal Kostenec committed
       
        sqlwhere = []
        sqltemp = {}

        if cat is not None and nocat is not None:
            raise Error("Unrealizable conditions. Choose cat or nocat option.", 500, method='getEvents', 
                        exc=sys.exc_info(), detail={'cat': cat, 'nocat' : nocat})

        if cat is not None or nocat is not None:
            if cat is not None:
Michal Kostenec's avatar
Michal Kostenec committed
                parent_cats = []
                sqltemp['cat'] = generateDynamicQuery(self, "Category", "category_id IN (%s)", json.loads(cat), parent_cats)
                for pcats in parent_cats:
                    sqltemp['cat'] += " %s category_id DIV %s = 1 " % (("OR" if sqltemp['cat'] else ""), pcats)
Michal Kostenec's avatar
Michal Kostenec committed
            if nocat is not None:
Michal Kostenec's avatar
Michal Kostenec committed
                parent_cats = []
                sqltemp['cat'] = generateDynamicQuery(self, "Category", "category_id NOT IN (%s)", json.loads(nocat), parent_cats)
                for pcats in parent_cats:
                    sqltemp['cat'] += " %s category_id DIV %s = 1 " % (("OR" if sqltemp['cat'] else ""), pcats)
                
Michal Kostenec's avatar
Michal Kostenec committed
            sqlwhere.append("e.id IN (SELECT event_id FROM event_category_mapping WHERE %s)" % sqltemp['cat'])
Michal Kostenec's avatar
Michal Kostenec committed

        if tag is not None and notag is not None:
            raise Error("Unrealizable conditions. Choose tag or notag option.", 500, method='getEvents', 
                        exc=sys.exc_info(), detail={'tag': cat, 'notag' : nocat})

        if tag is not None or notag is not None:
            if tag is not None:
                sqltemp['tag'] = generateDynamicQuery(self, "Tag", "tag_id IN (%s)", json.loads(tag))
            if notag is not None:
                sqltemp['tag'] = generateDynamicQuery(self, "Tag", "tag_id NOT IN (%s)", json.loads(notag))
            
Michal Kostenec's avatar
Michal Kostenec committed
            sqlwhere.append("e.id IN (SELECT event_id FROM event_tag_mapping WHERE %s)" % sqltemp['tag'])
Michal Kostenec's avatar
Michal Kostenec committed

     
        if group is not None and nogroup is not None:       
            raise Error("Unrealizable conditions. Choose group or nogroup option.", 500, method='getEvents', 
                        exc=sys.exc_info(), detail={'tag': cat, 'notag' : nocat})

        if group is not None or nogroup is not None:
            sqltemp['group'] = ""

            if group is not None:
                for identity in json.loads(group):
                    sqltemp['group'] += ("cl.identity LIKE '%s' AND " % (identity)) 
            if nogroup is not None:
                for identity in json.loads(nogroup):
                    sqltemp['group'] += ("cl.identity NOT LIKE '%s' AND " % (identity)) 

            # logging.debug(sqltemp['group'][:-4])
            sqlwhere.append(sqltemp['group'][:-4])

        sqlwhere_string = (" AND " . join(sqlwhere))
        # logging.debug(sqlwhere_string)
        # logging.debug(' AND ' . join(sqlwhere))
    
        #sqlwhere = sqlwhere[:-4]
        and_op = "" if not sqlwhere_string else "AND"

Michal Kostenec's avatar
Michal Kostenec committed
        # logging.debug("SELECT e.id, e.data FROM clients cl RIGHT JOIN events e ON cl.id = e.client_id WHERE e.id > %s AND %s %s e.valid = 1 LIMIT %s" % (str(id), sqlwhere_string, and_op, str(count)))
        self.crs.execute("SELECT e.id, e.data FROM clients cl RIGHT JOIN events e ON cl.id = e.client_id WHERE e.id > %s AND %s %s e.valid = 1 LIMIT %s" % (str(id), sqlwhere_string, and_op, str(count)))
Michal Kostenec's avatar
Michal Kostenec committed
        row = self.crs.fetchall()

Michal Kostenec's avatar
Michal Kostenec committed
        if row:
            maxid = max(r['id'] for r in row)
        else:
            maxid = self.getLastEventId()

        # logging.debug("MAX ID = %s", str(maxid))

Michal Kostenec's avatar
Michal Kostenec committed
            "lastid": maxid,
            # "lastid": row[-1]['id'] if row else str(id),
Michal Kostenec's avatar
Michal Kostenec committed
            "events": [row[i]['data'] for i in range(len(row))]
        }


    def store_events(self, client, events):
        errs = []   # See sendEvents and validation, should return something similar
Michal Kostenec's avatar
Michal Kostenec committed
        
        for event in events:
            try:
Michal Kostenec's avatar
Michal Kostenec committed
                # logging.debug("INSERT INTO events (detected,received,client_id,data) VALUES ('%s', NOW(), '%s', '%s')" % (event['DetectTime'], client['id'], self.con.escape_string(str(event))))
                self.crs.execute("INSERT INTO events (detected,received,client_id,data) VALUES ('%s', NOW(), '%s', '%s')" % (event['DetectTime'], client['id'], self.con.escape_string(str(event))))
Michal Kostenec's avatar
Michal Kostenec committed
                lastid = self.crs.lastrowid
                # logging.debug(str(lastid))
                for cat in event['Category']:
                    # logging.debug({'cat': cat})
                    cat_id = self.map_id('Category', cat) if self.map_id('Category', cat) else self.map_id('Category', 'Other.Other')
                    # logging.debug({'cat_id': cat_id})
Michal Kostenec's avatar
Michal Kostenec committed
                    # logging.debug("INSERT INTO event_category_mapping (event_id,category_id) VALUES ('%s', '%s')" % (str(lastid), str(cat_id)))
                    self.crs.execute("INSERT INTO event_category_mapping (event_id,category_id) VALUES ('%s', '%s')" % (str(lastid), str(cat_id)))
Michal Kostenec's avatar
Michal Kostenec committed
                    
                for tag in event['Node'][0]['Tags']:
                    tag_id = self.map_id('Tag', tag) if self.map_id('Tag', tag) else self.map_id('Tag', 'Other')
                    # logging.debug({'tag_id': tag_id})
Michal Kostenec's avatar
Michal Kostenec committed
                    # logging.debug("INSERT INTO event_tag_mapping (event_id,tag_id) VALUES ('%s', '%s')" % (str(lastid), tag_id))
                    self.crs.execute("INSERT INTO event_tag_mapping (event_id,tag_id) VALUES ('%s', '%s')" % (str(lastid), str(tag_id)))    
Michal Kostenec's avatar
Michal Kostenec committed

                self.con.commit()
            except:
                self.con.rollback()
                raise Error("Data storing error", 00, detail={'event': event})
                errs.append({"event": event})
    
Michal Kostenec's avatar
Michal Kostenec committed
    def insertLastReceivedId(self, client, id):
        logging.debug("INSERT INTO last_events(client_id, event_id, timestamp) VALUES(%s, %s, NOW())" % (str(client[0]['id']), id))
        self.crs.execute("INSERT INTO last_events(client_id, event_id, timestamp) VALUES(%s, %s, NOW())" % (str(client[0]['id']), id))

    def getLastEventId(self):
Michal Kostenec's avatar
Michal Kostenec committed
        self.crs.execute("SELECT MAX(id) as id FROM events")
Michal Kostenec's avatar
Michal Kostenec committed
        row = self.crs.fetchone()

        return row['id'] if row['id'] is not None else 0

    def getLastReceivedId(self, client):
        logging.debug("IN getLastReceivedId")
        client_id = client[0]['id'] 
        logging.debug("getLastReceivedId (client_id) = %s", str(client_id))

        logging.debug("SELECT MAX(event_id) as id FROM last_events WHERE client_id = %s" % (str(client_id)))
        self.crs.execute("SELECT MAX(event_id) as id FROM last_events WHERE client_id = %s" % (str(client_id)))
        row = self.crs.fetchone()

        logging.debug("getLastReceivedId - %s" % str(row['id']))

        return row['id'] if row is not None else 0


    def map_id (self, section, key, strict = False):
Michal Kostenec's avatar
Michal Kostenec committed
        # Should by placed in config file
        data = {}
        data['Tag'] = {
                "Connection" : 1, 
                "Datagram" : 2, 
                "Content" : 3, 
                "Data" : 4, 
                "File" : 5, 
                "Flow" : 6, 
                "Log": 7, 
                "Protocol" : 8, 
                "Host" : 9, 
                "Network" : 10, 
                "Correlation" : 11, 
                "External" : 12, 
                "Reporting" : 13,
                "Other" : 99
                }

Michal Kostenec's avatar
Michal Kostenec committed
        data['Category'] = {
                        "Abusive" : 100,  
Michal Kostenec's avatar
Michal Kostenec committed
                        "Abusive.Spam" : 101,
                        "Abusive.Harassment" : 102,
                        "Abusive.Child" : 103,
                        "Abusive.Sexual" : 104,
                        "Abusive.Violence" : 105,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Malware" : 200,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Malware.Virus" : 201,
                        "Malware.Worm" : 202,
                        "Malware.Trojan" : 203,
                        "Malware.Spyware" : 204,
                        "Malware.Dialer" : 205,
                        "Malware.Rootkit" : 206,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Recon.Scanning" : 3,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Recon.Scanning" : 301,
                        "Recon.Sniffing" : 302,
                        "Recon.SocialEngineering" : 303,
                        "Recon.Searching" : 304,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Attempt" : 400,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Attempt.Exploit" : 401,
                        "Attempt.Login" : 402,
                        "Attempt.NewSignature" : 403,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Intrusion" : 500,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Intrusion.AdminCompromise" : 501,
                        "Intrusion.UserCompromise" : 502,
                        "Intrusion.AppCompromise" : 503,
                        "Intrusion.Botnet" : 504,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Availability" : 600,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Availability.DoS" : 601,
                        "Availability.DDoS" : 602,
                        "Availability.Sabotage" : 603,
                        "Availability.Outage" : 604,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Information" : 700,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Information.UnauthorizedAccess" : 701,
                        "Information.UnauthorizedModification" : 702,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Fraud" : 800,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Fraud.UnauthorizedUsage" : 801,
                        "Fraud.Copyright" : 802,
                        "Fraud.Masquerade" : 803,
                        "Fraud.Phishing" : 804,
                        "Fraud.Scam" : 805,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Vulnerable" : 900,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Vulnerable.Open" : 901,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Anomaly" : 1000,
Michal Kostenec's avatar
Michal Kostenec committed
                        "Anomaly.Traffic" : 1001,
                        "Anomaly.Connection" : 1002,
                        "Anomaly.Protocol" : 1003,
                        "Anomaly.System" : 1004,        
                        "Anomaly.Application" : 1005,
                        "Anomaly.Behaviour" : 1006,
                        "Other" : 9998,
                        "Test" : 9999,
                    }

        try:
            return data[section][key]
        except:
Michal Kostenec's avatar
Michal Kostenec committed
            #Return 0 for strict mode (searching), otherwise map everything else to 'Other'
            return 0 if strict else data[section]['Other']


def generateDynamicQuery(self, section, query_string, variables, parent_cats = []):
    variables_id = []
    # parent_cats = []
Michal Kostenec's avatar
Michal Kostenec committed

Michal Kostenec's avatar
Michal Kostenec committed
    for v in variables:
        mapped_id = self.map_id(section, v, True)
        if mapped_id % 100 != 0:
            variables_id.append(mapped_id)
        else:
            parent_cats.append(mapped_id)

    # variables_id = [self.map_id(section, v) for v in variables if self.map_id(section, v) % 100 != 0]
Michal Kostenec's avatar
Michal Kostenec committed
    format_strings = ','.join(['\'%s\''] * len(variables_id))
    temp_string = query_string % format_strings
    
    return temp_string % tuple(variables_id)

def expose(meth):
    meth.exposed = True
    return meth


class Server(Object):

    def __init__(self, auth, handler):
        self.auth = auth
        self.handler = handler


    def __str__(self):
        return "%s(auth=%s, handler=%s)" % (type(self).__name__, type(self.auth).__name__, type(self.handler).__name__)


    def sanitize_args(self, path, func, args, exclude=["self", "_env", "_client"]):
        # silently remove internal args, these should never be used
        # but if somebody does, we do not expose them by error message
        intargs = set(args).intersection(exclude)
        for a in intargs:
            del args[a]
        if intargs:
            logging.info("%s called with internal args: %s" % (path, ", ".join(intargs)))

        # silently remove surplus arguments - potential forward
        # compatibility (unknown args will get ignored)
        badargs = set(args)-set(func.func_code.co_varnames[0:func.func_code.co_argcount])
        for a in badargs:
            del args[a]
        if badargs:
            logging.info("%s called with superfluous args: %s" % (path, ", ".join(badargs)))

        return args


    def wsgi_app(self, environ, start_response, exc_info=None):
        path = environ.get("PATH_INFO", "").lstrip("/")
        output = ""
        status = "200 OK"
        headers = [('Content-type', 'application/json')]
        exception = None

        try:
            try:
                injson = environ['wsgi.input'].read()
            except:
                raise Error("Data read error", 400, method=path, exc=sys.exc_info())

            try:
                method = getattr(self.handler, path)
                method.exposed    # dummy access to trigger AttributeError
            except Exception:
                raise Error("You've fallen of the cliff.", 404, method=path)

            client = self.auth.authenticate(environ)
            if not client:
Michal Kostenec's avatar
Michal Kostenec committed
                raise Error("I'm watching YOU. (Authenticate)", 403, method=path)

            try:
                events = json.loads(injson) if injson else None
            except Exception:
                raise Error("Deserialization error", 400, method=path,
                    exc=sys.exc_info(), detail={"args": injson})

            args = parse_qs(environ.get('QUERY_STRING', ""))
            for k, v in args.iteritems():
                args[k] = v[0]
            logging.debug("%s called with %s" % (path, str(args)))
            if events:
                args["events"] = events

Michal Kostenec's avatar
Michal Kostenec committed
            # if not self.auth.authorize(environ, client, path, args):
            #     raise Error("I'm watching YOU.", 403, method=path, detail={"client": client})

            args = self.sanitize_args(path, method, args)
            result = method(_env=environ, _client=client, **args)   # call requested method

            try:
            # 'default': takes care of non JSON serializable objects,
            # which could (although shouldn't) appear in handler code
                output = json.dumps(result, default=lambda v: str(v))
            except Exception as e:
                raise Error("Serialization error", 500, method=path,
                    exc=sys.exc_info(), detail={"args": str(result)})

        except Error as e:
            exception = e
        except Exception as e:
            exception = Error("Server exception", 500, method=path, exc=sys.exc_info())

        if exception:
            status = "%d %s" % (exception.error, exception.message)
            result = exception.to_dict()
            try:
                output = json.dumps(result, default=lambda v: str(v))
            except Exception as e:
                # Here all bets are off, generate at least sane output
                output = '{"error": %d, "message": "%s"}' % (
                    exception.error, exception.message)

            logging.error(str(exception))
            i = exception.info_str()
            if i:
                logging.info(i)
            d = exception.debug_str()
            if d:
                logging.debug(d)

        headers.append(('Content-Length', str(len(output))))
        start_response(status, headers)
        return [output]


    __call__ = wsgi_app



class WardenHandler(Object):

Michal Kostenec's avatar
Michal Kostenec committed
    def __init__(self, validator, db, auth,
            send_events_limit=100000, get_events_limit=100000,
            description=None):

Michal Kostenec's avatar
Michal Kostenec committed
        self.auth = auth
        self.db = db
        self.validator = validator
        self.send_events_limit = send_events_limit
        self.get_events_limit = get_events_limit
        self.description = description


    def __str__(self):
        return "%s(validator=%s, db=%s, send_events_limit=%s, get_events_limit=%s, description=\"%s\")" % (
            type(self).__name__, type(self.validator).__name__, type(self.db).__name__,
            self.get_events_limit, self.send_events_limit, self.description)


    @expose
    def getDebug(self, _env, _client):
Michal Kostenec's avatar
Michal Kostenec committed
        auth = self.auth.authorize(_env, _client, 'getDebug', None, None)
        if not auth:
            raise Error("I'm watching YOU. (Authorization)", 403, method='getDebug', detail={"client": _client})
        
Pavel Kácha's avatar
Pavel Kácha committed
        return {
            "environment": _env,
            "database": self.db.get_debug()
        }


    @expose
    def getInfo(self, _env, _client):
        info = {
            "version": VERSION,
            "send_events_limit": self.send_events_limit,
            "get_events_limit": self.get_events_limit
        }
        if self.description:
            info["description"] = self.description
        return info


    @expose
    def getEvents(self, _env, _client, id=None, count=None,
            cat=None, nocat=None,
            tag=None, notag=None,
            group=None, nogroup=None):

        try:
            id = int(id)
        except (ValueError, TypeError):
            id=0

Michal Kostenec's avatar
Michal Kostenec committed
        if id == 0: 
            try:
                id = self.db.getLastReceivedId(_client)
                # logging.debug("Last received ID for %s is %s" % (_client['hostname'], str(id)))
            except Exception, e:
                logging.error(e)
                id = 0
                
        if id == 0:
            try:
                id = self.db.getLastEventId()
            except Exception as e:
                raise Error("Last event id receiving error", 500, detail={"client": _client})
  
            self.db.insertLastReceivedId(_client, id)

            return {
                "lastid": id,
                "events": []
            }

        try:
            count = int(count)
        except (ValueError, TypeError):
            count = 1

        if self.get_events_limit:
            count = min(count, self.get_events_limit)

        logging.debug("getEvents - count: %s" % count)
        res = self.db.fetch_events(_client, id, count, cat, nocat, tag, notag, group, nogroup)
        logging.info("getEvents(%d, %d, %s, %s, %s, %s, %s, %s): sending %d events" % (
            id, count, cat, nocat, tag, notag, group, nogroup, len(res["events"])))

Michal Kostenec's avatar
Michal Kostenec committed
        self.db.insertLastReceivedId(_client, res['lastid'])
        logging.debug("lastid inserting: %s" % {'lastid': res['lastid'], 'client' : _client})

        return res


    @expose
    def sendEvents(self, _env, _client, events=[]):
        if not isinstance(events, list):
            raise Error("List of events expected", 400, method="sendEvents")

        if len(events)>self.send_events_limit:
            raise Error("Too much events in one batch", 400, method="sendEvents",
                detail={"limit": self.send_events_limit})

        # FIXME: Maybe just croak on first bad event, save good ones so far
        # and make client deal with the rest? Would simplify server error
        # handling greatly.
        okevents = []
        valerrs = []
        for event in events:
Michal Kostenec's avatar
Michal Kostenec committed
            auth_cl = self.auth.authorize(_env, _client, 'sendEvents', event, None)
            if not auth_cl:
                raise Error("I'm watching YOU. (Authorization)", 403, method='sendEvents', detail={"client": _client})

            verrs = self.validator.check(event)
            if verrs:
                valerrs.append({"errors": verrs, "event": event})
            else:
                okevents.append(event)

Michal Kostenec's avatar
Michal Kostenec committed
        dberrs = self.db.store_events(auth_cl, okevents)

        if valerrs or dberrs:
            raise Error("Event storage error", 500, method="sendEvents",
                detail=valerrs+dberrs)

        logging.info("sendEvents(...): Saved %i events" % len(okevents))

        return {"saved": len(okevents)}



def read_ini(path):
    c = ConfigParser.RawConfigParser()
    res = c.read(path)
    if not res or not path in res:
        # We don't have loggin yet, hopefully this will go into webserver log
        raise Error("Unable to read config: %s" % path)
    data = {}
    for sect in c.sections():
        for opts in c.options(sect):
            lsect = sect.lower()
            if not lsect in data:
                data[lsect] = {}
            data[lsect][opts] = c.get(sect, opts)
    return data


def read_cfg(path):
    with open(path, "r") as f:
        stripcomments = "\n".join((l for l in f if not l.lstrip().startswith("#")))
        conf = json.loads(stripcomments)

    # Lowercase keys
    conf = dict((sect.lower(), dict(
        (subkey.lower(), val) for subkey, val in subsect.iteritems())
    ) for sect, subsect in conf.iteritems())

    return conf


def fallback_wsgi(environ, start_response, exc_info=None):

    # If server does not start, set up simple server, returning
    # Warden JSON compliant error message
    error=503
    message="Server not running due to initialization error"
    headers = [('Content-type', 'application/json')]

    logline = "Error(%d): %s" % (error, message)
    status = "%d %s" % (error, message)
    output = '{"error": %d, "message": "%s"}' % (
        error, message)

    logging.critical(logline)
    start_response(status, headers)
    return [output]


def build_server(conf):

    # Functions for validation and conversion of config values
    def facility(name):
        return int(getattr(logging.handlers.SysLogHandler, "LOG_" + name.upper()))

    def loglevel(name):
        return int(getattr(logging, name.upper()))

    def natural(name):
        num = int(name)
        if num<1:
            raise ValueError("Not a natural number")
        return num

    def filepath(name):
        # Make paths relative to dir of this script
        return path.join(path.dirname(__file__), name)

    def objdef(name):
        return objects[name.lower()]

    obj = objdef    # Draw into local namespace for init_obj

    objects = {}    # Already initialized objects

    # List of sections and objects, configured by them
    # First object in each object list is the default one, otherwise
    # "type" keyword in section may be used to choose other
    section_def = {
        "log": ["FileLogger", "SysLogger"],
Pavel Kácha's avatar
Pavel Kácha committed
        "db": ["MySQL"],
        "auth": ["X509Authenticator", "NoAuthenticator"],
        "validator": ["JSONSchemaValidator", "NoValidator"],
        "handler": ["WardenHandler"],
        "server": ["Server"]
    }

    # Object parameter conversions and defaults
    param_def = {
        "FileLogger": {
            "filename": {"type": filepath, "default": path.join(path.dirname(__file__), path.splitext(path.split(__file__)[1])[0] + ".log")},
            "level": {"type": loglevel, "default": "info"},
        },
        "SysLogger": {
            "socket": {"type": filepath, "default": "/dev/log"},
            "facility": {"type": facility, "default": "daemon"},
            "level": {"type": loglevel, "default": "info"}
        },
        "NoAuthenticator": {},
        "X509Authenticator": {
            "db": {"type": obj, "default": "db"}
        },
        "NoValidator": {},
        "JSONSchemaValidator": {
            "filename": {"type": filepath, "default": path.join(path.dirname(__file__), "idea.schema")}
        },
Pavel Kácha's avatar
Pavel Kácha committed
        "MySQL": {
            "host": {"type": str, "default": "localhost"},
            "user": {"type": str, "default": "warden"},
            "password": {"type": str, "default": ""},
Michal Kostenec's avatar
Michal Kostenec committed
            "dbname": {"type": str, "default": "warden3"},
Pavel Kácha's avatar
Pavel Kácha committed
            "port": {"type": natural, "default": 3306}
        },
        "WardenHandler": {
            "validator": {"type": obj, "default": "validator"},
            "db": {"type": obj, "default": "DB"},
Michal Kostenec's avatar
Michal Kostenec committed
            "auth": {"type": obj, "default": "auth"},
            "send_events_limit": {"type": natural, "default": 10000},
            "get_events_limit": {"type": natural, "default": 10000},
            "description": {"type": str, "default": ""}
        },
        "Server": {
            "auth": {"type": obj, "default": "auth"},
            "handler": {"type": obj, "default": "handler"}
        }
    }

    def init_obj(sect_name):
        config = conf.get(sect_name, {})
        sect_name = sect_name.lower()
        sect_def = section_def[sect_name]

Michal Kostenec's avatar
Michal Kostenec committed
        #logging.debug("Testing %s" % sect_name)

        try:    # Object type defined?
            objtype = config["type"]
            del config["type"]
        except KeyError:    # No, fetch default object type for this section
            objtype = sect_def[0]
        else:
            if not objtype in sect_def:
                raise KeyError("Unknown type %s in section %s" % (objtype, sect_name))

        params = param_def[objtype]

        # No surplus parameters? Disallow also 'obj' attributes, these are only
        # to provide default referenced section
        for name in config:
            if name not in params or (name in params and params[name]["type"] is objdef):
                raise KeyError("Unknown key %s in section %s" % (name, sect_name))

        # Process parameters
        kwargs = {}
        for name, definition in params.iteritems():
            raw_val = config.get(name, definition["default"])
            try:
                val = definition["type"](raw_val)
            except Exception:
                raise KeyError("Bad value \"%s\" for %s in section %s" % (raw_val, name, sect_name))
            kwargs[name] = val

        cls = globals()[objtype]   # get class/function type
        try:
            obj = cls(**kwargs)         # run it
        except Exception as e:
            raise KeyError("Cannot initialize %s from section %s: %s" % (
                objtype, sect_name, str(e)))

        if isinstance(obj, Object):
            # Log only objects here, functions must take care of themselves
            logging.info("Initialized %s" % str(obj))

        objects[sect_name] = obj
        return obj

    # Init logging with at least simple stderr StreamLogger
    # Dunno if it's ok within wsgi, but we have no other choice, let's
    # hope it at least ends up in webserver error log
    StreamLogger()

    try:
        # Now try to init required objects
        for o in ("log", "db", "auth", "validator", "handler", "server"):
            init_obj(o)
    except Exception as e:
        logging.critical(str(e))
        logging.debug("", exc_info=sys.exc_info())
        return fallback_wsgi

    logging.info("Ready to serve")

    return objects["server"]


if __name__=="__main__":
    # FIXME: just development stuff
    srv = build_server(read_ini("warden3.cfg.wheezy-warden3"))