pampa_workers 0.0.38

data/lib/mycrawlprocess.rb

@@ -0,0 +1,49 @@
+module BlackStack
+  
+  # process class
+  class MyCrawlProcess < BlackStack::MyLocalProcess
+  
+    attr_accessor :nErrors, :nSuccesses, :browser, :proxy, :bot
+    
+    # constructor
+    def initialize(
+      the_worker_name, 
+      the_division_name, 
+      the_minimum_enlapsed_seconds=MyProcess::DEFAULT_MINIMUM_ENLAPSED_SECONDS, 
+      the_verify_configuration=true,
+      the_email=nil, 
+      the_password=nil
+    )
+      super(the_worker_name, the_division_name, the_minimum_enlapsed_seconds, the_verify_configuration, the_email, the_password)
+      self.nErrors = 0
+      self.nSuccesses = 0
+      self.browser = nil
+      self.proxy = nil
+      self.bot = nil
+    end
+    
+    def canRun?()
+      super &&
+      nErrors < Params.getValue("crawl.company.discretion.max_errors") && 
+      nSuccesses < Params.getValue("crawl.company.discretion.max_successes")
+      #(Params.getValue("crawl.company.use_proxy")==false || Company.availableProxiesWithDiscretionForCrawl() > 0)
+    end
+    
+    def whyCantRun()
+      ret = super
+      if (ret.to_s.size == 0)
+        if (self.nErrors >= Params.getValue("crawl.company.discretion.max_errors")) 
+          return "Reached the max number of errors (#{self.nErrors.to_s})"
+        end
+  
+        if (self.nSuccesses >= Params.getValue("crawl.company.discretion.max_successes"))      
+          return "Reached the max number of successes (#{self.nSuccesses.to_s})"
+        end
+      end
+      return ret
+    end
+    
+  end # class MyCrawlProcess
+
+
+end # module BlackStack

data/lib/mylocalprocess.rb

@@ -0,0 +1,164 @@
+module BlackStack
+
+  # Process located in the same LAN than the Database Server
+  class MyLocalProcess < BlackStack::MyChildProcess
+    
+    # constructor
+    def initialize(
+      the_worker_name, 
+      the_division_name, 
+      the_minimum_enlapsed_seconds=BlackStack::MyProcess::DEFAULT_MINIMUM_ENLAPSED_SECONDS, 
+      the_verify_configuration=true,
+      the_email=nil, 
+      the_password=nil
+    )
+      super(the_worker_name, the_division_name, the_minimum_enlapsed_seconds, the_verify_configuration, the_email, the_password)
+    end
+  
+    def division()
+      if (self.division_name != "local")
+        d = BlackStack::Division.where(:name=>self.division_name).first
+        if (d!=nil)
+          return BlackStack::Division.where(:db_name=>d.db_name, :home=>true).first
+        else
+          return nil
+        end
+      else
+        return BlackStack::Division.where(:central=>true).first
+      end
+    end
+  
+    def worker()
+      BlackStack::Worker.where(:name=>self.fullWorkerName).first
+    end
+      
+  
+    # update worker configuration in the division
+    def updateWorker()
+      w = BlackStack::Worker.where(:name=>self.fullWorkerName).first
+      if (w==nil)
+        w = BlackStack::Worker.new
+        w.id = guid()
+        w.process = ''
+        w.last_ping_time = now()
+        w.name = self.fullWorkerName
+        w.assigned_process = self.assigned_process
+        w.id_object = self.id_object
+        w.division_name = self.division_name
+        w.save
+      end
+      if (w!=nil)
+        w.assigned_process = self.assigned_process
+        w.id_object = self.id_object
+        w.division_name = self.division_name
+        w.id_division = self.id_division
+        w.save
+      end
+    end
+  
+    def run()
+        super
+  
+        # creo el objeto logger
+        self.logger = RemoteLogger.new(
+          "#{self.fullWorkerName}.log",
+          BlackStack::Pampa::api_protocol, 
+          BlackStack::Pampa::api_domain, 
+          BlackStack::Pampa::api_port, 
+          BlackStack::Pampa::api_key
+        )
+
+        # announcing my in the log
+        logger.log "Child process is alive!"
+  
+        # obtengo los parametros del worker
+        logger.logs "Update from central (1)... "
+        self.get
+        logger.done
+  
+        # actualizo los datos de este worker (parent process)
+        logger.logs "Update worker (1)... "
+        self.updateWorker
+        logger.done
+  
+        while (self.canRun?)
+          # reseteo en contador nested del logger
+          self.logger.reset()
+  
+          # announcing my in the log
+          logger.log "Going to Run Local"
+          logger.log "Process: #{self.assigned_process.to_s}."
+          logger.log "Object: #{(self.id_object.to_s.size==0)? 'n/a' : self.id_object.to_s})"
+  
+          # obtengo la hora de inicio
+          start_time = Time.now
+  
+          begin
+            # libero recursos
+            logger.logs "Release resources... "
+            GC.start
+            DB.disconnect
+            logger.done
+  
+            # cargo el objeto worker
+            logger.logs "Load the worker... "
+            the_worker = self.worker
+            logger.done
+  
+            # actualizo el valor del proceso que corre actualmente para este worker
+            logger.logs "Update current process... "
+            the_worker.process=self.assigned_process
+  		      the_worker.active = true
+            the_worker.save()
+            logger.done
+            
+            logger.logs "Ping... "
+            the_worker.ping()
+            logger.done
+            
+            # corro el procesamiento
+            self.process(ARGV)
+            
+          rescue => e
+            puts ""
+            logger.log "Local Process Error: " + e.to_s + "\r\n" + e.backtrace.join("\r\n").to_s
+          end
+  
+          # obtengo los parametros del worker
+          logger.logs "Update from central (2)... "
+          self.get
+          logger.done
+  
+          # actualizo los datos de este worker (parent process)
+          logger.logs "Update worker (2)... "
+          self.updateWorker
+          logger.done
+  
+          # sleep
+          logger.logs "Sleep... "
+          self.doSleep(start_time)
+          logger.done
+  
+          logger.log "-------------------------------------------"
+  
+          DB.disconnect
+          GC.start
+        end # main while
+  
+        # 
+        logger.log "Process Finish!"
+        logger.log "Finish Reason: " + self.whyCantRun.to_s
+  
+        #
+        logger.logs "Disconnect to Database... "
+        begin
+          DB.disconnect()
+          logger.done
+        rescue => e
+          logger.error(e)
+        end
+    end # run
+    
+  end # class MyLocalProcess
+
+end # module BlackStack

data/lib/myparentprocess.rb

@@ -0,0 +1,141 @@
+module BlackStack
+
+  # es un proceso sin conexion a base de datos, que itera infinitamente.
+  # en cada iteracion saluda a la central (hello), obtiene parametros (get)
+  class MyParentProcess < BlackStack::MyProcess
+    def run()
+        super
+        
+        # creo el objeto logger
+        self.logger = BlackStack::RemoteLogger.new(
+          "#{self.fullWorkerName}.log",
+          BlackStack::Pampa::api_protocol, 
+          BlackStack::Pampa::api_domain, 
+          BlackStack::Pampa::api_port, 
+          BlackStack::Pampa::api_key
+        )
+  
+        # 
+        pid = nil
+        while (true)
+          begin
+            GC.start # 331 - avoid lack of memory
+            #DB.disconnect # este proceso esta desacoplado de la conexion a la base de datos
+  
+            # reseteo en contador nested del logger
+            self.logger.reset()
+  
+            # get the start time 
+            start_time = Time.now
+            
+            # consulto a la central por la division asignada
+            url = "#{BlackStack::Pampa::api_url}/api1.3/pampa/hello.json"
+            logger.logs("Hello to the central... ")
+            res = BlackStack::Netting::call_post(url, {
+              'api_key' => BlackStack::Pampa::api_key, 
+              'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+            )
+            parsed = JSON.parse(res.body)
+            if (parsed['status'] != BlackStack::Netting::SUCCESS)
+              self.logger.logf("Error: " + parsed['status'].to_s)
+            else
+              self.logger.done
+  
+              logger.logs("Get worker data... ")
+              url = "#{BlackStack::Pampa::api_url}/api1.3/pampa/get.json"
+              res = BlackStack::Netting::call_post(url, {
+                'api_key' => BlackStack::Pampa::api_key, 
+                'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+              )
+              parsed = JSON.parse(res.body)
+              if (parsed['status'] != BlackStack::Netting::SUCCESS)
+                self.logger.logf("Error: " + parsed['status'].to_s)
+              else
+                # map response
+                self.id                 = parsed['id']
+                self.assigned_process   = parsed['assigned_process']
+                self.id_object          = parsed['id_object']
+                self.id_division        = parsed['id_division']
+                self.division_name      = parsed['division_name']
+                self.ws_url             = parsed['ws_url']
+                self.ws_port            = parsed['ws_port']
+                self.logger.logf "done (#{self.division_name})"
+  
+                # 
+                self.logger.logs "Notify division... "
+                if self.division_name.to_s.size == 0
+                  self.logger.logf "no division assigned"                                
+                else
+                  self.notify # notifico a la division
+                  self.logger.done
+  
+                  # 
+                  self.logger.logs "Spawn child process... "    
+                  # lanzo el proceso
+                  if self.assigned_process.to_s.size > 0
+                    command = "ruby #{self.assigned_process} name=#{self.worker_name} division=#{self.division_name}"
+                    pid = Process.spawn(command)
+                    logger.logf "done (pid=#{pid.to_s})" 
+                    
+                    logger.log("Wait to child process to finish.")
+                    Process.wait(pid)
+                  else
+                    if self.assigned_process.to_s.size == 0
+                      self.logger.logf "no process assigned"                                
+                    end
+                  end # if self.assigned_process.to_s.size > 0
+                end # if self.division_name.to_s.size == 0
+              end # if (parsed['status'] != "success") <-- #{BlackStack::Pampa::api_url}/api1.3/pampa/get.json
+            end # if (parsed['status'] != "success") <-- #{BlackStack::Pampa::api_url}/api1.3/pampa/hello.json
+  
+            #  
+            logger.logs "Sleep... "
+            self.doSleep(start_time)
+            logger.done
+  
+            logger.log "-------------------------------------------"
+          
+          rescue Interrupt => e
+            logger.reset
+            
+            logger.log "Interrupt signal!"
+    
+            logger.logs "Kill process... "        
+            if (pid!=nil)
+              system("taskkill /im #{pid.to_s} /f /t >nul 2>&1")
+            end
+            logger.done
+            
+            logger.logs "Disconnect to Database... "
+            begin
+    #          DB.disconnect()
+              logger.done
+            rescue => e
+              logger.error(e)
+            end
+            
+            logger.log "Process is out."
+            exit(0)
+    
+          rescue => e
+            begin
+              logger.log "Unhandled exception: #{e.to_s}\r\n#{e.backtrace.join("\r\n").to_s}"
+              logger.logs "Sleep #{self.minimum_enlapsed_seconds.to_s} seconds... "
+              sleep(self.minimum_enlapsed_seconds)
+              logger.done
+            rescue => e
+              puts "Fatal error: #{e.to_s}"
+              print "Sleep #{self.minimum_enlapsed_seconds.to_s} seconds... "
+              sleep(self.minimum_enlapsed_seconds)
+              puts          
+            end
+    
+          end # rescue
+  
+        end # while
+  
+    end # def run()
+    
+  end # class MyParentProcess
+
+end # module BlackStack

data/lib/myprocess.rb

@@ -0,0 +1,264 @@
+module BlackStack
+
+  class MyProcess
+    DEFAULT_MINIMUM_ENLAPSED_SECONDS = 60
+    
+    attr_accessor :assigned_process_changed, :assigned_division_changed, :verify_configuration
+    attr_accessor :logger, :id, :worker_name, :division_name, :minimum_enlapsed_seconds, :assigned_process, :id_object, :id_division, :ws_url, :ws_port
+    attr_accessor :email, :password
+  
+    # constructor
+    def initialize(
+        the_worker_name, 
+        the_division_name, 
+        the_minimum_enlapsed_seconds=MyProcess::DEFAULT_MINIMUM_ENLAPSED_SECONDS, 
+        the_verify_configuration=true,
+        the_email=nil, 
+        the_password=nil
+    )
+      self.assigned_process_changed = false
+      self.assigned_division_changed = false
+      self.assigned_process = File.expand_path($0)
+      self.worker_name = "#{the_worker_name}" 
+      self.division_name = the_division_name
+      self.minimum_enlapsed_seconds = the_minimum_enlapsed_seconds
+      self.verify_configuration = the_verify_configuration
+      self.email = the_email
+      self.password = the_password
+    end
+  
+    # retrieves the id of the current process
+    def pid()
+      Process.pid.to_s
+    end
+  
+    # Retorna un array de hashes.
+    # => Cada elemento del hash tiene la forma: {:executablepath, :pid, :ppid},
+    # => donde imagename es el patch completo del proceso, pid es el id del proceso
+    # => y ppid es el id del proceso padre.
+    def list()
+      a = []
+      s = `wmic process get executablepath,processid,parentprocessid`
+      s.split(/\n+/).each { |e|
+        aux = e.strip.scan(/^(.+)\s+(\d+)\s+(\d+)$/)[0]
+        if (aux!=nil)
+          if (aux.size>=3)
+            a << {
+              :executablepath => aux[0].strip.to_s,
+              :pid => aux[2].to_s, # TODO: deberia ser aux[1], pero por algo que no entiendo ahora el pid viene en aux[2]
+              :ppid => aux[1].to_s, # TODO: deberia ser aux[2], pero por algo que no entiendo ahora el pid viene en aux[1]
+            }
+          end
+        end
+      }
+      a
+    end
+  
+    # ejecuta TASKKILL /F /PID #{the_pid} y retorna el output del comando
+    def self.kill(the_pid)
+      system("TASKKILL /F /PID #{the_pid}")
+    end
+  
+    # obtiene la diferencia en segundos entre la hora actual y el parametro the_start_time.
+    # si la diferencia es mayor al atributo minimum_enlapsed_seconds, entonces duerme el tiempo restante.
+    def doSleep(the_start_time)
+      # si el proceso tardo menos del minimum_enlapsed_seconds, entonces duermo el tiempo restante
+      end_time = Time.now
+      elapsed_seconds = end_time - the_start_time # in seconds
+      if (elapsed_seconds < self.minimum_enlapsed_seconds)
+        sleep_seconds = self.minimum_enlapsed_seconds - elapsed_seconds 
+        sleep(sleep_seconds)
+      end
+    end
+  
+    # This function works in windows only
+    # TODO: Esta funcion no retorna la mac address completa
+    # TODO: Validar que no se retorne una macaddress virtual, con todos valores en 0
+    def self.macaddress()
+      BlackStack::SimpleHostMonitoring.macaddress
+    end
+  
+    def self.fullWorkerName(name)
+      "#{Socket.gethostname}.#{MyProcess.macaddress}.#{name}"
+    end
+  
+    def fullWorkerName()
+      MyProcess.fullWorkerName(self.worker_name)
+    end
+  
+    # saluda a la central
+    def hello()
+      # me notifico a la central. obtengo asignacion si ya la tenia
+      url = "#{BlackStack::Pampa::api_url}/api1.3/pampa/hello.json"
+      res = BlackStack::Netting::call_post(url, {
+        'api_key' => BlackStack::Pampa::api_key, 
+        'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+      )
+      parsed = JSON.parse(res.body)
+      if (parsed['status'] != BlackStack::Netting::SUCCESS)
+        raise parsed['status'].to_s
+      end
+    end # hello()
+  
+    # notifico mis parametros (assigned_process, id_object) a la division asignada
+    def set(new_assigned_process, new_id_object)
+      if (self.ws_url.to_s.size > 0 && self.ws_port.to_s.size > 0)
+        url = "#{BlackStack::Pampa::api_protocol}://#{self.ws_url.to_s}:#{self.ws_port.to_s}/api1.3/pampa/notify.json"
+        res = BlackStack::Netting::call_post(url, {
+          'api_key' => BlackStack::Pampa::api_key, 
+          'name' => self.fullWorkerName,
+          'assigned_process' => new_assigned_process,
+          'id_object' => new_id_object }.merge( BlackStack::RemoteHost.new.poll )
+        )
+      end    
+    end
+  
+    # obtiene sus parametros de la central
+    def get()
+      # me notifico a la central. obtengo asignacion que tenga
+      url = "#{BlackStack::Pampa::api_url}/api1.3/pampa/get.json"
+      res = BlackStack::Netting::call_post(url, {
+        'api_key' => BlackStack::Pampa::api_key, 
+        'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+      )
+      parsed = JSON.parse(res.body)
+      if (parsed['status'] != BlackStack::Netting::SUCCESS)
+        raise parsed['status'].to_s
+      else 
+        if self.verify_configuration
+          # si ya tenia un proceso asignado, y ahora se le asigna un nuevo proceso 
+          if self.assigned_process.to_s.size > 0
+            a = File.expand_path(self.assigned_process)
+            b = File.expand_path(parsed['assigned_process'])
+            if a != b
+              self.assigned_process_changed = true
+            else
+              self.assigned_process_changed = false
+            end
+          end
+    
+          # si ya tenia un proceso asignado, y ahora se le asigna un nuevo proceso 
+          if self.id_division.to_s.size > 0
+            if self.id_division.to_guid != parsed['id_division'].to_guid
+              self.assigned_division_changed = true
+            else
+              self.assigned_division_changed = false
+            end
+          end
+        end # verify_configuration
+              
+        # si ya tenia asignada una division, entonces le notifico mi nueva configuracion
+        self.set(parsed['assigned_process'], parsed['id_object'])
+  
+        self.id                 = parsed['id']
+        self.assigned_process   = parsed['assigned_process']
+        self.id_object          = parsed['id_object']
+        self.id_division        = parsed['id_division']
+        self.division_name      = parsed['division_name']
+        self.ws_url             = parsed['ws_url']
+        self.ws_port            = parsed['ws_port']      
+  
+        # le notifico a la nueva division asignada mi nueva configuracion
+        self.set(parsed['assigned_process'], parsed['id_object'])
+      end
+    end # get()
+  
+  
+    # update worker configuration in the division
+    def updateWorker()
+      raise "Abstract Method."
+    end
+  
+    # ping the central database
+    def ping()
+      # me notifico a la central.
+      url = "#{BlackStack::Pampa::api_url}/api1.3/pampa/ping.json"
+      res = BlackStack::Netting::call_post(url, {
+        'api_key' => BlackStack::Pampa::api_key, 
+        'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+      )
+      parsed = JSON.parse(res.body)
+      if (parsed['status'] != BlackStack::Netting::SUCCESS)
+        raise parsed['status'].to_s
+      end
+  
+      # me notifico a la division.
+      if (self.ws_url != nil && self.ws_port != nil)
+        url = "#{BlackStack::Pampa::api_protocol}://#{self.ws_url.to_s}:#{self.ws_port.to_s}/api1.3/pampa/ping.json"
+        res = BlackStack::Netting::call_post(url, {
+          'api_key' => BlackStack::Pampa::api_key, 
+          'name' => self.fullWorkerName }.merge( BlackStack::RemoteHost.new.poll )
+        )
+        parsed = JSON.parse(res.body)
+        if (parsed['status'] != "success")
+          raise parsed['status'].to_s
+        end
+      end # if
+    end # ping()
+  
+    # se notifica al dispatcher de la division
+    def notify()
+      if (self.ws_url==nil || self.ws_port==nil)
+        raise "Cannot notify. Worker has not parameters."
+      end
+              
+      # me notifico a la division. obtengo trabajo
+      url = "#{BlackStack::Pampa::api_protocol}://#{self.ws_url}:#{self.ws_port}/api1.3/pampa/notify.json"
+      res = BlackStack::Netting::call_post(url, 
+        {
+        'api_key' => BlackStack::Pampa::api_key, 
+        'name' => self.fullWorkerName, 
+        'assigned_process' => self.assigned_process,
+        'id_object' => self.id_object }.merge( BlackStack::RemoteHost.new.poll )
+      )
+      parsed = JSON.parse(res.body)
+      if (parsed['status'] != "success")
+        raise parsed['status'].to_s
+      end
+    end
+  
+    # Get the data object of the divison assigned to this worker.
+    # Needs database connections. So it's available for ChildProcess only.
+    def division()
+      raise "This is an abstract method."
+    end
+  
+    # Get the data object of worker linked to this process.
+    # Needs database connections. So it's available for ChildProcess only.
+    def worker()
+      raise "This is an abstract method."
+    end
+    
+    # retorna true si el proceso hijo (child) esta habilitado para trabajar.
+    def canRun?()
+      self.assigned_process_changed == false && 
+      self.assigned_division_changed == false
+    end
+  
+    def whyCantRun()
+      if self.assigned_process_changed == true
+        return "Assigned process has changed." 
+      elsif self.assigned_division_changed == true
+        return "Assigned division has changed." 
+      else
+        return "unknown"
+      end
+    end
+  
+    # este metodo 
+    # ejecuta el trabajo para el que fue creado el objeto.
+    def process(argv)
+      raise "This is an abstract method."
+    end
+  
+    # ejecuta el proceso, en modo parent, bot o child segun la clase que se implemente.
+    # en modo parent, hace un loop infinito.
+    # en modo bot o child, hace un loop hasta que el metodo canRun? retorne false.
+    # en modo bot o child, invoca al metodo process() en cada ciclo.
+    def run()
+      #raise "This is an abstract method"
+    end # run
+    
+  end # class MyProcess
+
+end # module BlackStack