Пример #1
0
    def __init__(self, verbose=False, quiet=True):

        ETL.__init__(self, verbose=verbose)

        self.read_configfiles()

        self.config["plugins"] = []
Пример #2
0
    def __init__(self, verbose=False):

        ETL.__init__(self, verbose=verbose)

        self.verbose = verbose

        self.read_configfiles()

        # Watched events
        #
        # We need IN_MOVE_SELF to track moved folder paths
        # pyinotify-internally. If omitted, the os instructions
        # mv /docs/src /docs/dest; touch /docs/dest/doc.pdf
        # will produce a IN_MOVED_TO pathname=/docs/dest/ followed by
        # IN_CLOSE_WRITE pathname=/docs/src/doc.pdf
        # where we would like a IN_CLOSE_WRITE pathname=/docs/dest/doc.pdf
        self.mask = (pyinotify.IN_DELETE
                     | pyinotify.IN_CLOSE_WRITE
                     | pyinotify.IN_MOVED_TO
                     | pyinotify.IN_MOVED_FROM
                     | pyinotify.IN_MOVE_SELF)

        self.watchmanager = pyinotify.WatchManager()  # Watch Manager

        self.handler = EventHandler()

        self.notifier = pyinotify.Notifier(self.watchmanager, self.handler)
	def __init__(self, verbose=False, quiet=True):

		ETL.__init__(self, verbose=verbose)

		self.read_configfiles()

		self.config["plugins"] = []
Пример #4
0
    def __init__(self, verbose=False, quiet=True):

        ETL.__init__(self, verbose=verbose)

        self.quiet = quiet

        self.set_configdefaults()

        self.read_configfiles()
	def __init__(self, verbose=False, quiet=True):

		ETL.__init__(self, verbose=verbose)

		self.quiet = quiet

		self.set_configdefaults()
		
		self.read_configfiles()
	def __init__(self, verbose = False ):

		ETL.__init__(self, verbose=verbose)

		self.verbose=verbose

		self.read_configfiles()
		
		self.mask = pyinotify.IN_DELETE | pyinotify.IN_CLOSE_WRITE | pyinotify.IN_MOVED_TO | pyinotify.IN_MOVED_FROM  # watched events

		self.watchmanager = pyinotify.WatchManager()  # Watch Manager

		self.handler = EventHandler()
		
		self.notifier = pyinotify.Notifier(self.watchmanager, self.handler)
Пример #7
0
    def __init__(self, verbose=False):

        ETL.__init__(self, verbose=verbose)

        self.verbose = verbose

        self.read_configfiles()

        self.mask = pyinotify.IN_DELETE | pyinotify.IN_CLOSE_WRITE | pyinotify.IN_MOVED_TO | pyinotify.IN_MOVED_FROM  # watched events

        self.watchmanager = pyinotify.WatchManager()  # Watch Manager

        self.handler = EventHandler()

        self.notifier = pyinotify.Notifier(self.watchmanager, self.handler)
Пример #8
0
	def __init__(self, 	plugins=[], verbose=False ):


		ETL.__init__(self, plugins=plugins, verbose=verbose)




		self.read_configfile ('/etc/etl/config')
		self.read_configfile ('/etc/opensemanticsearch/etl')
		self.read_configfile ('/etc/opensemanticsearch/enhancer-rdf')

		self.fields = self.getfieldnames_from_plugins()


		# init exporter	(todo: exporter as extended PySolr)	
		self.export_solr = export_solr.export_solr()


		# init PySolr
		solr_uri = self.config['solr']
		if not solr_uri.endswith('/'):
			solr_uri += '/'
		solr_uri += self.config['index'] 

		self.solr = pysolr.Solr( solr_uri )

		self.threads_max = None

		# if not set explicit, autodetection of count of CPUs for amount of threads
		if not self.threads_max:
			import multiprocessing
			self.threads_max = multiprocessing.cpu_count()
			if self.verbose:
				print ( "Setting threads to count of CPUs: " + str(self.threads_max) )

		self.rows_per_step = 100
		if self.rows_per_step < self.threads_max * 2:
			self.rows_per_step = self.threads_max * 2


		self.work_in_progress = []
		self.delete_from_work_in_progress_lock = threading.Lock() 

		self.delete_from_work_in_progress_after_commit = []
		self.work_in_progress_lock = threading.Lock() 

		self.e_job_done = threading.Event()
Пример #9
0
    def __init__(self, verbose=False, quiet=True):

        ETL.__init__(self, verbose=verbose)

        self.quiet = quiet

        self.set_configdefaults()

        self.read_configfiles()

        # read on what DB or search server software our index is
        export = self.config['export']

        # call delete function of the configured exporter
        module = importlib.import_module(export)
        objectreference = getattr(module, export)
        self.connector = objectreference()
    def __init__(self, verbose=False, quiet=True):

        ETL.__init__(self, verbose=verbose)

        self.quiet = quiet

        self.set_configdefaults()

        self.read_configfiles()

        # read on what DB or search server software our index is
        export = self.config['export']

        # call delete function of the configured exporter
        module = importlib.import_module(export)
        objectreference = getattr(module, export)
        self.connector = objectreference()
Пример #11
0
    def __init__(self, verbose=False, quiet=True):

        ETL.__init__(self, verbose=verbose)

        self.quiet = quiet

        self.set_configdefaults()

        self.read_configfiles()

        # if not set explicit, autodetection of count of CPUs for amount of threads
        if not self.threads_max:
            import multiprocessing
            self.threads_max = multiprocessing.cpu_count()
            if self.verbose:
                print("Setting threads to count of CPUs: " +
                      str(self.threads_max))

        self.e_job_done = threading.Event()