Example #1
0
    def _save_configuration(self, browser_params):
        """ Saves crawl configuration details to db and logfile"""
        cur = self.db.cursor()

        # Get git version and commit information
        openwpm_v, browser_v = get_version()

        # Record task details
        cur.execute(("INSERT INTO task "
                     "(manager_params, openwpm_version, browser_version) "
                     "VALUES (?,?,?)"),
                (json.dumps(self.manager_params), openwpm_v, browser_v))
        self.db.commit()
        self.task_id = cur.lastrowid

        # Record browser details for each brower
        for i in xrange(self.num_browsers):
            cur.execute("INSERT INTO crawl (task_id, browser_params) VALUES (?,?)",
                        (self.task_id, json.dumps(browser_params[i])))
            self.db.commit()
            browser_params[i]['crawl_id'] = cur.lastrowid

        # Print the configuration details
        self.logger.info(get_configuration_string(self.manager_params,
                                                  browser_params,
                                                  (openwpm_v, browser_v)))
Example #2
0
    def _save_configuration(self, browser_params):
        """ Saves crawl configuration details to db and logfile"""
        cur = self.db.cursor()

        # Get git version and commit information
        openwpm_v, browser_v = get_version()

        # Record task details
        cur.execute(("INSERT INTO task "
                     "(manager_params, openwpm_version, browser_version) "
                     "VALUES (?,?,?)"),
                    (json.dumps(self.manager_params), openwpm_v, browser_v))
        self.db.commit()
        self.task_id = cur.lastrowid

        # Record browser details for each brower
        for i in xrange(self.num_browsers):
            cur.execute(
                "INSERT INTO crawl (task_id, browser_params) VALUES (?,?)",
                (self.task_id, json.dumps(browser_params[i])))
            self.db.commit()
            browser_params[i]['crawl_id'] = cur.lastrowid

        # Print the configuration details
        self.logger.info(
            get_configuration_string(self.manager_params, browser_params,
                                     (openwpm_v, browser_v)))
Example #3
0
    def __init__(self, manager_params, browser_params, process_watchdog=False):

        # Make paths absolute in manager_params
        for path in ['data_directory', 'log_directory']:
            if manager_params[path] is not None:
                manager_params[path] = os.path.expanduser(manager_params[path])
        manager_params['database_name'] = os.path.join(
            manager_params['data_directory'], manager_params['database_name'])
        manager_params['log_file'] = os.path.join(
            manager_params['log_directory'], manager_params['log_file'])
        manager_params['screenshot_path'] = os.path.join(
            manager_params['data_directory'], 'screenshots')
        manager_params['source_dump_path'] = os.path.join(
            manager_params['data_directory'], 'sources')
        self.manager_params = manager_params
        self.browser_params = browser_params

        # Create data directories if they do not exist
        if not os.path.exists(manager_params['screenshot_path']):
            os.makedirs(manager_params['screenshot_path'])
        if not os.path.exists(manager_params['source_dump_path']):
            os.makedirs(manager_params['source_dump_path'])

        # check size of parameter dictionary
        self.num_browsers = manager_params['num_browsers']
        if len(browser_params) != self.num_browsers:
            raise Exception("Number of <browser_params> dicts is not the same "
                            "as manager_params['num_browsers']")

        # Flow control
        self.closing = False
        self.failure_status = None
        self.threadlock = threading.Lock()
        self.failurecount = 0
        if manager_params['failure_limit'] is not None:
            self.failure_limit = manager_params['failure_limit']
        else:
            self.failure_limit = self.num_browsers * 2 + 10

        self.process_watchdog = process_watchdog

        # sets up logging server + connect a client
        self.logging_status_queue = None
        self.loggingserver = self._launch_loggingserver()
        # socket location: (address, port)
        self.manager_params['logger_address'] = self.logging_status_queue.get()
        self.logger = MPLogger.loggingclient(
            *self.manager_params['logger_address'])

        # Initialize the data aggregators
        self._launch_aggregators()

        # sets up the BrowserManager(s) + associated queues
        self.browsers = self._initialize_browsers(browser_params)
        self._launch_browsers()

        # start the manager watchdog
        thread = threading.Thread(target=self._manager_watchdog, args=())
        thread.daemon = True
        thread.start()

        # Save crawl config information to database
        openwpm_v, browser_v = get_version()
        self.data_aggregator.save_configuration(openwpm_v, browser_v)
        self.logger.info(
            get_configuration_string(self.manager_params, browser_params,
                                     (openwpm_v, browser_v)))