예제 #1
0
def test_load_tar_file(tmp_path):
    """Test that load_profile does not delete or modify the tar file."""
    tar_path = Path("profile.tar.gz")
    profile_path = tmp_path / "browser_profile"
    browser_params = BrowserParamsInternal(browser_id=1)
    modified_time_before_load = tar_path.stat().st_mtime
    load_profile(profile_path, browser_params, tar_path)
    assert modified_time_before_load == tar_path.stat().st_mtime
예제 #2
0
    def __init__(
        self,
        manager_params_temp: ManagerParams,
        browser_params_temp: List[BrowserParams],
        structured_storage_provider: StructuredStorageProvider,
        unstructured_storage_provider: Optional[UnstructuredStorageProvider],
        logger_kwargs: Dict[Any, Any] = {},
    ) -> None:
        """Initialize the TaskManager with browser and manager config params

        Parameters
        ----------
        manager_params_temp : ManagerParams
            TaskManager configuration parameters
        browser_params_temp : list of BrowserParams
            Browser configuration parameters. It is a list which
            includes individual configurations for each browser.
        logger_kwargs : dict, optional
            Keyword arguments to pass to MPLogger on initialization.
        """

        validate_crawl_configs(manager_params_temp, browser_params_temp)
        manager_params = ManagerParamsInternal.from_dict(manager_params_temp.to_dict())
        browser_params = [
            BrowserParamsInternal.from_dict(bp.to_dict()) for bp in browser_params_temp
        ]

        manager_params.screenshot_path = manager_params.data_directory / "screenshots"

        manager_params.source_dump_path = manager_params.data_directory / "sources"

        self.manager_params = manager_params
        self.browser_params = browser_params
        self._logger_kwargs = logger_kwargs

        # 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

        # Parse and flesh out js_instrument_settings
        for a_browsers_params in self.browser_params:
            js_settings = a_browsers_params.js_instrument_settings
            cleaned_js_settings = clean_js_instrumentation_settings(js_settings)
            a_browsers_params.cleaned_js_instrument_settings = cleaned_js_settings

        # Flow control
        self.closing = False
        self.failure_status: Optional[Dict[str, Any]] = None
        self.threadlock = threading.Lock()
        self.failure_count = 0

        self.failure_limit = manager_params.failure_limit
        # Start logging server thread
        self.logging_server = MPLogger(
            self.manager_params.log_path,
            str(structured_storage_provider),
            **self._logger_kwargs
        )
        self.manager_params.logger_address = self.logging_server.logger_address
        self.logger = logging.getLogger("openwpm")

        # Initialize the storage controller
        self._launch_storage_controller(
            structured_storage_provider, unstructured_storage_provider
        )

        # 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.name = "OpenWPM-watchdog"
        thread.start()

        # Save crawl config information to database
        openwpm_v, browser_v = get_version()
        self.storage_controller_handle.save_configuration(
            manager_params, browser_params, openwpm_v, browser_v
        )
        self.logger.info(
            get_configuration_string(
                self.manager_params, browser_params, (openwpm_v, browser_v)
            )
        )
        self.unsaved_command_sequences: Dict[int, CommandSequence] = dict()
        self.callback_thread = threading.Thread(
            target=self._mark_command_sequences_complete, args=()
        )
        self.callback_thread.name = "OpenWPM-completion_handler"
        self.callback_thread.start()