ioannis@67: """ This is a script that allows interaction with the SCC through the command line. ioannis@67: ioannis@67: It is based on the requests module for accessing the server. ioannis@67: ioannis@67: Most of the interactions are done through the web interface, i.e. by mimicking user interaction with the ioannis@67: SCC website (i.e. user login, data submission, etc.). In few cases, the SCC API is also used. ioannis@67: ioannis@67: Most of the functionality is included in the SCC class. The class is used to login into the SCC website and automate ioannis@67: interaction with the site (i.e. upload a file, get measurement status, etc.). ioannis@67: ioannis@67: Two other classes (Measurement, AncillaryFile) are used in some cases to handle the output of the SCC API. ioannis@67: ioannis@67: Several shortcut functions are defined to perform specific tasks using the SCC class (e.g. process_file, delete_measurements etc). ioannis@67: """ ioannis@43: import sys ioannis@43: victor@7: import requests victor@7: ioannis@43: try: ioannis@43: import urllib.parse as urlparse # Python 3 ioannis@43: except ImportError: ioannis@17: import urlparse # Python 2 ioannis@43: victor@7: import argparse madrouin@24: import datetime madrouin@24: import logging victor@7: import os victor@7: import re ioannis@32: from io import BytesIO ioannis@45: victor@7: import time i@40: victor@7: from zipfile import ZipFile madrouin@24: i@14: import yaml victor@7: ioannis@43: import netCDF4 as netcdf ioannis@43: ioannis@43: requests.packages.urllib3.disable_warnings() i@14: logger = logging.getLogger(__name__) victor@7: victor@7: # The regex to find the measurement id from the measurement page victor@7: # This should be read from the uploaded file, but would require an extra NetCDF module. i@36: regex = "

Measurement (?P.{12,15}) " # {12, 15} to handle both old- and new-style measurement ids. victor@7: victor@7: victor@7: class SCC: ioannis@67: """A class that will attempt to interact SCC server. i@14: ioannis@67: Most interactions are by simulating a normal browser session. In the current ioannis@67: version few checks are performed before upload a file, and no feedback is given in case the upload ioannis@67: fails. victor@7: """ victor@7: i@14: def __init__(self, auth, output_dir, base_url): moritz@29: victor@7: self.auth = auth victor@7: self.output_dir = output_dir i@14: self.base_url = base_url victor@7: self.session = requests.Session() moritz@29: self.session.auth = auth moritz@29: self.session.verify = False victor@7: ioannis@67: # Setup SCC server URLS for later use i@14: self.login_url = urlparse.urljoin(self.base_url, 'accounts/login/') i@31: self.logout_url = urlparse.urljoin(self.base_url, 'accounts/logout/') i@31: self.list_measurements_url = urlparse.urljoin(self.base_url, 'data_processing/measurements/') i@31: i@14: self.upload_url = urlparse.urljoin(self.base_url, 'data_processing/measurements/quick/') ioannis@54: self.measurement_page_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/') ioannis@43: self.download_hirelpp_pattern = urlparse.urljoin(self.base_url, ioannis@67: 'data_processing/measurements/{0}/download-hirelpp/') ioannis@43: self.download_cloudmask_pattern = urlparse.urljoin(self.base_url, ioannis@67: 'data_processing/measurements/{0}/download-cloudmask/') ioannis@34: ioannis@43: self.download_elpp_pattern = urlparse.urljoin(self.base_url, ioannis@67: 'data_processing/measurements/{0}/download-preprocessed/') ioannis@43: self.download_elda_pattern = urlparse.urljoin(self.base_url, ioannis@67: 'data_processing/measurements/{0}/download-optical/') ioannis@45: self.download_plots_pattern = urlparse.urljoin(self.base_url, moritz@29: 'data_processing/measurements/{0}/download-plots/') ioannis@43: self.download_elic_pattern = urlparse.urljoin(self.base_url, ioannis@67: 'data_processing/measurements/{0}/download-elic/') i@14: self.delete_measurement_pattern = urlparse.urljoin(self.base_url, 'admin/database/measurements/{0}/delete/') i@31: ioannis@67: # Setup API URLs for later use i@14: self.api_base_url = urlparse.urljoin(self.base_url, 'api/v1/') i@31: self.api_measurement_pattern = urlparse.urljoin(self.api_base_url, 'measurements/{0}/') i@31: self.api_measurements_url = urlparse.urljoin(self.api_base_url, 'measurements') i@38: self.api_sounding_search_pattern = urlparse.urljoin(self.api_base_url, 'sounding_files/?filename={0}') i@38: self.api_lidarratio_search_pattern = urlparse.urljoin(self.api_base_url, 'lidarratio_files/?filename={0}') i@38: self.api_overlap_search_pattern = urlparse.urljoin(self.api_base_url, 'overlap_files/?filename={0}') ioannis@43: i@14: def login(self, credentials): ioannis@67: """ Login to the SCC. ioannis@67: ioannis@67: Parameters ioannis@67: ---------- ioannis@67: credentials : tuple or list ioannis@67: A list or tuple in the form (username, password). ioannis@67: """ victor@7: logger.debug("Attempting to login to SCC, username %s." % credentials[0]) moritz@29: login_credentials = {'username': credentials[0], moritz@29: 'password': credentials[1]} victor@7: i@14: logger.debug("Accessing login page at %s." % self.login_url) victor@7: ioannis@67: # Get login form moritz@29: login_page = self.session.get(self.login_url) i@14: i@31: if not login_page.ok: i@31: raise self.PageNotAccessibleError('Could not access login pages. Status code %s' % login_page.status_code) victor@7: i@31: logger.debug("Submitting credentials.") victor@7: # Submit the login data i@14: login_submit = self.session.post(self.login_url, moritz@29: data=login_credentials, victor@7: headers={'X-CSRFToken': login_page.cookies['csrftoken'], moritz@29: 'referer': self.login_url}) victor@7: return login_submit victor@7: victor@7: def logout(self): ioannis@67: """ Logout from the SCC """ i@31: return self.session.get(self.logout_url, stream=True) victor@7: ioannis@67: def upload_file(self, filename, system_id, force_upload, delete_related, delay=0, rs_filename=None, ioannis@67: ov_filename=None, lr_filename=None): ioannis@67: """ Upload a file for processing. ioannis@67: ioannis@67: If the upload is successful, it returns the measurement id. ioannis@67: ioannis@67: ioannis@67: Parameters ioannis@67: ---------- ioannis@67: filename : str ioannis@67: File path of the file to upload ioannis@67: system_id : int ioannis@67: System id to be used in the processing ioannis@67: force_upload : bool ioannis@67: If True, if a measurement with the same ID is found on the server, it will be first deleted and the ioannis@67: file current file will be uploaded. If False, the file will not be uploaded if the measurement ID is ioannis@67: already present on the SCC server. ioannis@67: delete_related : bool ioannis@67: Answer to delete related question when deleting existing measurements from the SCC. ioannis@67: rs_filename, ov_filename, lr_filename : str ioannis@67: Ancillary files pahts to be uploaded. ioannis@67: """ ioannis@67: # Get the measurement ID from the netcdf file ioannis@43: measurement_id = self.measurement_id_from_file(filename) ioannis@43: ioannis@67: # Handle possible existing measurements with the same ID on the SCC server. ioannis@43: logger.debug('Checking if a measurement with the same id already exists on the SCC server.') ioannis@50: existing_measurement, _ = self.get_measurement(measurement_id) ioannis@43: ioannis@43: if existing_measurement: ioannis@43: if force_upload: ioannis@43: logger.info( ioannis@43: "Measurement with id {} already exists on the SCC. Trying to delete it...".format(measurement_id)) ioannis@43: self.delete_measurement(measurement_id, delete_related) ioannis@43: else: ioannis@43: logger.error( ioannis@43: "Measurement with id {} already exists on the SCC. Use --force_upload flag to overwrite it.".format( ioannis@43: measurement_id)) ioannis@67: # TODO: Implement handling at the proper place. Exiting here does not allow the SCC class to be ioannis@67: # used by external programs. Instead an exception should be raised. ioannis@43: sys.exit(1) ioannis@43: ioannis@67: # Upload the file(s) ioannis@67: victor@7: # Get submit page moritz@29: upload_page = self.session.get(self.upload_url) victor@7: victor@7: # Submit the data ioannis@54: upload_data = {'system': system_id, ioannis@54: 'delay': delay} ioannis@54: ioannis@54: logger.debug("Submitted processing parameters - System: {}, Delay: {}".format(system_id, delay)) ioannis@54: victor@7: files = {'data': open(filename, 'rb')} victor@7: ioannis@67: # Add ancillary files to be uploaded madrouin@20: if rs_filename is not None: i@38: ancillary_file, _ = self.get_ancillary(rs_filename, 'sounding') i@38: i@38: if ancillary_file.already_on_scc: ioannis@67: logger.warning( ioannis@67: "Sounding file {0.filename} already on the SCC with id {0.id}. Ignoring it.".format(ancillary_file)) i@38: else: i@38: logger.debug('Adding sounding file %s' % rs_filename) i@38: files['sounding_file'] = open(rs_filename, 'rb') madrouin@20: ioannis@32: if ov_filename is not None: i@38: ancillary_file, _ = self.get_ancillary(ov_filename, 'overlap') i@38: i@38: if ancillary_file.already_on_scc: ioannis@67: logger.warning( ioannis@67: "Overlap file {0.filename} already on the SCC with id {0.id}. Ignoring it.".format(ancillary_file)) i@38: else: i@38: logger.debug('Adding overlap file %s' % ov_filename) i@38: files['overlap_file'] = open(ov_filename, 'rb') i@31: i@31: if lr_filename is not None: i@38: ancillary_file, _ = self.get_ancillary(lr_filename, 'lidarratio') i@38: i@38: if ancillary_file.already_on_scc: i@38: logger.warning( ioannis@67: "Lidar ratio file {0.filename} already on the SCC with id {0.id}. Ignoring it.".format( ioannis@67: ancillary_file)) i@38: else: i@38: logger.debug('Adding lidar ratio file %s' % lr_filename) i@38: files['lidar_ratio_file'] = open(lr_filename, 'rb') i@31: ioannis@67: # Upload the files i@14: logger.info("Uploading of file %s started." % filename) victor@7: i@14: upload_submit = self.session.post(self.upload_url, victor@7: data=upload_data, victor@7: files=files, victor@7: headers={'X-CSRFToken': upload_page.cookies['csrftoken'], moritz@29: 'referer': self.upload_url}) victor@7: victor@7: if upload_submit.status_code != 200: i@14: logger.warning("Connection error. Status code: %s" % upload_submit.status_code) victor@7: return False victor@7: ioannis@67: # Check if there was a redirect to a new page. If not, something went wrong i@14: if upload_submit.url == self.upload_url: victor@7: measurement_id = False ioannis@32: logger.error("Uploaded file(s) rejected! Try to upload manually to see the error.") victor@7: else: ioannis@67: # TODO: Check if this is needed. This was used when the measurement ID was not read from the input file. ioannis@67: measurement_id = re.findall(regex, upload_submit.text)[0] # Get the measurement ID from the output page ioannis@43: logger.info("Successfully uploaded measurement with id %s." % measurement_id) ioannis@67: logger.info("You can monitor the processing progress online: {}".format( ioannis@67: self.measurement_page_pattern.format(measurement_id))) ioannis@43: return measurement_id ioannis@43: ioannis@43: @staticmethod ioannis@43: def measurement_id_from_file(filename): ioannis@67: """ Get the measurement id from the input file. ioannis@67: ioannis@67: Parameters ioannis@67: ---------- ioannis@67: filename : str ioannis@67: File path of the input file. ioannis@67: """ ioannis@43: ioannis@43: if not os.path.isfile(filename): ioannis@43: logger.error("File {} does not exist.".format(filename)) ioannis@43: sys.exit(1) ioannis@43: ioannis@43: with netcdf.Dataset(filename) as f: ioannis@43: try: ioannis@43: measurement_id = f.Measurement_ID ioannis@43: except AttributeError: ioannis@43: logger.error( ioannis@43: "Input file {} does not contain a Measurement_ID global attribute. Wrong file format?".format( ioannis@43: filename)) ioannis@43: sys.exit(1) victor@7: victor@7: return measurement_id victor@7: victor@7: def download_files(self, measurement_id, subdir, download_url): victor@7: """ Downloads some files from the download_url to the specified ioannis@67: subdir. ioannis@67: ioannis@67: This is a general method used to download preprocessed file, optical ioannis@67: files by other, file-specific, methods. victor@7: """ ioannis@32: # TODO: Make downloading more robust (e.g. in case that files do not exist on server). victor@7: # Get the file moritz@29: request = self.session.get(download_url, stream=True) moritz@29: moritz@29: if not request.ok: moritz@29: raise Exception("Could not download files for measurement '%s'" % measurement_id) victor@7: victor@7: # Create the dir if it does not exist victor@7: local_dir = os.path.join(self.output_dir, measurement_id, subdir) victor@7: if not os.path.exists(local_dir): victor@7: os.makedirs(local_dir) victor@7: victor@7: # Save the file by chunk, needed if the file is big. ioannis@43: memory_file = BytesIO() victor@7: victor@7: for chunk in request.iter_content(chunk_size=1024): victor@7: if chunk: # filter out keep-alive new chunks victor@7: memory_file.write(chunk) victor@7: memory_file.flush() victor@7: victor@7: zip_file = ZipFile(memory_file) victor@7: victor@7: for ziped_name in zip_file.namelist(): victor@7: basename = os.path.basename(ziped_name) victor@7: victor@7: local_file = os.path.join(local_dir, basename) victor@7: victor@7: with open(local_file, 'wb') as f: victor@7: f.write(zip_file.read(ziped_name)) victor@7: ioannis@43: def download_hirelpp(self, measurement_id): ioannis@43: """ Download hirelpp files for the measurement id. """ ioannis@43: # Construct the download url ioannis@43: download_url = self.download_hirelpp_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'hirelpp', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download HiRElPP files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) ioannis@43: ioannis@43: def download_cloudmask(self, measurement_id): ioannis@43: """ Download cloudmask files for the measurement id. """ ioannis@43: # Construct the download url ioannis@43: download_url = self.download_cloudmask_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'cloudscreen', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download cloudscreen files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) ioannis@43: ioannis@43: def download_elpp(self, measurement_id): victor@7: """ Download preprocessed files for the measurement id. """ victor@7: # Construct the download url ioannis@43: download_url = self.download_elpp_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'elpp', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download ElPP files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) victor@7: ioannis@43: def download_elda(self, measurement_id): victor@7: """ Download optical files for the measurement id. """ victor@7: # Construct the download url ioannis@43: download_url = self.download_elda_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'elda', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download ELDA files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) victor@7: ioannis@43: def download_plots(self, measurement_id): victor@7: """ Download profile graphs for the measurement id. """ victor@7: # Construct the download url ioannis@45: download_url = self.download_plots_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'elda_plots', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download ELDA plots. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) victor@7: ioannis@43: def download_elic(self, measurement_id): ioannis@43: """ Download ELIC files for the measurement id. """ ioannis@43: # Construct the download url ioannis@43: download_url = self.download_elic_pattern.format(measurement_id) ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'elic', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download ELIC files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) ioannis@43: ioannis@43: def download_eldec(self, measurement_id): ioannis@43: """ Download ELDEC files for the measurement id. """ ioannis@43: # Construct the download url ioannis@43: download_url = self.download_elda_pattern.format(measurement_id) # ELDA patter is used for now ioannis@43: try: ioannis@50: self.download_files(measurement_id, 'eldec', download_url) ioannis@43: except Exception as e: ioannis@43: logger.error("Could not download EDELC files. Error message: {}".format(e)) ioannis@43: logger.debug('Download exception:', exc_info=True) ioannis@43: ioannis@43: def rerun_elpp(self, measurement_id, monitor=True): ioannis@45: logger.debug("Started rerun_elpp procedure.") ioannis@45: ioannis@45: logger.debug("Getting measurement %s" % measurement_id) madrouin@26: measurement, status = self.get_measurement(measurement_id) victor@7: victor@7: if measurement: ioannis@45: logger.debug("Attempting to rerun ElPP through %s." % measurement.rerun_all_url) ioannis@45: request = self.session.get(measurement.rerun_elpp_url, stream=True) victor@7: victor@7: if request.status_code != 200: i@14: logger.error( i@14: "Could not rerun processing for %s. Status code: %s" % (measurement_id, request.status_code)) ioannis@45: else: ioannis@45: logger.info("Rerun-elpp command submitted successfully for id {}.".format(measurement_id)) victor@7: victor@7: if monitor: victor@7: self.monitor_processing(measurement_id) victor@7: victor@7: def rerun_all(self, measurement_id, monitor=True): victor@7: logger.debug("Started rerun_all procedure.") victor@7: victor@7: logger.debug("Getting measurement %s" % measurement_id) madrouin@26: measurement, status = self.get_measurement(measurement_id) victor@7: victor@7: if measurement: victor@7: logger.debug("Attempting to rerun all processing through %s." % measurement.rerun_all_url) victor@7: moritz@29: request = self.session.get(measurement.rerun_all_url, stream=True) victor@7: victor@7: if request.status_code != 200: victor@7: logger.error("Could not rerun pre processing for %s. Status code: %s" % victor@7: (measurement_id, request.status_code)) ioannis@45: else: ioannis@45: logger.info("Rerun-all command submitted successfully for id {}.".format(measurement_id)) victor@7: victor@7: if monitor: victor@7: self.monitor_processing(measurement_id) victor@7: ioannis@67: def process(self, filename, system_id, monitor, force_upload, delete_related, delay=0, rs_filename=None, ioannis@67: lr_filename=None, ov_filename=None): victor@7: """ Upload a file for processing and wait for the processing to finish. victor@7: If the processing is successful, it will download all produced files. victor@7: """ victor@7: logger.info("--- Processing started on %s. ---" % datetime.datetime.now()) victor@7: # Upload file ioannis@50: logger.info("Uploading file.") ioannis@45: measurement_id = self.upload_file(filename, system_id, force_upload, delete_related, ioannis@54: delay=delay, ioannis@32: rs_filename=rs_filename, ioannis@32: lr_filename=lr_filename, ioannis@32: ov_filename=ov_filename) ioannis@43: ioannis@54: if monitor and (delay > 0): ioannis@54: logger.warning("Will not start monitoring, since a delay was specified: {} hours.".format(delay)) ioannis@54: return None ioannis@54: ioannis@32: if measurement_id and monitor: ioannis@54: logger.info("Monitoring processing.") moritz@29: return self.monitor_processing(measurement_id) victor@7: moritz@29: return None victor@7: ioannis@61: def monitor_processing(self, measurement_id, retry_max=2, time_sleep=2, exit_if_missing=True): victor@7: """ Monitor the processing progress of a measurement id""" victor@7: madrouin@24: # try to deal with error 404 ioannis@61: attempts_count = 0 ioannis@61: max_attempts = retry_max + 1 madrouin@24: ioannis@67: # try to wait for measurement to appear in API. A user has reported that this does not happen immediately. madrouin@24: measurement = None ioannis@50: logger.info("Looking for measurement %s on the SCC.", measurement_id) ioannis@62: ioannis@61: while attempts_count < max_attempts: ioannis@61: attempts_count += 1 madrouin@26: measurement, status = self.get_measurement(measurement_id) ioannis@61: if status != 200: ioannis@61: logger.warning("Measurement not found.") ioannis@61: if attempts_count < max_attempts: ioannis@61: logger.warning("Waiting %ds.", time_sleep) ioannis@61: time.sleep(time_sleep) madrouin@26: else: madrouin@26: break ioannis@62: print("Measurement: {}".format(measurement)) madrouin@24: ioannis@62: if measurement is None: ioannis@61: logger.error("Measurement %s doesn't seem to exist.", measurement_id) ioannis@61: if exit_if_missing: ioannis@61: sys.exit(1) ioannis@61: else: ioannis@61: return measurement madrouin@26: ioannis@45: logger.info('Measurement %s found.', measurement_id) ioannis@62: while not measurement.has_finished: ioannis@62: measurement.log_processing_status() ioannis@62: time.sleep(10) ioannis@62: measurement, status = self.get_measurement(measurement_id) madrouin@24: ioannis@62: logger.info("Measurement processing finished.") ioannis@62: measurement.log_detailed_status() ioannis@53: ioannis@62: if measurement.hirelpp == 127: ioannis@62: logger.info("Downloading HiRElPP files.") ioannis@62: self.download_hirelpp(measurement_id) ioannis@62: if measurement.cloudmask == 127: ioannis@62: logger.info("Downloading cloud screening files.") ioannis@62: self.download_cloudmask(measurement_id) ioannis@62: if measurement.elpp == 127: ioannis@62: logger.info("Downloading ElPP files.") ioannis@62: self.download_elpp(measurement_id) ioannis@62: if measurement.elda == 127: ioannis@62: logger.info("Downloading ELDA files.") ioannis@62: self.download_elda(measurement_id) ioannis@62: logger.info("Downloading ELDA plots.") ioannis@62: self.download_plots(measurement_id) ioannis@62: if measurement.elic == 127: ioannis@62: logger.info("Downloading ELIC files.") ioannis@62: self.download_elic(measurement_id) ioannis@67: if measurement.is_calibration and measurement.eldec == 0: ioannis@62: logger.info("Downloading ELDEC files.") ioannis@62: self.download_eldec(measurement_id) ioannis@62: logger.info("--- Processing finished. ---") ioannis@43: victor@7: return measurement victor@7: ioannis@43: def get_measurement(self, measurement_id): ioannis@67: """ Get a measurement information from the SCC API. victor@7: ioannis@67: Parameters ioannis@67: ---------- ioannis@67: measurement_id : str ioannis@67: The measurement ID to search. ioannis@67: ioannis@67: Returns ioannis@67: ------- ioannis@67: : Measurement object or None ioannis@67: If the measurement is found, a Measurement object is returned. If not, it returns None ioannis@67: """ ioannis@67: # TODO: Consider to homogenize with get_ancillary method (i.e. always return a Measurement object). ioannis@67: ioannis@67: if measurement_id is None: # TODO: Is this still required? victor@7: return None victor@7: ioannis@67: # Access the API i@31: measurement_url = self.api_measurement_pattern.format(measurement_id) i@31: logger.debug("Measurement API URL: %s" % measurement_url) victor@7: moritz@29: response = self.session.get(measurement_url) victor@7: ioannis@43: response_dict = None ioannis@45: ioannis@43: if response.status_code == 200: ioannis@43: response_dict = response.json() ioannis@45: elif response.status_code == 404: ioannis@43: logger.info("No measurement with id %s found on the SCC." % measurement_id) ioannis@45: else: i@14: logger.error('Could not access API. Status code %s.' % response.status_code) victor@7: ioannis@55: # TODO: Implement better handling for status 401. ioannis@55: victor@7: if response_dict: ioannis@43: measurement = Measurement(self.base_url, response_dict) victor@7: else: ioannis@45: measurement = None victor@7: ioannis@45: return measurement, response.status_code victor@7: ioannis@45: def delete_measurement(self, measurement_id, delete_related): victor@7: """ Deletes a measurement with the provided measurement id. The user ioannis@67: should have the appropriate permissions (i.e. access to the admin site). madrouin@20: victor@7: The procedures is performed directly through the web interface and victor@7: NOT through the API. victor@7: """ victor@7: # Get the measurement object i@31: measurement, _ = self.get_measurement(measurement_id) victor@7: victor@7: # Check that it exists victor@7: if measurement is None: victor@7: logger.warning("Nothing to delete.") victor@7: return None victor@7: victor@7: # Go the the page confirming the deletion moritz@29: delete_url = self.delete_measurement_pattern.format(measurement_id) ioannis@43: moritz@29: confirm_page = self.session.get(delete_url) victor@7: victor@7: # Check that the page opened properly victor@7: if confirm_page.status_code != 200: victor@7: logger.warning("Could not open delete page. Status: {0}".format(confirm_page.status_code)) victor@7: return None victor@7: ioannis@43: # Get the delete related value ioannis@43: if delete_related: ioannis@43: delete_related_option = 'delete_related' ioannis@43: else: ioannis@43: delete_related_option = 'not_delete_related' ioannis@43: victor@7: # Delete the measurement victor@7: delete_page = self.session.post(delete_url, ioannis@43: data={'post': 'yes', ioannis@43: 'select_delete_related_measurements': delete_related_option}, victor@7: headers={'X-CSRFToken': confirm_page.cookies['csrftoken'], victor@7: 'referer': delete_url} victor@7: ) i@31: if not delete_page.ok: victor@7: logger.warning("Something went wrong. Delete page status: {0}".format( victor@7: delete_page.status_code)) victor@7: return None victor@7: victor@7: logger.info("Deleted measurement {0}".format(measurement_id)) victor@7: return True victor@7: victor@7: def available_measurements(self): ioannis@67: """ Get a list of available measurement on the SCC. ioannis@67: ioannis@67: The methods is currently not used, could be merged with list_measurements. ioannis@67: """ i@31: response = self.session.get(self.api_measurements_url) victor@7: response_dict = response.json() victor@7: victor@7: if response_dict: victor@7: measurement_list = response_dict['objects'] i@14: measurements = [Measurement(self.base_url, measurement_dict) for measurement_dict in measurement_list] victor@7: logger.info("Found %s measurements on the SCC." % len(measurements)) victor@7: else: victor@7: logger.warning("No response received from the SCC when asked for available measurements.") ioannis@43: measurements = None victor@7: victor@7: return measurements victor@7: ioannis@65: def list_measurements(self, id_exact=None, id_startswith=None): ioannis@67: """ Get the response text from the API. """ moritz@29: ioannis@67: # TODO: Add some error handling, e.g. as per available_measurements method i@31: moritz@29: # Need to set to empty string if not specified, we won't get any results ioannis@65: params = {} i@31: ioannis@65: if id_exact is not None: ioannis@65: params['id__exact'] = id_exact ioannis@65: else: ioannis@65: params['id__startswith'] = id_startswith moritz@29: ioannis@65: response_json = self.session.get(self.api_measurements_url, params=params).text moritz@29: ioannis@65: return response_json moritz@29: ioannis@43: def measurement_id_for_date(self, t1, call_sign, base_number=0): victor@7: """ Give the first available measurement id on the SCC for the specific madrouin@20: date. victor@7: """ ioannis@67: # TODO: Check if this method needs updating to handle all measurement_ID formats. victor@7: date_str = t1.strftime('%Y%m%d') i@31: base_id = "%s%s" % (date_str, call_sign) i@31: search_url = urlparse.urljoin(self.api_base_url, 'measurements/?id__startswith=%s' % base_id) victor@7: moritz@29: response = self.session.get(search_url) victor@7: victor@7: response_dict = response.json() victor@7: victor@7: measurement_id = None victor@7: victor@7: if response_dict: victor@7: measurement_list = response_dict['objects'] i@31: i@31: if len(measurement_list) == 100: i@31: raise ValueError('No available measurement id found.') i@31: victor@7: existing_ids = [measurement_dict['id'] for measurement_dict in measurement_list] victor@7: victor@7: measurement_number = base_number i@31: measurement_id = "%s%02i" % (base_id, measurement_number) victor@7: victor@7: while measurement_id in existing_ids: victor@7: measurement_number = measurement_number + 1 i@31: measurement_id = "%s%02i" % (base_id, measurement_number) victor@7: victor@7: return measurement_id victor@7: i@40: def get_ancillary(self, file_path, file_type): i@38: """ i@38: Try to get the ancillary file data from the SCC API. i@38: i@38: The result will always be an API object. If the file does not exist, the .exists property is set to False. i@38: i@38: Parameters i@38: ---------- i@40: file_path : str i@40: Path of the uploaded file. i@38: file_type : str i@38: Type of ancillary file. One of 'sounding', 'overlap', 'lidarratio'. i@38: i@38: Returns i@38: : AncillaryFile i@38: The api object. i@38: """ i@38: assert file_type in ['sounding', 'overlap', 'lidarratio'] i@38: i@40: filename = os.path.basename(file_path) i@40: i@38: if file_type == 'sounding': i@38: file_url = self.api_sounding_search_pattern.format(filename) i@38: elif file_type == 'overlap': i@38: file_url = self.api_overlap_search_pattern.format(filename) i@38: else: i@38: file_url = self.api_lidarratio_search_pattern.format(filename) i@38: i@38: response = self.session.get(file_url) i@38: i@38: if not response.ok: i@38: logger.error('Could not access API. Status code %s.' % response.status_code) i@38: return None, response.status_code i@38: i@38: response_dict = response.json() i@38: object_list = response_dict['objects'] i@38: i@38: logger.debug("Ancillary file JSON: {0}".format(object_list)) i@38: i@38: if object_list: i@38: ancillary_file = AncillaryFile(self.api_base_url, object_list[0]) # Assume only one file is returned i@38: else: i@38: ancillary_file = AncillaryFile(self.api_base_url, None) # Create an empty object i@38: i@38: return ancillary_file, response.status_code i@38: ioannis@43: def __enter__(self): ioannis@43: return self victor@7: ioannis@43: def __exit__(self, *args): ioannis@43: logger.debug("Closing SCC connection session.") ioannis@43: self.session.close() ioannis@43: i@31: class PageNotAccessibleError(RuntimeError): i@31: pass ioannis@43: victor@7: ioannis@17: class ApiObject(object): victor@7: """ A generic class object. """ victor@7: i@14: def __init__(self, base_url, dict_response): i@14: self.base_url = base_url victor@7: victor@7: if dict_response: victor@7: # Add the dictionary key value pairs as object properties victor@7: for key, value in dict_response.items(): ioannis@17: # logger.debug('Setting key {0} to value {1}'.format(key, value)) ioannis@17: try: ioannis@17: setattr(self, key, value) ioannis@17: except: ioannis@17: logger.warning('Could not set attribute {0} to value {1}'.format(key, value)) victor@7: self.exists = True victor@7: else: victor@7: self.exists = False victor@7: victor@7: victor@7: class Measurement(ApiObject): ioannis@43: """ This class represents the measurement object as returned in the SCC API. ioannis@43: """ victor@7: i@14: def __init__(self, base_url, dict_response): victor@7: ioannis@45: # Define expected attributes to assist debugging ioannis@53: ioannis@53: self.hirelpp = None ioannis@53: self.hirelpp_exit_code = None ioannis@43: self.cloudmask = None ioannis@53: self.cloudmask_exit_code = None ioannis@53: self.elpp = None ioannis@53: self.elpp_exit_code = None ioannis@43: self.elda = None ioannis@53: self.elda_exit_code = None ioannis@43: self.elic = None ioannis@53: self.elic_exit_code = None ioannis@53: self.eldec = None ioannis@53: self.eldec_exit_code = None ioannis@53: self.elquick = None ioannis@53: self.elquick_exit_code = None ioannis@53: ioannis@43: self.id = None ioannis@43: self.is_calibration = None ioannis@43: self.is_running = None ioannis@53: ioannis@43: self.resource_uri = None ioannis@43: self.start = None ioannis@43: self.stop = None ioannis@43: self.system = None ioannis@43: self.upload = None ioannis@43: ioannis@45: super().__init__(base_url, dict_response) victor@7: ioannis@57: @property ioannis@57: def has_finished(self): ioannis@57: """ Temporary implementation for SCC version 5.2.0, until the API returns a flag indicating if the ioannis@57: processing measurement has finished. """ ioannis@57: if (self.is_running is False) and (self.hirelpp != 0 or self.elpp != 0): ioannis@57: return True ioannis@57: else: ioannis@57: return False ioannis@57: ioannis@53: def log_processing_status(self): ioannis@53: """ Log module status. """ ioannis@53: logger.info("Measurement is being processed. Status: {}, {}, {}, {}, {}, {}). Please wait.".format( ioannis@53: self.upload, ioannis@53: self.hirelpp, ioannis@53: self.cloudmask, ioannis@53: self.elpp, ioannis@53: self.elda, ioannis@53: self.elic)) ioannis@53: ioannis@53: def log_detailed_status(self): ioannis@53: """ Log module exit and status codes.""" ioannis@53: logger.info("Measurement exit status:".format(self.id)) ioannis@53: if self.is_calibration: ioannis@53: self._log_module_status('ElPP', self.elpp, self.elpp_exit_code) ioannis@53: self._log_module_status('ElDEC', self.eldec, self.eldec_exit_code) ioannis@53: else: ioannis@53: self._log_module_status('HiRElPP', self.hirelpp, self.hirelpp_exit_code) ioannis@53: self._log_module_status('CloudScreen', self.cloudmask, self.cloudmask_exit_code) ioannis@53: self._log_module_status('ElPP', self.elpp, self.elpp_exit_code) ioannis@53: self._log_module_status('ELDA', self.elda, self.elda_exit_code) ioannis@53: self._log_module_status('ELIC', self.elic, self.elic_exit_code) ioannis@53: self._log_module_status('ELQuick', self.elquick, self.elquick_exit_code) ioannis@53: ioannis@53: def _log_module_status(self, name, status, exit_code): ioannis@53: if exit_code: ioannis@54: if exit_code['exit_code'] > 0: ioannis@54: logger.warning("{0} exit code: {2[exit_code]} - {2[description]}".format(name, status, exit_code)) ioannis@54: else: ioannis@54: logger.info("{0} exit code: {2[exit_code]} - {2[description]}".format(name, status, exit_code)) ioannis@53: else: ioannis@53: logger.info("{0} exit code: {2}".format(name, status, exit_code)) ioannis@53: ioannis@43: @property ioannis@43: def rerun_elda_url(self): ioannis@43: url_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/rerun-elda/') ioannis@43: return url_pattern.format(self.id) victor@7: victor@7: @property ioannis@43: def rerun_elpp_url(self): ioannis@43: url_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/rerun-elpp/') i@14: return url_pattern.format(self.id) victor@7: victor@7: @property victor@7: def rerun_all_url(self): i@14: ulr_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/rerun-all/') i@14: return ulr_pattern.format(self.id) victor@7: victor@7: def __str__(self): ioannis@43: return "Measurement {}".format(self.id) victor@7: victor@7: i@38: class AncillaryFile(ApiObject): i@38: """ This class represents the ancilalry file object as returned in the SCC API. i@38: """ ioannis@67: i@38: @property i@38: def already_on_scc(self): i@38: if self.exists is False: i@38: return False i@38: i@38: return not self.status == 'missing' i@38: i@38: def __str__(self): i@38: return "%s: %s, %s" % (self.id, i@38: self.filename, i@38: self.status) i@38: i@38: ioannis@67: # Methods that use the SCC class to perform specific tasks. ioannis@45: def process_file(filename, system_id, settings, force_upload, delete_related, ioannis@54: delay=0, monitor=True, rs_filename=None, lr_filename=None, ov_filename=None): ioannis@32: """ Shortcut function to process a file to the SCC. """ ioannis@32: logger.info("Processing file %s, using system %s" % (filename, system_id)) victor@7: ioannis@43: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@43: scc.login(settings['website_credentials']) ioannis@45: measurement = scc.process(filename, system_id, ioannis@45: force_upload=force_upload, ioannis@45: delete_related=delete_related, ioannis@54: delay=delay, ioannis@45: monitor=monitor, ioannis@45: rs_filename=rs_filename, ioannis@45: lr_filename=lr_filename, ioannis@45: ov_filename=ov_filename) ioannis@43: scc.logout() victor@7: return measurement victor@7: victor@7: ioannis@45: def delete_measurements(measurement_ids, delete_related, settings): moritz@29: """ Shortcut function to delete measurements from the SCC. """ ioannis@45: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@45: scc.login(settings['website_credentials']) ioannis@45: for m_id in measurement_ids: ioannis@54: logger.info("Deleting %s." % m_id) ioannis@45: scc.delete_measurement(m_id, delete_related) ioannis@45: scc.logout() moritz@29: moritz@29: moritz@29: def rerun_all(measurement_ids, monitor, settings): moritz@29: """ Shortcut function to rerun measurements from the SCC. """ i@14: ioannis@43: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@43: scc.login(settings['website_credentials']) ioannis@45: for m_id in measurement_ids: ioannis@54: logger.info("Rerunning all products for %s." % m_id) ioannis@45: scc.rerun_all(m_id, monitor) ioannis@43: scc.logout() victor@7: victor@7: moritz@29: def rerun_processing(measurement_ids, monitor, settings): victor@7: """ Shortcut function to delete a measurement from the SCC. """ i@14: ioannis@43: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@43: scc.login(settings['website_credentials']) ioannis@45: for m_id in measurement_ids: ioannis@45: logger.info("Rerunning (optical) processing for %s" % m_id) ioannis@45: scc.rerun_elpp(m_id, monitor) ioannis@45: scc.logout() moritz@29: moritz@29: ioannis@65: def list_measurements(settings, id_exact=None, id_startswith=None): moritz@29: """List all available measurements""" ioannis@45: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@45: scc.login(settings['website_credentials']) ioannis@65: ioannis@65: results_json = scc.list_measurements(id_exact=id_exact, id_startswith=id_startswith) ioannis@65: print(results_json) ioannis@65: ioannis@43: scc.logout() victor@7: victor@7: ioannis@61: def download_measurements(measurement_ids, max_retries, exit_if_missing, settings): moritz@29: """Download all measurements for the specified IDs""" ioannis@43: with SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) as scc: ioannis@43: scc.login(settings['website_credentials']) ioannis@45: for m_id in measurement_ids: ioannis@61: scc.monitor_processing(m_id, retry_max=max_retries, time_sleep=3, exit_if_missing=exit_if_missing) ioannis@61: ioannis@43: scc.logout() i@14: i@14: moritz@29: def settings_from_path(config_file_path): i@14: """ Read the configuration file. i@14: i@14: The file should be in YAML syntax.""" i@14: i@14: if not os.path.isfile(config_file_path): moritz@29: raise argparse.ArgumentTypeError("Wrong path for configuration file (%s)" % config_file_path) i@14: i@14: with open(config_file_path) as yaml_file: i@14: try: i@14: settings = yaml.safe_load(yaml_file) i@14: logger.debug("Read settings file(%s)" % config_file_path) moritz@29: except Exception: moritz@29: raise argparse.ArgumentTypeError("Could not parse YAML file (%s)" % config_file_path) i@14: i@14: # YAML limitation: does not read tuples i@14: settings['basic_credentials'] = tuple(settings['basic_credentials']) i@14: settings['website_credentials'] = tuple(settings['website_credentials']) i@14: return settings i@14: i@14: moritz@29: # Setup for command specific parsers moritz@29: def setup_delete(parser): moritz@29: def delete_from_args(parsed): ioannis@45: delete_measurements(parsed.IDs, ioannis@45: delete_related=False, ioannis@45: settings=parsed.config) moritz@29: moritz@29: parser.add_argument("IDs", nargs="+", help="measurement IDs to delete.") moritz@29: parser.set_defaults(execute=delete_from_args) moritz@29: moritz@29: moritz@29: def setup_rerun_all(parser): moritz@29: def rerun_all_from_args(parsed): moritz@29: rerun_all(parsed.IDs, parsed.process, parsed.config) moritz@29: moritz@29: parser.add_argument("IDs", nargs="+", help="Measurement IDs to rerun.") moritz@29: parser.add_argument("-p", "--process", help="Wait for the results of the processing.", moritz@29: action="store_true") moritz@29: parser.set_defaults(execute=rerun_all_from_args) moritz@29: moritz@29: ioannis@45: def setup_rerun_elpp(parser): moritz@29: def rerun_processing_from_args(parsed): moritz@29: rerun_processing(parsed.IDs, parsed.process, parsed.config) moritz@29: moritz@29: parser.add_argument("IDs", nargs="+", help="Measurement IDs to rerun the processing on.") moritz@29: parser.add_argument("-p", "--process", help="Wait for the results of the processing.", moritz@29: action="store_true") moritz@29: parser.set_defaults(execute=rerun_processing_from_args) moritz@29: moritz@29: ioannis@45: def setup_upload_file(parser): ioannis@45: """ Upload but do not monitor processing progress. """ ioannis@67: ioannis@45: def upload_file_from_args(parsed): ioannis@45: process_file(parsed.filename, parsed.system, parsed.config, ioannis@54: delay=parsed.delay, ioannis@45: monitor=parsed.process, ioannis@45: force_upload=parsed.force_upload, ioannis@45: delete_related=False, # For now, use this as default ioannis@32: rs_filename=parsed.radiosounding, ioannis@32: ov_filename=parsed.overlap, ioannis@32: lr_filename=parsed.lidarratio) moritz@29: ioannis@54: def delay(arg): ioannis@54: try: ioannis@54: int_arg = int(arg) ioannis@54: except ValueError: ioannis@54: raise argparse.ArgumentTypeError("Could not convert delay argument {} to integer.".format(arg)) ioannis@54: ioannis@54: if 0 <= int_arg <= 96: ioannis@54: return int_arg ioannis@54: else: ioannis@54: raise argparse.ArgumentTypeError("Delay should be an integer between 0 and 96.") ioannis@54: moritz@29: parser.add_argument("filename", help="Measurement file name or path.") moritz@29: parser.add_argument("system", help="Processing system id.") ioannis@54: parser.add_argument("--delay", help="Delay processing by the specified number of hours (0 to 96).", ioannis@54: default=0, type=delay) ioannis@45: parser.add_argument("-p", "--process", help="Wait for the processing results.", ioannis@45: action="store_true") ioannis@45: parser.add_argument("--force_upload", help="If measurement ID exists on SCC, delete before uploading.", ioannis@45: action="store_true") i@30: parser.add_argument("--radiosounding", default=None, help="Radiosounding file name or path") ioannis@32: parser.add_argument("--overlap", default=None, help="Overlap file name or path") ioannis@32: parser.add_argument("--lidarratio", default=None, help="Lidar ratio file name or path") i@30: moritz@29: parser.set_defaults(execute=upload_file_from_args) moritz@29: moritz@29: moritz@29: def setup_list_measurements(parser): moritz@29: def list_measurements_from_args(parsed): ioannis@65: list_measurements(parsed.config, id_exact=parsed.id_exact, id_startswith=parsed.id_startswith) moritz@29: ioannis@65: group = parser.add_mutually_exclusive_group() ioannis@65: group.add_argument("--id_exact", help="Exact measurement id.") ioannis@65: group.add_argument("--id_startswith", help="Initial part of measurement id.") moritz@29: moritz@29: parser.set_defaults(execute=list_measurements_from_args) moritz@29: moritz@29: moritz@29: def setup_download_measurements(parser): moritz@29: def download_measurements_from_args(parsed): ioannis@61: download_measurements(parsed.IDs, parsed.max_retries, parsed.ignore_errors, parsed.config) moritz@29: moritz@29: parser.add_argument("IDs", help="Measurement IDs that should be downloaded.", nargs="+") ioannis@67: parser.add_argument("--max_retries", help="Number of times to retry in cases of missing measurement id.", default=0, ioannis@67: type=int) ioannis@67: parser.add_argument("--ignore_errors", help="Ignore errors when downloading multiple measurements.", ioannis@67: action="store_false") moritz@29: parser.set_defaults(execute=download_measurements_from_args) moritz@29: moritz@29: victor@7: def main(): victor@7: # Define the command line arguments. victor@7: parser = argparse.ArgumentParser() moritz@29: subparsers = parser.add_subparsers() moritz@29: moritz@29: delete_parser = subparsers.add_parser("delete", help="Deletes a measurement.") ioannis@67: rerun_all_parser = subparsers.add_parser("rerun-all", ioannis@67: help="Rerun all processing steps for the provided measurement IDs.") ioannis@45: rerun_processing_parser = subparsers.add_parser("rerun-elpp", ioannis@45: help="Rerun low-resolution processing steps for the provided measurement ID.") ioannis@67: upload_file_parser = subparsers.add_parser("upload-file", ioannis@67: help="Submit a file and, optionally, download the output products.") ioannis@32: list_parser = subparsers.add_parser("list", help="List measurements registered on the SCC.") moritz@29: download_parser = subparsers.add_parser("download", help="Download selected measurements.") moritz@29: moritz@29: setup_delete(delete_parser) moritz@29: setup_rerun_all(rerun_all_parser) ioannis@45: setup_rerun_elpp(rerun_processing_parser) ioannis@45: moritz@29: setup_upload_file(upload_file_parser) moritz@29: setup_list_measurements(list_parser) moritz@29: setup_download_measurements(download_parser) victor@7: victor@7: # Verbosity settings from http://stackoverflow.com/a/20663028 victor@7: parser.add_argument('-d', '--debug', help="Print debugging information.", action="store_const", victor@7: dest="loglevel", const=logging.DEBUG, default=logging.INFO, victor@7: ) victor@7: parser.add_argument('-s', '--silent', help="Show only warning and error messages.", action="store_const", victor@7: dest="loglevel", const=logging.WARNING victor@7: ) victor@7: ioannis@32: # Setup default config location moritz@29: home = os.path.expanduser("~") moritz@29: default_config_location = os.path.abspath(os.path.join(home, ".scc_access.yaml")) moritz@29: parser.add_argument("-c", "--config", help="Path to the config file.", type=settings_from_path, moritz@29: default=default_config_location) victor@7: victor@7: args = parser.parse_args() ioannis@43: victor@7: # Get the logger with the appropriate level victor@7: logging.basicConfig(format='%(levelname)s: %(message)s', level=args.loglevel) victor@7: moritz@29: # Dispatch to appropriate function moritz@29: args.execute(args) moritz@29: i@14: moritz@29: # When running through terminal moritz@29: if __name__ == '__main__': moritz@29: main()