victor@7: import requests moritz@29: victor@7: requests.packages.urllib3.disable_warnings() victor@7: victor@7: import urlparse victor@7: import argparse victor@7: import os victor@7: import re victor@7: import time victor@7: import StringIO victor@7: from zipfile import ZipFile victor@7: import datetime victor@7: import logging i@14: import yaml victor@7: 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. victor@7: regex = "

Measurement (?P.{12}) " victor@7: victor@7: victor@7: class SCC: victor@7: """ A simple class that will attempt to upload a file on the SCC server. i@14: victor@7: The uploading is done by simulating a normal browser session. In the current victor@7: version no check is performed, and no feedback is given if the upload victor@7: was successful. If everything is setup correctly, it will work. 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 i@14: self.login_url = urlparse.urljoin(self.base_url, 'accounts/login/') i@14: self.upload_url = urlparse.urljoin(self.base_url, 'data_processing/measurements/quick/') moritz@29: self.download_preprocessed_pattern = urlparse.urljoin(self.base_url, moritz@29: 'data_processing/measurements/{0}/download-preprocessed/') moritz@29: self.download_optical_pattern = urlparse.urljoin(self.base_url, moritz@29: 'data_processing/measurements/{0}/download-optical/') moritz@29: self.download_graph_pattern = urlparse.urljoin(self.base_url, moritz@29: 'data_processing/measurements/{0}/download-plots/') i@14: self.delete_measurement_pattern = urlparse.urljoin(self.base_url, 'admin/database/measurements/{0}/delete/') i@14: self.api_base_url = urlparse.urljoin(self.base_url, 'api/v1/') moritz@29: self.list_measurements_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/') i@14: i@14: def login(self, credentials): victor@7: """ Login the the website. """ 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: victor@7: # Get upload form moritz@29: login_page = self.session.get(self.login_url) victor@7: victor@7: logger.debug("Submiting 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): victor@7: pass victor@7: victor@7: def upload_file(self, filename, system_id): victor@7: """ Upload a filename for processing with a specific system. If the victor@7: upload is successful, it returns the measurement id. """ victor@7: # Get submit page moritz@29: upload_page = self.session.get(self.upload_url) victor@7: victor@7: # Submit the data victor@7: upload_data = {'system': system_id} victor@7: files = {'data': open(filename, 'rb')} victor@7: 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: victor@7: # Check if there was a redirect to a new page. i@14: if upload_submit.url == self.upload_url: victor@7: measurement_id = False i@14: logger.error("Uploaded file rejected! Try to upload manually to see the error.") victor@7: else: victor@7: measurement_id = re.findall(regex, upload_submit.text)[0] i@14: logger.error("Successfully uploaded measurement with id %s." % measurement_id) 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 victor@7: subdir. This method is used to download preprocessed file, optical victor@7: files etc. victor@7: """ 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. victor@7: memory_file = StringIO.StringIO() 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: victor@7: def download_preprocessed(self, measurement_id): victor@7: """ Download preprocessed files for the measurement id. """ victor@7: # Construct the download url i@14: download_url = self.download_preprocessed_pattern.format(measurement_id) victor@7: self.download_files(measurement_id, 'scc_preprocessed', download_url) victor@7: victor@7: def download_optical(self, measurement_id): victor@7: """ Download optical files for the measurement id. """ victor@7: # Construct the download url moritz@29: download_url = self.download_optical_pattern.format(measurement_id) victor@7: self.download_files(measurement_id, 'scc_optical', download_url) victor@7: victor@7: def download_graphs(self, measurement_id): victor@7: """ Download profile graphs for the measurement id. """ victor@7: # Construct the download url i@14: download_url = self.download_graph_pattern.format(measurement_id) victor@7: self.download_files(measurement_id, 'scc_plots', download_url) victor@7: victor@7: def rerun_processing(self, measurement_id, monitor=True): victor@7: measurement = self.get_measurement(measurement_id) victor@7: victor@7: if measurement: moritz@29: request = self.session.get(measurement.rerun_processing_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)) victor@7: return 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) victor@7: measurement = 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)) victor@7: return victor@7: victor@7: if monitor: victor@7: self.monitor_processing(measurement_id) victor@7: moritz@29: def process(self, filename, system_id, monitor): 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 victor@7: measurement_id = self.upload_file(filename, system_id) victor@7: moritz@29: if monitor: moritz@29: return self.monitor_processing(measurement_id) moritz@29: return None victor@7: victor@7: def monitor_processing(self, measurement_id): victor@7: """ Monitor the processing progress of a measurement id""" victor@7: victor@7: measurement = self.get_measurement(measurement_id) victor@7: if measurement is not None: victor@7: while measurement.is_running: victor@7: logger.info("Measurement is being processed (status: %s, %s, %s). Please wait." % (measurement.upload, i@14: measurement.pre_processing, i@14: measurement.processing)) victor@7: time.sleep(10) victor@7: measurement = self.get_measurement(measurement_id) victor@7: logger.info("Measurement processing finished (status: %s, %s, %s)." % (measurement.upload, i@14: measurement.pre_processing, i@14: measurement.processing)) victor@7: if measurement.pre_processing == 127: victor@7: logger.info("Downloading preprocessed files.") victor@7: self.download_preprocessed(measurement_id) victor@7: if measurement.processing == 127: victor@7: logger.info("Downloading optical files.") victor@7: self.download_optical(measurement_id) victor@7: logger.info("Downloading graphs.") victor@7: self.download_graphs(measurement_id) victor@7: logger.info("--- Processing finished. ---") victor@7: return measurement victor@7: victor@7: def get_status(self, measurement_id): victor@7: """ Get the processing status for a measurement id through the API. """ i@14: measurement_url = urlparse.urljoin(self.api_base_url, 'measurements/?id__exact=%s' % measurement_id) victor@7: moritz@29: response = self.session.get(measurement_url) victor@7: victor@7: response_dict = response.json() victor@7: victor@7: if response_dict['objects']: victor@7: measurement_list = response_dict['objects'] i@14: measurement = Measurement(self.base_url, measurement_list[0]) i@14: return measurement.upload, measurement.pre_processing, measurement.processing victor@7: else: victor@7: logger.error("No measurement with id %s found on the SCC." % measurement_id) victor@7: return None victor@7: victor@7: def get_measurement(self, measurement_id): i@14: measurement_url = urlparse.urljoin(self.api_base_url, 'measurements/%s/' % measurement_id) victor@7: moritz@29: response = self.session.get(measurement_url) victor@7: moritz@29: if not response.ok: i@14: logger.error('Could not access API. Status code %s.' % response.status_code) moritz@29: return None i@14: victor@7: response_dict = response.json() victor@7: victor@7: if response_dict: moritz@29: measurement = Measurement(self.base_url, response_dict) victor@7: return measurement victor@7: else: victor@7: logger.error("No measurement with id %s found on the SCC." % measurement_id) victor@7: return None victor@7: victor@7: def delete_measurement(self, measurement_id): victor@7: """ Deletes a measurement with the provided measurement id. The user victor@7: should have the appropriate permissions. victor@7: 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 victor@7: 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) victor@7: 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: victor@7: # Delete the measurement victor@7: delete_page = self.session.post(delete_url, victor@7: data={'post': 'yes'}, victor@7: headers={'X-CSRFToken': confirm_page.cookies['csrftoken'], victor@7: 'referer': delete_url} victor@7: ) victor@7: if delete_page.status_code != 200: 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): victor@7: """ Get a list of available measurement on the SCC. """ i@14: measurement_url = urlparse.urljoin(self.api_base_url, 'measurements') moritz@29: response = self.session.get(measurement_url) victor@7: response_dict = response.json() victor@7: victor@7: measurements = None 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.") victor@7: victor@7: return measurements victor@7: moritz@29: def list_measurements(self, station=None, system=None, start=None, stop=None, upload_status=None, moritz@29: processing_status=None, optical_processing=None): moritz@29: moritz@29: # Need to set to empty string if not specified, we won't get any results moritz@29: params = { moritz@29: "station": station if station is not None else "", moritz@29: "system": system if system is not None else "", moritz@29: "stop": stop if stop is not None else "", moritz@29: "start": start if start is not None else "", moritz@29: "upload_status": upload_status if upload_status is not None else "", moritz@29: "preprocessing_status": processing_status if processing_status is not None else "", moritz@29: "optical_processing_status": optical_processing if optical_processing is not None else "" moritz@29: } moritz@29: resp = self.session.get(self.list_measurements_pattern, params=params).text moritz@29: tbl_rgx = re.compile(r'(.*?)
', re.DOTALL) moritz@29: entry_rgx = re.compile(r'(.*?)', re.DOTALL) moritz@29: measurement_rgx = re.compile( moritz@29: r'.*?]*>(\w+).*?.*?([\w-]+ [\w:]+).*